Upload
leminh
View
220
Download
3
Embed Size (px)
Citation preview
Modelling crops and cropping systems – evolving purpose, practice and prospects.
Brian Keating and Peter Thorburn, CSIRO Australia
Outline
1. Evolution of crop-soil-systems models over the last Century
2. Patterns of model use: 2000-2015
3. Looking forward – what real world impact is arising from the modelling activity?
Classical Plant Growth Analysis in the 1920’s
1. Early plant physiologists Gregory 1917
VH Blackman 1919
West et al 1920
Fisher 1921
Briggs 1928
Relative Growth Rate = Net Assimilation Rate x Leaf Area Ratio
Advances in quantifying crop canopy process – 1940’s
1. Watson (1947) First used the concept of Leaf
Area Index (LAI)
2. Monsi and Saeki (1953) Modified Beer’s Law to model
radiation capture in crop canopies
20th Century soil – plant science
1. EJ (Sir John) Russell 1st Edition in 1912
“In all ages the growth of plants has interested thoughtful men” [and women].
2. Focus was on how plants responded to soil processes.
3. 36 chapters in 8th Edition (1950) No integration
No predictive modelling apart from Penman
1950’s & 60’s - the De Wit contribution
1. Mathematical analysis of complex plant and soil processes
2. First dynamic mechanistic simulation models of canopy photosynthesis (1965) and crop growth (1968)
Kees (Cornelis) De Wit
1950’s - The van Bavel contribution
1. 1953 – was calculating daily water balances on an IBM machine for different regions of the USA
One punch-card per day!
2. Applied to calculating drainage design and irrigation requirements
Cornelius HM van Bavel 1921 - 2014
1960’s - Australian water balance models – R.O. Slatyer et al.
Computer based water balance models from 1962
WATBAL operational in land and agricultural assessment for Australian continent from 1968 (Nix and colleagues)
1960’s and 70’s - The WG (Bill) Duncan contribution
First published model for maize in 1967
Used an IBM 7044 machine – 6 seconds to calculate one day’s photosynthesis
Interested in integration
"one way of putting what we know about the parts of a system back together to see how it functions as a whole"
W.G. Duncan
1971 Agricultural Productivity – Ann. Review Plant Physiol. RS Loomis, WA Williams and AE Hall
“need for integrative research by plant physiologists and to
show how techniques of modelling and simulation are
a powerful aid to such research.”
RS (Bob) Loomis 1928-2015
1970’s and 80’s - The emergence of the CERES, GRO, IBSNAT, and DSSAT effort
1. Crop-Environment-Resource Synthesis (CERES) from Texas
2. GRO models from Florida
3. Decision Support System Agrotechnology Transfer (DSSAT)
Joe Ritchie
James W Jones
1984-1995 -Systems Analysis and Simulation for Rice Production (SARP) Systems Approaches to Agricultural Development (SAAD)
• Initially IRRI and Wageningen Agricultural University
• Fritz Penning de Vries, Martin Kropff
• Applications of the WAU models (MACROS)
• ORYZA Rice model emerged
• Evolved into the SAAD Forum during early 1990’s
• Facilitated the Dutch – American – Australian modelling connections via ICASA – and wider CGIAR connections
1984-1992 Australia-Kenya Dryland Farming Systems Research Project
Benson Wafula with 8086 chip Olivetti M21 in 1985
A very challenging environment for these early crop-soil models - Low yielding low input farming systems - Complexities of low plant populations, intercrops, weeds, manure, residues - Highly erratic dryland farming – models needed to be configured to enable
tactical management - Eventually decided we had to fundamentally reinvent the cropping systems
platform
1990 – APSIM emerged driven by farming systems needs
Early process routines for maize growth, water balance and nitrogen balance came from Ceres Maize and elsewhere.
APSIM’s key innovation was it’s “farm systems” conceptualization - with supporting modular software engineering
“Crops [pastures, animals, seasons and farm managers] come and go, each finding the soil in a particular
state and leaving it in an altered state.”
…. McCown et al 1996
A biophysical (or code) lens through time
Presentation title | Presenter name | Quantitative description of plant and soil processes Page 23
Cropping systems models
Soil process models
Crop-soil yield
models
Plant Process models
Quantitative description of plant and soil processes 1910 - 50
1970-80’s
1990’s
1960-70’s
1950-60’s
Foundations
1st Generation
2nd Generation
3rd Generation
Next Generation ??
2003 – 2nd International Symposium: Modelling Cropping Systems
1. A major milestone for cropping systems models??
Reference publication for;
– DSSAT,
– APSIM,
– CropSyst,
– STICS,
– Wageningen Crop Models
2. Opportunity to examine patterns of model use 2000 – 2015
Citations* of the 2003 EJA Special issue papers (all years since 2003)
Paper/Model Citations EJA
2003 paper
Wageningen crop models 246
DSSAT 927
APSIM 853
CropSyst 446
STICS 326
Total+ 2798
* Citation search from Thompson Reuters Web of Science on 16/02/2016
X
Citation rates continue to increase …
Citations per annum for 5 model overview papers from EJA vol 18 special issue.
Presentation title | Presenter name | Page 27
0
50
100
150
200
250
300
350
400
450
2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015
Nu
mb
er o
f C
itat
ion
s p
.a.
Wageningen crop
models
DSSAT
APSIM
CropSyst
STICS
Papers that have model name in title or topics – all years (including abstract)
Paper/Model Total number
of papers *
Total citations *
Wageningen crop models 313# 4519#
DSSAT 590 5929
APSIM 701 9909
CropSyst 203 3085
STICS 437 6747
Total+ 2253 30255
* Papers from all years with model name in title or topics (including abstracts). # Based on three searchable models, WOFOST, SUCROS and LINTUL + There may be some double counting of papers that report on more than one model (<5%)
0
20
40
60
80
100
120
140
160
2000 2005 2010 2015
Nu
mb
er
of
pap
ers
Year
DSSAT and APSIM(n=280)
Agronomy Methods
Climate change Model development and testing
Hydrology SOM dynamics
Precision Agriculture Weed Management
Agro-forestry LCA
Soil compaction Hydrology
Conservation Ag Crop breeding
Regional yield prediction Crop-livestock systems
Grain quality Pest/Disease
Yield gap Plant breeding
Conservation Agriculture GHG emissions
What are these papers doing with the models ?
DSSAT and APSIM (n = 280)
Agronomy Agronomy
Climate Change
Methods
Model testing
0
20
40
60
80
100
120
140
160
2000 2005 2010 2015
Nu
mb
er
of
pap
ers
Year
DSSAT and APSIM(n=280)
Agronomy Methods
Climate change Model development and testing
Hydrology SOM dynamics
Precision Agriculture Weed Management
Agro-forestry LCA
Soil compaction Hydrology
Conservation Ag Crop breeding
Regional yield prediction Crop-livestock systems
Grain quality Pest/Disease
Yield gap Plant breeding
Conservation Agriculture GHG emissions
• Agronomic and climate change applications dominate model use
• Significant on-going activity in modelling methods, development and testing
• Rapid growth in climate change applications over last 5 years
APSIM and DSSAT papers sum of 2000, 2005, 2010 and 2015 (n = 280)
China Australia global
USA India southern Africa
Canada Iran Italy
Kenya Brazil NZ
Ghana Other (38 countries)
Geographic distribution of model use
Australia
China
Global USA
Other
India
• APSIM and DSSAT used in 50 countries
• Australia, China, USA and India represent over half of applications
Geographic distribution over time
Presentation title | Presenter name | Page 31
0
20
40
60
80
100
120
140
160
2000 2005 2010 2015
Nu
mb
er o
f p
aper
s
APSIM and DSSAT Country of Application
China Australia global USA
India southern Africa Canada Iran
Italy Kenya Brazil NZ
Ghana Other (38 countries)
China
Australia
Global
USA
India • Australian applications have been strong for >10 years
• China, USA and India papers growing in recent years
What is driving model use in Australia ?
1. Long history of modelling in strongly water limited agriculture
2. Some of the most variable rainfall environments
3. APSIM factor – availability of elaborated “systems” capabilities for two decades
Beyond “white peg” agronomy at Australia’s biennial Agronomy Meeting
…from Robertson et al 2015
Use of simulation modelling in Australian agronomy
10-20% of all papers use models
Closer examination of 70 papers that used APSIM in 2015
Evidence of “real world” impact
Consideration of model relevance to a real world
decision problem
No consideration of an impact pathway beyond
research domain 61 papers
8 papers
1 paper
Impact from agronomy model application?
“Other” includes: Precision Ag, Grain Quality, Yield Gap, Pest/Disease, Breeding, Food Security and crop-livestock
0%
5%
10%
15%
20%
25%
30%
35%
climate change agronomy modeldevelopmentand methods
other
Pe
rce
nt
of
pap
ers
2015 papers in WoS using APSIM (n=70)
Case studies of model impact in Australian grains Industry
New agronomic strategies
Spring sown mungbeans
Skip-row sorghum
Soil water monitoring
Summer fallow management
Early sown wheat
Influence on breeding programs
Long season canola
Early maturing peanuts
Tactical within season decision
support
Yield Prophet
(FARMSCAPE)
Aflatoxin management
in peanuts
from Robertson et al 2015
Yield Prophet
(FARMSCAPE)
Timely and specific data have limited model use in real world decision making.
Expect the “big data” revolution will start to change this.
Impact from climate change model applications ?
“Other” includes: Precision Ag, Grain Quality, Yield Gap, Pest/Disease, Breeding, Food Security and crop-livestock
0%
5%
10%
15%
20%
25%
30%
35%
climate change agronomy modeldevelopmentand methods
other
Pe
rce
nt
of
pap
ers
2015 papers in WoS using APSIM (n=70)
What about the climate change papers ?
Our WoS sample was imperfect ….
- not yet captured in 2015 Web of Science - 2014 ahead of IPPC AR5 may have given different results
Research on decision context
and impact pathway
Evidence base & Data Platforms
Future Farm
Practice
Potential impact pathways for agricultural climate change analysis
Climate Change
Modelling Activity
Impact Analysis
Adaptation Options
Mitigation Options &
Consequences
Methods and Models
Current Farm
Practice
Global Policy (IPCC,COP etc)
National or Regional Policy
Agri-industry Strategy
Research Strategy
(eg. Breeding)
Research on decision context
and impact pathway
Evidence base & Data Platforms
Future Farm
Practice
Potential impact pathways for agricultural climate change analysis
Climate Change
Modelling Activity
Impact Analysis
Adaptation Options
Mitigation Options &
Consequences
Methods and Models
Current Farm
Practice
Global Policy (IPCC,COP etc)
National or Regional Policy
Agri-industry Strategy
Research Strategy
(eg. Breeding)
Andy Challinor – Time of Emergence analysis for crop breeding in sub Saharan Africa
Research on decision context
and impact pathway
Evidence base & Data Platforms
Future Farm
Practice
Potential impact pathways for agricultural climate change analysis
Climate Change
Modelling Activity
Impact Analysis
Adaptation Options
Mitigation Options &
Consequences
Methods and Models
Current Farm
Practice
Global Policy (IPCC,COP etc)
National or Regional Policy
Agri-industry Strategy
Research Strategy
(eg. Breeding)
Australia’s Emissions Reduction Fund / Carbon Farming Initiative
Research on decision context
and impact pathway
Evidence base & Data Platforms
Future Farm
Practice
Potential impact pathways for agricultural climate change analysis
Climate Change
Modelling Activity
Impact Analysis
Adaptation Options
Mitigation Options &
Consequences
Methods and Models
Current Farm
Practice
Global Policy (IPCC,COP etc)
National or Regional Policy
Agri-industry Strategy
Research Strategy
(eg. Breeding)
Many papers on climate change impacts and adaptation in agriculture
Research on decision context
and impact pathway
Evidence base & Data Platforms
Future Farm
Practice
Potential impact pathways for agricultural climate change analysis
Climate Change
Modelling Activity
Impact Analysis
Adaptation Options
Mitigation Options &
Consequences
Methods and Models
Current Farm
Practice
Global Policy (IPCC,COP etc)
National or Regional Policy
Agri-industry Strategy
Research Strategy
(eg. Breeding)
Many papers on climate change impacts and adaptation in agriculture
Some reflections - using models from agronomy to climate change
Models need to address the key biophysical processes in credible and transparent ways.
Ensemble approaches may have their place but not as a remedy for bad models or modelling practice.
Continually question model performance and parameterization and look for evidence emergent behaviours are predicted in sensible ways.
Engage directly with the target decision makers to understand their world and needs and build trust.
Collect baseline data and monitor measures of impact. Include a stronger “real world impact” focus in future AgMIP activity?