19
Coastal and Watershed Resources and Hazards Modeling Dr. Danang Sri Hadmoko, M.Sc DETERMINISTIC MODELLING ON LANDSLIDE HAZARD ZONATION

Landslide Hazard Modelling using ILWIS 3.x (Aji Putra Perdana-MPPDAS)

Embed Size (px)

DESCRIPTION

Landslide Hazard Modelling using ILWIS 3.x (Aji Putra Perdana-MPPDAS)Hasil praktikum pemodelan longsor dengan ILWIS 3.x semasa kuliah di MPPDAS

Citation preview

Page 1: Landslide Hazard Modelling using ILWIS 3.x (Aji Putra Perdana-MPPDAS)

Coastal and Watershed Resources and Hazards Modeling

Dr. Danang Sri Hadmoko, M.Sc

DETERMINISTIC MODELLING ON LANDSLIDE HAZARD ZONATION

Page 2: Landslide Hazard Modelling using ILWIS 3.x (Aji Putra Perdana-MPPDAS)

Contents

Table of Figures

à

Page 3: Landslide Hazard Modelling using ILWIS 3.x (Aji Putra Perdana-MPPDAS)

IntroductionLandslides are frequently responsible for considerable loss of money and

lives and the most widespread and damaging hazard in the world. Landslide canbe triggered by intense rainfall, earthquake, ground water level change, stormwaves or rapid stream erosion. Landslides may be corrected or controlled by oneor more combinations of four principle measures: drainage, slope geometrymodification, retaining structures and internal slope reinforcement. Landslides andrelated slope stability problems disturb many parts of the world but experienceindicates that understanding, recognition and treatment of landslide hazards is stillfragmentary. A particular area requiring attention concerns the selection anddesign of appropriate, cost-effective remedial measures, which in turn require aclear understanding of the conditions and processes that caused the landslides.Much progress has been made in developing techniques to minimize the impact oflandslides, although more efficient, quicker and cheaper methods could wellemerge in the future (Hadmoko 2007, Yilmas and Keskin 2009).

Recently, developing techniques and methods are managed and analyzeinformation and data using geospatial information and technology which knownas Geographic Information System (GIS) and it is widely used as tool formapping. Landslide mapping has become an important tool for risk assessment,prediction, and management (Hadmoko 2007, Mantovani, F et al 2009)

Synthetic maps showing slopes at risk of movement have proven to bevaluable for planning, risk analysis, and design of infrastructure. Whengeomorphological and geological surveys are imported into a GeographicInformation System environment, thematic maps can be generated to meet theneeds of a wide variety of end users. Applications for products derived from thesesurveys include: thematic cartography implementation and improvement; creationof geological and Geomorphological databases; thematic mapping aimed atmitigating geoenvironmental risks such as landsliding and flooding; remotesensing for geological and geomorphological mapping; land management andhazard mitigation; hydrological modeling; water resources protection andmanagement; numerical modeling; environmental impact assessment; andmanaging environmental emergencies (Turner and Schuster 1996 in Mantovani,Franco. et al 2009).

Landslides hazard assessment has been playing an important role in spatialplanning due to the increasing of landslide disasters occurring in many regions inIndonesia. Various methods of landslide hazard assessments have been proposedand conducted by earth scientists from the simplest method by using the simpleparameter that caused landslide to very complex methods by using the largenumber of terrain parameters and also the complex quantitative landslide hazard

Page 4: Landslide Hazard Modelling using ILWIS 3.x (Aji Putra Perdana-MPPDAS)

zonation. One methods of landslide hazard assessment is deterministic method(Hadmoko, 2009).

This exercise deals with the spatial deterministic landslide hazardassessment by using GIS as the main tool of the analysis. This analysis is based onthe slope stability analysis with infinite slope stability model. This model basedon theory that landslide will be occur when the mobilized shear stress in the soilincrease and the available shear strength in the soil is decrease, so the ratio isknown as safety factor (Fs). The safety factor can be calculated pixel by pixel forindividual slope, and for given area, the combination of each pixel can begeneralized as the area value.

ObjectivesBased on question in the exercise, we can see that the objectives of this

exercise are as follow:1. To compare for each scenario, and to explain the pixel distributions for all

stability classes (stable, critical and unstable) for each scenario.2. To know why the distributions of the pixels in the hazard maps have no

pattern and distributed irregularly.3. To calculate the number of pixels for all maps and make charts showing the

number of pixel for all the stability classes (one chart for one scenario).4. To calculate the area of each hazard level for all scenarios and make a chat

showing the influence of soil water content to the area of each hazard level.5. To make simple statistical regression based on your results to predict the

unstable area as a function of soil water content.6. To know the advantages and disadvantages this methods in landslide hazard

simulation.

Page 5: Landslide Hazard Modelling using ILWIS 3.x (Aji Putra Perdana-MPPDAS)

Material and MethodThe materials needed in this exercise are:1. ILWIS software package which is installed in your system2. Elevation data of our study area : Girimulyo and its surrounding area3. PC / Laptop

The hazard degree can be expressed by the Safety Factor, which is theratio between the forces that make the slope fail and those that prevent the slopefrom failing. The value of safety factor (FS) can be classified into three classes i.e.FS-values larger than 1 indicate of the slope in stable condition, FS-values smallerthan 1 represent the unstable condition. At F=1 the slope is at the point of failure.Many models of safety factor calculation are available but we use the simplestmethod we can call it infinite slope models. The main principle of this method isto understand the failure mechanism by the change of soil moisture content(degree of saturation), the groundwater level and seismic acceleration. The modelis conducted as an ILWIS function.

a. Creation of DEMDigital Elevation Model (DEM) is the digital map which contains the

elevation data recorded in the spatial data as z value. DEM is a main importantinput variable in this exercise. In this exercise DEM will be extracted from digitaltopographical map that contain point elevation data. The format of DEM is theraster-based DEM. The elevation data recorded in each pixel or regular grid,which is grid defined as a square cell with constant size and an elevation valueassociated to it. The digital elevation model will be created using pointinterpolation in the ILWIS software. Point interpolation performs an interpolationon randomly distributed point values and returns regularly distributed pointvalues. In ILWIS, the output values are raster values. The input map on thisexercise is a point map in which the points themselves are values (point map witha value domain). For each pixel in the output map, a value is calculated by aninterpolation on input point values. Moving average methods will be used ininterpolation process. Moving average assigns to pixels weighted averaged pointvalues.

Create DEM from elevation point data: OperationmenuàInterpolationàpoint interpolationàmoving average or we can right clickon point data and choose interpolationàmoving average. We can see in Fig 1below.

Page 6: Landslide Hazard Modelling using ILWIS 3.x (Aji Putra Perdana-MPPDAS)

b. Creation of slope angle mapSlope angle map is derived from elevation map in raster-based DEM.

Slope angle map can be calculated by using filter operation applied to DEM inILWJS environment. The slope angle map will be classified into degree system.The conceptual of calculating slope angle can be done by using filter operationwith 3x3 windows size which applied to all pixels in the map.The schematic operation of 3x3 windows filter can be seen in Fig 2.

Page 7: Landslide Hazard Modelling using ILWIS 3.x (Aji Putra Perdana-MPPDAS)

Create Slope Map by using the following instruction:

a. Create the map of height difference in x and y direction (DX and DYmap) by applying the filter function before we create the slope map.DX and DY can be calculated using DFDX and DFDY linear filteroperation which standard linear filter. It calculates the first derivativein x and y-direction (df/dx and df/dy) per pixel.The result of DFDY filters on a DEM:Positive values of DX in the output map mean the terrain goes up fromleft to right (West to East); negative values in the output map mean theterrain goes down.Positive value of DY in the output map mean that the terrain goes upfrom bottom to up (South to North); Negative value in the output mapmean the terrain goes down.PIXSIZE is the pixel size used in DEM, in this exercise used 20x20meters in pixel size.

Click Operation à Image Processing à Filterà Select DEMas input then choose linear filter. Select the DFDX for the DX Mapand DFDY for the DY Map. Use output name DX and DY. We can seein Fig 3 and Fig 4.

Page 8: Landslide Hazard Modelling using ILWIS 3.x (Aji Putra Perdana-MPPDAS)

DFDX DXDY

b. Create the slope map by typing this formula in command line:

SLOPEPCT=100*HYP (DX, DY)/PIXSIZE(DEM)

SLOPEPCT is the slope in percent (%)

HYP is the function to finds the side opposite the right angle in a right-angled triangle (an internal Mapcalc/Tabcalc function).

Page 9: Landslide Hazard Modelling using ILWIS 3.x (Aji Putra Perdana-MPPDAS)

To transform slope in percent to degree can be calculated by usingformula:

SLOPEDEG=RADDEG(ATAN(SLOPEPCT/100)) where,

SLOPEDEG : slope in degree

RADDEG : converter radians to degree

ATAN : map calculation for the inverse tanget (tan-1)

SLOPEPCT : slope in percent

The classification of slope angle can be listed at Table 1.

Table 1. Slope classification in degree

Slope_Id Slope (in degree) Slope_Id Slope (in degree)1 0 – 10 6 50 – 602 10 – 20 7 60 – 703 20 – 30 8 70 – 804 30 – 40 9 80 – 905 40 – 50

c. Reorganize slope map using three parameters as follow:SinSlope = the sine of the slopeCosSlope = the cosine of the slopeCos2Slope = cos(slope)*cos(slope)

NOTE: All slope data have to be converted into radians fromdegree before appling trigonometric equations (degrad)

So, to calculate above formula on the command line:

SinSlope=sin(degrad(SLOPEDEG))

CosSlope=cos(degrad(SLOPEDEG))

Cos2Slope= cos(degrad(SLOPEDEG))

Page 10: Landslide Hazard Modelling using ILWIS 3.x (Aji Putra Perdana-MPPDAS)

d. After calculation the parameters of slope map, now calculating theslope stability by applying scenario. Built new user-defined function =Function Hydrology.

Double click the New Function in operational list. Type thefunction name FS_Hydrology. Type the expression below:

(Cohesion+(Gamma*Gammaw)*Z*Cos2Slope*Tanphi)/(Gamma*Z*SinSlope*CosSlope)

In dialog above then to be edited the expression as follow:

So, we need only two variables to be changed: Value Gamma andValue M, because the value of the unit weight of soil and groundwaterdepth above the slip surface change and depend on the degree ofsaturation. The fully saturated condition lets us to know the slopestability condition when the soil is completely filled by water. This isalso not a very realistic situation, but it will give us the mostpessimistic estimation of slope stability, with only one triggering factorinvolved (rainfall leading to high water tables) (van Westen, 1993 onHadmoko 2009).Below we can use the gamma and m for the calculation scenario asfollow (Table 2).

Page 11: Landslide Hazard Modelling using ILWIS 3.x (Aji Putra Perdana-MPPDAS)

Table 2.Value Gamma and zw for each scenario

No. Degree of saturation Gamma value(N/m2)

Zw

1 Dry condition 5000 0

2 25% saturated condition 7000 0,25

3 50% saturated condition 9000 0,5

4 75% saturated condition 11000 0,75

5 Fully saturated 13000 1

e. To calculate the scenarios mentioned above, we can type thefollowing command in the command field in ILWIS:Table 3. Scenario formula

No Degree of Saturation Command

1 Dry condition Fs_dry=Fs_hydrology(5000,0)

2 25% saturated condition Fs_25=Fs_hydrology(7000,0.25)

3 50% saturated condition Fs_50=Fs_hydrology(9000,0.50)

4 75% saturated condition Fs_75=Fs_hydrology(9000,0.75)

5 Fully saturated Fs_sat=Fs_hydrology(13000,1)

Don’t forget to define the Georeference, use elevation point data.

f. Create classes for slope stability

After finish then classify of each map into three classes by using mapslicing and create classes by create a domain with the name:Fs_classes.Then create the representation color as follow.

Table 4. Boundaries used for each class of slope stability

No Upper boundary Slope Stability Representation Color1 100 Stable Green2 1,5 Critical Slope Yellow3 0,99 Unstable Slope Red

g. Create final map for all scenarios, click Image Processing àSlicing.

Page 12: Landslide Hazard Modelling using ILWIS 3.x (Aji Putra Perdana-MPPDAS)

Result and DiscussionIn this exercise the final result are the stability classes (stable, critical and

unstable) for each scenario, Pixel chart for each scenario and Statistic regressionas a function of soil water content. This discussion is an answer of question in lastof this exercise. We can compare maps for each scenario (question 1) to explainthe pixel distributions for all stability classes (stable, critical and unstable) foreach scenario as follow (Figure 7).

Page 13: Landslide Hazard Modelling using ILWIS 3.x (Aji Putra Perdana-MPPDAS)

Start from the dry condition, in this condition we can see that unstableslope and critical slope distributed in hills/ridge and almost are in stable condition.In 25% saturated condition, the unstable area increase due to the value of the unitweight of soil and groundwater depth above the slip surface change and depend onthe degree of saturation (increase 25%). Drastic changes start in 50% saturatedcondition, where stable area decrease due to the increase of saturation. Only inriverbank and alluvial plain at condition of stable in fully saturated condition. Thatkind of condition because the factor we use is slope and we know that rain waterthat fulfilled water content will increase the unstable condition area. Let’s wecompare to the slope classes (Figure 8) to see more clearly relationship betweendifferent scenario with slope condition. Shown with number 1 are area with stablecondition in almost all scenario, number 2 is unstable area even though in drycondition and number 3 is start unstable un 25% saturated condition.

The distribution of the pixels in the hazard maps have no pattern anddistributed irregularly (qeustion2) because the results are depend on slopesteepness where each pixel has different value and the calculation was conductedfor each pixel. The group of pixels can be shown as an area, but sometimes thereis individual pixel with different value with its surroundings. If we would like toget more general result or see better distribution we may use filter majority toeliminate single pixel or more with differs.

Page 14: Landslide Hazard Modelling using ILWIS 3.x (Aji Putra Perdana-MPPDAS)

To calculate the number of pixels for all maps and make charts showingthe number of pixel for all the stability classes (question 3), we can directly get bysee the histogram. Below is the chart for scenarios Figure 9.

Page 15: Landslide Hazard Modelling using ILWIS 3.x (Aji Putra Perdana-MPPDAS)

Besides, the existing scenario we also do two more scenario 30% saturatedand 62,5 % saturated condition with fulfilled the value of Gamma and Zw basedon range of scenario before.

Page 16: Landslide Hazard Modelling using ILWIS 3.x (Aji Putra Perdana-MPPDAS)

Additional data above will be included in the calculation of pixel numberto create chart and showing the influence of soil water content to the area of eachhazard level (question 4). In histogram data above we can see that there is Areathat calculated from number of pixel x pixel size where pixel size is 20m, so Npixmultiply with 400m2. Below is the table of area to be used as chart to see theinfluence:

Table 5. Area of each hazard level in each scenario condition (in meter square)

From table, we can see that unstable area is decreasing in line with watercontent until 75% and decrease at fully saturated condition. The number of criticalarea changes into unstable area, meanwhile stable area changes into criticalcondition. We can see more clearly using chart (Figure 11) below that showingthe influence of water content in slope stability. Stable shown with green colordecreasing along the saturation, yellow as critical condition also decrease,meanwhile unstable is increase but start decrease in fully saturated condition. Inrainy condition the soil water content will increase and that cause unstable areaincrease as what we can see in the figure and table in this scenario.

Page 17: Landslide Hazard Modelling using ILWIS 3.x (Aji Putra Perdana-MPPDAS)

Besides showing the result in above chart, we can also make a simplestatistical regression to predict the unstable area as a function of soil water content(question 5).

Page 18: Landslide Hazard Modelling using ILWIS 3.x (Aji Putra Perdana-MPPDAS)

Analysis result above (Figure 12) show mathematical model; regression topredict unstable area that is y = 25688x + 7547 with R² = 0.933. It can be explainthat every increasing value of saturated condition (FS) will followed with theincreasing of number of pixels in unstable condition. Determinant value R² is93.3% show positive relation and show the influence of soil water content thatcaused by rainfall as the most triggering factor in this model.We can see thelinkage between parameters and their influence in landslide process. So, thisdeterministic model can shown and help us to more understand the processes,mechanism that occur in landslide hazard.

Landslide hazard simulation with deterministic model also known as atwhite-box model or physically-based model, because it based on physicalprocesses involve in the system. Deterministic approaches are based on slopestability analysis, physical process in landslide triggering and propagationmechanism. The landslide deterministic modeling is suitable for spatial landslidehazard zonation moreover temporal landslide hazard prediction. (Hadmoko,2007).

Deterministic model as model with the techniques of white box modelfactor of safety combined with hydrological model has the advantages anddisadvantages in landslide simulation. The disadvantages are cannot be applied inRegional, Medium and Large scale, deterministic model only applicable fordetailed scale and data required for this model is sometimes really difficult toobtain. The advantages of deterministic model as mentioned above let us tounderstand the landslide triggering processes; hazard degree by using factor ofsafety (FS) value; detailed geotechnical soil properties can be known, suitable forsmall area particularly for engineering work and provide the best quantitativeinformation on landslide hazard.

Information for landslide hazard assessment can be manage into databasein Geospatial software known as Geographic Information System (GIS) andcombined with spatial analysis, modeling help us in deterministic model based onslope stability approach to produce landslide hazard map. In other hand, thedevelopment of Remote Sensing data that can be used to develop spatial model inidentifying areas of high hazard of landslide occurrence. Even though there aresources of errors and uncertainties in GIS-based landslide hazard assessment, anexample in this exercise is the existence of undefined value in DEM data derivedfrom elevation point. But according to van Westen in Hadmoko 2007, slope angleand slope direction are low in degree of uncertainty because most of uncertaintiesare caused by the human factor such as qualitative data interpretation fromsatellite images.

Page 19: Landslide Hazard Modelling using ILWIS 3.x (Aji Putra Perdana-MPPDAS)

ConclusionLandslide hazard assessment is very important in spatial planning due to

the increasing of landslide disaster. Many methods are used to determine landslidehazard, one of the methods is deterministic model for small area or detailedresearch area. The advantages of this model are let us to understand the processes;mechanism occurs and hazard degree of safety factor. Deterministic modelsimulation is based on slope data that can show result of slope stability thatinfluenced by soil water content to the area of each hazard level.

This exercise use elevation point data to get Digital Elevation Model(DEM) and then derived as slope data. Slope stability scenarios in research areashow mathematical model; regression to predict unstable area that is y = 25688x +7547 with R² = 0.933. Using GIS software as tool for mapping and its spatialanalysis helps us to process and analyze or applied this white-box model to becreated as landslide hazard map.

References

Hadmoko, D. 2007. Toward GIS-based itergrated landslide hazardassessment:critical overview. Indonesian Journal of Geography, Vol 39.No 1 June 2007 pp 55-77.

Hadmoko, D. 2009. Practical Guideline Deterministic Modelling On LandslideHazard Zonation. Faculty Of Geography.

Is¸ k Yilmaz and Inan Keskin. 2009. GIS based statistical and physical approachesto landslide susceptibility mapping (Sebinkarahisar, Turkey). Bull EngGeol Environ (2009) 68:459–471. DOI 10.1007/s10064-009-0188-z.

Mantovani , F., Gracia, F. J., Domenico de Cosmo, P. Andrea Suma. 2009. Anew approach to landslide geomorphological mapping using the OpenSource software in the Olvera area (Cadiz, Spain). Landslides (2010)7:69–74.DOI 10.1007/s10346-009-0181-4