European Journal of Geography · 2018. 6. 20. · Editorial 4.3 . The publication of the European Journal of Geography (EJG) European is based on the . Association of Geographers’

European Journal of Geography

Volume 4 • Number 3 • October 2013 • ISSN 1792-1341

E u r o p e a n A s s o c i a t i o n o f G e o g r a p h e r s

European Journal of Geography The publication of the EJG (European Journal of Geography) is based on the European Association of Geographers’ goal to make European higher education a worldwide reference and standard. Thus, the scope of the EJG is to publish original and innovative papers that will substantially improve, in a theoretical, conceptual or empirical way the quality of research, learning, teaching and applying geography, as well as in promoting the significance of geography as a discipline. Submissions should have a European dimension. Contributions to EJG are welcomed. They should conform to the Notes for authors and should be submitted to the Editor, as should books for review. The content of this journal does not necessarily represent the views or policies of EUROGEO except where explicitly identified as such. Editor

Kostis C. Koutsopoulos Professor, National Technical University of Athens, Greece [email protected]

Assistant Editor Yorgos N. Photis Associate Professor, University of Thessaly, Volos Greece [email protected]

Editorial Advisory Board

Bailly Antoine, Prof., University of Geneva, Geneva Switzerland Bellezza Giuliano, Prof., University of Tuscia, Viterbo, Italy Buttimer Anne, Prof., University College Dublin, Ireland Chalkley Brian, Prof., University of Plymouth, Plymouth UK Martin Fran, S. Lecturer, Graduate School of Education Exeter S. Vice President of the Geographical Association Gosar Anton, Prof., University of Primorska, Koper, Slovenia Haubrich Hartwig, Prof., University of Education Freiburg, Germany Nazmiye Ozguc, Prof., Istanbul University, Istanbul Turkey Strobl Josef, Prof., University of Salzburg, Salzburg Austria Van der Schee Joop, Prof., VU University, Amsterdam The Nederlands

© EUROGEO, 2013

ISSN 1792-1341 The European Journal of Geography is published by EUROGEO - the European Association of Geographers (www.eurogeography.eu).

1

mailto:[email protected]


http://www.eurogeography.eu/

European Journal of Geography

Volume 4 Number 3

2013

C O N T E N T S

4 Letter from the Editor

6 FROM THE HUMAN-ENVIRONMENT THEME TOWARDS SUSTAINABILITY – DANISH GEOGRAPHY AND EDUCATION FOR SUSTAINABLE DEVELOPMENT Thomas SKOU GRINDSTED

21 ASSESSING THE TRIGGERING RAINFALL-INDUCED LANDSLIP EVENTS IN THE SHIVKHOLA WATERSHED OF DARJILING HIMALAYA, WEST BENGAL. Sujit MANDAL, Ramkrishna MAITI

38 “DTH 1.0”: TOWARDS AN ARTIFICIAL INTELLIGENCE DECISION SUPPORT SYSTEM FOR GEOGRAPHICAL ANALYSIS OF HEALTH DATA Dimitris KAVROUDAKIS, Phaedon C. KYRIAKIDIS

50 URBAN CHANGE DETECTION BASED ON REMOTE SENSING AND GIS STUDY OF SALEM REVENUE DIVISION, SALEM DISTRICT, TAMIL NADU, INDIA. Shanmugam TAMILENTHI, Rajagopalan BASKARAN

60 NEO-INDUSTRIALIZATION OF FORMER INDUSTRIAL REGIONS OF RUSSIA (THE CASE OF “TITANIUM VALLEY”) Anatoly V. STEPANOV, Maria Y. ILYUSHKINA

65 DETERMINANTS OF STRUCTURAL DIMENSION OF DAILY BEHAVIOUR IN A TRADITIONAL AFRICAN CITY: A CASE STUDY OF ILORIN, NIGERIA. Moses Olutoyin ADEDOKUN

2

3

Editorial 4.3 The publication of the European Journal of Geography (EJG) is based on the European Association of Geographers’ goal to make European Geography a worldwide reference and standard. As a result, the papers published in the EJG, including those on this issue, are focused in promoting the significance of geography as a discipline, in resolving global issues or applying geography, complementing, of course, the fundamental goals of improving the quality of research, learning and teaching of Geography. In other words with the EJG the European Association of Geographers provides a forum for geographers worldwide to communicate on all aspects of research and applications of geography with a European dimension, but not exclusive. As a result, every issue of the EJG provides a glimpse of the important role Geography can play in helping researchers, academics, professionals as well as decision makers and politicians in resolving a wide spectrum of problems. In other words, EJG following Geography which connects the physical, human and technological sciences is aiming at enhancing teaching, research, and of interest to decision makers, problem solving. That is, in every issue of the journal a reader can find answers of how aspects of these sciences are interconnected and are forming spatial patterns and processes that impact on global issues and thus effecting present and future generations. The goal of the editorial team, which up to now has been achieved to a great extent, is that the papers of the EJG by dealing with places, people and cultures, will explore those issues ranging from physical, urban and rural environments and their evolution to climate, pollution, development and political-economy. Thus, your contributions to the EJG are not only desirable, but necessary for Geography and Science as a whole. Kostis C. Koutsopoulos

Editor EJG

4



European Journal of Geography, 4:3 Copyright © European Association of Geographers, 2013 ISSN 1792-1341

5

European Journal of Geography Volume 4, Issue 3:6-20 October 2013 © Association of European Geographers

FROM THE HUMAN-ENVIRONMENT THEME TOWARDS SUSTAINABILITY – DANISH GEOGRAPHY AND EDUCATION FOR SUSTAINABLE DEVELOPMENT

Thomas SKOU GRINDSTED

Roskilde University, Department of Environmental, Social and Spatial Change, Universitetsvej 1, DK-4000, Roskilde, Denmark.

[email protected], http://www.ruc.dk/en/departments/department-of-environmental-social-and-spatial-change/

Abstract Research on geography in relation to education for sustainable development (ESD), has only recently climbed the research agenda. The geopolitics of intended learning outcomes in the ESD debate, carries policy that produce dilemmas and challenges confronted with disciplinary traditions. In this article it is examined dialectically how the changing climate and the paradigm of sustainability have been dealt with in Danish geographical university education. It is shown how curriculum programs in higher geographical education have taken different approaches to address issues of sustainability and climate change and how geographers articulate their role and function as knowledge on human-environment interactions changes. The analysis of the geographical education reveal that geographers’ find their discipline contribute considerably to ESD, and thus the human environment theme seems to be associated more closely with sustainability issues. Keywords: ’higher geographical education’, ’climate change’, ’human-environment theme’, ’education for sustainable development’.

1. INTRODUCTION

Imagining climate change often transforms into remarkable geopolitical agendas. In general consensus exists among scientific and geographical communities as to cause and effect of climate change and unsustainable production patterns. Controversy arises when dealing with strategic actions and solutions in educational settings since a given position legitimizes a given political agenda (Morgan, 2011). The geopolitics of education for sustainable development (ESD) therefore produces complex dilemmas, that immutable tends to frame education as a change agent that socializes students to accept new kinds of explanations and pre-analytic assumptions to deal with an academic problem than the previous ones (Lambert & Morgan, 2009). Yet, “Global warming presents an enormous threat to humanity, but the response from academia, including geography, has been relatively slow (…). I find this surprising, indeed astonishing, for there could hardly be a more important geographical topic” (Sayer, 2009:350). Sayer’s statement is remarkable. The world’s population is growing by 900 million people each decade and is expected to reach 9.1 billion when it stabilizes by 2050. The Rome Club (1962) prognoses still seem remarkably precise and even European Journal of Geography - ISSN 1792-1341 © All rights reserved


http://www.ruc.dk/en/departments/department-of-environmental-social-and-spatial-change/

Grindsted Skou T./ European Journal of Geography 3 2011:78). Education is needed to deal with future challenges and over the years education for sustainable development has gradually been reflected in educational policy. Since the Stockholm Conference on the Human Environment (1972) that first established a relation between education and sustainable development, the Rio Declaration (1992) and a number of subsequent declarations, policies and national strategies have promoted the idea of integrating ESD into all disciplines and academic traditions (Jahn et al., 2011). By way of example the European Ministers of Education decided to integrate sustainability into the Bologna process during a meeting in Bergen (2005), and in Louvain-la-Neuve (2009) it was repeated for the next decade. Also, the EU Commission has encouraged member states to use the UN Decade of Education for Sustainable Development (UNDESD) 2005-2014 as a point of reference in the development of national plans (EU Commission, 2009). In a Nordic context national strategies have been put forward and the Nordic Council of Ministers (NCM, 2011) has proclaimed its commitment to ESD. Also academia has responded to the challenges of climate change and more than 31 declarations on sustainability in higher education have been made signed by more than 1400 universities worldwide (Grindsted, 2011). In 2007 also the International Geographical Union (IGU) committed itself to the ESD agenda and developed the “Lucerne Declaration on Geographical Education for Sustainable Development” (IGU-CGE, 2007). How these developments influence various disciplines and academic traditions is in particular interesting in geographical education due to its strong tradition in the human-environment theme. Much research on society-environment interactions on climate change reduces human behaviour to economic rationality when construed in sophisticated climate models (GCM) and sometimes even in non-geographical representations (Urry, 2011). Geographical representations do matter and the need to comprehensively take methodological approaches into consideration in the interface of society-environment interactions seem highly relevant to contemporary and future challenges (Yarnal & Neff, 2004). But how are these problems dealt with in the case of Danish geography? What does it look like in practice and have these developments crystalized in geographical curricula? The aim of this paper is to examine to what extent Danish higher geographical education is organized around sustainability in accordance with the Lucerne Declaration. To do so, the following questions have been addressed. 1) Is the human- environment theme under reconfiguration in geography? 2) How do geographical education programs contribute to ESD? 3) How are issues of sustainability addressed in curricula? and 4) What is the influence of the Lucerne Declaration on Geographical Education for Sustainable Development? Answering these questions will reveal if and how sustainability themes have condensed into curricula. In the following we will first reflect upon what I shall call spatio-temporal tides and waves of the human environment theme to examine the hypothesis that the human-environment theme is under configuration in geography towards sustainability.

2. FROM THE HUMAN ENVIRONMENT THEME TOWARDS SUSTAINABILITY

Geography may be one of the most prominent and oldest disciplines in the conceptualization of human-environment interactions that integrates elements from natural and social sciences (Rasmussen & Arler, 2010). In fact, “The theme of man-environment relation has never been far from the heart of geographical research, and for many it has functioned as the overriding theme” (Harvey, 1969:115). Alexander von Humboldt may be one of the immediate forefathers e.g. by his advocacy of geognocy (todays Earth Science) and later Joachim Schouw, Vidal de la Blache, Carl O. Sauer, Harlan Barrows among others advocated that geographers should study human beings in relation to their geographical environment (Turner, 2002). The history of the human environment theme however, has taken multiple

European Journal of Geography - ISSN 1792-1341 © All rights reserved

Grindsted Skou T./ European Journal of Geography 3 forms and methodological approaches over the years. Some geographers conceptualize the human-environment theme more or less ad hoc, implicitly or explicitly whereas others organize it in constructs separating human and nature or build certain interfaces. Though assumptions of the human environment theme are sometimes implicit they hold “information” that is mediated through scientific and educational practice. Although implicit information neither is logical, consistent nor reflected methodologically it carries huge amounts of tacit knowledge that exists in the interface between subject and object, between human and nature (Demeritt, 2002). Therefore we must never ignore the nexus between (tacit) knowledge and power e.g. in the construction and use of climate models and future sustainability scenarios. Following Harvey geographers build explanation in the way a theme is constituted; “A theme acts as a directive by indicating the sort of facts the geographer ought to collect and by suggesting a mode of organization of those facts” (Harvey, 1969:116). A “theme gives rise to theorize” and how the human-environment theme is considered implicitly or explicitly in geographical educations can be examined through the way different kinds of explanations are perpetuated (Harvey, 1969).

Dialogue about these issues in ESD may not only better prepare students for dealing with wicked and controversial problems like sustainability, resilience or climate change, but may also make students better understand the geopolitics of scientific and educational practices that constitute the “object of study” and fundamentally shape questions asked and data collected. Turner (2002) illustrates how the interface between the spatial chorological approach and the human-environment theme has been the dominating source of (often) conflicting identities in geography for which reason we will briefly draw attention to how determinism, the quantitative revolution and the cultural turn reconfigured the human-environment theme. For centuries there has been much controversy, enthusiasm and vigor around core dimensions of human-environment interaction. Debates in the eighteen and early nineteen centuries have largely concerned environmental determinism (roots from Darwin) versus possibilitism and whether culture or nature played a determining part in the relationship (Christiansen, 1967). Explanations of determinism are often causal and seek to demonstrate how bio-physical factors such as climate, soil and altitude determine social and economic activity. Nature is external, and the domination of nature thesis was in particular inscribed in the enlightenment tradition (Harvey, 1996:121). However, in its strictest form the historical-genetic model conceiving nature as the overruling determinant, never gained full recognition in geography partly because the human-environment theme is poorly understood within isolated and fixed categories that tend to form dichotomies of culture or economy opposed to nature. Environmental determinism suffers from the bias of dualist thinking as does much Western philosophy and such antagonisms has political implications because they involve an attitude of detachment while at the same time holding a perspective of scientific objectivism (Birkeland, 1998). Legitimation of geographical knowledge relate to how geographers construct their object of study. Within the enlightenment tradition geographers build explanations that objectified nature to be instrumentally used, tamed and exploited (Harvey, 1996:124-125). In the context of ESD the domination of nature thesis can hardly address anthropogenic processes or changes in socio-natural systems, but remain undisputed in the tradition of climate skepticism since we are assumed to live on a finite planet.

Another shift in the way geographers have dealt with human-environment interactions relates to the descriptive tradition and the quantitative revolution (Turner, 2002). During the early and mid-20th century positivism became a platform to combat what was regarded speculative science. Universal regularities became a focal point of study (Hansen & Simonsen, 2005:57). The move from ideographic toward a nomothetic approach in geography reconfigured the human-environment theme, since it could not also encompass environmental determinism. “This obviously implied that the traditional focus of Geography


Grindsted Skou T./ European Journal of Geography 3 on Human-Environment relationships lost its defining status” (Rasmussen & Arler, 2010:38). Subsequently, the Kantian geography emphasizing spatial or chorological topographies gained support in favor of the human-environment theme though focus on environmental problems from the 1960 gave inspiration to system ecology and human ecology (Rasmussen & Arler, 2010). Commitment to a theory of knowledge, according to which any phenomenon natural or social, is to be explained through systems of laws and causalities do not fit well with the social dimension of climate change or sustainability. To frame research as unbiased observation of nature as external is to ignore power relations that are inherent in any research agenda. Power relations form the very interpretative categories research questions is designed from or presented in curricula texts (Demeritt, 2002). In the context of ESD the gradualist perspective of climate change carries references to the quantitative revolution e.g. climate models. According to this perspective climate change can be instrumentally adjusted as a form of global technocratic climate management (Urry, 2011:28).

The cultural or linguistic turn is yet another spatio-temporal wave that influenced human-environment debates, though it never gained much attention in physical geography. According to Birkeland (1998) rediscovery of the importance of language led to a shift in the relationship between nature and culture favoring socio-spatial formations so that “cultural geography has lost touch with its basic relationship to the concept of nature” (Birkeland,1998:230). According to Fitzsimmons (1989) only few geographers show interest in the human-environment relationship during the 1980s, and by comparing conceptions of space with conceptions of nature Fitzsimmons demonstrates how geographical thought is imbalanced, not emphasizing the latter. Though discursive constructions favored the spatio-chorological tradition the theme never died as geography fractioned. Discursive constructions do share concern for the effects of power for which reason constructionism tend to be engaged in the critic of the way nature is construed. Constructions of nature nevertheless face the dilemma of the prison of language: that we can never know if our conceptual construction of nature corresponds to how nature actually appears. Yet, conceptualizations of culture are fundamental to environmental challenges (Demeritt, 2002). In the context of ESD discursive constructionists seem to be much more critical to technosciences and therefore call for a precautionary approach to deal with them. Though nature also becomes internalized in socio-environmental interaction, one of the problems of constructionism in climate change is how to deal with irreversible processes and tipping points - that cannot be construed back. This kind of mutual construction implies a rejection of classical divides of subject/object and society/nature dualisms central to anthropocentrism and essentialist assumptions of ESD. In recent years the human-environment theme seems to create much debate about sustainability agendas as will be demonstrated in the following section. In a historical perspective it is interesting to observe that invitations to upscale ecological themes have been numerous:”the view of geography as human ecology has quite a long history” (Harvey, 1969:115) and since Vidal de la Blache or Humboldt geographers like Fosberg (1962), Moss & Morgan (1965) or Stoddart (1967) have argued for up scaling ecological themes in geography, particularly around the concept of system ecology and human ecology (Christiansen, 1967). Today, this seems to be organized around sustainability due to the study of human-environmental interactions (Huckle, 2002; Yarnal & Neff, 2004; Whitehead, 2006; Morgan, 2008; Westaway, 2009; Zimmerer, 2010; Clarke & Button, 2011; Firth, 2011).

3. METHODOLOGY

A critical history-geographical approach is followed to analyze if the human environment theme is being transformed toward sustainability and in what ways sustainability themes has materialized in Danish geographical education (Harvey, 1996). Therefore the issue of human-


Grindsted Skou T./ European Journal of Geography 3 environment reconfigurations is addressed in the interface between identities that have dominated in geography seen as complex spatio-temporal tides and waves intersecting, overlapping and conflicting. The point of departure seeing geographical representations as a mosaic of understandings often in opposition to other representations, serve as a framework to examine the hypothesis that the human environment theme is under reconfiguration in geography. Smith (1982) formulates “the production of nature” as a concept that extends spatial theoretical work of “the production of space” and amalgamates the spatial chronological theme with the human environment theme. Massey grasps such methodological reflections and shows how nature and society interactions must be studied as “endlessly, mobile, restless, given to violence and unpredictability (Massey, 2006:38). Within such a meta-theoretical framework, epistemology is based on an intra- and extra-discursive reality. This implies that elements independent of human perception are also formed through human practice. At the same time the framework rests on the assumption that habits of thought and societal development and habits of thought and the material world are closely interrelated (Elling, 2003). Methodological assumptions of determinism, possibilism or descriptive connotations provide radical different answers to such questions and hence different answers to analyse the same phenomena. To overcome such barriers, human-environment interactions in curricula can be studied (1) by asking what is related and how, (2) by unfolding concepts and (3) how they are formed and give rise to other interpretations (Harvey, 1996:69). The strength of a dialectical framework is that it splinters binary thinking, essentialism, and absolutism of human-environmental conceptualizations as illustrated in the history-theoretical section (Birkeland, 1998). Against this background key concepts of the human-environment theme and human environment interactions were first inductively identified in the Geographic Education Declaration (1992), International Declaration on Geographical Education for Cultural Diversity (2000) and Lucerne Declaration, (2007) to see if issues of sustainability have become important. Second, on the basis of the declaration analysis key concepts in geographical study regulations were categorized, coded, and condensed to analyze if and what “sustainability” themes have been introduced in curricula. Study regulations can be acknowledged as the “law” that constitutes the legal and administrative basis of the study. Therefore they are considered to be a useful analytical object because the development of courses, curricula and the core academic content are reflected in them (Roskilde University, 2006; Copenhagen University, 2009a, 2009b; Aalborg University, 2010a, 2010b, 2011). Study regulations are reformed independently of each other, why the method enables us to explore how the programs have taken different directions to deal with sustainability issues and makes it possible to comprehend how socially constructed ecosystems are represented in the study regulations. However, the topics covered were also categorized and condensed through an abductive method to explore if the human environment theme differentiates from the Lucerne Declaration. In this way it is possible to comprehend inclusion/exclusion of features and themes as a practice to build geographical explanation and how it affects analyses, practices or policies (Harvey, 1969:116).

The method of analyzing curricula was joined by in depth interviews of the chairs of the study boards on the basis of the findings of the Declaration analysis. Direct experiences are to be considered valuable contribution than purely comparison of curricula texts if also to see if the themes are joined in practices. During 2012, all fulltime associate professors and professors allocated the education of geographers at Copenhagen, Roskilde and Aalborg University were interviewed, except researchers that were either on fieldwork, visiting other universities or attending conferences. Thus, 31 of 42 research geographers were interviewed. In the interview the responders were asked if you find climate change and sustainability issues important to geographical education, what you think of the sustainability concept in geography education, if you consider the dynamism of changes in the human environment


Grindsted Skou T./ European Journal of Geography 3 theme to be associated sustainability challenges, and if you include sustainability themes in their own teaching. The interviews were processed by the same methodology as described above. Finally a literature study was conducted to discuss the findings against similar studies and Scandinavian geographical journals were reviewed if they have featured ESD.

4. IS THE HUMAN-ENVIRONMENT THEME UNDER RECONFIGURATION IN GEOGRAPHY?

Three interesting perspectives substantiate the hypothesis that the environment theme is under reconfiguration in geography, but do also suggest a discrepancy between the “role(s)” of ESD in geography. However this does not imply that all geographers working on human-environment interactions conceive their research activities with sustainability. Nothing could be more contradictory. Geography is much else than sustainability but, the following three tendencies suggest changes in discourse coalitions toward being associated sustainability. Firstly, the development of geographical declarations demonstrates a remarkable shift. Over the years the International Geographical Union (IGU), have developed three declarations. The International Charter on Geographical Education (1992), The International Declaration on Geographical Education for Cultural Diversity (2000), and The Lucerne Declaration (2007). The latter differs from the former in several respects. First of all it stresses a shift of the role of geography in a shrinking and globalized world, with spatial transformations of economic, social and political significance, toward a discipline performing a key role in solving sustainable challenges on Earth. By way of example The International Charter of Geographical Education (1992) represents the spatio-temporal tides and waves scarcely paying attention to the human-environment theme. Though human-environmental interactions are mentioned once “concern for the quality and planning of the environment and human habitat for future generations” (Geographical Charter, 1992) issues of globalization related to human rights remain the central focus. The Lucerne Declaration however, refers to the UN Decade of Education for Sustainable Development (UNDESD) 2005-2014, and expresses that the themes of the UNDESD are very much in common with geography’s objects of study; why “the paradigm of sustainable development should be integrated into the teaching of Geography at all levels” (Lucerne Declaration, 2007:243). In the declaration it is claimed that nearly all topics of the UNDESD 2005-2014 possess a geographic dimension, for which reason geography is bound to integrate the concept of ESD. In the below table a word search condenses key aspects of the human-environment theme in the Geographic Education Declaration (1992), International Declaration on Geographical Education for Cultural Diversity (2000) and Lucerne Declaration, (2007) sketching the role of geography.

Table 1. The human-environment theme in International Geographical Union Declaration International Charter on Geographical Education (1992)

International Declaration on Geographical Education for

Cultural Diversity (2000)

Lucerne Declaration on Geographical Education for

Sustainable Development (2007)

Sustainable: 0 Pollution, Contamination, Hazards: 0 Climate change/ global warming: 0 Ecology: 0 Environment: 2 Emission, greenhouse gas: 0 Nature: 0 Energy: 0 Biodiversity: 0 Human-nature interaction: 1

Sustainable: 1 Pollution, Contamination, Hazards: 1 Climate change/ global warming: 1

Ecology: 0 Environment: 13

Emission, greenhouse gas: 0 Nature: 1 Energy: 0

Biodiversity: 0 Human-nature interaction: 6

Sustainable: 60 Pollution, Contamination, Hazards: 1 Climate change/ global warming: 2

Ecology: 7 Environment: 13

Emission, greenhouse gas: 0 Nature: 8 Energy: 3

Biodiversity: 1 Human-nature interaction: 14


Grindsted Skou T./ European Journal of Geography 3 As can be seen (table 1) sustainability was not mentioned in 1992, once in 2000 and 60 times in 2007 and marks a noteworthy difference in the quantification of “sustainable related content”. This illustrates that social-ecological and political-economic processes are not only intertwined, but also that the identity of geography is under reconfiguration. These three declarations illustrate a significant shift. For whatever reasons it might be, the content analysis of declarations illustrates the down scaling of the human environment theme during the late 1980s and early 1990s, as Birkeland (1998), Fitzsimmons (1989) and Stoddart (1987) among others have argued. Over a period of 15 years the human environment theme transforms significantly integrate human environment themes reflecting sustainability challenges (Firth, 2011; Morgan, 2011) to what Birkeland (1998) hypothesizes is the core of geography though in a context of sustainability: “the relationship between culture and nature, the human world and the natural world, society and nature, represents the core of geography, both for human and physical geography” (Birkeland, 1998:230).

Secondly, though the interviews demonstrate great variety they also reveal that sustainability issues are considered huge importance to geography. “The concept of sustainability has huge importance to geography at Copenhagen University, but also related concepts as resilience, vulnerability or ecology” (Interview, 2012). Thus, nearly all geographers interviewed found sustainability issues essential to geography, but remain critical to the concept itself. A major part of the teachers claimed that sustainability is important to geography as an implicit notion, but as explicitly articulated many related concepts may better address particular phenomena (Interview, 2012). Precisely the tacit or implicit underlying basis suggests that the human-environment theme is in transition. Correspondingly, the international review suggests it is not hard to find geographers pushing the agenda for up scaling sustainability though it was not to be identified as an immanent notion. This is featured in research politics, e.g. by the U.S. National Research Council Report (2010) that promotes the relevance of the nature-society issues, merely in the context of sustainability, as well as a huge number of geographers articulate that geography has a particular role for dealing with issues of climate change and sustainability. Bednarz (2006:239) is exemplary: “It seems that non-geographers also think that geography has an important role to play in environmental education (…) many geographers have defined geography as a discipline with a major, if not primary, interest in human – environmental interactions”. Also in the context of ESD a dozen of researchers e.g. (Huckle, 2002; Yarnal & Neff, 2004; Whitehead, 2006; Westaway, 2009; Firth, 2011; Grahn, 2011; Morgan, 2011), build an explanation why geography plays an indispensable role in ESD. Chalkley for example states that “In pursuing this agenda [ESD red.], disciplines such as geography (with its strong interest in society—environment relationships) will obviously be expected to play a leading role; and geographers may also be encouraged to work outside the subject in trans- or inter-disciplinary initiatives” (Chalkley, 2006:235). Also Jahn et al., (2011) and Clarke & Button (2011) may serve as examples of how geographers find the human-environment theme to be a platform for linking ESD and geography, “The subject geography is of importance in the context of ESD due to the analyses of human-environment-interactions and their implications on a geographic area” (Jahn et. al. 2011:22). The recognition that geographical knowledge has pivotal importance for sustainable development makes Westaway (2009:9) state that geography has a special role, maybe even above other disciplines: “Sustainable development is the extrinsic educational purpose that geography is best, indeed almost uniquely, equipped to serve (…) There is little doubt that geography is the best place to take the lead on sustainable development in schools.” Such claims are indeed controversial, but authenticate the human environment theme gives geography its raison d’étre in the struggle for having a share in sustainability issues. The nature- society as well as


Grindsted Skou T./ European Journal of Geography 3 the spatial dimension of sustainability becomes a major pillar that geography seeks to patentee, due to its historical tradition.

Thirdly, environmental and sustainability issues are widely represented in Scandinavian Journals, as well as other international journals, from the late 1980s and have grown significantly since then. Zimmerer (2010) find that 175 articles have been published within the society-nature or human-environment theme in Annals of the Association of American Geographers during the past two decades and that “The number of nature–society articles is estimated to have increased more than twofold in the 1990s alone” (Zimmerer, 2010). Additionally, Lui (2011) and Kidman & Papadimitriou (2012) demonstrate how geographers’ research on sustainability has increased exponentially. However, I found no articles that dealt with ESD neither in the Norwegian, Danish nor Swedish Journals of Geography, nor in the Annals of the Association of American Geographers. However, Molin (2006), Sætre (2009), Grahn (2011) in their Ph.D. dissertations serve as examples of how ESD has been dealt with in secondary school and higher education geography in a Scandinavian context. Lui (2011) reveal how geographers’ research on sustainability has increased exponentially and contrasts a slow growth in integrating sustainable development into curriculum within the US. Thus Lui (2011) identifies an imbalance in the relationship between ESD in geography and geography’s contribution to sustainable research; “An examination of publications in sustainability education journals also reveals geography’s lack of participation in sustainable education” (Lui, 2011). This suggests a discrepancy between statements of the “role of ESD in geography” and geographers’ research activity. Hence, there appear to be little evidence that ESD is recognized a central concern in geography within the US (Haggitt et al., 2005). Also Bednarz (2006), Turner (2002), Yarnal & Neff (2004) among others argue that geography courses lag behind the growing environmental and sustainable research. Thus they identify difference between geographers’ involvement in environmental and sustainability research and education, concluding that the latter is not given priority. The declarations, research statements and review of journals indicate that the human-environment theme is under configuration toward issues of sustainability and climate change. But, what does it look like at Danish geographical education, and might a similar tendency be identified in curricula texts?

5. HOW DO GEOGRAPHICAL EDUCATION PROGRAMS CONTRIBUTE TO ESD?

Geography education shows its commitment to ESD in various ways, reflected in curriculum, interviews and courses. Examination of the study regulations and interviews with chairs of the study boards and researchers reveal that the importance of geography to ESD demonstrates similarities to the Lucerne Declaration: “Geography has a major role on sustainability. Many disciplines are experts on relatively narrowed subjects, whereas geography possesses the broadness which is an important dimension of sustainability. Geography is particularly potent because of its interdisciplinary approach as many other disciplines do not encompass. Moreover, geography merges the natural sciences and social sciences” (Interview, 2012). Interdisciplinary approaches which integrate economic, social and physical aspects of sustainability are not only well suited for geography that deals with interactions between the human ecosystem and the earth system, but a pre-condition to understand its multiple dimensions. Phenomena at global scales are caused by cumulative small scale activities in local places, and the impact of global processes exacerbates phenomena in specific localities. We may not appropriately understand sustainability issues if we ignore the climatological, hydrological or environmental processes that work in nature. Likewise, we misguide explanations of sustainability problems if we ignore social dynamics


Grindsted Skou T./ European Journal of Geography 3 and economic activities. Geography knowledge is important to ESD and distinguished from other disciplines, because a narrow disciplinary focus may not unfold problems of sustainability that operates at multiple scales (Interview, 2012). Another aspect found to be critical is that geography has a role in integrating perspectives from the natural and social sciences. “Geography can contribute in a unique way to sustainable development, especially regarding the integration of knowledge between social and natural sciences. In this way, geography plays a crucial role in dealing with sustainable challenges that you do not find in the tradition of many other disciplines, e.g. Sociology. Secondly, sustainability has an immanent spatial dimension” (Interview, 2012). Thus it is argued that current environmental problems not only call for research and education that epistemologically transcend traditional disciplinary divides, challenging the problems in its complexity, but also seek to bridge the gap between natural and social sciences in dealing with sustainability. To this may be added that geography has a distinct role being able to enrich related disciplinary discussions on ESD. “I find that geography has a responsibility to deal with issues of sustainability. We range competences and skills from the social and natural sciences – a holistic approach is imperative for dealing with sustainability” (Interview, 2012). Additionally, complex interaction between nature and society and the spatio-temporal dimension of sustainability, requires methodological approaches to grasp such interactions that may even be impossible without geographical knowledge. ”Before specialization, all students will acquire a holistic and broad basis of knowledge and approaches, about soil science, climate change, society and urban development. [1.5 years of study red.]. This broad foundation enables students to think critically and analyze side effects of a given phenomenon or human action. This body of knowledge is vital for sustainability, in order to understand side effects in very different areas and scales. Such questions I would say are only possible to deal with through geographical skills” (Interview, 2012). Integrating the production of space and nature as a fundamental perspective of abstraction in e.g. area differentiation generates geographical knowledge and methodologies that make it possible to manage risks involved in the spatial distribution of problems. Non geographical methodologies fail to understand such dynamics of spatial distribution and thus, sustainability may even be seen as a common ground bridging the spatial chorological approach and the human-environment theme. The analysis of the study regulations and the interviews reveal that geographers find their discipline particularly contribute to ESD along three distinct dimensions: 1) human-environment theme that merges the natural and social sciences 2) spatio-temporal dimension of sustainability issues 3) and geographical methodologies and interdisciplinary approaches. These dimensions will be examined in further detail in curricula texts. Though the interviews reveal recognition of the importance of sustainability issues to geography it also demonstrates that most geography teachers remain critical to the concept of sustainability and/or find it better as an implicit basis for educating geographers (Interview, 2012). “When I teach in accessibility for instance, then a sustainability aspect lies behind. If sustainability is there depends how explicit it should be mentioned. I do rarely mention the term, but implicitly sustainability is the main objective for what we do and why we study it in this way. Sustainability is part of all geographers mindset I would say; sometimes so penetrated that one may not need to explicate it” (Interview, 2012). This may be one of the reasons why the analysis of study regulation reveals that sustainability has a limited status in geography educations in practice.

6. HOW ARE ISSUES OF SUSTAINABILITY ADDRESSED IN CURRICULA?

Examination of the preamble of the Aalborg, Copenhagen and Roskilde curricula indicates a methodological foundation in which interconnectedness, processes and flows are given a primary status rather than fixed objects, direct causalities and permanencies. It is obvious that


Grindsted Skou T./ European Journal of Geography 3 emphasis on processes and dynamics in study regulations undermines debates on environmental determinism e.g. in the context of finite resource constrains. In this way dealing with issues of sustainability reformulates previous deterministic concepts into ecological principles and balances, such as carrying capacity and environmental footprint (Interview, 2012). Ecological approaches often focus on environmental problems from an interdisciplinary angel, and are often normative and problem oriented (Rasmussen & Arler, 2010). By way of example the study regulation at Aalborg University requires that “students should acquire knowledge on human influences on ecosystems and the most important anthropogenic changes in history. They should be able to critically reflect on different philosophical views upon nature and its implications (…) understand concepts of sustainability and ecosystems in relation to elasticity and robustness to be able to analyze interactions between human activity (demands) and nature’s capacity and limits” (Study regulation, Aalborg University, 2010a:29).

Second, there seems to be only little distinction between encouragements of the Lucerne Declaration and geographical educations as to interdisciplinary approaches. Thus study regulation requires problem based group work so that ”Students should be able to look upon scientific problems and solutions in an interdisciplinary approach – not only from particular disciplinary premises, but also by including relevant theories, methods and philosophical interpretations from related disciplines” (Study regulation, Roskilde University, 2006:23). Though study regulations do not go into detail in describing which methods to be used, it gives at least a basis for enriching methodological reflection. A holistic understanding of wicked problems like sustainability or climate change requires a reflective rather than an explanatory or commercial ambition (Interview, 2012). In terms of sustainability critical thinking is essential to understand different practices and agendas in play to be able to find possible solutions to sustainability challenges. However, though student driven project on sustainability themes have been limited as well as in teaching (Interview, 2012) similarities in learning objective and interdisciplinary approaches correspond to the Lucerne Declaration.

Third, the spatial-temporal dimensions of sustainability in curricula focus on processes in different time scales and spatial contexts. By way of example the graduate geographical qualification profile in Ecological climatology and climate change, causes, effects, limitation and adaption at Copenhagen University seeks to integrate the range from geological to economic time scales in its very complexity, processes and dynamics. The aim is to gain fundamental knowledge on climate change in history, relations between climate systems, ecosystems and land use, as well as relations between climate and the content of GHG gasses in the atmosphere. Thus, the learning outcome is to be able to work with climate data and environmental observation in various scales to grasp complex relationship between physical and economic activities affecting the global climate. Moreover curricula favor analysis of mechanisms used to prevent climate changes, modeling of GCM’s to construe previous and future climate scenarios, and its consequences and limitations (Study regulation, Copenhagen University, 2009b:8). This specialization illustrates how curricula seeks to explain contemporary sustainability challenges and climate change (e.g. decertification and deforestation) in its interconnectedness, complexity spatio-temporal processes, between the earth system and the human ecosystem, that most profoundly are aligned with the gradual perspective on climate change. If it is assumed that a dialectic approach comprehends the complexity of socio-spatial and economic-ecological processes, this, in turn, will make us recognize that environmental/social problems mutually interact, are spatially distributed, and produce different effects in different spatial scales. Relational time-space configurations vary considerably in different sciences, from geologists (million to billions of years) to economists (month, years and decades), even between physical and human geography (Rasmussen & Arler, 2010). Geographical education programs emphasize how the matter of scale and the


Grindsted Skou T./ European Journal of Geography 3 analysis of environmental problems are inseparable. These pre-analytic assumptions are fundamental to the ESD debate and the interview suggest that geographers find their education programs emphasize how the matter of scale and the analysis of environmental problems are inseparable (Interview, 2012). Although the study regulation at Roskilde University, as the only one, does not mention neither sustainability nor climate change, the Chair of the Study Board explains “students are given a body of knowledge on economic geography, natural processes and planning – also in an context of sustainability – so that they will be able to analyze the impact of a given plan or action in ecological, economic and social terms” (Interview, 2012). However, it seems that Aalborg and Copenhagen Universities attach greater importance to the human environment theme as sustainability and environmental challenges than Roskilde University where the academic traditions to some extent favor the spatial-chorological approach. The reason may be that urban geography is dominant and has a strong spatial tradition. Whereas Aalborg and Copenhagen Universities require various courses and projects in which sustainability or climate change is central, Roskilde University does not address sustainability as a mandatory theme in its curriculum, except from one course in physical geography (Study regulation, Roskilde University, 2006:4). Aalborg University offers the most proactive geographical program dealing with issues of sustainability at BA level. At BA level, 80 of 180 ECTS (and similarly at MA level) are allocated to various themes of the human-environment theme, mostly within issues of sustainability, climate change and anthropogenic and environmental stress. Copenhagen University offers a similar curriculum in terms of sustainability, both at BA and MA level. Thus, it is only the one of six qualification profiles (transformation of cities) that does not explicate sustainability, climate nor energy and resource themes (Study regulation, Copenhagen University, 2006:5). Here too, the reason may be that urban geography traditionally has identified itself closer to the spatial chorological theme.

Overall the curricula demonstrate a move from a mechanical view of nature as external toward more dynamic conceptualizations of human-nature interaction that claims a strong ontological position about the materiality of nature’s construction. Mutual construction implies a rejection of classical divides of subject/object and society/nature dualisms central to anthropocentrism. Though sustainability themes take various forms in geography at Copenhagen and Aalborg University the chairs of the study boards do not find it is necessary to enhance sustainability further; ”Well, I see no need for improving the sustainable content quantitatively. I would say it is saturated. Sustainability issues are filling out much of the education already, so I see no need of giving it more room, but sustainability will stay as a core dimension of geography also in the long run”( Interview, 2012). Also at Aalborg University geography seems to be saturated: “Sustainability is already integrated at many levels and has a strong critical mass. I do not see a point in enhancing it further. Environmental sustainability has the strongest position though, so in a way we could enhance aspects of social sustainability” (Interview, 2012). Despite ESD at geography in Roskilde is given less priority, reconfiguration of the human-environment theme seems to undergo a process toward a resource turn whereby geographical analysis of society is also (certainly not definite) carried through in terms of resource dependency and the character of sustainability challenges. Thus the gradualist perspective of climate change is easily identified. In contrast the Lucerne Declaration resembles discourses of climate catastrophes and ESD as moral development as a form of environmental romantics (Elling, 2003).

7. WHAT IS THE INFLUENCE OF THE LUCERNE DECLARATION IN GEOGRAPHY AT DANISH UNIVERSITIES?


Grindsted Skou T./ European Journal of Geography 3 Complementary elements have been identified between the Lucerne Declaration and study regulations. The interviews revealed that the declaration was unknown for geographers at Copenhagen, Roskilde and Aalborg Universities and has neither been dealt with in the study boards nor on any other occasion (Interview, 2012). “I have never heard of the declaration and it has neither been discussed in the study board nor at teacher meetings” (Interview, 2012). It therefore goes without saying that the Lucerne Declaration has not produced an impact on the development of geography at the Danish universities. Instead the declaration (and curricula) can be considered to be a product of specific history-geographical circumstances signalling geographical representations of human environment theme. The recognition that the carbon dependent society transforms the earth systems has gained acceptance and replaces the “finite construction of human-nature system, by another geomorphological and geographical period (Crutzen, 2002). Reconfiguration of the human environment-theme toward the paradigm of sustainability or related concepts seems further to form a culturally embedded understanding of human environment interactions as implicit educational preconditions. As the Chair of the Study Board at Copenhagen University explains;” sustainability is not something we discuss, it is not necessary to discuss what you agree upon and which already is there” (Interview, 2012). However, this does not imply that the human-environment theme undergoes transformations toward uniform conceptualizations, but that normative and methodological horizons changes as new knowledge emerge. Thus, the underlying basis to study a given phenomenon seems to change from a mechanistic toward an ecological worldview with a low-carbon society as the normative horizon. However, this varies significantly between geography at Aalborg, Copenhagen and Roskilde. While Aalborg and Copenhagen explicitly address ecological principles in their study regulations (as the Lucerne Declarations suggests) they are not to be identified explicitly in curricula at Roskilde. Nevertheless, this does only signify that the study regulations differ and that it to a smaller extent orchestrates an underlying basis of sustainability as a platform for human-environment analysis.

8. CONCLUSION

Declarations as well the analysis of geographical education programs suggest that the human-environment theme is under reconfiguration toward being associated with sustainability. Interviews as well as analysis of study regulations reveal that Danish geography find the discipline to contributes to ESD in three different ways. First, geography’s strong tradition in the human-environment theme provides a methodological basis for dealing with issues of sustainability. Second, the spatio-temporal dimensions of sustainability call for geographical approaches to be able to understand the dynamics, complexity and interactions in various scales. Third, it is widely accepted that geography contributes by its interdisciplinary approaches to bridge the social and natural sciences, though it does not seem to be the case in practice. Although issues of sustainability and climate change have been materialized in Danish geographical curricula differently, sustainability is mainly an immanent contested representation. Danish geography education is reluctant to activate the concept of sustainability as an explicit notion, partly because of the criticism attached to it. Still nearly all geographers interviewed find sustainability themes or related concepts important to geography. At Aalborg and Copenhagen University sustainability themes are set to be saturated, whereas at Roskilde University curricula does not require students to deal with sustainability and student driven activities are limited.


Grindsted Skou T./ European Journal of Geography 3 Acknowledgement I would like to thank Associate Professor, Henrik Toft Jensen and Professor Bo Elling at ENSPAC, Department of Environmental, Social and Spatial Change, Roskilde University, Denmark; and Annette Marie Grindsted, Associate Professor, Institute of Language and Communication, University of Southern Denmark, for insightful and valuable contributions.

REFERENCES

Aalborg University. 2011. Study regulation for geography, Bachelor, Department of Development and Planning.

Aalborg University. 2010a. Study regulation for geography, Master, Department of Development and Planning.

Aalborg University. 2010b. Study regulation for integrative geography, Master, Department of Development and Planning.

Bednarz, R. 2006. Environmental research and education in US Geography, Journal of Geography in Higher Education, 30 (2): 237-250.

Birkeland, I. J. 1998. Nature and the ‘cultural turn’ in human Geography, Norwegian Journal of Geography, 52 (4): 221-228.

Chalkley, B. 2006. Education for sustainable development: Continuation, Journal of Geography in Higher Education, 30(2), pp.235-236.

Christiansen, S. (1967). Geografi som „human økologi”, Danish Journal of Geography, Bind 66.

Clark, B. & Button, C. 2011. Sustainability transdisciplinary education model: Interface of arts, science, and community (STEM), International Journal of Sustainability in Higher Education, 12 (1): 41-54.

Copenhagen University. 2009a. Study regulation for geography, Bachelor, Department for Geography and Geology.

Copenhagen University. 2009b. Study regulation for geography, Master, Department for Geography and Geology.

Crutzen, P.J. 2002. Geology of mankind, Nature, 415:23.

Demeritt, D. 2002.What is the 'social construction of nature'? A typology and sympathetic critique, Progress in Human Geography, 26 (6): 767–790.

Elling, B. 2003. Modernitetens Miljøpolitik, Doctoral dessitation, Roskilde Universitetsforlag.

EU Commission. 2009. The Bologna Process 2020 – The European Higher Education Area in the new decade, Communiqué of the Conference of European Ministers Responsible for Higher Education, Leuven and Louvain-la- Neuve, April, 2009.

Fitzsimmons, M. 1989. Matter of nature, Antipode, (21):106-120.


Grindsted Skou T./ European Journal of Geography 3 Firth, R. 2011. The nature of ESD through geography, Teaching Geography, 36 (1): 14-16.

Grahn, A. 2011. Fakta, normativitet eller pluralism? Didaktiske typologier inom gymnasieskolans geografiundervisning om klimatförändringer”, Licentiatuppsats, Uppsala Universitet, Forskerskolan i Geografi.

Grindsted, T.S. 2011. Sustainable universities – from declarations on sustainability in higher education to national law, Environmental Economics, 2 (2): 29-36.

Hansen, F. & Simonsen, K. 2004. Geografiens Videnskabsteori – en introducerende diskussion, Roskilde Universitetsforlag.

Harvey, D. 1969. Explanation in Geography, London, Edward Arnold.

Harvey, D. 1996. Justice, Nature and the Geography of Difference, Oxford, Blackwell.

Huckle, J. 2002. Reconstructing nature: Towards a geographical education for sustainable development, Geography, 87 (1): 64-72.

International Charter on Geographical Education (1992). The Commission on Geographical Education of the International Geographical Union (IGU CGE).

International Declaration on Geographical Education for Cultural Diversity (2000). The Commission on Geographical Education of the International Geographical Union (IGU CGE).

Jahn, M., Haspel, M., Siegmund, A. 2011. Glokal Change: Geography meets remote sensing in the context of the education for sustainable development, European Journal of Geography, 2 (2):21-34.

Kidman, G., Papadimitriou, F. 2012. Content analysis of international research in geographical and environmental education: 18 years of academic publishing, International Research in Geographical and Environmental Education, 21 (1): 3-10.

Lambert, D., Morgan, J. 2009. Corrupting the curriculum? The case of geography, London Review of Education, 7 (2): 147–157.

Lui, L. (2011).” Where in the world of sustainability education is US geography” Journal of Geography in Higher Education 35 (2), pp. 245-263.

Lucerne Declaration. 2007. International Geographical Union Commission on Geographical Education (IGU CGE), Geographiedidaktische Forschungen, 42: 243—250.

Massey, D. 2006. Landscape as provocation. Reflections on moving mountains, Journal of Material Culture 11(1/2): 33-48.

Molin, L. 2006. Rum, frirum och moral: en studie av skolgeografins innehållsvalg. Acta Universitatis Upsaliensis, Department of Social and Economic Geography, Uppsala University, Geografiska Regionsstudier, 0432-2023:69.

Morgan, J. 2008. Curriculum development in 'new times, Geography, 93 (1): 17-24.


Grindsted Skou T./ European Journal of Geography 3 Morgan, A. 2011. Sustaining ESD in geography, Teaching Geography, 36 (1): 6-8.

Nordic Council of Ministers. 2011. Kunnskap for grønn vekst og velferd (MR-U) strategi for utdannings- og forskningsområdet i perioden 2011–2013, Köpenhamn.

Rasmussen, K., Arler, F. 2010. Interdisciplinarity at the Human-Environment Interface, Danish Journal of Geography, 110(1):37-45.

Roskilde University. 2006. Study regulation for geography, Bachelor and Master, Department of Environmental, Social and Spatial Change.

Sayer, A. 2009. Geography and global warming: can capitalism be greened?, Royal Geographical Society, 41 (3): 350–353.

Smith, N. 2010. Uneven development, Nature, Capital and the Production of Space, Third Edition, Verso, London.

Stoddard, D. R. 1987. “To claim the high ground: Geography for the end of the century”, Institute of British Geographers, N.S. 12:327–36.

Sætre, J. P. 2009. Geografi i tekst og bilde, Ph.d.-avhandling, NTNU, Norwegian University of Science and Technology.

Turner, B. 2002. Contested identities: Human-environment geography and disciplinary implications in a restructuring academy, Annals of the Association of American Geographers, 92 (1): 52-74.

Urry, J. 2011. Climate Change and Society, Polity Press, Cambridge.

U.S. National Research Council. 2010. Understanding the changing planet: Strategic directions for the geographical sciences. Washington, DC, National Academy Press.

Westaway, J. 2009. A sustainable future for geography?, Geography, 94 (1): 4-12.

Whitehead, M. 2006. Spaces of Sustainability, Geographical Perspectives on the Sustainable Society, London, Routledge.

Yarnal, B., Neff, R. 2004. Whither parity? The need for a comprehensive curriculum in human-environment geography, The Professional Geographer, 56 (1): 28-36.

Zimmerer, K. S. 2010. Retrospective on Nature–Society Geography: Tracing Trajectories (1911–2010) and Reflecting on Translations, Annals of the Association of American Geographers, 100 (5): 1076-1094.


European Journal of Geography Volume 4, Issue 3: 21-37 October 2013 © Association of European Geographers

ASSESSING THE TRIGGERING RAINFALL-INDUCED LANDSLIP EVENTS IN THE SHIVKHOLA WATERSHED OF DARJILING HIMALAYA, WEST BENGAL.

Sujit MANDAL

Raja N. L. Khan Women’s College, Department of Geography, Paschim Medinipur, West Bengal, 721102, India [email protected]

Ramkrishna MAITI

Vidyasagar University, Department of Geography and Environment Management, Paschim Medinipur, West Bengal, 721102, India

[email protected]

Abstract Landslip events are triggered by the rainfall which causes a great damage to human lives and properties worldwide. The present study, encompasses the relationship between rainfall and landslide, the determination of the critical rain and its return period and the assessment of temporal probability of rainfall that triggers landslip events. Landslide inventory statistics were used to pertain the relationship between rainfall and landslip events. The continuous and uniform rate of minimum amount of rainfall (approx. less than 80 mm/day) for few consecutive days can cross the geomorphic threshold and can introduce slope instability in the Study area of Shivkhola Watershed. The critical rainfall for two major landslide locations i.e. Paglajhora and Tindharia were estimated incorporating geo-technical parameters such as angle of internal friction (φ), slope angle (Ѳ), upslope contributing area (UCA), transmissivity (T), wet soil density (ps), and density of water (pw). The return period of critical rain at various recurrence intervals were being assessed. At Lower Paglajhora the critical rainfall is 88.93mm which is less than the estimated rainfall of 90.54 mm/day at the recurrence interval of 1.01 year with 99% probability. The temporal probability of the landslide events were estimated applying Binomial and Poisson Probability Distribution Model based on historical landslip events since 1968. The probability model suggests that occurrences of major landslide events with more than 90 percent certainty could be expected in every 7.5 years. Keywords: Shivkhola Watershed; Landslides; Critical rainfall; Return period; Antecedent rainfall; Probability Model.

1. INTRODUCTION

Geomorphic threshold is the significant parameter in analyzing the stability condition of a particular spatial unit in a quantitative way. According to White et al., (1996) ‘the minimum or maximum level of some quantity needed for a process to take place or a state to change is generally defined as threshold’. Brunsden et al., 1981; Wagner 1983; Manandhar and Khanal, 1988; Dhital et al., 1993; JICA, 1993; Upreti and Dhital, 1996; Gerrard and Gardner,





Mandal S.- Maiti R. / European Journal of Geography

2000; Dhital, 2003; Dahal et al. 2006a, while other works, such as Caine and Mool (1983), Dhakal et al. (1999), and Scott Wilson (2003) focused mainly on landslide risk assessment in Himalayan terrain assessing physical properties of landslides and debris flows, effects of regional and local geological settings, and recommendations for environmental-friendly preventive measures. Varnes (1978) studied the role of minimum intensity and duration of rainfall to cause a landslide of shallow soil slips, debris flows, debris slides or slumps. Crozier (1997) opined a maximum threshold, beyond which there is 100% chances of occurrences of the process at any time when the threshold value is exceeded. Starkel (1972) for the first time, observed the geomorphic effects of an extreme rainfall event in the eastern Himalaya (Darjiling). Froehlich et al., (1990) investigated the same area (Darjiling Himalaya) and found that shallow slides and slumps on steep slope segments occur when 24 hours rainfall reaches 130-150 mm or continuous three days rainfall totals 180-200 mm. Campbell, 1975; Cotecchia, 1978; Caine, 1980; Innes, 1983; Pomeroy, 1984; Canon and Ellen, 1985; Neary and Swift, 1987; Keefer et al., 1987; Kim et al., 1991; Li and Wang, 1992; Larsen and Simon, 1993; Wilson et al., 1995; Wieczorek, 1987, 1996, 2000; Terlien, 1997, 1998; Crosta, 1998; Crozier, 1999; Glade et al., 2000; Aleotti, 2004; Guzzetti et al., 2004, 2007; Hong et al., 2005; and Zezere et al., 2005 tried to establish rainfall-intensity thresholds for predicting the slope failure accurately. Caine (1980) for the first time established worldwide rainfall threshold values for landslides. Recently Guzzetti et al., (2007) reviewed rainfall thresholds for the initiation of landslides worldwide and proposed new empirical thresholds based on the statistical analysis of the relationship between rainfall and landslide occurrences. They defined intensity-duration threshold as: Where, I is the hourly rainfall intensity in millimeters (mm hr-1) and D is duration in hours.

The history of the landslide events in the Shivkhola watershed, Darjiling Himalaya shows that most of the landslide occurred as a result of heavy and continuous downpour for few days. Researchers (Ghosh, 1950; Nautiyal, 1951, 1966; Dutta et al, 1966; Roy and Sensharma, 1967; Basu, 1985, 1987 and 2001; Verma, 1972; Paul, 1973; Sinha, 1975; Chatterjee, 1983; Sengupta, 1995; Basu and De, 2003; Pal, 2006; Maiti, 2007; Ghosh, 2009b; and Sarkar, 2011) carried out a demand oriented studies in Darjiling Himalaya and identified the causes and consequences of major landslide occurrences phenomena. Only one concentrated 50 mm showers/hour during monsoon may initiate slope failure and that endangers innumerable people and their properties. The instability of landslide increases due to progressive absorption of moisture from excessive rainfall and cutting of hill slope both artificially and naturally together that makes the drainage inefficient.

Since 1968, the Shivkhola watershed faced 128 approachable landslide events till 2011 and among of them 76 events had been treated as reactivated (not 70m away from old slided area) and 52 as fresh events (70m away from the old slided area). The total landslide events took place in 16 years and out of which 12 years were recognized as the major landslide event years. Study suggested that except earthquake-induced landslide all the events occurred during the monsoon period with continuous and heavy showers which were more than the critical rainfall calculated after Borga et.al. (1998). Most of the landslide events were also closely associated with physiographic configuration, proximity to the weaknesses planes and weak lithological compositions. Yet, there is no ‘universal threshold value’ for the initiation of landslide phenomena. Because, landslide not only depends on one single parameter but also on others geomorphic and geo-hydrologic attributes. The study identified that only 105.88mm and 88.928mm daily rainfall was the critical rain for initiation of slide at two



major landslide prone areas i.e. Tindharia and Lower Paglajhora respectively. So there is an every possibility for the generation of geomorphic threshold for initiation of landslide and there is a frequent occurrence of debris slide which will reduce the slope angle on landslide scar face to that of repose angle to attain temporary stability through internal feed back in a process of homeostatic adjustment. The study thrusts on the relationship between antecedent rainfall and landslide events, estimation of critical rainfall after Borga et al. (1998) and its return period for Paglajhora and Tindharia, and the application of the ‘poisson’ and ‘Binomial’ probability distribution model to estimate the temporal probability of landslide events in the Shivkhola Watershed. The study area, Shivkhola Watershed (Fig.1) is located in the southern escarpment slope of Darjiling Himalaya and is attributed with the landslide dynamicity (Fig.2, 3 & 4), mainly at the places of Paglajhora and Tindharia.

Figure 1. Locational map of the Shiv-khola Watershed

2. MATERIALS AND METHODS

2.1. Relationship Between Rainfall and Landslide Events Drainage basin is a proper spatial scale for analyzing hydrological parameters like input of rainfall and resultant output of discharge in a systematic interactive combination with other topographic and geometric attributes (Chorley, 1969; Strahler, 1957). Amount of rainfall is one of the triggering factors for slope instability because it affects surface run-off, infiltration, depth of the saturated soil and thus influences soil-moisture condition, cohesion and angle of internal friction. Infiltration and evapotranspiration are considered as the important hydrological parameters that determine slope instability. The hydrologic factors like daily rainfall threshold in connection with slope angle and regolith thickness (Gabet et.al.2004), rain fall intensity, infiltration (Schumm, 1983) etc. are given due importance in the analysis of slope instability. In the Shivkhola watershed the amount of rainfall increases from the month of May and it reaches peak in July then it starts decreasing and reaches minimum in December-January.



Figure 2, 2a, 2b, 2c, 3, 3a, 3b, 4 Analysis of monthly average rainfall since 1979 to 2010 reveals that the months of June,

July, August and September were registered with rainfall of more than normal (between third quartile-Q3 and first quartile-Q1) and average. On the other hand below average rainfall was being found mostly in the months of October, November, December, January, February and March. July and August were the most consistent rainfall months [Table.1] of the year where the values of co-efficient of variation were very low (37 and 30). These two months were also characterized by the catastrophic rainfall months because of the frequent occurrences of landslide events due to few days’ continuous rainfall. The picturesque slope failure took place in the Shivkhola watershed due to catastrophic rainfall in the month of July of 1985, 1989, 1992, 1993, 1998, 2002, 2003, 2004 and 2006. Rainfall induced slope failure also occurred in September 1980, 2006, July 2007, August 2007, August 2010.

Table.1. Statistical analysis of Monthly average rainfall since 1979-2010. Months J F M A M J J A S O N D Mean 13 17 43 95 302 728 996 784 588 183 12 23 S.D. 23 20 48 60 173 304 364 236 170 154 18 61 C.V. 176 117 111 63 57 41 37 30 28 84 150 97

The yearly average rainfall since 1979 to 2009 also stated that the year 1978, 1980, 1981,

1984, 1985, 1989, 1990, 1991, 1998, 1999, 2000, 2003, 2005, 2006 and 2007 were the year of above average rainfall (317.75 mm). The symbolic red line in Fig.5 shows successive three years average rainfall conditions of 32 years which also helps to understand the effects of climate on ground-water condition, because the latter are influenced by the rainfall of the preceding two years. The devastating landslide occurrence in the 1984 and 85, 1987, 1988, 1999, 2000, 2003, 2005, 2006, and 2009 were associated with the preceding years increasing trend of successive three years average rainfall (Fig.5). So, the landslide phenomena in the Shivkhola watershed were closely related with the cumulative effects of precipitation and the changing nature of ground-water condition. In the present work, an simple assessment was made on antecedent rain prior to the landslides date for the month of July’ 1993, 1998, 2003, 2007 and 2010 on the basis of collected day wise rainfall from nearby Selim Hill Tea Estate.



Figure 5.

2.2. Estimation of Critical Rainfall Campbell, 1975; Caine, 1980; Larsen and Simon, 1993 established that the empirical threshold condition to initiate landslide refers to relational value based on statistical analysis of the relationship between rainfall and landslide occurrences whereas the physical thresholds were usually determined with the help of hydrologic and stability models that take into consideration of various attributes such as transmissivity (T), wet soil density (ps), density of water (pw), slope angle (θ), angle of internal friction (φ), upslope contributing area (b), relation between rainfall and pore-water pressure etc. In the absolutely unstable condition the role of rainwater to initiate the threshold for sliding could be determined. If the hydrological factors like rain fall and seepage flow are considered the threshold condition for absolute instability that can be predicted. The critical rainfall (rcr) was calculated after Borga et.al. (1998) using equation no.2.

Angle of internal friction was measured by tri-axial compression test (Fig.6) following Mohr stress Diagram. All the tests were carried out under drained condition using 100 mm diameter and 25 mm thick specimen in Geotechnical Laboratory, GSI, Kolkata. The major stress (σ1), minor stress (σ3) and cohesion (c) were estimated through tri-axial soil testing mechanism from Geo-technical Laboratory of GSI, Kolkata (22/com/soil/GTL/ER/O6-07) by Geologists Sufiyan, Sengupta, Ghosh and Pramanik (2007). On the basis of these three major attributes a Mohr stress circle was developed to obtain angle of internal friction and angle of rupture. At first, a circle was drawn through σ3 and σ1 with the centre on the horizontal axis; the centre of the circle was obviously (σ1 + σ3)/2 and the radius was (σ1 –σ3)/2. The values of confining pressure, σ3, and compressive stress, σ1 were plotted on horizontal axis where stress difference is σ1 - σ3. On a plane parallel to the greatest principal stress axis (2α=0) the normal stress across the plane was σ3 and the shearing stress was 0. If the plane makes an angle of 450 with the greatest principal stress axis (2α=90), the shearing stress is at a maximum and the normal stress is (σ1 + σ3)/2. If the plane makes an angle of 900 with the greatest principal stress axis (2σ =1800), the shearing stress is 0 and the normal stress is σ1

(Billings, 1987). Experiments were done with different values of confining pressure (σ3). The Mohr Circle

shows that as the confining pressure is increased, the stress as well as the stress difference must be increased to produce rapture. A line which is the tangent of the ‘Mohr Circle’ is called as the ‘Mohr Envelope’. The angle that this line makes with the horizontal axis of the diagram is the angle of internal friction, φ (Fig.7).



Figure 6.

The saturated conductivity of the sail varies from 10 –2 m s –1 for the soil depth less than

0.5m to 10-5 m s-1 for soil depth between 1 to 2 m (Fenti, 1992). Based on these and other data, Matteotti (1996) estimated the transmissivity (T) of saturated soil to lie between 5 and 30 m-2 day –1, with a mean value of 15 m-2 day –1 (Borga et.al 1998). Specific unit weight of water (γw) and unit weight of the soil (γs) were estimated through laboratory test (Keen Box Method). The density of soil and water varies from place to place due to in-situ geo-hydrologic condition.

Figure 7.

The saturated soil density of rock was also consulted and adopted from the field

experiences done by Deoja (Mountain Risk Engineering Handbook, 1991). Specific contributing area (total contributing area divided by the contour length) was computed by distributing flow from a pixel among its entire lower elevation neighbour pixel (Borga et.al, 1998). Quinn et al. (1991) proposed Fraction of Flow (Fi) allocated to each lower neighbour was determined by:

Where the summation is for the entire lower neighbour; S is the directional slope, and L is an effective contour length that acts as the weighting factor. The value of L used here is 10 m of the pixel size of the cardinal neighbour and 14.14m of the pixel diagonal for diagonal neighbour.

2.3. Estimation of Average Catastrophic Rainfall to Obtain Return Period and Probabilistic Recurrence Interval of the Critical Rain

The Selim Hill Tea Estate situated 250m North West of Tindharia registered 52 days having more than the critical rain fall to initiate threshold condition during the years 2005-2010 [Table.2]. Average day wise rainfall for catastrophic days in 2005, 2006, 2007, 2008, 2009, and 2010, are 120.7 mm., 127 mm., 128.5 mm., 161.12 mm., 141.53 mm and 102.4 mm respectively which are greater than the estimated threshold rainfall for initiating slide at Tindharia and Lower Paglajhora. This indicates high possibility of frequent slide in those places.



Table.2. Analysis of catastrophic rainfall event during 2006 –2011. 2006 Rain in

mm 2007 Rain

in mm

2008 Rain in

mm

2009 Rain in mm

2010 Rain in mm

2011 Rain in mm

20th June

95.5 23rd May

103.5

10th June

222.72

7th June

125 3rd June

146.5 24th May

88.9

26th June

183.5 28th July

150 28th June

93.5 9th June

100 19th June

133 28th May

101.6

31st July 134.5 29th July

160 29th June

120.5 23rd June

203 7th July 175 16th June

111.7

3rd Oct 100 30th July

112.5

10th July

120 26th June

179 12th July

200.5 14th July

103.6

4th Oct 90 19th Aug

150 17thJuly

124.5 29th June

196.5 9th Sep 98.7 18th July

102

31st Aug

120.5

18th July

100 7th July

273.5 27th Sep 95.5 25th July

92.4

17th Sep.

89.5 27thJuly

120.5 8th July

162.5 5th Aug 114.3

3rd Oct

130 18th Aug

145.5 21st July

146.5 25th Aug

89.4

23rd Sep.

106.2 28th July

148.5 28th Aug

102.6

30th July

100 16thSep.

115.5

10th Aug

191.5 25thSep.

116.3

31st

Aug 107.5 26thSe

p. 90.5

No of Days

5 8 9 12 6 12

Total 603.5 1016 1153.42

1933.5 849.2 1228.8

Average 120.7 127 128.15

161.12 141.53 102.4

Source: Selim Hill Tea Estate, 2011. (1/2 km. Crow fly dist. from Tindharia).

2.4. Application of the ‘Poisson’ and ‘Binomial’ Probability Distribution Models to Estimate the Temporal Probability of Landslide Events

The frequent occurrence of landslides in the unstable terrain of Skivkhola watershed and their continuous monitoring through intensive field investigation and in consultation with others research works made by Starkel and Basu (1985), Basu and Sarkar (1985 and 1988) Basu and Ghosh (1993), Basu and Maiti (2001), Maiti (2007 and 2011) Ghosh (2009b), Sarkar (2011) and author himself have provided most reliable earlier landslide frequency data since 1968 [Table.3].

To determine the temporal probability of rain-induced landslide events the exceedance probability of one or more landslides were attempted by considering the landslide as random point events. Two major discrete probability distribution models such as the ‘Poisson distribution’ and the ‘Binomial distribution’ were mostly applied to calculate the exceedance probability of landslide (Coe et al., 2004; Crovelli, 2000). According to ‘Poisson distribution model’ the occurrences of landslide events that is experiencing ‘n’ landslides during the time‘t’ could be expressed by

…………………………………(eq.4) Where, λ= average rate of landslide occurrence. n= 0, 1, 2, 3……………..….n. So, the exceedance probability/the probability of experiencing landslide events during the time‘t’ could be expressed as



…………….(eq.5) Where, and 𝝁𝝁 = mean recurrence interval between successive landslide events. In the same way, the exceedance probability could be assessed by using the binomial probability distribution model with the help of following expression.

…….(eq.6)

To estimate the temporal probability of the landslide events in the Shivkhola watershed, the mean recurrence interval of known landslide events year was deduced that is 2.75 (16 known events year in 44 years). In the same way, mean recurrence interval of known major landslide events was also deduced that that is 3.66 (12 major events year in 44 years). Then, both Poisson and Binomial distribution models were being applied to determine the exceedance probability.

3. RESULT AND DISCUSSION

The thickness of the soil and that of the saturated soil during monsoon were measured to be 4.5m (Tindharia T.E.) and 7.25m (Lower Paglajhora). The wet soil buck density was measured to be 1.96 g/cc and density of water was 1.07 g/cc. The angle of internal friction varies from 210 to 260 with an average of 240. The upslope contributing area and contour length was 968 m2 and 22.00 m of Paglajhora and 1404 m2 and 27.00 m of Tindharia respectively. The slope angle at Paglajhora was 48020′ and Tindharia was 53020′. The basic requirement for the short term stability of the slope at marginal escarpment of Tindharia and Lower Paglajhora are to maintain the slope angle to be around 240. A steep slope will decline by slope failure to an angle of repose slope to attain short term stability. This concept leads to the concept of limiting or Threshold slope angle. The calculated critical/threshold rainfall of two major landslide prone parts of the Shivkhola watershed wer 105.88mm/day (Tindharia T.E.) and 88.93mm/day [Lower Paglajhora]. The friction angle (φ) in connection to the critical rainfall was assessed from the following two Stress circles (Fig.8 & 9).

A relationship between antecedent cumulative rainfall and landslide vents of 1993, 1998, 2003, 2007 and 2010 was established on the basis of the data recorded from earlier research work done by Ghosh et al. (2009b); Basu et al. (2000) and the collection of rainfall data from nearby Selim Hill Tea Estate by author himself. Only two days antecedent cumulative rainfall of 211.3 mm invited the slope failure at the places of Tindharia and Gayabari and Mahanadi. The 1998 landslide event took place due to 300-600 mm cumulative rainfall in the past 2/3 days only. The two days’ antecedent cumulative rainfall of 390 mm was responsible for 1998 landslide events. The major event of 2003 happened due to incessant rainfall of 500 mm in 2 days. 17th and 18th July, 2007 received rainfall of 124.5 mm and 100 mm respectively.



Figure 8.

Figure 9.

These two days’ antecedent cumulative rainfall of 224.5 mm caused havoc slope failure

at Tindharia and Upper and Lower Paglajhora. Again 2007 faced landslide events on 8th September when 6th, 7th and 8th September’s antecedent cumulative rainfall amount was 275 mm. In 2010, major and prominent landslide events happened as a result of 5 days’ rainfall of 345 mm at 14 Mile near lower Paglajhora, Nurbong, Gitingia, Shiviter [Table.4]. Antecedent Cumulative rainfall induced landslide analysis shows that the continuous and uniform rate of minimum amount of rainfall (approx. less than 80 mm/day) for few consecutive days can cross the geomorphic threshold and can introduce slope instability condition.

Table.4. Major landslide events and 5 days antecedent cumulative rainfall. Landslide events Location

1 day 2 day 3 day 4 day 5 day

Tindharia and Gayabari and Mahanadi.

110mm

211.3 mm 265 mm 305 mm 340 mm

2nd July, 1993 3RD July, 1993 [Landslide]

Chunabhati, Tidharia, 150 390 450 485 520



Paglajhora, Mahanadi (along NH-55), Jogmaya

6th July, 1998 7th July [Landslide]

8th July [Landslide]

Gayabari T.E., Along NH-55, Tindharia and Shiviter. Paglajhora,

197 mm. 500 mm. 527 565 590 8th July, 2003 9th July, 2003

[Landslide]

Tindharia and Upper and Lower Paglajhora.

124.5 224.5 255.5 300 315 17 th July, 2007 18 th July, 2007

[Landslide]

Tindharia, Lower Paglajhora and Shiviter.

91 mm 187 mm 275 mm 6th Sept. 2007 7th Sept. 2007 8th Sept. 2007

[Landslide]

14 Mile near lower Paglajhora, Nurbong, Gitingia, Shiviter

111.7 mm 15 th June, 2010

345 mm 16th June, 2010

[Landslide]

245 295 365

The return period of the Total of the Catastrophic Rainfall and Average daily rain of

catastrophic days was calculated on the basis of the duration of the period 2005 – 2010 following Gumbel, 1954 [Table.5] and using the following formula. T= (N+ 1)/m--------------------------- ------------ (eq.6). [Gumbel, 1954]

The derived recurrence interval of the Total of the Catastrophic Rainfall and Average daily catastrophic rain of days recording more than the calculated threshold rain which is done by log probability law following Chow, 1951, 1954 and Schwab et.al. 2002. Xc = x (1 + CvK) --------------------- (eq.7).

Table.5. Return Period of Catastrophic Rainfalls after Gumbel, 1954. Year No of Days

of Catastrop

hic Rainfall

Total of the

Catastrophic

Rainfall

Arranged in

descending order

Rank

T= (N+ 1)/m

(Gumbel ,1954)

Average daily rain

Arranged in

descending order

Rank T= (N+ 1)/m

(Gumbel ,1954)

2005 5 603.5 1933.5 1 7 120.7 161.12 1 7 2006 8 1016 1228.8 2 3.5 127 141.53 2 3.5 2007 9 1153.42 1153.42 3 2.33 128.15 128.15 3 2.33 2008 12 1933.5 1016 4 1.75 161.12 127 4 1.75 2009 6 849.2 849.2 5 1.4 141.53 120.7 5 1.4

T- Return Period N-No of years m- Rank in ascending order.

Xc – Calculated Rainfall x - mean value Cv – Coefficient of variation K- Log Probality Frequency Factor (calculated from the table of Chow, 1954)



2010 12 1228.8 603.5 6 1.16 102.4 102.4 6 1.16 Mean 1130.7367 130.1500 Std. Deviation-S.D.

452.7617 19.8087

Coefficient of Variation-C.V.

0.4000415 0.152199

The daily average catastrophic rain (more than the calculated threshold) that can be

experienced at a recurrence interval of 20 years with 5% probability) was 164.97 and that at a recurrence interval of 5 years (with 20% probability) was 131.793 [Table.6]. Table.6. Amount of Rain fall at Certain Probability and with specific return period (After Chow, 1951

and 1954). P % T (Years) K Xc (mm) 99 1.01 -2.001 90.539 50 2 -0.083 128.507 20 5 0.083 131.793 5 20 1.759 164.971 1 100 2.669 182.985

The calculated 105.88mm and 88.928mm daily rainfall were the threshold rain for

Tindharia and Paglajhora respectively and the analysis of return period shown that 120.7 mm daily rainfall can occur at a recurrence interval of 1.4 years following Gumbel, 1954 [Table.5] and 128.507 mm daily rain had a recurrence interval of 2 years with 50% probability following Chow, 1951 and 1954 [Table.6]. That means there is every possibility for the generation of geomorphic threshold for initiation of slide due to hydrologic factor. At Paglajhora the critical rainfall is 88.93mm which is less than the estimated rainfall of 90.54 mm/day at the recurrence interval of 1.01 year with 99% probability. So it can be inferred that Paglajhora is a place of higher probability of rainfall triggering landslide phenomena in every rainy season. On the other hand at Tindharia the threshold rainfall is 105.88mm/day which is 15 mm more than Paglajhora and less than the derived rain of 128.507 mm/day at the recurrence interval of 2 years with 50% probability.

The occurrences of major landslide events with more than 90 percent certainty could be expected in every 7.5 years in case of Poisson distribution model. If we consider landslide event, then we can say that it can be expected in every 13 years with 100 percent certainty. In case of Binomial distribution the 100 percent certainty of the major landslide events are to be expected at the return period of 19 years (Fig.10).

Figure 10.



4. CONCLUSION

In the Shivkhola watershed, physical and anthropogenic processes are active on slope in an interactive combination. Construction of settlement, road and associated deforestation destabilize soil and slope. Slope is steepened, soil becomes loose and friable, lateral support is removed, soil becomes saturated by hydrological intervention. All these together leads to instability and threshold condition are achieved. Ultimately slope failure occurs and that helps to achieve temporary stability.

The study was conducted to establish the role of rainfall, antecedent and critical rainfall that can generate geomorphic threshold and introduce slope failure. The intensive field investigation depicts that the rainfall in connection to the weak lithological composition [Gneiss, mica-schist & granulitic rocks, Mylonitised granite with sub-parallel thrust, Phyllite, silvery-mica-chlorite-schist, grey sericite, and Slate phyllite with quartzite, quartz-schist & greywake schist] permits easy saturation and reduces cohesion of the slope soil and initiate down slope movement at most of the places in the Shivkhola watershed. The study area receives the orographic rainfall and all the landslides are triggered by few days and one or two day’s continuous and heavy showers. Such rain introduces the geomorphic threshold by increasing pore-water pressure and reducing the cohesion and angle of internal friction. The determined critical rainfall at Paglajhora and Tindharia called for a warning system that can be used to warn the people or to rehabilitate the people living in the hazardous areas. Study concludes that there is a greater probability of frequent occurrence of debris slide which will reduce the slope angle on landslide scar face to that of repose angle to attain temporary stability through internal feed back in a process of homeostatic adjustment.

REFERENCES

Aleotti, P. 2004. A warning system of rainfall-induced shallow failure. Engineering Geology 73, 247–265.

Alford, D. 1992. Hydrological aspects of the Himalayan region. Occasional Paper No. 18. ICIMOD Kathmandu, 68 pp.

Bloom, A. L. 1991. Geomorphology, a systematic analysis of the Cenozoic Landforms, New Delhi, Prentice Hall of India Pvt. Ltd. (pp. 76-177).

Borga et al. 1998. Shallow Landslide hazard assessment using a physically based model and digital elevation data. Journal of Environmental Geology, 35(2-30): 81-88.

Brunsden, D., Jones, D.K.C., Martin, R.P., Doornkamp, J.C. 1981. The geomorphological character of part of the Low Himalaya of Eastern Nepal. Z. Geomorph. N.F. Suppl.-Bd. 37, 25–72.

Caine, N. 1980. The rainfall intensity–duration control of shallow landslides and debris flows. Geografiska Annaler 62A, 23–27.

Caine, N., Mool, P.K. 1982. Landslides in the Kolpu Khola Drainage, Middle Mountain, Nepal. Mountain Research and Development 2, 157–173.

Campbell, R.H. 1975. Soil slips, debris flows, and rainstorms in the Santa Monica Mountains and vicinity, Southern California. US Geol. Surv. Prof. Paper, vol. 851, pp. 1–20.

Cannon, S.H., Ellen, S.D. 1985. Rainfall conditions for abundant debris avalanches, San Francisco Bay region, California. Geology 38 (12), 267–272.

Cannon, S.H. 1988. Regional rainfall-threshold conditions for abundant debris-flow activity. In: Ellen, S.D., Wieczorek, G.F. (Eds.), Landslides, floods, and marine effects of the



storm of January 3–5, 1982, in the San Francisco Bay Region, California. US Geol. Surv. Prof. Paper, vol. 1434, pp. 27–34.

Cardinali, M., Galli, M., Guzzetti, F., Ardizzone, F., Reichenbach, P., Bartoccini, P. 2006. Rainfall induced landslides in December 2004 in Southwestern Umbria, Central Italy. Nat. Hazards Earth Syst. Sci. 6, 237–260.

Carson, M.A. 1977. Application of the concepts of Threshold slopes to the Laramie Mountains, Wyoming, Inst. Br. Geogr. Spec. Publ. 3, 31-48.

Carson, M.A. 1977. Threshold and characteristic angles of straight slopes, proceedings of the 4th Guelph Symposium on Geomorphology, Norwich Geo Books, 19-34.

Carson, M.A. 1977. Angle of repose, angles of shearing resistance at angle of talus slopes, Earth Surface Processes (2), 263-380.

Ceriani, M., Lauzi, S., Padovan, N. 1992. Rainfall and landslides in the Alpine area of Lombardia Region, central Alps, Italy. In: Interpraevent Int. Symp, Bern, vol. 2, pp. 9–20.

Cotecchia, V. 1978. Systematic reconnaissance mapping and registration of slope movements. Bull. Int. Ass. Eng. Geol. 17, 5–37.

Crosta, G. 1998. Regionalization of rainfall threshold: an aid to landslide hazard evaluation. Environmental Geology 35 (2-3), 131–145.

Crosta, G., Frattini, P. 2001. Rainfall thresholds for triggering soil slips and debris flow. Proc.of EGS 2nd Plinius Conference 2000, Mediterranean Storms, Siena, pp. 463–488.

Crozier, M.J. 1997. The climate-landslide couple: a southern hemisphere perspective. In: Matthews, J.A., Brunsden, D., Frenzel, B., Glaeser, B., Weiss, M.M. (Eds.), Rapid mass movement as a source of climatic evidence for the Holocene. Gustav Fischer Verlag, Stuttgart, pp. 333–354.

Crozier, M. J. 1999. Prediction of rainfall-triggered landslides: a test of the antecedent water status model. Earth Surface Processes and Landforms 24, 825–833.

Culmann, C. 1866. Graphische Statik, Zurich.

Dahal, R.K. 1997. Engineering and environmental geological study of Kharidhunga–Tokarpa– Balephi area in the Lesser Himalaya, Central Nepal, M.Sc. Dissertation, Central Department of Geology, Tribhuvan University, Nepal, unpublished, 120 pp.

Dahal, R.K., Hasegawa, S., Masuda, T., Yamanaka, M. 2006a. Roadside slope failures in Nepal during torrential rainfall and their mitigation, In: Marui, H., Marutani, T., Watanabe, N., Kawabe, H., Gonda, Y., Kimura, M., Ochiai, H., Ogawa, K., Fiebiger, G., Heumader, J., Rudolf-Miklau, F., Kienholz, H. and Mikos, M. (Eds.), Proc. Interpraevent Int. Symp, Niigata 2006, Disaster mitigation of debris flow, slope failures and landslides, Universal Academy Press, Tokyo, vol. 2, pp. 503–514.

Dahal, R.K., Hasegawa, S., Yamanaka, M., Nishino K. 2006b. Rainfall triggered flow-like landslides: Understanding from southern hills of Kathmandu, Nepal and northern Shikoku, Japan. Proc 10th Int Congr of IAEG, The Geological Society of London, IAEG2006 Paper number 819 :1–14 (CD-ROM).

Dahal, R.K. 2006c. Geology for Technical Students - A textbook for Bachelor Level Students, Brikuti Academic Publication, Exhibition Road, Kathmandu, Nepal, 756 pp.



De Vleeschauwer, C. &. De Smedt, F. 2002. Modeling slope stability using GIS on a regional scale, Proceedings of the first Geological Belgica International Meeting, Leuven, 11-15 September 2002. Aardkundige Mededelingen, 12: 253-256.

De Smedt, F. 2005. Slope Instability analysis using GIS on a regional scale: a case study of Narayanghat-Mungling highway section, Nepal, a dissertation report presented at Universiteit Gent, Vrije Universiteit Brusssel, Belgium.

Dhakal, A.S., Amada, T.K., Aniya M. 1999. Landslide hazard mapping and application of GIS in the Kulekhani Watershed, Nepal, Mountain Research and Development 19(1), 3–16.

Dhital, M.R., Khanal, N., Thapa, K.B. 1993. The role of extreme weather events, mass movements, and land use changes in increasing natural hazards, A Report of the preliminary field assessment and workshop on causes of recent damage incurred in southcentral Nepal, July 19-20 1993. ICIMOD, Kathmandu, 123 pp.

Dhital, M. R. 2003. Causes and consequences of the 1993 debris flows and landslides in the Kulekhani watershed, central Nepal. In: Rickenmann, D., Chen, C.-L. (Eds.), Proc. 3rd Intl. conf. debris-flow hazards mitigation: mechanics, prediction and assessment, Millpress, Rotterdam, Netherlands, vol. 2, pp. 931–942.

Fenti, V. 1992. Indagini geologic-techniche sull’ area del dispositivo di misura (in Italian) in Marchi L. (ed) II Basino attrezzato del Rio Cordon, Quadevni di Ricerca, n. 13, Ragione Veneto. Dipartimento Foreste, Venezia Mestre: (pp. 109-122).

Froehlich, W. Gil, E., Kasza I., Starkel, L. 1990).Thresholds in the transformation of slopes and river channels in the Darjeeling Himalaya, India. Mountain Research and Development 10(4), 301–312.

Froehlich, W., Starkel, L. 1993. The effects of deforestation on slope and channel evolution in the tectonically active Darjeeling Himalaya. Earth Surface Processes and Landforms 18(3), 285–290.

Gabet, E.J., Burbank, D.W., Putkonen, J.K., Pratt-Sitaula, B.A., Oiha, T. 2004. Rainfall thresholds for landsliding in the Himalayas of Nepal. Geomorphology 63, 131–143.

Gerrard, J., Gardner, R.A.M. 2000. Relationships between rainfall and landsliding in the Middle Hills, Nepal. Norsk geogr. Tidsskr. 54, 74–81.

Giannecchini, R. 2006. Relationship between rainfall and shallow landslides in the southern Apuan Alps (Italy). Nat. Hazards Earth Syst. Sci. 6, 357–364.

Glade, T 1998. Establishing the frequency and magnitude of landslide-triggering rainstorm events in New Zealand, Environmental Geology, 35, 160-174.

Glade, T., Crozier, M., Smith, P. 2000. Applying probability determination to refine landslidetriggering rainfall thresholds using an empirical Antecedent Daily Rainfall Model. Pure and Applied Geophysics 157, 1059–1079.

Guzzetti, F., Cardinali, M., Reichenbach, P., Cipolla, F., Sebastiani, C., Galli, M., Salvati, P. 2004. Landslides triggered by the 23 November 2000 rainfall event in the Imperia Province, Western Liguria, Italy. Engineering Geology 73, 229–245.

Guzzetti, F., Peruccacci, S., Rossi, M., Stark, C.P. 2007. Rainfall thresholds for the initiation of landslides in central and southern Europe. Meteorol. Atmos. Phys., online first version, DOI 10.1007/s00703-007-0262-7 (accessed on 2007–08–07).

Innes, J.L. 1983 Debris flows. Prog Phys Geog 7, 469–501.



Jibson, R.W. 1989 Debris Flow in Southern Porto Rico. In: Schultz, A.P., Jibson, R.W. (Eds.), Landslide processes of the Eastern United States and Puerto Rico. Geological Society of America Special Paper, vol. 236, pp. 29–55.

JICA, 1993 Report of Japan Disaster Relief Team (Expert Team) on heavy rainfall and floods in Nepal. Japan International Cooperation Agency (JICA), JR1JDC/93-03, unpublished, 125pp.

Keefer, D.K., Wilson, R.C., Mark, R.K., Brabb, E.E., Brown, W.M., Ellen, S.D., Harp, E.L., Wieczorek, G.F., Alger, C.S., Zatkin, R.S. 1987 Real-time warning during heavy rainfall. Science 238(4829), 921–925.

Khanal, N.R., Watanabe, T. 2005 Landslide and debris flow hazards induced by heavy precipitation in Nepal. Proc. Intl. symposium on landslide hazard in orogenic zone from the Himalaya to Island Arc in Asia, pp. 111–118.

Khanal, R.K. 1991 Historic landslides of Nepal during 1902-1990 A.D., extent and economic significance. M.Sc. Dissertation, Central Department of Geology, Tribhuvan University, Nepal, unpublished, 94 pp.

Kim, S.K., Hong, W.P., Kim, Y.M. 1991 Prediction of rainfall triggered landslides in Korea. In: Bell, D.H. (Ed.), Landslides, Rotterdam: A.A. Balkema, vol. 2, pp. 989–994.

Larsen, M.C., Simon, A. 1993 A rainfall intensity-duration threshold for landslides in a humid tropical environment, Puerto Rico. Geografiska Annaler 75(1–2), 13–23.

Li, T. 1990. Landslide management in the mountain area of China. ICIMOD Kathmandu, Occasion Paper No. 15, 50 pp.

Li, T., Wang, S. 1992. Landslide hazards and their mitigation in China, Science Press, Beijing, 84 pp.

Maiti, R. 2007. Identification of Potential Slope Failure Zones of Shiv-Khola Watershed; Darjiling Himalaya, Through Critical Analysis of Slope Instability- A Step towards Rational and Scientific Management of Land, Soil and Water, UGC Sponsored Minor Research Project [F.31-210/2005(31.03.2007)]

Manandhar, I.N., Khanal, N.R. 1988. Study on landscape process with special reference to landslides in Lele watershed, central Nepal. Report submitted to Research Division, Tribhuvan University, unpublished, 53 pp.

Manandhar, S. 2000. Engineering geological studies of the landslides in the Kulekhani catchment. central Nepal, Lesser Himalaya, M.Sc. Dissertation, Central Department of Geology, Tribhuvan University, Nepal, unpublished, 130 pp.

Metteotti, G. 1996. Valutazione del rischio di franosita per unbacino di tipo alpino (in Italian), Ph. D. Dissertation, University of Padova, Italy.

Melnikov, M. and Chensokov, M. 1969. Safety in open cast minning, Moscow: Mir Publications.

MoHA, (2004) Disaster scenario of Nepal (2000–2003). Ministry of Home Affair, Government of Nepal, 157 pp.

Montgomery, D.R., Dietrich, W.E. 1994. A physically based model for the topographic control on shallow landsliding. Water Resources Research 30 (4), 1153–1171.

Neary, D.G., Swift, Jr., L.W. 1987. Rainfall thresholds for triggering a debris avalanching event in the southern Appalachian Mountains. In: Costa, J.E., Wieczorek, G.F. (Eds.),



Debris flows/avalanches: process, recognition and mitigation. Geological Society of America, Reviews in Engineering Geology 7, 81–92.

Pethick, J. 1984. An introduction to Coastal Geomorphology, (pp. 73-74) Arnold Pub. London.

Polloni, G., Ceriani, M., Lauzi, S., Padovan, N., Crosta, G. 1992. Rainfall and soil slipping events in Valtellina. Proc. of the VI International Symposium on Landslides, Christchurch, vol. 1, pp.183–188.

Polloni, G., Aleotti, P., Baldelli, P., Nosetto, A., Casavecchia, K. 1996. Heavy rain triggered landslides in the Alba area during November 1994 flooding event in the Piemonte Region (Italy). Proc. VII International Symposium on Landslides, Trondheim, vol. 3, pp. 1955–1960.

Pomeroy, J.S. 1984. Storm-induced slope movements at East Brady, northwestern Pennsylvania. US Geological Survey Bull, vol. 1618, 16 pp.

Schumm, S.A. 1977. The Fluvial System, John Willy and Sons, New York.

Schumm, S.A. 1977. Drainage Basin Morphology, Bechmark Papers in Geology, 4, Hutchinsons and Ross, Pennsylvannia: Dowden.

Schwab et al. 2002. Soil and water conservation Engineering (pp.18-47) John Wiley and Sons, Inc. New York.

Skempton, A.W. and Hutchinson, J.N. 1969. Stability of natural slope and embankment section, Proceedings 7th Int. Cong. Soil. Mech. Eng. Mexicom, 291-340.

Starkel, L. 1972. The role of catastrophic rainfall in the shaping of the relief of the Lower Himalaya (Darjeeling Hills). Geogr. Polonica 21, 103–147.

Terlien, M.T.J. 1997. Hydrological landslide triggering in ash covered slopes of Manizales (Columbia). Geomorphology 20, 165–175.

Terlien, M.T.J. 1998. The determination of statistical and deterministic hydrological landslide triggering thresholds. Environmental Geology 35 (2-3), 124–130.

Terzaghi, K. 1962. Stability of steep slopes on Hard Unweathered Rock, Geothnique, 12, 251-270.

Upreti, B.N. 1999. An overview of the stratigraphy and tectonics of the Nepal Himalaya. Journal of Asian Earth Sciences 17, 577–606.

Upreti, B.N., Dhital, M.R. 1996. Landslide studies and management in Nepal. ICIMOD, Nepal, 87 pp.

Van Burkalow, A. 1945. Angle of repose and angle of sliding friction; An Experimental study, Geol. Soc. America. Bull. 56, 669-707.

Varnes, D.J. 1978. Slope movement types and process. In Schuster R.L., Krizek, R.J. (Eds.), Landslides Analysis and Control. Special Report 176, Transportation Research Board, National Academy of Sciences, Washington D.C, pp.12–33.

White, I.D., Mottershead, D.N., Harrison J.J. 1996. Environmental systems, 2nd ed. London, Chapman and Hall, 616 pp.

Wieczorek, G.F. 1987. Effect of rainfall intensity and duration on debris flows in central Santa Cruz Mountains, California. In: Crosta, G., Wieczorek G.F. (Eds.), Debris flows/



avalanches: processes, recognition and mitigation. Reviews in Engineering Geology, vol. 7. Geological Society of America, pp. 23–104.

Wieczorek, G.F. 1996. Landslide triggering mechanisms. In: Turner, A.K., Schuster, R.L. (Eds.), Landslides: Investigation and Mitigation, Transportation Research Board, Special Report 247. National Research Council, Washington, pp. 76–79.

Wieczorek, G.F., Morgan, B.A., Campbell, R.H. 2000. Debris flow hazards in the Blue Ridge of Central Virginia. Environmental and Engineering Geoscience 6 (1), 3–23.

Wilson, R.C., Torikai, J.D., Ellen, S.D. 1992. Development of rainfall warning thresholds for debris flows in the Honolulu District, Oahu. USGS Open-File, vol. 92–521, 45 pp.

Wilson, R.C., Wieczorek, G.F. 1995. Rainfall threshold for the initiation of debris flow at La Honda, California. Environmental and Engineering Geoscience 1(1), 11–27.

Wilson, R.C. 1997. Normalizing rainfall/debris-flow thresholds along the U.S. Pacific coast for long-term variations in precipitation climate. In: Chen, C.L., (Ed.), Proc. 1st Int. Conf. on Debris-Flow Hazard Mitigation, San Francisco, American Society of Civil Engineers, pp.32–43.

Zezere J.L., Trigo R.M., Trigo I.F. 2005. Shallow and deep landslides induced by rainfall in the Lisbon region (Portugal): assessment of relationships with the North Atlantic Oscillation. Nat. Hazards Earth Sys. Sc


European Journal of Geography Volume 4, Issue 3: 38-49 October 2013 © Association of European Geographers

“DTH 1.0”: TOWARDS AN ARTIFICIAL INTELLIGENCE DECISION SUPPORT SYSTEM FOR GEOGRAPHICAL ANALYSIS OF HEALTH DATA

Dimitris KAVROUDAKIS

University of the Aegean, Geography Department, University Hill, Mytilene, Lesvos, 81100, Greece http://www.geo.aegean.gr/intro_en.htm, [email protected]

Phaedon C. KYRIAKIDIS

University of the Aegean, Geography Department, University Hill, Mytilene, Lesvos, 81100, Greece http://www.geo.aegean.gr/intro_en.htm, [email protected]

Abstract The complexity of modern scientific research requires advanced approaches to handle and analyse rich and dynamic data. Organizations such as hospitals, hold a great number of health datasets which may consist of many individual records. Artificial Intelligence methodologies incorporate approaches for knowledge retrieval and pattern discovery, which have been proven to be useful for data analysis in various disciplines. Decision trees methods belong to knowledge discovery methodologies and use computational algorithms for the extraction of patterns from data. This work describes the development of an autonomous Decision Support System (“Dth 1.0”) for the real-time analysis of health data with the use of decision trees. The proposed system uses a patient's dataset based on the patients’ symptoms and other relevant information and prepares reports about the importance of the characteristics that determine the number of patients of a specific disease. This work presents the basic concept of decision trees, describes the design of a tree-based system and uses a virtual database to illustrate the classification of patients in a hypothetical intra-hospital case study. Keywords: Decision making, geographical analysis, artificial intelligence, data mining, health geography, decision trees.

1. INTRODUCTION

The complexity of modern data analysis is constantly increasing as the number of variables involved increases. Modern scientific problems, require even greater computational power to handle and analyse available data in order to produce meaningful outputs for analysis and informed decision making. The more the variables and the characteristics of a problem, the greater the complexity in the association between problem and its characteristics. A modern group of methodologies is Artificial Intelligence (AI) which was introduced during the last decades in computational sciences. This group of methods include Machine Learning (ML) which is a category of methods for the extraction of knowledge from data to “train” a system which will later accumulate knowledge for analysis and prediction. The adaptation of AI methods to problem solving and data analysis is valuable to modern scientists as the complexity of scientific problems increases. Moreover, geographical applications require


http://www.geo.aegean.gr/intro_en.htm


http://www.geo.aegean.gr/intro_en.htm


Kavroudakis D. – Kyriakidis Ph. / European Journal of Geography 4 (2013)

advanced tools for spatial analysis. The variety, type and computational intensity involved in spatial data analysis, make AI methods valuable to the modern scientific arsenal of geographical analysis methods.

Data mining is a methodological group of AI which extracts information from data. Often the type, extent and complexity of datasets hides the underlying information and trends which are crucial for scientific analysis. Those trends and associations can be extracted with data mining operations. “Decision trees” is a common modern data mining methodology dealing with training and prediction. This approach predicts the value of a variable by knowing other available attributes and can be applied to variables of categorical nature. For example when information is available about individuals concerning: car ownership, house ownership, income category and age, then by training a decision tree model with data, a prediction model can be build to predict the car ownership (binomial variable) according to the values of the other variables. A very good exposure to the challenges and capabilities of the scientific field of data mining is the work of Witten et. al. [1] which analyses ML and data mining and provides practical recommendations.

Health geography is a field of geography focusing on the spatial characteristics of health related problems. Some of the topics of health geography can be approached by the use of AI methods. For example decision trees can be implemented towards the development of a health decision support system to understand and illustrate possible development of a disease by a patient via the use of patient's health record database. In other words a decision tree can be implemented for understanding the levels of a variable (development of disease) by parsing other data on the system. The underlying mechanism associates habits with a disease and offers a statistical model for the analysis of the development of the disease. This work illustrates the use of decision tree models for the understanding of health related datasets. After the presentation of the theory of decision trees, a model is constructed. Following the generation of a random health dataset (“arth2000”) the model is trained to predict the development of a disease in a sample of 1000 patients. Finally we discuss the potential use and extensions of the proposed decision support system.

2. DECISION TREE CLASSIFICATION

Classification is a training methodology in ML. It assigns class labels to cases based on models linking known class labels with attribute levels. Some of the most common data classification techniques to date are: neural networks [2], [3] , Bayesian networks [4–7] and Support Vector Machines [8–11]. Those approaches have their advantages and disadvantages and are suitable for different types of data analysis. Decision tree methodology is mostly suitable for classifying datasets with nominal variables and exploring relationships between the standardized variations of their attributes. Some of the advantages of this methodology are the fast learning algorithms that can be used such as ID3 [12] and C4.5 [13] and the robustness of the methodology to noise such as missing values and attribute noise. Some of the disadvantages of decision trees are the difficulty to represent the parity of values in a relationship and proportional complexity of the output diagram which sometimes can be misleading if not followed by expert analysis. In other words when data are complex and the output graph has a substantial amount of nodes and leafs, the human eye can be misled. This will not happen if the tree graph is followed by a detailed explanation of each part of the tree. A decision tree accepts a dataset of nominal data and produces a dendrogramatic representation [14] of the data variables according to their levels. The generic algorithm used for such a classification is the following:

1. A is the best decision attribute for the next node 2. Assign A as decision attribute for the node



3. For each value of A, create a new descendant of the node 4. Sort data by leaf nodes 5. Iterate over leaf nodes, until data are classified The result of this algorithm produces a number of nodes and leafs illustrating the number

of potential decisions from the attributes of the data. This categorization of potential decisions groups all possibilities, and counts the most prominent ones. The modelling process splits the data into two subsets: one for learning and the other for prediction. The first subset of the data is used as a trainee for the model and teaches the model to understand the relationship between attributes. The second subset is used for evaluating this knowledge. The error calculation is based on the number of successful predictions of the model. The splitting of the data depends on the number of cases. The error quantification is also associated with the number of cases in the data, the number of variables and the number of levels in the ordinal scale, discretizing the range of variability of continuous explanatory variables.

Decision trees can process data involved in various disciplines and develop a knowledge discovery tool to predict levels of ordinal dependent variables. In the broader scientific field of health sciences there have been some interesting attempts to use decision tree methodologies. The work by Andreescu et.al. [15] illustrates the use of decision trees in the prediction of patients respond to treatment of late-life depression. With a number of 461 records, the authors developed a hierarchy of predictors with decision trees. Additionally, the work of Mann et. al [16] aims of determining the risk for a suicide attempt in psychiatric patients with the analysis of multiple risk factors. Decision trees method has been used in a health dataset of 408 patients with mood schizophrenia or personality disorders to distinguish possible attempters. Another interesting use of decision trees in health sciences is the work by Zhang et.al [17] which is an attempt to demonstrate the effectiveness of one treatment against another with respect to pregnancy in poly-cystic ovary syndrome (PPCOS). That work used a dataset of 445 women who ovulated in response to treatments among a dataset of 626 participants. Decision trees was used to reflect treatment results between types of the syndrome. Furthermore, Koko et. al. [18] described the evaluation of various decision tree methods on problems of orthopaedic fracture data and concluded that there are some limitations on the accuracy of the model and the sensitivity of the decision tree size. Tsien et. al. [19] in their research about classification trees for diagnosing myocardial infraction, concluded that ML methods such as decision trees can be used in medicine for supporting early diagnostic decisions. Jones et. al. [20], illustrate the use of decision trees in the identification of signals of possible drug reactions and concluded that data mining methods, such as decision trees can be a promising tool for identifying new patterns in medical datasets. Dantchev et. al. [21] argue that decision trees are still in experimental stages and remain difficult to apply to clinical practice in psychiatry. Nevertheless in the same report they argue that those tools allow researchers to see epidemiological data from a more generalized perspective and focus on new priorities. Letourneau et. al. [22] focused on decision making techniques for chronic wound care and concluded that decision trees can help decision making by guiding trained personnel through assessment and treatment options. Another notable example of use of decision trees in health sciences can be seen in the work of Alemi and Gustason [23] , who describe some analytical tools that aid decision making such as decision trees, also includes a number of examples with decision making scenarios.

3. USE OF DECISION TREES METHOD

This part describes the use of Decision Trees in the proposed health decision support system. An artificial dataset is used for the training and validation of the method. The dataset consists of five independent variables (city, age, sex, activity, the milk) and one dependent variable



(hypothetical disease AD). The cases of the dataset represent visits of individuals to a health care facility (hospital). It is assumed that during each visit the medical personnel examined the individuals and recorded information about those variables. The total cases in the dataset is n=2000 (1000 training, 1000 evaluating) as can be seen in table 1. Table 2 shows the probabilities for each variable's level, used for the generation of the “arth2000” dataset. Those probabilities are also used for the evaluation of the accuracy of the classification. This work assumes that independent variables influence the dependent variable and via the decision tree methodology illustrates the type and amount of this influence. Subsequently, the model will be able to predict the probability of existence of disease AD by processing the levels of the five dependent variables of an individual. The process of prediction can be complex and may also depend on the type and quality of the available dataset. The variables of the data are: the city variable which indicates the area of living (5 levels) of the patient, the age variable is a categorization of the age group of the patient (4 levels). Activity indicates the type of patient's active or passive type of living (2 levels) and milk variable indicates the daily consumed milk units (3 levels).

We prepare a model which uses information gain as a quality measure to populate a dendrogram. Information gain is represented as the entropy value of the data passed from the model. If the data values are new and haven't been processed earlier from the model, then they add to the overall information scheme. This information gain is calculated by counting the number of previous occurrences of the particular combination of data values, in the dataset. For example if the model processes a list of 10 individuals consisting of 9 males and one female, the information gain increases by one when the model is processing the 10th individual because up until the 9th individual the model only knew about the existence of just one sex. After handling the new sex-level (female) the model creates a new category of individuals and assigns all new females to that.

Table 1. Overview of the "arth2000" dataset

City of residence Age category Sex Daily activity Milk

consumption Suffer from AB

disease

Athens mature female high high Yes

Mytilini old female average average Yes

Mytilini mature male low low No

... ... ... ... ... ...

... ... ... ... ... ...

Additionally the model uses no pruning mechanism as this would limit the extent of the tree and because the number of cases in the “arth2000” dataset is limited. When the model processes the individuals, the minimum records per node is set to 5. In other words, the information gain weight should reach 5 individuals before creating a new leaf. This is an empirical value and varies depending on the type of data or the type of analysis and the required complexity for the results.



Table 2. Probabilities of levels for the variables of the "arth2000" dataset

city age activity milk AD disease

Athens 0.1 middle 0.2 active 0.5 average 0.3 No 0.4

Chios 0.2 old 0.4 passive 0.4 high 0.2 Yes 0.6

Crete 0.2 young 0.1 low 0.5

Mytilini 0.3 mature 0.3

Rodos 0.2

Figure 1 depicts the overall structure of the developed model. Initially, health data are

inserted via a csv reader. Those data could be in principle fetched from the examination rooms of a hospital and following the model processing procedure, they can be presented to decision makers in real time. The proposed procedure then splits the database in two parts for cross validation. In other words, the model will learn from the first part of the data and then evaluate the quality of the knowledge upon the second half of the data. The process of learning is held in module number 3 and the linkage of the dataset is taking place in module number 4. The linkage evaluates the leaning ability of the model with the remaining data.

Figure 1. Work-flow process diagram for data mining of the arth2000 dataset

Decision Trees approach, helps solve a problem, which in this case-study is the understanding of possible future existence of disease AB to a number of patients by knowing a limited number of information about each patient. This approach is used to represent the various decision points along the examination of a potential patient. As can be seen in table 3, (Rule 3) according to the arth2000 dataset, if the patient is from Chios city, it has an increased probability to suffer from AB disease. The examination personnel, then needs to ask the patient about his/her age as in Rule 6, patients of mature, old and young age, have an additional increased probability to suffer from disease AB. The resulted decision tree, offers a list of characteristics which have increased probability over the possible existence of AB disease, according to the arth2000 dataset. It describes the logical steps required for determining whether an individual has increased probability to suffer from disease AB (dependent variable) by knowing the value of a number of other variables (independent variables).

The decision tree methodology consists of a root node split by a single variable into partitions. In turn those partitions become nodes to be split further. This divide-and-conquer approach continues until no further splitting would improve the performance of the model. The performance is the ability of the model to understand the possible categorization of a case, based on its attributes. This ability of the model, increases as the statements incorporate



additional knowledge about the training dataset. In other words, the more the information gain from a categorization, the higher the ability of the model to categorize cases with less information. The categorization statements of the produced decision tree are depicted in Table 3.

Table 3. The categorization statements for the “arth2000” dataset

Rule Appearance of disease AB cases percentage categories

1 No 450/1000 45%

2 No 348/791 43% milk=high,low

4 No 239/569 42% city=Athens,Crete,Mytilini,

Rodos milk=high,low

5 No 109/222 49% city=Athens,Crete,Mytilini,

Rodos milk=average

10 No 51/120 42% city=Crete,Rodos milk=average

11 Yes 44/102 43% milk=average city=Athens,Mytilini

3 No 102/209 48% city=Chios

6 No 83/180 46% city=Chios age=mature,old,young

7 Yes 10/29 34% city=Chios age=middle

For each statement (row) the first column is the id of the statement. The disease AB

column indicates that this statement can categorize cases that may suffer from disease AB. The “cover” column shows the number of cases that have been categorized with this rule. The percentage column shows the percentage of the categorized cases with respect to the total number of cases. Finally, the “categories” column shows the information gain from each rule. For example rule 11 indicates that cases from Athens or Mytilene with average consumption of milk, have 43% probability to suffer from disease AB. The combination of those rules can categorize this particular “arth2000” dataset with 100% accuracy.

In order for the model to judge how good a potential split (node-leaf) is, the information gain rule is used, which creates a new split at the attribute with the highest information gain. This approach creates new splits only when they will create concrete partitions of the dataset. The split function strategy for this model is the entropy reduction strategy. The greater the information from a categorization, the greater the knowledge of the model for future categorizations. Finally the minimum number of observations in a node before attempting a split (splitting factor) is 100 cases for 1000 cases. This reduces the tree complexity and produces a more readable representation. The selection of the splitting factor depends on the number of total cases and the amount of detail required in the results. The more complex the results, the more the leafs of the decision tree.

The very same categorization process can be used with any other categorical dataset to prepare categorization rules. This makes the approach generalizable and flexible. A factor that determines the accuracy of the results is the number of levels for each variable. Binomial variables can be categorized with less rules than variables with 4 or 5 levels. This is making sense if we consider that entropy increases for variables with larger number of levels. This forces the model to produce more rules in order to fully categorize a dataset. On the other hand, binomial variables can be easily categorized and require less rules. Finally, a fully



randomized dataset, where there is no relationship between variables, may require a great number of rules to be fully categorized. The tendency of the dataset towards a random distribution is directly associated with the inability of the decision tree to categorize all cases with less rules.

The methodology of data categorization discussed, can be potentially useful to health facilities, such as hospitals for categorizing patient records and present statistics based on the patient profile. For example a large hospital with a great number of daily visits, can produce a good amount of data related to patient characteristics and health problems. Those data can be obtained from a patient upon his/her arrival and stored in a database. Then, a centralized computational decision support system can process this dataset and prepare correlations between patient's characteristics. A decision tree can be implemented in R statistical language, with the use of the “part” library [24], to prepare a statistical description for such a database. The administration of a hospital may access the results of the process and use visualizations for decision support. The results may indicate trends and patters that are not initially visible and can offer a centralized and categorized view over the characteristics of each disease.

4. AUTOMATED HEALTH DECISION SYSTEM

The proposed system makes decisions based upon health related datasets and evaluates the possibilities of occurrence of a control variable. The application of the proposed system is the statistical analysis of health databases from non-expert uses such as managers and directors of hospitals. This analysis can also be helpful in early examination procedures or in decision making in epistemological analysis.

Considering the lack of centralized statistical tool in the Greek Health Sector, this work is innovative because it describes and proposes the use of such a tool and advocates the use of not only descriptive statistics but also the use of AI for advanced statistical analysis. Sometimes it is difficult to fully understand the big picture of a cause-effect relationship especially when it is hidden deep into a great amount of data. The proposed system is using data mining methods and databases to construct a “hospital-oriented” computer system that will prepare on the fly data statistics. More specifically, this work describes the design and the details of a proposed system that can be installed in a health facility such as hospitals, and prepare on-the-fly statistics about the patient's database. As illustrated in figure 2 the flow of information starts from the examination room (point 1) where the basic description of the characteristics and the medical record are transformed into a digital record (point 2) and inserted into a database (point 3). This process can be facilitated by modern palmtops with live html forms that can either create a new medical record or update an existing one with new medical information. The database of the system can retrieve data with queries. Instead for the researchers/managers (point 5) to use traditional SQL queries to retrieve tabular results, they can use a series of predefined actions that will call a number of queries. The pre-defined actions will present the results in graphical form (point 4). One of those predefined actions may include the use of a decision tree to analyse the data and observe the relationship between variables.

Figure 2. The abstract design of the proposed decision system



This proposed system requires a number of “examination-room computers” which will be used for data entry during examination. The software required for those machines, should be just basic intra-net browsing and an html compatible client (web browser). The main database will be hosted on an in-house server which will be an average computer with an installation of a database (PostGreSQL, MySql). In this computer the freely available statistical environment R is required which will prepare statistical analysis, and graphs. Finally the managerial and administrative staff will need average computers with basic intra-net browsing for selection of predefined actions and visualization of results.

The preparation of predefined actions saves time and effort and provides a relatively error-free environment for the generation of statistics. The following code snippet (code 1) shows an example of a predefined action which is triggered by a managerial staff to prepare a description of the age categories of the patients in the database. The following code snippet (code 1) is written in the R statistical programming language and its output is depicted in figure 3.

• Lines 1-3 import the required libraries and packages for the analysis. • Lines 5 and 6 activate the database driver and connect to the database. • Lines 8 and 11 discover the available tables in the database and list its fields. • Lines 14 and 15 retrieve all available data from the database table • Lines 17-20 convert the data according to age categories • Lines 22-24 prepare the diagram variables • Lines 26-28 generate the bar-plot diagram

The generated bar-plot show the absolute number of patients and the percentage for each age group. This predefined statistical analysis generates output that informs the user (managerial and admin staff) about the age groups of the patients that visit the hospital.

Code 1. Snippet of R code for the generation of bar-plot with the age structure of the patients. The generated plot is figure 3.



Figure 3. Result of an example predefined action showing the age groups of patients in the dataset

Another predefined statistical “action” could be the profiling of the dataset with decision trees methodology. This methodological approach may serve as an information categorization process that generates a dendrogram of groups of cases. Code 2 depicts the R code of such a process which may be used in the proposed computational system.

Code 2. Snippet of R code for the generation of sample data and the subsequent construction of a decision tree from them.

After the installation and calibration of this autonomous health decision support system, the users can use it without any knowledge on statistics or artificial intelligence. The predefined actions will run code snippets that will prepare various statistical analysis. It is an autonomous system as it uses only the predefined statistical actions and does not require any input by the user. This enables the end user to focus only on a number of important and error-free statistical analyses. Of course as with other similar systems, the quality of the imported data will influence the quality of the exported statistics.

5. GEOGRAPHICAL ANALYSIS OF HEALTH DATA

The data that can be collected form the heath care facilities, include place of residence of the patient. This is important as can provide spatial attributes to health records for further analysis. The health data that have spatial records can be presented in maps and analysed by area in order to provide a better insight on the spatial distribution of health events along with other information about the patient such as proximity to health facilities, socio-economic characteristics etc. Those type of information could be used for geographical analysis of a disease outspread and map the areas which have significant amount of incidents. The understanding of spatial characteristics of a disease outspread can help decision makers to provide better health services and information to the general public for the protection of public health. Geographical epidemiological studies aim to understand the spatial characteristics of health data and formulate hypotheses regarding the spatial causes and effects of a disease [25]. Some of the different branches of spatial epidemiology are disease mapping, cluster identification and spatial socio-economic analysis of a disease[26]. Understanding the greater spatial trends of a disease as well as mapping the spatial distribution of a disease from health records of a hospital, can be a challenging task especially



due to privacy and ethical reasons. Nevertheless, if handled with care, health data records that have spatial attributes can be very useful in early warning epidemiology systems and provide a different approach on understanding the causes and effects of a disease outspread in an area. Automated systems of identifying underlying trends such as decision trees, provide the necessary lay of data mining, which extracts knowledge from individual records. The use of such an autonomous system can spare financial and other resources by avoiding the manual examination of data records and by quickly indicating the geographical characteristics of an event.

6. DISCUSSION AND CONCLUSIONS

The automated system discussed in this work may provide a basic mechanism for real-time analysis of health related data. The statistical analysis of databases may not substitute the actual medical diagnosis from specialized personnel in any case. Nevertheless it is a first step towards the automation of diagnosis and a valuable tool for decision makers and researchers to understand and identify data patterns.

The arth2000 dataset is but an arbitrary dataset consisting of just 2000 cases. No real data have been used in this stage of the research as there is lack of free health database for Greece. In a future state of this research, there is the possibility to use anonymized data from Greek Health Public System. One of the benefits of using decision tree methodology is that by scaling up the number of the cases and using a real medical dataset consisting of more than 10000 cases, the accuracy of the model will increase, and the model will be able to learn faster. Additionally, the geographical analysis of the model results can be facilitated by analysing the effects of disease AB by city which is the second determinant factor of disease AB. This may indicate that disease AB is more common in some geographical areas than others. This may be the basis for a more advanced spatial analysis and may lead to more sophisticated results, such as the identification of the reasons and effects of disease AB on other characteristics of daily life of individuals in those areas. Another interesting point is that this analysis can be automated. This means that an intelligent system can be constructed with very basic hardware such as a home-range personal computer which accepts data from a hospital and prepares instantly (in real time) statistical reports about the profile of the patients. Finally, This proposed system may be used for research and decision support processes and act as a data exploration tool.

With the use of ML methods such as decision trees, one can better reveal information which is hidden in data. Complex relationships that may exist in very large datasets are sometimes difficult to understand and may require a great number of computations and cross-tabulations. Analysis of rich and complex datasets can be valuable to the health sector as it may reveal underlying geographical patterns for diseases, symptoms and characteristics of patients. Finally, decision trees in general may act as a framework to consider the probability of events and pay-offs of decisions in various data analyses, not only in health-related sectors but also from in sectors such as geography marketing and logistics.

7. SUMMARY

• What was already known to the topic ◦ Artificial intelligence techniques can give a very good insight in medical data and

provide good understanding in related variables ◦ Health geography is a scientific area which under good knowledge of the

underlying mechanisms, can provide meaningful information about geographical distribution of health-related data



◦ Decision trees is an artificial intelligence approach which has been successfully used in a number of scientific areas and provides a good understanding on the relationship of variables in a dataset.

• What this study added to our knowledge ◦ Computerised systems with the incorporation of AI, may give medical personnel a

very good understanding regarding the topic in research ◦ Decision trees can be used in such systems in order to facilitate the understanding

of the relationship between patients characteristics ◦ Hospitals in Greek health sector do not yet have expert centralized computer

systems for aggregate depiction of patients data. ◦ A novel expert system in Greek health Sector may give a very good understanding

on patients data and interconnect patient's characteristics such as geographical area of residence.

REFERENCES

[1] I. H. Witten and E. Frank, Data Mining: Practical Machine Learning Tools and Techniques, 3rd ed. Morgan Kaufmann Pub, 2011.

[2] S. Haykin, Neural Networks: A Comprehensive Foundation. Prentice Hall PTR Upper Saddle River, NJ, USA, 1994.

[3] M. T. Hagan, H. B. Demuth, and M. H. Beale, Neural Network Design. PWS Boston, MA, 1996.

[4] D. B. Rubin, “The bayesian bootstrap,” The Annals of Statistics, pp. 130–134, 1981. [5] F. V. Jensen, An Introduction to Bayesian Networks, vol. 210. UCL press London, 1996. [6] F. V. Jensen and T. D. Nielsen, Bayesian networks and decision graphs. Springer Verlag,

2007. [7] A. Darwiche and E. Corporation, Modeling and Reasoning with Bayesian Networks.

Citeseer, 2009. [8] N. Cristianini and J. Shawe-Taylor, Support Vector Machines. Cambridge University

Press, 2000. [9] M. A. Hearst, S. T. Dumais, E. Osman, J. Platt, and B. Scholkopf, “Support vector

machines,” Intelligent Systems and their Applications, IEEE, vol. 13, no. 4, pp. 18–28, 2002.

[10] A. W. Moore, “Support Vector Machines,” http://jmvidal. cse. sc. edu/csce883/svm14. pdf, vol. 31, p. 2001, 2007.

[11] I. Steinwart and A. Christmann, Support Vector Machines. Springer Verlag, 2008. [12] J. Cheng, U. M. Fayyad, K. B. Irani, and Z. Qian, “Improved decision trees: A

generalized version of ID3,” in Proceedings of the Fifth International Conference on Machine Learning: June 12-14, 1988, University of Michigan, Ann Arbor, 1988, p. 100.

[13] J. R. Quinlan, C4.5: Programs For Machine Learning. Morgan Kaufmann, 1993. [14] J. B. Phipps, “Dendrogram topology,” Systematic Biology, vol. 20, no. 3, p. 306, 1971. [15] C. Andreescu, B. H. Mulsant, P. R. Houck, E. M. Whyte, S. Mazumdar, A. Y.

Dombrovski, B. G. Pollock, and C. F. Reynolds, “Empirically Derived Decision Trees for the Treatment of Late-Life Depression,” Am J Psychiatry, vol. 165, no. 7, pp. 855–862, 2008.

[16] J. J. Mann, S. P. Ellis, C. M. Waternaux, L. XINHUA, M. A. Oquendo, K. M. Malone, B. S. Brodsky, G. L. Haas, and D. Currier, “Classification trees distinguish suicide attempters in major psychiatric disorders: a model of clinical decision making,” The Journal of clinical psychiatry, vol. 69, no. 1, pp. 23–31, 2008.



[17] H. Zhang, R. S. Legro, J. Zhang, L. Zhang, X. Chen, H. Huang, P. R. Casson, W. D. Schlaff, M. P. Diamond, S. A. Krawetz, C. Coutifaris, R. G. Brzyski, G. M. Christman, N. Santoro, and E. Eisenberg, “Decision trees for identifying predictors of treatment effectiveness in clinical trials and its application to ovulation in a study of women with polycystic ovary syndrome,” Human Reproduction, vol. 25, no. 10, pp. 2612 –2621, 2010.

[18] P. Kokol, M. Zorman, M. Stiglic, and I. Malcic, “The limitations of decision trees and automatic learning in real world medical decision making,” in Proc. 9th World Congr. Med. Inform. (MEDINFO-98), 1998, vol. 52, pp. 529–533.

[19] C. L. Tsien, H. S. Fraser, W. J. Long, and R. L. Kennedy, “Using classification tree and logistic regression methods to diagnose myocardial infarction,” Stud Health Technol Inform, vol. 52 Pt 1, pp. 493–497, 1998.

[20] J. K. Jones, “The role of data mining technology in the identification of signals of possible adverse drug reactions: value and limitations,” Current Therapeutic Research, vol. 62, no. 9, pp. 664–672, Sep. 2001.

[21] N. Dantchev, “Decision trees in psychiatric therapy,” Encephale, vol. 22, no. 3, pp. 205–214, 1996.

[22] S. Letourneau and L. Jensen, “Impact of a decision tree on chronic wound care,” J Wound Ostomy Continence Nurs, vol. 25, no. 5, pp. 240–247, 1998.

[23] F. Alemi and D. H. Gustafson, Decision Analysis for Healthcare Managers. Health Administration Press, 2006.

[24] T. M. Therneau, B. Atkinson, and B. Ripley, “Rpart: recursive partitioning,” R package version, vol. 3, pp. 1–23, 2005.

[25] M. Rezaeian, G. Dunn, S. S. Leger, and L. Appleby, “Geographical epidemiology, spatial analysis and geographical information systems: a multidisciplinary glossary,” J Epidemiol Community Health, vol. 61, no. 2, pp. 98–102, Feb. 2007.

[26] P. Elliott and D. Wartenberg, “Spatial epidemiology: current approaches and future challenges,” Environmental health perspectives, vol. 112, no. 9, p. 998, 2004.



URBAN CHANGE DETECTION BASED ON REMOTE SENSING AND GIS STUDY OF SALEM REVENUE DIVISION, SALEM DISTRICT, TAMIL NADU, INDIA.

Shanmugam TAMILENTHI

Tamil University, Department of Earth science, Thanjavur - 613010, India http://www.tamiluniversity.ac.in/, [email protected]

Rajagopalan BASKARAN

Tamil University, Department of Earth science, Thanjavur - 613010, India http://www.tamiluniversity.ac.in/, [email protected]

Abstract Rapid urbanization has significant impact on resources and urban environment. With increased availability and improved quality of multi-spatial and multi-temporal remote sensing data, it is now possible to detect urban changes in a timely and cost-effective way. This study aims to quantify changes in urban area of Salem revenue division, Tamil nadu located in India, using Land sat and IRS-LISS-III image. Urban changes were detected by satellite images of Land sat MSS in 1973, Land sat and IRS-LIS-III in 2010 using a geographic information system (GIS). The settlement area has grown 2.07 times during the past 37 years as a result of industrial development, population growth and permanent migration was the main driving forces for settlement area expansion. According to the results, the extent of urban areas 69.899sqkm and 144.97sqkm in the years 1973 and 2010 respectively. Keywords: Settlement, Salem revenue division, Remote sensing, Change detection, and GIS

1. INTRODUCTION

Land use is influenced by economic, cultural, political, and historical and land – tenure factors at multiple scales. Land use referred to as man’s activities and the various uses which are carried on land. Urbanization is inevitable, when pressure on land is high, agriculture incomes are low and population increases are excessive, as is the case in most of the developing countries of the world. In a way urbanization is desirable for human development. However, uncontrolled urbanization has been responsible for many of the problems, our cities experiences today, resulting in substandard living environment, acute problems of drinking water, noise and air pollution, disposal of waste, traffic congestion etc. To improve these environmental degradations in and around the cities, the technological development in relevant fields have to solved these problems caused by rapid urbanization, only then the fruits of development will reach most of the deprived ones. European Journal of Geography - ISSN 1792-1341 © All rights reserved

http://www.tamiluniversity.ac.in/


http://www.tamiluniversity.ac.in/


Tamilenthi S. – Baskaran R./ European Journal of Geography 3 3

Recent technological advances made in domain of spatial technology, cause considerable impact on planning activities. This domain of planning is of prime importance for a country like India with varied geographic patterns, cultural activities etc. The purpose of using GIS is that, maps provide an added dimension to data analysis, which brings us one step closer to visualizing the complex patterns and relationships that characterize real-world planning and policy problems. Visualization of spatial patterns also supports change analysis, which is important in monitoring of social indicators. This in turn should result in improving need assessment.

According to Macleod and Congalton (1998), in general, remote sensing considers following four aspects of change detection (a) detect the changes, (b) identify the nature of change, (c) measure the aerial extent of change and (d) assess the spatial pattern of change.

The objectives of this paper are to explain remote sensing and GIS applications in various stages of planning, implementation and monitoring of the urban area.

1.1. Overview of Detect Changes in Urban A variety of change detection techniques are available for monitoring land use/land cover changes. These techniques can be grouped into two main categories: post classification comparison techniques and enhancement change detection techniques (Nelson, 1998).

(a). Post classification techniques The post classification technique involves the independent production and subsequent comparison of spectral classifications for the same area at two different time periods (Mas, 1999). Post classification techniques have the advantage of providing direct information on the nature of land cover changes. The classification process used with these techniques can be either supervised or unsupervised.

Sohl (1999) reported accuracies of 96 percent for the identification of new forest land and 62 percent for new agricultural land using a post classification technique in a semi-arid environment. Furthermore, Sohl (1999) noted the strength of the method for providing users with a complete descriptive comparison between images. Pilon et al. (1988) employed post classification in combination with a simple enhancement technique to differentiate areas of human induced change from areas of natural change. Mas (1999) also obtained the highest accuracy with this technique in a study comparing six different techniques.

(b).Enhancement change detection techniques Enhancement techniques involve the mathematical combination of images from different dates which, when displayed as a composite image, show changes in distinctive colors (Pilon et al. 1988). The enhancement change detection techniques have the advantage of generally being more accurate in identifying areas of spectral change (Singh, 1989). However, these techniques often require additional analysis to characterize the nature of the spectral change, and also require more accurate image normalization and co-registration.

(i). Image differencing Image differencing is a technique by which registered images acquired at different times have pixel DN values for one band subtracted from the corresponding pixel DN values from the same band in the second image to produce a residual image, which represents the change between the two dates (Mas, 1999).

Ridd and Liu (1998) reported image differencing was fairly effective in its ability to detect change in an urban environment, with TM band 3 producing the highest accuracies. Sunar (1998)



and Sohl (1999) reported that the image differencing technique was extremely straightforward, but with the qualification that image differencing technique becomes slightly more complicated when using multiple bands, instead of single bands, due to the difficulty of interpreting the colors of multiband false color composites.

(ii) Principal component analysis Principal component analysis (PCA) is a commonly used statistical method for many aspects of remote sensing image analysis, including estimation of the underlying dimensions of remotely sensed data, data enhancements for geological studies, and land cover change detection (Fung and Le Drew, 1987). The PCA technique for change detection requires the separate images first be stacked in a multi-temporal composite image (Sunar, 1998). The major strength of this technique is its ability to reduce the dimensionality of the data with relatively minor loss of overall information content. The major weakness of this technique is that it can be difficult to interpret.

Li and Yeh (1998) compared principal component analysis to post classification techniques and concluded that principal component analysis was much more accurate than post classification techniques and therefore suggested it as an accurate alternative for detecting land use change.

(iii) Normalized difference vegetation index (NDVI) The Normalized Difference Vegetation Index (NDVI) estimates the vitality of vegetation by exploiting the known gap in vegetation reflectance between the visible and near infrared channels. Common change detection methods include the comparison of land cover classifications, multi-date classification, band arithmetic, simple rationing, vegetation index differencing and change vector analysis (Jomaa, 2003). The NDVI is calculated as a normalized ratio (ranging from -1 through 1) from the NIR and the red band and emphases apparent vegetation (Sabins, 1996).

2. STUDY AREA

Salem is an interior district of Tamil Nadu in India with an area of 8634.23 Km2 (Fig.1) and is bounded by Dharmapuri district on the North, Coimbatore on the West, South Arcot on the northeast and Tiruchirapalli on the South and South-West. The district lying between latitudes N 11°00’ and 12°00 and longitudes E 77°40’ and 78°50’. Study area is extracted from the distric map with area covering 1737 sqkms.It is The lying between latitudes N 11°25’ and 11°55 and longitudes E 77°48’ and 78°32’.

The soils of Salem District can be assorted into the main types viz., Red Calcareous, Red non-calcareous, brown soil calcareous, Red collurial careareous, Red collurial non calcreous, Black soils, Alluvial calcareous, Brown soil non calcareous.

Salem district is underlain entirely by Archaean Crystalline formations with Recent alluvial and Colluvial deposits of limited areal extents along the courses of major rivers and foothills respectively. Weathered and fractured crystalline rocks and the Recent Colluvial deposits constitute the important aquifer systems in the district. Ground water occurs under phreatic conditions and is developed by means of dug wells. They are important from ground water development point of view in the hilly terrain.

The District has a hot tropical climate with temperature ranging from 18.9° C (Minimum) to 37.9° C (Maximum) and the relative humidity is high at 79% with an average ranging from 80% to 90%. European Journal of Geography - ISSN 1792-1341 © All rights reserved


The major source for groundwater in the study area is rainfall during monsoonal season. The average 10 years annual rainfall is about 759.03 mm.

Figure 1. Study area- Salem Revenue Division.

3. MATERIALS AND METHODS

In the present study, for assessing the temporal changes in the Urban/Settlement cover Landsat 1 Panchromatic RBV,MSS- 154/52, 57X57, 1973 from GLCF and IRS LISS III (P6)- 101/65, 23.5 x 23.5, 2010 from NRSC were used. As a time series data, the Survey of India (SOI) toposheet of 1972 was also used. Moreover, the urban planning reports and administrative maps were also taken into account. (Fig.2)

Figure 2. Methodology flow chart

The study area map was prepared from SOI topographical sheets on 1:50,000 scale. The

settlement in the study area, during 1973 and 2010 were derived from the Satellite images were



compared with one another and were used for carrying out change detection studies for the period 1973 and 2010.

The same classes were then visually interpreted from the 1973 satellite data by using the common image interpretation elements. Necessary field checks were carried out and correction were made at required places. Then, the software such as Arc GIS9.3 and Erdas imagine9.2 were used to prepare the urban/settlement cover changes during 1973-2010.

3.1. Change Detection Methods Adopted in this Study The change detection techniques will be discussed, using the two main categories, post-classification comparison techniques and enhancement change detection techniques described in the literature section.

4. IMAGE ANALYSIS

The acquired images of 1973 and 2010 are used to prepare the cluster of study area in order to better analysis purpose which clearly shown all categories (fig.3)

Figure 3. Cluster analysis for 1973 and 2010 data.

4.1. Post Classification Techniques For the classification-based approach a supervised classification was done on both the 1973s and 2010s images for both test sites. The main advantage of using a supervised classification is that was able to produce the same number of classes for each set of images.

The classifications were then combined using the to produce a map of all class transitions (Erdas Imagine 8.4 Tour Guides, 1999).The transitions were then assigned colors, based on the type of change that was occurring, to produce the final output image.

The maps clearly express the spatial cover of settlements in 1973 and 2010(Fig.4) and it occupied 69.89sqkm in 1973 and 144.97sqkm in 2010.The settlements/urban change area is 75.07 sqkms(Fig.5)



Figure 4. Settlements cover in 1973 and 2010.

Figure 5. Settlements cover change in 1973 and 2010.

4.2. Enhancement Change Detection Technique

4.2.1. Principal component analysis PCA was performed using the Imagine Principal Component program in the standardized form (Erdas Imagine 8.4 Tour Guides, 1999). Out of all four bands of each of the 1973s and 2010s images the settlement showing is extracted for both and others were made false colour composite to create the major change classes. Here in the high shows that settlement occupied area with dark red colour(Fig.6)



Figure 6. Principal component analysis -1973 and 2010

4.2.2. Normalized difference vegetation index (NDVI) NDVI is calculated from the visible and near-infrared light reflected by vegetation. Calculations of NDVI for a given pixel always result in a number that ranges from minus one (-1) to plus one (+1);0 means no vegetation NIR=Red +1 (0.8 -0.9) indicates the highest possible density of green leaves. 0 to -1 Indicated higher red reflectance than NIR.

The NDVI analysis reveals that 1)Decrease in NDVI between two scenes will be the result of new development 2)Increase in NDVI between two scenes will be the result of forest re-growth 3)Urban changes in red signal may be unrelated to vegetation(Source: NSAS Earth Observatory)

The two NDVI results shows there is a increase in vegetation due to government policy on aforestation and encourage of social forest besides permanent pasture and tree type of crop commercial cultivation. The settlement area is also increased in the same side by side vegetal cover too along with their residential area (Fig.7).



Figure 7. Normalized Difference Vegetation Index

4.3. Change Detection The analysis reveal the following information and changes also shown in clearly in the table.(Table 1.) This will help the planners and other researchers for further research at micro level and macro level. The changes is mostly cause of human inference which affects the natural ecosystem one or other way. The normal temperature raised significantly compare with last 3 decades this result of urbanization and settlement expansion.

Table 1. Showing spatial changes.

Sl. No Components 1973 2010 Changes in

area(Sqkm)* Changes in

%

1 Settlements 69.89 144.977 +75.07 +207.4

2 Mining activities 17.44 18.96 +1.52 +8.71

3 Forest 611.95 594.92 -17.95 -2.93

4 Rock expose 31.18 29.11 -2.07 -6.63

5 Grass land and others 714.04 604.12 -109.92 -15.39

6 Open land 292.5 344.99 +52.49 +17.94

(*+ is Increase, - is Degrease of area)

5. RESULTS AND DISCUSSION

This section deals with the settlements, urban and human occupancy land covers in the study area over the period of 37 yrs. Settlement and urban cover categories in 1973 and 2010: The spatial cover of 1973 and 2010 is vividly shown in Fig. 3.

In the year 1973, Settlement /urban occupied by 69.89 km2, which is about 4.02 % of the total study area. Mining occupied 17.44Km2 with 1%, forest 611.95 km2 with 35.23%, rock European Journal of Geography - ISSN 1792-1341 © All rights reserved


expose is 31.18 Km2 with 1.79%, Grass and others is 714.04 km2 with 41.1% and Open land is 292.5 km2 with 16.83%.

In the year 2010, Settlement /urban occupied by 144.977 km2, which is about 8.34 % of the total study area. Mining occupied 18.96 Km2 with 1.1%, forest 594.92 km2 with 34.24%, rock expose is 29.11 Km2 with 1.67 %, Grass and others is 604.12 km2 with 34.77% and Open land is 344.99 km2 with 19.86%.

The thirty seven years (1973-2010) of urban and settlement/human interference variations has been precisely analyzed.

The study reveals that the rate of settlement and urban occupancy is more during the period 2010 with doubled the settlement area of 144.977 km2. The result shows that the population increases and migration taken place. Mining area has increased with 1.52 km2 8.71%. Forest area degreased 17.03 km2 with 2.93%. Rock expose has decreased 2.07 km2 with 6.6%. Grass land and others has decreased 109.92km2 with 15.39% and Open land is 52.49 km2 with increase of 17.94 %.

Acknowledgments The authors are thankful to the Department of Earth science, Tamil University, Thanjavur, Tamilnadu, India to carry out this research work.

REFERENCES

Erdas Inc. 1999. Erdas Imagine Tour Guides, Erdas Imagine Version 8.4. Erdas, Atlanta, Georgia : 636.

Fung, T., ,. LeDrew E. 1987. Application of Principle Component Analysis to Change Detection. Photogrammetric Engineering & Remote Sensing :53: 1649-1658.

Jomaa, I. , Kheir, R. Bou. 2003. “Multitemporal unsupervised classification and NDVI to monitor Land cover change in Lebanon (1987-1997)” National Council for Scientific Research/National Center for Remote Sensing, Beirut, Lebanon

Li, X., ,. Yeh , A. 1998. Principle Component Analysis of Stacked Multi-Temporal Images for the Monitoring of Rapid Urban Expansion in the Pearl River Delta. International Journal of Remote Sensing: 19 (8): 1501-1518.

Macleod R.D , Congalton R.G.1998.A quantitative comparison of change detection algarithems for monitoring Eelgrass from remotely sensed data. Photogramatic Engineering& Remote sensing ,64 (3): 207-216.

Mas, J. F. 1999. Monitoring land-cover change: a comparison of change detection techniques. International Journal of Remote Sensing: 20: 139–152.

NSAS Earth Observatory http://earthobservatory.nasa.gov/.

Nielson , A., Conradsen, K., , Simpson, J. 1998, Multivariate alteration detection (MAD) and MAF post processing in multi-spectral bi-temporal image data: new approaches to change detection studies. Remote Sensing of Environment: 64: 1–19.


http://earthobservatory.nasa.gov/


Pilon, P. G., Howarth, P. J., Bullock, R. A., , Adeniyi, P. O. 1988, An enhanced classification approach to change detection in semi-arid environments. Photogrammetric Engineering and Remote Sensing: 54: 1709–1716.

Ridd, M. , Liu J. 1998. A Comparison of Four Algorithms for Change Detection in an Urban Environment. Remote Sensing of Environment: 63: 95-100.

Sabins, F. 1996. Remote Sensing: Principles and Interpretation. New York, W H Freeman & Co.

Sohl, T.L. 1999. Change Analysis in the United Arab Emirates: An Investigation of Techniques. Photogrammetric Engineering and Remote Sensing: 65(4): 475 – 484.

Singh, A. 1989. Digital change detection techniques using remotely-sensed data. International Journal of Remote Sensing: 10: 989–1003.

Sunar, F. 1998. An Analysis of Change in a Multi-date Data Set: A Case study in The Ikitelli Area, Istanbul, Turkey. International Journal of Remote Sensing: 19: 225-235.



NEO-INDUSTRIALIZATION OF FORMER INDUSTRIAL REGIONS OF RUSSIA (THE CASE OF “TITANIUM VALLEY”)

Anatoly V. STEPANOV

Ural Federal University, 620002, ul. Mira, 19 Yekaterinburg Russia www.urfu.ru, [email protected]

Maria Y. ILYUSHKINA

Ural Federal University, 620002, ul. Mira, 19 Yekaterinburg Russia www.urfu.ru, [email protected]

Abstract Complex transformational processes in the Russian economy in the ХХ century, which exited a 70-year long experience of a planned-administrative economic system and transitioned to the market model, have led to unprecedented experimentation with the world economy. Having introduced the principles of a market economy, Russia has found itself subjected to the twin pressures of globalization and neo-industrialization, the results of which have proven to be very effective for its industrial structure. Globalization has expanded state borders and opened the gates for Russian entrepreneurs to conquer the world oil and gas markets, ferrous and non-ferrous metals sectors, engineering and chemical industries. The Urals region of Russia is developing in a very intensive way. The “Titanium Valley” production cluster was formed in this region. This article examines some of the special features of the functioning of this cluster. Keywords: cluster, titanium, globalization, transition economies, neo-industrialization, post-communist economies. The process of globalization has had a dramatic influence on the development of the world economy. Commercial production, the development of financial markets, the distribution of the same consumer goods across countries, and the considerable movement of the labour force have all spread beyond national borders. Amongst these factors, independent transnational capital can be singled out. Based on the achievements of scientific and technical progress, globalization has opened new horizons for development and multi-level cooperation between different countries and regions. The ever-increasing pace of direct foreign investment together with advanced technologies, management and professional marketing is accelerating this process.

Transnational corporations are often named as the main regulators of production capacities’ location and the development of the service sphere on a global scale. Global actors have worked out effective strategies for entrepreneurship and constructed new business structures for production with high added value, coordinating the production process, financial and commercial activities of subsidiaries world-wide under the auspices of


http://www.urfu.ru/


http://www.urfu.ru/


Stepanov V. A. - . Ilyushkina Y. M./ European Journal of Geography 3

international business cooperation and developing markets in separate regions and the world of business.

As a result, globalization has led to a high level of competition in places of maximum concentration of production capacity in specific areas of the industrial world. Nowadays, specialists in economic geography and regional economics with increasing frequency, state the existence of rivalry between world industrial regions for foreign direct investment as global actors. As a result, we can speak about the phenomenon of “neo-industrialization”, with its simultaneous processes of transformation and diversification of former industrial regions. Innovative developmental research in the XX-XXI centuries showed that the transfer to a new “technological vector” of development would be impossible without the support of so-called “cluster policies”. 1 If we take the definition given by foreign experts, we can stress the fact that “clusters” and “cluster policy” can be named as the “phenomena” of intensive acceleration and development of countries and world regions. The genesis and evolution of “Silicone Valleys” and “Bioengineering valleys” has proven that strategies for the economic development of prosperous regions are realized through implementing cluster policies. Cluster policies can frequently become tools for the innovative re-positioning of particular regions under conditions of global rivalry.

Russian scientists began to show an interest in clusters only in the first decade of the XXI century, later than their overseas colleagues. Theoretically speaking, we should note that the word “cluster” has lost its original meaning and has been transformed into a brand, used by local authorities for raising investment attractiveness and changing the image of a particular region.

Clusters are the means for spatial regional development. The use of this tool corresponds to world trends of cluster development, such as the connection of industrial activity to world networks, the intensification of the role of transnational companies, and the priority collaboration of cluster enterprises with national and global delivery chains.

As far as the cluster policy model of different countries is concerned, we can distinctly observe diverse vectors of the so called “continental” policy of cluster development (which is typical of Japan, Sweden, France and Italy), where the state worked through a complex system of controls, starting with the choice of priority clusters and finishing with the financial support of developing strategies and programs and key factors of successful functioning, and the “Anglo-Saxon” model (which is used in the USA, Great Britain and Australia). Under this model, genesis and cluster evolution are considered to be the result of market competition with the obligatory presence of spin-off processes from global entrepreneurs who are able to start up their own business.

The accepted innovative conception for the long-term economic development of the Russian Federation (named “Strategy - 2020”)2 takes into consideration the necessity to construct territorial clusters in various Russian federal regions [1]. We can include “Titanium Valley” in the row of already famous projects such as “Skolkovo” and “Zhukovsky” which have been developed in the Moscow area. “Titanium Valley” is a high-technology cluster within the space and rocket sector, located in the Sverdlovsk region.

With the importance of natural resources as a factor in the development and allocation of production capacities in world industry and particular countries steadily decreasing, their

1 We define the term “cluster policy” in a traditional way as a set of activities aimed at raising country competitiveness through cluster development stimulation. See: Regional development and regional policy in Russia during the transition period/ S.S.Artobolevsky, Moscow State Technical University, Moscow, 2011, page 195. 2 “Strategy 2020” is a development perspective Russian program which was launched in 2008 and supported by social organizations. This strategy throws light on priority aspects of development in different spheres of economic and social life of the country.



influence on the allocation of new organizational and production models, social factors and public premises is becoming more significant. Different interregional prices and the professional qualifications of the staff have always been extremely important, but currently, this factor is being taken into account to an even greater extent. Capital now flows beyond national borders influencing the allocation of production capacities all over the world, including such previously closed regions as the Central Urals in the Russian Federation. There follows a detailed breakdown of the situation in one of the Ural regions.

When selecting the territory for forming the “Titanium Valley” cluster, Russia’s leading enterprise in this sector, “VSMPO-AVISMA”,3 was singled out (this enterprise is responsible for 100 % of domestic titanium and titanium composites production and 20 % of world production) [2]. This enterprise is at the heart of this “proto-cluster”. During the era of the planned-administrative economy it was constituent in the military-industrial complex, and under the terms of the process of conversion, “VSMPO-AVISMA” became a competitive enterprise possessing unique resources used in titanium production, with a wide network of subcontractors and customers.

The investment attractiveness in locating high-technology manufacturing for foreign and local space engineering producers on the grounds of this enterprise is defined by the availability of existing markets, technological facilities for developing vertically integrated production and its corporate culture. In case of unfavourable macroeconomic trends in the sphere of titanium and magnesium production, the potential resources of the region give rise for hopes of possible diversification of the “Titanium Valley”.

The progressive development of the appointed enterprise in the first half of the XXI century was a necessary condition for the positioning of “VSMPO-AVISMA” by the authorities in the Central Urals as the “growth pole” in this former industrial region with the aim of additional investment inflow from potential investors from the international space and rocket sector [5].

Touching again on the historical process of creating this proto-cluster, the “Titanium Valley” was formed on the territory of Verkhnaya Salda district in Sverdlovsk region with the aim of attracting foreign investors in 2005 to this special economic zone, which is located 180 km from Yekaterinburg between the cities of Nizhny Tagil and Verkhnaya Salda. 4 Its aim is to attract huge world producers from innovative industries, to develop new modern high-technology enterprises and conditions for the further integration of the Russian economy into the world economy5. Enterprises producing more than 40 % of all metallurgical products in the Russian Federation are located around the “Titanium Valley”, and the government intends to invest about 16.5 billion rubles in this infrastructure, creating a total volume of investment of 64.5 billion rubles in 2020. According to the agreement all participating companies are only permitted to function as industrial producers, and must invest 3 million Euros into the project during the first three years. Several strategies of the “Titanium Valley”, connected with the development of raw materials base of titanium production, are planned for implementation:

1. The raw materials base development will be aimed at the advancement of titanium and alloyed composites6.

3 VSMPO – Verkhnaya Salda metallurgical production association (city of Verkhnaya Salda, Sverdlovsk region); AVISMA – titanium and magnesium plant (city of Berezniki, Perm region). 4 A special economic zone is a territory with a legal position, customs and tax preferences for entrepreneurs, developed industrial, business and social infrastructure. 5 Launching the project in the Sverdlovsk region will create more than 17000 new jobs. 6 In the first instance, attention will be paid to the Kachkanar fields of titanium, magnesium, tungsten and molybdenum ores which contain alloying metals.



2. Production development of semi-finished titanium goods for aviation, medicine as well as wiring and thin-walled wrought pipes production from titanium composites.

3. Component production for the aircraft industry,7 including the manufacturing of spare parts for aircrafts and other forms of air transport.

4. Fabricated production assumes the creation of a machine-building complex which will be involved in making titanium combined equipment for chemical engineering, nuclear power engineering, mechanical engineering and nonferrous metallurgy.

As a work incentive, residents of the “Titanium Valley” special economic zone receive: - guaranteed provision of production, social, business and cultural infrastructure; - personnel training for professional activity; - plots of land8.

The unique nature of “VSMPO-AVISMA” is quite well-known in the world titanium and magnesium sectors. In intense competition with American, Canadian and Japanese titanium and composite producers, this corporation is the biggest producer of pressed large-sized items of aluminum composites, semi-finished goods from alloyed steel and nickel-based heat-resistant composites. . The corporation exports 70 % of all its titanium products, with 30 % used in the domestic market. The competitive advantage of the “Titanium Valley” cluste, is considered to be the concept of development which is based not on the traditional “Greenfield” model, whereby available empty spaces with infrastructure is inhabited by residents, but, vice versa, based on the basis of a prosperous enterprise and attracting companies already involved in titanium projects.

Thus, the “Titanium Valley” cluster, according to its evolutionary type, should be classified as a regional industrial cluster which represents a group of neighbouring interdependent companies and organizations linked in titanium and composites production and spare parts for space engineering. The synergy of these companies in terms of market competition leads to a general increase in efficiency promotes innovation and intensifies the international, transnational division of labour.

The cluster policy of the “Titanium Valley” represents a whole complex of measures carried out by federal and regional authorities, aimed at cluster identification and implementing the general policy of cluster support in the country and in the region.

REFERENCES

Bazhenova Y. V. 2009. Dissertation: World experience in cluster attitude to the development of free economic zones and Russian perspectives, 24. Moscow.

Lavrikova Y. G. 2008. Clusters: strategies of their formation and development in the regional economy, 232.Yekaterinburg.

Cluster development. Its essence, approaches and foreign experience, 2008./ed.S.N.Pyatinkina. Moscow Press.

2011. Regional development and regional policy of Russia in the transition period,./ ed. S.S.Artobolevsky, O.B.Glezer, 237. Moscow Press.

7 Special attention to the cluster is being paid by UAC (United Aircraft Corporation), which plans to have a routine production of the “Super Jet 100” passenger aircraft in Komsomolsk-on-the-Amur. 8 Average price for lots in a special economic zone “Titanium Valley” is 75000 rubles for renting hectors, 250000 rubles per hector if purchased [2011]. If foreign goods are imported on the territory of “Titanium Valley”, VAT is not paid.



http://www.midural.ru/news/100520/100521/

http://titanium-valley.com/

http://www.oao-oez.ru/special_economic_zones/titanium-valley/


http://www.midural.ru/news/100520/100521/

http://titanium-valley.com/

http://www.oao-oez.ru/special_economic_zones/titanium-valley/


DETERMINANTS OF STRUCTURAL DIMENSION OF DAILY BEHAVIOUR IN A TRADITIONAL AFRICAN CITY:

A CASE STUDY OF ILORIN, NIGERIA.

Moses Olutoyin ADEDOKUN Federal College of Education, Department of Geography, P.M.B.1041, Zaria, Kaduna-State Nigeria

[email protected], [email protected]/[email protected]

Abstract This paper analyses the determinants of daily temporal behavioral pattern of residents of Ilorin, especially the length of time spent on their activities. Data were collected from 500 residents of Ilorin, each of whom completed a time budget diary over one week. Descriptive statistics were used to summarize the data while Principal Component Analysis was used to test the research hypotheses. The result shows that the activities were fixed in time and three variables: age, income and occupation were the major determinants of the time spent on the activities. The study also shows clearly that the temporal structure of activity in Ilorin is different from what obtains in Western cities where there is flexibility in the usage of time. Keywords: behavioral pattern, activity data, descriptive statistics, Africa.

1. INTRODUCTION

The study of how a given population spends their times on various activities is a complex one with many dimensions. These dimensions include the location, timing, duration, sequence and type of activities and or trips. These characteristics of time behaviour have made the analysis of its many dimensions imperative (Kwan, 2002, 2003). Previous researches have either focused on spatial dimensions (Main, 1982) or have completely ignored time element as if it is not important, though geographers view places in a two dimensional ways: space and time or spatio-temporal, they often tend to ignore the time element. This paper, thus, discusses the temporal pattern of activities, that is, the time spent on various categories of activities, the determinants of the time spent and its implication on urban infrastructural planning.

The approach of integrating individual spatial behaviour overtime was pioneered by Hãggertrand (1969). He used a simple diagram to illustrate his concept of space-time dimensions (Figure 1) Hãggertrand postulated the geographers’ two-dimensional space on the surface of the earth or on the surface of a map. A line on this surface indicated movement in space but not in time. He suggested a third dimension to signify time.



mailto:[email protected]/[email protected]

Adedokun O.M../ European Journal of Geography 4 1

Source: Adapted from Haggerstrand (1969)

Figure1 represents a very simple working day. Solid lines represent the path of all obligatory activities and dotted lines the prism or feasible regions of movement in periods for which there are no fixed activities. The worker is assumed to be effectively fixed at home until 7:30am to 8:00am where he can conveniently sleep and take breakfast. He must then take a direct route to work, where he is obliged to stay until lunchtime. During the lunch hour he has a certain amount of freedom; he must be back in the office exactly an hour. From 2pm until about 5pm he is again

Time

9Pm 8Pm 7Pm 6Pm 5Pm 4Pm 3Pm 2Pm 1Pm 12Non 11am 10am 9am 8am 7am 6am

Space Office Home

Figure 1. Man’s daily space-time dimensions



expected to stay at work. But after 5pm he has no need to be home until 7pm for supper. In this period, 5-7pm, he can stay on at work or he can go somewhere near or stop off on the way back for a drink or visit. The main feature implicit in this model of daily behaviour is the idea that certain activities are fixed in both space and time.

2. THE STUDY AREA

When the present city of Ilorin was founded is not very clear. Indeed, little is known about its pre-jihad political development. Ilorin is today the capital of Kwara State. It is located on latitude 80.30N and Longitude 40.35’E. It lies on the southern fringes of the savanna region and north of the forest zone. Ilorin is located in the Guinea savanna grassland belt of middle belt region of Nigeria (Figure 2). The main river in Ilorin is the Asa which flows in the south-north direction. It divides Ilorin into two parts: a western part representing the core or indigenous area and the eastern part where the Government Reservation Area (GRA) is located.

Ilorin has experienced a rapid growth in its population over the years. The first population census in 1911 put the population of Ilorin at 36,343 while the 1953 population census put the town’s population at 40,994. The 1963 and 1991 censuses recorded the population of the town as 208,546 and 532,088 respectively. The projected population of Ilorin in 2005 when this research was carried out was 748,150 based on an assumed annual growth of 3.5 percent.

Figure 2. Map of study Area showing the Sampled Wards

3. METHODOLOGY

3.1. Source of Data The Activity Network Approach (ANA) was adopted for this study. ANA is a micro-behavioural, inductive approach that makes ‘predictions about the whole from disaggregate data of the behaviour of individuals using Time Budget Diary (TBD). TBD questionnaire focuses on the socio-economic attributes of the individual, types of activities, location of activities, beginning and end time of activities, number of participants in each activity, extent to which each activity was arranged and whether an individual could have done: - anything else at the time of this activity,



- this activity at any other time, - this activity elsewhere, and whether, - been anywhere else at the time of his activity is taking place.

3.2. Sampling Procedure The 20 electoral wards in Ilorin formed the spatial framework for primary data collection. The use of these wards was based on the fact that it makes it easier to obtain data on population. The sample size was 500 literate individuals. This number was proportionally distributed among the 20 wards based on their 1991 population projected to 2006, using 3.5 percent annual growth rate. Number of respondents to be interviewed from each ward was randomly selected. This sample is considered adequate for the study of this nature because of the complexity of completing the questionnaire, the time and cost involved in administering the questionnaire, monitoring the respondents, and more importantly, because researches involving Time Budget Diary do not normally accommodate large samples (Timmermanns, 2000; Kwan, 2005). Each respondent was issued seven copies of the TBD questionnaire, one for each day of the week. Research assistants monitored the respondents at home and work places.

4. MAJOR FINDINGS

4.1. Temporal Fixity of Activities To establish the temporal nature (fixity) of the respondents’ activities, they were asked whether they could have done anything else at the time they did a particular activity. The result shows that 280 respondents (94.6 percent) said they could not do anything else at the time while 16 (5.4 percent) indicated that they could do something else at the time. The temporal fixity of activities was further established when it was asked if these activities could be done at some other time The result shows that 8.5 percent of the respondent do not have their activity fixed in time. This include 7.4 percent, 0.7 percent in office/work place activities. On the other hand, majority of the respondents (91.5 percent) could not have done their activities at some other time. This includes 40.20 percent who are engaged in home based activities, 51.70% in office/work place activity and 0.7 percent in outdoor activities.

4.2. Relationship between Structural and Temporal Activity Fixity Result shows that there is a relationship between the nature of activity, that is, whether an activity is arranged, planned, routine or unplanned and its temporal fixity. Structural fixity of an activity determines its temporal fixity. For instance, respondent could not perform “arranged” and “planned” activities at any other time. The few respondents (9.5 percent and 1.4 percent) who were engaged in arranged and planned activities could not have done them at any other time. Even in the case of routine activities, 248 respondents or 83.8% had their activities fixed in time.

Temporal fixity of activity also varies with the location of activity, that is, where the activity is based. 288 (97.3) of the respondents concerned could not have done anything else at that time i.e. they had their activity fixed in time. This comprises of 45.6 percent who were engaged in home based activities, 51.7 percent who were engaged in office/work place-based activities, and 0.7 percent who were engaged in outdoor activities. Only 8.0 percent respondents did not have their activities fixed in time, these were 6.0 percent and 0.2.0 percent who were engaged in home based and office/work place-based activities.



4.3. Temporal Pattern of Respondents’ Daily Activities This section discusses the temporal pattern, that is, the time spent on various categories of activities. Mean figures in minutes were used. The daily variation in the allocation of time by the location of activities shows that the respondents spent more time on home based activities. On the average, the weekly mean time spent on home based activity is 910 minutes while the weekly mean time spent on office/workplace based activity is about is 410 minutes. It is evident that although the location of respondent activities were almost equally shared between home and office/workplace based activities, more time were devoted to home-based activities with a mean of 910 minutes as against the mean of 410 minutes for work place based activities.

4.4. Temporal Fixity of Activities To establish the temporal nature (fixity) of the respondents’ activities, they were asked whether they could have done anything else at the time they did a particular activity. The result shows that 280 respondents (94.6 percent) said they could not do anything else at the time while 16 (5.4 percent) indicated that they could do something else at the time. The temporal fixity of activities was further established when it was asked if these activities could be done at some other time The result shows that 8.5 percent of the respondent do not have their activity fixed in time. This include 7.4 percent, 0.7 percent in office/work place activities. On the other hand, majority of the respondents (91.5 percent) could not have done their activities at some other time. This includes 40.20 percent who are engaged in home based activities, 51.70% in office/work place activity and 0.7 percent in outdoor activities.

4.5. Relationship between Structural and Temporal Activity Fixity Result shows that there is a relationship between the nature of activity that is, whether an activity is arranged, planned, routine or unplanned and its temporal fixity. Structural fixity of an activity determines its temporal fixity. For instance, respondent could not perform “arranged” and “planned” activities at any other time. The few respondents (9.5 percent and 1.4 percent) who were engaged in arranged and planned activities could not have done them at any other time. Even in the case of routine activities, 248 respondents or 83.8% had their activities fixed in time.

Temporal fixity of activity also varies with the location of activity, that is, where the activity is based. 288 (97.3) of the respondents concerned could not have done anything else at that time i.e. they had their activity fixed in time. This comprises of 45.6 percent who were engaged in home based activities, 51.7 percent who were engaged in office/work place-based activities, and 0.7 percent who were engaged in outdoor activities. Only 8.0 percent respondents did not have their activities fixed in time, these were 6.0 percent and 0.2.0 percent who were engaged in home based and office/work place-based activities.

Attempt is also made in the research to determine if there is any discernable group of behavioural pattern in any typical day. That is what are those variables that account for time devoted to activities each day of the week. To achieve this, factor analysis (principal component analysis) was used by subjecting a matrix of time which individual spent on each activity, each day to a factor analysis.

Both factor and principal component analysis are multivariate statistical techniques that reduce the dimensionality of a variable. Principal component analysis maximizes the variance accounted for in the original variable.( Abumere, 2001) From the data, there are three dependent variables i.e. (activity location, home, office/work place and outdoor). Out of these three only two (Home and office/work place) based activities were significant and extracted by the principal component analysis. There are six independent variables (gender, marital status, age



(years) religion, education qualification and occupation) of the six only two (age and income) are significant and extracted. For home based activities, the principal component analysis is presented inTable1

Table 1. Principal Component Analysis Derived from the Time Devoted to Home Based Activity.

Component Initial Eigevalues Extraction sum of squared loadings

Rotation sum of squared loadings

Total % of variance

Cumulative % Total % of

variance Cumulative

% Total % of variance

Cumulative %

1 4.521s 64.585 64.585 4.521 64.585 64.585 3.666 52.370 52.370 2 1.380 19.717 84.303 1.380 19.717 84.303 2.235 31.933 84.303

Source: Authors analysis, 2006.

From Table 1, component 1 has an initial equivalent of 4.521 and a percentage of variance value and extraction sum of squared loading percentage of variance value of 64.585 respectively and a sum of squares loading percentage variance of 51.650. While component 2 has an initial eigeun value of 1.380, extraction sum of squared loading percentage variance of 19.717 and rotation sum of squared loading percentage of variance as 31.933. Cumulatively, components 1 and 2 jointly account for 84.303 percent of total variance explained. But component 1 is more significant in explaining the variance (52.370).

However to properly explained dimensions of daily behaviour the components were rotated and transformed into coefficient of concurrence by squaring the component scores for each day as presented in the Table 2

Table 2. Transformed Component Score Matrix Time devoted to activity

Component Score 1 2

Rotated score Coeffiencient of concurrence

Rotated score Coeffiencient of concurrence

Day 1 .833 0.694 .167 0.028 2 .934 0.872 .204 0.042 3 .946 0.895 .158 0.025 4 .777 0.604 .326 0.106 5. .718 0.516 .461 0.213 6. .212 0.045 .958 0.918 7 .200 0.040 .961 0.924


From Table 2, it is clear that for days 1 to 7, the structural dimension of daily behaviour as related to time devoted to home based activities, it is explained by component 1 (age of the respondents. This varies from 51.6 percent on day 5 to 60.4 percent for day 4, 69.4 percent for day 1, 87.2 percent day 2 and 89.5 percent on day 3). On days 6 and 7 variance is explained by component 2 (estimated annual income). This account for 91.8 percent on day 6 and 92.4 percent (the highest) on day 7. For office work place the variance explained by principal component analysis is presented in Table 3.



Table 3. Principal Component Analysis Derived for Time Devoted to office/workplace based

Component Initial Eigevalues Extraction sum of squared

loadings Rotation sum of squared loadings

Total % of variance

Cumulative % Total % of

variance Cumulative

% Total % of variance

Cumulative %

1 5.856 83.655 83.655 5.856 83.655 83.655 5.850 83.577 83.577

2 1.001 14.294 97.949 1.001 14.294 97.949 1.006 14.372 97.949 Source: Authors analysis, 2006

From Table 3 the variance explained or the initial eiugenvalue and extraction sum or the

squared loading are 5.856 respectively while initial eiugenvalue percentage of variance and the extraction sums of squared loading percentage variance is 83.655 respectively and the rotation sums of squared loadings percentage variance is 83,577. This is in contrast to component 2 initial eiugenvalues total of 1.001, percentage variance of 14.294, extraction sums of squared loadings percentage variance and rotation sum of squared loading percentage variance of 14.372 percentage variance of 14.372.

Overall, the two components 1 and 2 accounted for about 98 percent of total variance explained; however component 1 (age) is very significant in explaining the behavioral pattern in any typical day in the office/workplace because it contributed or accounted for about 84 percent while component 2 (estimated annual income contributed only 14 percent). The contribution of these two components in a typical day is displayed in the Table 4.

Table 4. Transformed Component Score Matrix for Office/workplace based activity.

Time devoted to activity Component Score 1 2

Rotated score Coeffiencient of concurrence Rotated score Coeffiencient of

concurrence Day 1 - - 1.00 .100 2 .986 0.97 0.173 .030 3 .992 0.98 0.141 .020 4 .978 0.96 0.200 .040 5. .989 0.96 0.141 .020 6. .992 0.98 0.141 .020 7 .987 0.97 0.173 .030


Table, 4 shows that for all days except day 1, the structural dimension of daily behaviour is explained by component 1. The component score and its transformation, the coefficient of concurrence varies from 96 percent for days 4 and 5, to 97 percent for days 2 and 7 and 98 percent for days 3 and 6 respectively. Component 2 (estimated annual income) only account for structural dimension of behaviour on day 1 and this is 100 percent.

As for outdoor activity, the computer analysis (result) shows a very insignificant account or contribution. The computer result states that there are fewer than two cases, and that at least one of the variables has zero variance. There is only one variable in the analysis, or correlation coefficient could not be computed for all pairs of variables, hence no further statistics could be computed for the file. This is because outdoor activities are a reflection of affluence. Due to poverty, people have to work from morning till evening of time without provision for leisure.

The results of the regression, ANOVA and factor analyses show that three variables: age, income and occupation affect time spent on activities. The explanation for this is that, most people are self-employed or they are having multiple occupations/combining many occupations



due to lack of formal employment, also most of the people are low-income earners. This is a major characteristic of Africa urban centers. It is an indication of the nature of activities in an emerging nation with a weak industrial base where informal activities dominate urban economic landscape. Also most people are self employed or they own their businesses hence these is no time table for their activities. Furthermore they have no structured work scheduled they work from morning till evening to make ends meet because their still within the active working population.

Based on the above findings in the study area, a generalized model of land use planning and facility location in a traditional medium size urban center, using Ilorin as a case study (Figure 3).

Figure 3. A Model of Urban Neighbourhood Activity Centers

Given a medium size urban centre with its population, there would emerge various types of

urban activity located in different parts of the city. The location of these activities would in turn generate activity pattern with space and time (or spatio-temporal) dimensions. The activity pattern itself would generate human spatio-temporal behavior. In the study area, the human behaviour in space and time was fixed. The fixity in human spatial behaviour is shared between home and office/work place. The urban neighbourhood activity centre model is of the view,

Space

Human (Spatio-Temporal) Behaviour

Office/Workplace

Neighbourhood Activity Centres

Fixed

Home

Activity Pattern Time

Activity Location

Urban Activities

Urban Centre Urban Population



therefore, that in planning for a medium size urban centre in developing world; there may be the need to adopt a strategy that would incorporate the behaviour of the people. Instead of strict land use zonation approach, facilities may be located closely to or around neighborhoods where people are fixed to. In this case and as demonstrated, facilities and infrastructures should be located between homes and work places. Obviously, if there is a demonstrable linkage between two activities in time, it makes sense to locate the facilities housing them in the same space so as to eliminate time and energy consuming travel.(Adedokun, 2008, 2009)

REFERENCES

Abumere, S.I. 2001. ‘Data Sources, Acquisition and Analysis’ Paper presented at the 44th Annual Conference of Nigerian Geographical Association. University of Ibadan.

Adedokun, O. M. 2008. A Spatio-Temporal Analysis of Activity Linkages in Ilorin, Kwara-State. Savanna 2 (1): 41-52.

Adedokun, O. M. 2009. A Spatio-Temporal Analysis of Urban Activity Linkages: A Case Study of Ilorin, Nigeria. Unpublished Ph.D Dissertation, Ahmadu Bello University, Zaria.

Haggerstand, T. 1969. ‘What about People in regional science’ Regional Science Association 24.

Kwan, Mei -Po. 2007a. ‘The Interaction between Information Communication Technology (ICT) and Human Activity Travel’ Transportation Research A (Special issue) 41 (2).

Kwan, Mei - Po. 2007b. ‘Geovisualization of Human hybrid – travel Patterns’ Transactions on Geographical Information System (GIS) 11 (5): 721-744.

Kwan, Mei -Po. 2005. Measuring Activity and Action Space/time. In Integrated Land-Use and Transportation Models: Behavioural Foundaations, eds Martins E.H.,Lee-Gosselin and T. D. Seen. Oxford: Pergamon-Elsevier, 101-132.

Kwan, Mei -Po. 2003a. New Information Technologies, Human Behaviour In Space Time and the Urban Economy. Paper Presented at the 82nd Annual Meeting of Transportation Research Board (TRB) Washington D.C. Jan 12-16.

H.A.C. 1982. ‘Time – Space Study of Daily Activity in Urban Sokoto’ Nigeria Unpublished Ph.D Thesis, University of Liverpool, U.K.

Timmermans, H.J.P. 2000 ‘Theories and Models of Activity Patterns’ Albataros , 6 – 70.


EUROPEAN JOURNAL OF GEOGRAPHY

Volume 4 • Number 3 October 2013 • ISSN 1792-1341

E u r o p e a n A s s o c i a t i o n o f G e o g r a p h e r s

Documents

European Journal of Geography · 2018. 6. 20. · Editorial 4.3 . The publication of the European Journal of Geography (EJG) European is based on the . Association of Geographers’