- Email: [email protected]
Options for Victorian agriculture in a “new” climate: Pilot study linking climate change and land suitability modelling
Environmental Modelling & Software 21 (2006) 1280e1289 www.elsevier.com/locate/envsoft
Options for Victorian agriculture in a ‘‘new’’ climate: Pilot study linking climate change and land suitability modelling Adam Hood a,*, Bob Cechet b, Hemayet Hossain c, Kathryn Sheffield c a
Land and Catchments Division, Department of Sustainability and Environment, 2/8 Nicholson Street, East Melbourne, Victoria 3002, Australia b Risk Research Group, Geohazards Division, Geosciences Australia, Australia c Strategic Resources Planning Unit, Spatial Sciences, Department of Primary Industries (DPI), Victoria, Australia Received 16 April 2004; received in revised form 5 April 2005; accepted 5 April 2005 Available online 7 November 2005
Abstract This pilot project determines the potential for growing Cool Climate Grapes, high yield pasture and blue gum across Gippsland (Victoria, Australia), and the likely shift in that potential as projected climate change occurs in the region. It provides an indication of the information that can be generated combining Land Suitability Analysis and climate change scenarios. By considering the changing suitability of a few key agricultural enterprises in this region, the project offers an important direction for the agricultural elements of the Victorian Greenhouse Strategy. This pilot project utilises Land Suitability Analysis (LSA) developed by the former Victorian Department of Natural Resources and Environment (DNRE) under the National Land and Water Resources Audit of Australia and further enhanced with Local Government and Catchment Management Authorities across Victoria. This modelling technique allows for an assessment of impacts on a wide range of commodities beyond the historic focus of cropping and pasture models. Following a series of workshops with agronomists, soil scientists, climate scientists, strategic planners and growers, a series of maps/overlays were prepared for Gippsland showing potential suitability of blue gum plantations, Cool Climate Grapes and high yield pasture, as an initial illustration of the methodology. This potential suitability has since been adjusted/modified using a variety of climate change scenarios developed by CSIRO Climate Impacts Group, resulting in an assessment of the change in potential suitability for each enterprise at years 2020 and 2050. The resultant work can be analysed using Geographic Information Systems (GIS) to consider change over time and space in the region. Combined with changes in water availability, this may act as an innovative and powerful decision support tool for policy makers in government and industry. Ó 2005 Elsevier Ltd. All rights reserved. Keywords: Climate change; Expert systems; Modelling climate change; Land Suitability Analysis; Agricultural commodities; Victorian Greenhouse Strategy; Ecologically sustainable agriculture initiative; Climate impact scenarios
Software availability Name of software: OzClim Australian climate change scenario generator, version 2.3 Web page: http://www.dar.csiro.au/publications/ozclim. htm
* Corresponding author. Tel.: C61 3 96379021; fax: C61 3 96378485. E-mail address: [email protected] (A. Hood). 1364-8152/$ - see front matter Ó 2005 Elsevier Ltd. All rights reserved. doi:10.1016/j.envsoft.2005.04.022
Developer: CSIRO Atmospheric Research, PMB 1, Aspendale, Vic 3195, Australia (contact: rogerjones@ csiro.au or [email protected]) Year first available: 2001 Program language: Borland Delphi Operating system: Windows 2000/XP Program size: 40 MB approx (Program, GUI & 0.25 degree horizontal resolution data) Minimum hardware required: 500 MHz Pentium III processor, 128 MB RAM, 20 GB HDD, and CD-ROM
A. Hood et al. / Environmental Modelling & Software 21 (2006) 1280e1289
Hardware used: 1 GHz Pentium III processor, 1 GB RAM Availability and cost: Free (0.25 degree horizontal resolution data). Higher horizontal resolution data (range of geophysical fields available including percentile occurrence for daily and monthly data) available at a cost. Please contact developer (details above). Name of software: Land Suitability Analysis tool for agriculture (includes models for grapes, pasture and blue gum) Web page: Developer: Primary Industries Research, Department of Industries, Government of Victoria, Australia Year first available: 1992 Program language: Developed using Model Builder tools in ArcView GIS package and Macro language in IDRISI32 GIS package. Operating system: Windows 2000 Program size: 174 MB for ArcView and 95 MB for IDRISI32 Minimum hardware required: PC Pentium III, 512 MB RAM, 64 MB Graphics RAM, 60 GB HD Hardware used: Pentium 4 Xeon PC with 1 GB RAM, 128 MB Graphics RAM, 120 GB HD Availability and cost: Availability and Price on request
1. Introduction
1281
of the ‘‘Gippsland Model’’, a broader regional strategic planning model developed under the National Land and Water Resources Audit of Australia (NLWRA, 2002). The LSA is considered useful at this early phase of regional impact assessment for agriculture for several reasons: 1. It enables assessment at regional and strategic levels where data paucity and computing costs often limit more complex dynamic plant, catchment or farm models. 2. It is based on an ‘‘expert systems’’ approach and the associated communication effort delivers a range of peripheral outcomes for the project including raising awareness about climate change. 3. It is highly flexible and can be adjusted depending on the specific question being asked. 4. The tool is simple to manipulate and interrogate, providing a highly transparent modelling framework throughout the process. For this pilot study, LSA uses discrete spatial data sets of soil, landscape and climate attributes. The climate factors introduced into the model are developed using OzClim, a climate change scenario generator developed by the Australian Commonwealth Scientific and Industrial Research Organisation (CSIRO) Division of Atmospheric Research. Fig. 1 outlines the full extent of the study. For the purposes of this paper only the numbered boxes will be discussed in detail, while the dashed sections indicate future directions of the project.
1.1. Background The Greenhouse Strategy of the Australian State of Victoria (DNRE, 2002) highlighted the need for a range of response actions to deal with the goals of raising awareness of greenhouse issues, limiting greenhouse emissions, living in a carbon-constrained society, building capacity within industry and better understanding climate change impacts and the development of adaptation options. This paper examines part of a pilot project that sets out to consider some climate change issues facing agricultural production in parts of Victoria and some options for response. The modelling was seen as a starting point for opening dialogue between state and regional policy makers and specific agricultural industries. The project considers long-term averages and suitability of production. Full documentation of these projects and the methodologies used are available from the corresponding author. 1.2. Project design The pilot study utilises a specific Land Suitability Analysis (LSA) tool that was created as an integral part
2. Climate scenario tools Climate scenarios are generated using a combination of greenhouse gas emission scenarios and global climate general circulation models (GCMs).
2.1. SRES e emissions scenarios This pilot study employs the latest set of emissions scenarios issued by the Intergovernmental Panel on Climate Change (IPCC) (IPCC, 2000). The Special Report on Emission Scenarios (SRES) produced 40 future emission scenarios for greenhouse gases and sulfate aerosols. These scenarios cover a wide range of the main driving forces of future emissions, from demographic to technological and economic developments. Six of the 40 SRES emissions scenarios (A1B, A1T, A1F1, A2, B1 and B2 representing the families of technologyepopulatione economy futures) are shown in Fig. 2. Further details of these can be found at the SRES website (http://sres. ciesin.org).
1282
A. Hood et al. / Environmental Modelling & Software 21 (2006) 1280e1289
Climate Scenario Tools
Land Suitability Analysis Commodities for Analysis
IPCC Climate Change Evidence
• Rapid Assessment • GVAP Industry vs. Regional • Risk and Opportunity
• Recent Rapid Change - Industrial Society • Link to Greenhouse Gas Emissions • Relationship CO2 - Climate Change
Multi-criteria Evaluation 3.2
3.3
• AHP - IDRISI DSS Module • Weighted Growth Factors • Soil Attributes • Landform • Climate
General Circulation Models
2.1
SRES - Emissions Scenarios
2.2
Expert Systems Modelling • Workshops
OzClim - Climate Scenario Generator • Long-Term Averages
2.3
• Commodity Specific Attributes GIS - Spatial Modelling 3.4
• GIS Coverages to 2100 by Scenario and GCM
• ESRI Model Builder • Refine data coverage
Regional Commodity Suitability Mapping • 2020, 2050 •A1F1, B1 scenarios (Extreme, Moderate)
4.1
Biophysical Impact Analysis Change Detection Maps
4.2
Economic and Social Implications Industry and Regional Adaptation Options Fig. 1. Project design (relevant text sections indicated).
2.2. General circulation models The IPCC (2001) provides estimates of global-average warming based on 35 of the 40 SRES emission scenarios (Fig. 3). These estimates were determined using simple models tuned to match the global performance of
GCMs. The two inner lines represent the variation in results according to emission scenarios averaged across all models, while the two outer lines allow for model to model variation. The GCMs are the best available tools to study possible future climate and include complex equations
6
35 30 25 20
Temperature change (°C)
CO2 emissions (Gt/year)
A1F1 A2 A1B B2 A1T B1
15
IS92a
10 5 0 1980
2000
2020
2040
2060
2080
2100
Year Fig. 2. Six of the SRES emission scenarios (Giga-tonnes per year) for carbon dioxide (CO2), plus the IS92a mid-case scenario used by the IPCC in 1996 (IPCC, 2001).
5
Range due to emissions uncertainty and climate response uncertainty
4 Range due to emissions uncertainty
3 2 1 0 1980
2000
2020
2040
2060
2080
2100
Year Fig. 3. Projections for global-average warming relative to 1990 for an envelope of SRES greenhouse gas and sulfate aerosol emission scenarios (IPCC, 2001).
A. Hood et al. / Environmental Modelling & Software 21 (2006) 1280e1289
to represent processes that govern atmospheric and oceanic circulations, temperature, precipitation etc. Most GCMs satisfactorily simulate observed temperature, precipitation and mean sea-level pressure patterns over south-eastern Australia (Whetton et al., 2002).
2.3. OzClim climate scenario generator OzClim is a PC-based regional climate scenario generator for Australia that simplifies the process of calculating future climate change for application to impact models. It is chiefly used as a research tool for climate sensitivity studies, and it features a graphical user interface and point and click technology for ease of use, fast calculations and visualisation capabilities (Page and Jones, 2001). OzClim also allows comparison of output from a range of climate models and construction of future climate change scenarios for risk assessment. The details of the modelling science driving OzClim can be found at http://www.dar.csiro.au/impacts/ozclimscience. html. CSIRO Atmospheric Research has ‘‘tailored’’ a version of the OzClim scenario generator for the Victorian Government to describe how the Victorian climate may change during the next 100 years. This has been achieved by combining a range of projected global warming as given by IPCC (2001) (lowemiddleehigh climate sensitivity derived from Fig. 3) with projected regional climate changes (Victorian region) obtained from a range of atmosphereeocean general circulation models (AOGCMs) and also high horizontal resolution atmospheric general circulation models (AGCMs) forced by the output of the AOGCMs. OzClim is grid-based, and in this pilot study it employs an array of about 20,000 points (Victorian region), producing a grid spacing of about 5 km over Victoria. Climatic variables such as maximum and minimum temperature, and rainfall were augmented by evaporation, solar radiation and relative humidity. In addition, deciles for daily and monthly maximum and minimum temperature, and rainfall are available and can be modified according to selected emission scenarios. OzClim also has the capacity to ‘plug in’ a range of impact models to explore the sensitivity of impacts to different emissions scenarios and climatic ranges. Impact models for the Victorian region developed for this pilot study included simple water balance (precipitation minus potential evaporation), Heating Degree Days, Branas index (estimation of likely disease pressure on grapes caused by temperature and rain and is calculated by multiplying mean monthly temperature by mean monthly rainfall and summing the totals for the period October through March) and frost frequency. Models of effective rainfall, heat stress and chilling units are currently under development.
1283
3. Land Suitability Analysis 3.1. The land suitability framework The land suitability framework was developed by the Food and Agriculture Organisation of the United Nations (FAO). Unlike land capability assessment, which essentially focuses on the limitations to land use in an area, LSA assesses the suitability of land for a specific use or commodity (Conacher and Conacher, 2000). LSA can be used to assess potential climate change impacts by using spatial data sets and analysing the various elements based on a function of interaction between them and/or through a multi-criteria analysis method. Due to its probabilistic nature, LSA lends itself well to assessing the impacts of climate change as it can provide assessments of changing potential for production. 3.2. Multi-criteria evaluation tools and concepts This pilot study used LSA developed by the Victorian Government under the National Land and Water Resources Audit of Australia (NLWRA). The methodology incorporates a Multiple Criteria Evaluation (MCE) within a Geographic Information Systems (GIS) environment. MCE has been developed to improve spatial decision making when considering multiple objectives and conflicting preferences or priorities (Malczewski, 1999). MCE can be an effective decision-making tool for complex issues, using both qualitative and quantitative information. It combines the information from several criteria to form a single composite index of evaluation, and has been utilised around the world for Land Use Suitability modelling (Burrough et al., 1992; Ghafari et al., 2000; Ceballos-Silva and Lopez-Blanco, 2003). The MCE is implemented using the experts systems modelling approach of Analytical Hierarchy Process (AHP) developed by Saaty (1980). AHP provides a framework that allows hierarchical combination of criteria and incorporates experts participation in the decision-making process. Compared to empirical models based purely on the correlation amongst the data, the experts modelling system incorporates the knowledge of the experts that can address many of the other issues and understandings that may be important to decision making. The experts model is also better suited when access to good data is limited. The AHP orders critical factors into a hierarchy of importance. This improves the representativeness of the mapping because not all factors have an equal weighting of importance. It also allows criteria to trade-off with each other depending on the importance weights assigned to them. Furthermore, AHP can deal with criteria that are interdependent, both from the effect on land and in the interaction between spatial units.
1284
A. Hood et al. / Environmental Modelling & Software 21 (2006) 1280e1289
the quantitative aspects of human thought: the qualitative to define the problem and its hierarchy and the quantitative to express judgements and preferences concisely. The process itself is designed to integrate these dual properties.’’ (Saaty, 1995, p. 18). AHP also provides an effective structure for group decision making by imposing a discipline on the group thought’s processes. The necessity of assigning a numerical value (i.e., weight) to each variable of the problem helps decision makers to maintain cohesive thought patterns and to reach a conclusion. In addition, the consensual nature of group decision making improves the ‘‘consistency of judgements and enhances the reliability of the AHP as a decision-making tool’’ (Saaty, 1995, p. 5). Although a variety of techniques exist for the development of criterion weights, one of the most promising is that of pair-wise comparison in AHP. In the procedure for multi-criteria evaluation using Weighted Linear Combination (WLC), it is necessary that the weights sum to one. In AHP, these weights are calculated by taking the principal eigenvector of a square reciprocal matrix of pair-wise comparison between the criteria. All possible pairs of criteria are compared by experts
Weighted Linear Combination Model
Max
Trade off
Traditional Boolean Model
Min Risk averse (AND)
Risk
Risk-taking (OR)
Fig. 4. Decision strategy space (Eastman, 1999).
The Analytic Hierarchy Process breaks down a complex unstructured situation into its component parts. These parts, or variables, are arranged into a hierarchical order with numerical values assigned to the subjective judgements based on the relative importance of each variable. These judgements are synthesised to determine which variables have the highest priority and should be acted upon to influence the outcome of the situation. In this respect, ‘‘AHP incorporates both the qualitative and
Perennial Ryegrass/White Sub-Clover Pasture
Land Suitability Victoria
Climate Change Scenario A1F1 (2050) Gippsland Region
Suitability Index Restricted 0 (Very low) 1 (Very low) 2 (Low) 3 (Low) 4 (Low) 5 (Moderate)
This map is suitable for strategic planning purposes, rather than specific site investigation. Further detailed site analysis should be carried out prior to site-specific development proceeding.
Copyright © Department of Natural Resources and Environment (DNRE), 2002. Map prepared by the Strategic Resources Planning Unit, Agriculture Victoria, Department of Natural Resources and Environment. This data can not be guaranteed to be without flaw of any kind, and therefore AVS disclaims all liability of error or inappropriate use of the map data. Data source: Climate data - CSIRO-OzClim (Data Scale: 1:250,000); Soils and DEM data - DNRE, 2001/2002 (Data Scale: 1:25,000). Base data source: Corporate Geospatial Data Library, DNRE, 2001 (Data Scale: 1:25,000); Auslig Geodata Topo-250K.
10
0
10
20
30
40
50
Map Scale: 1:1,100,000
60
70
80 Kilometers
Legend
7 (Moderate)
Gippsland CMA and LGA boundary Public Land
8 (High)
State Border
9 (High)
Principal Road
10 (High)
Lake
No Data
River
6 (Moderate)
N
Prepared for: Options for Agriculture in a "New Climate" Edition 3, June 2002, SRP Project V12CCa; Map Number M24CCb Model No. A1F1 2050 land AMG Zone 55 (Transverse Mercator) Projection
Fig. 5. Future land suitability (year 2050) for Cool Climate Grapes under an extreme emissions scenario (A1F1; see Fig. 2).
1285
A. Hood et al. / Environmental Modelling & Software 21 (2006) 1280e1289
Climate Suitability Change 2000 - 2050 Cool Climate Grapes Climate Change Scenario B1 (2000) - A1F (2050)
Victoria
Gippsland Region
Legend Decreased Unchanged Increased Gippsland CMA and LGA boundary Public Land State Border
This map is suitable for strategic planning purposes, rather than specific site investigation. Further detailed site analysis should be carried out prior to site-specific development proceeding.
Copyright © Department of Natural Resources and Environment (DNRE), 2002. Map prepared by the Strategic Resources Planning Unit, Agriculture Victoria, Department of Natural Resources and Environment. This data can not be guaranteed to be without flaw of any kind, and therefore AVS disclaims all liability of error or inappropriate use of the map data. Data source: Climate data - CSIRO-OzClim (Data Scale: 1:250,000); Soils and DEM data - DNRE, 2001/2002 (Data Scale: 1:25,000). Base data source: Corporate Geospatial Data Library, DNRE, 2001 (Data Scale: 1:25,000); Auslig Geodata Topo-250K.
10
0
10
20
30
40
50
Principal Road Lake River
60
Map Scale: 1:1,100,000
70
80 Kilometers
N
Prepared for: Options for Agriculture in a "New Climate" January 29, 2003, SRP Project V12CCa Model No. B1 AMG Zone 55 (Transverse Mercator) Projection
Fig. 6. Change in climate suitability (year 2050 compared to 2000) for Cool Climate Grapes under the A1F1 extreme emissions scenario.
on a nine-point continuous scale for their relative importance in determining the suitability of the stated objective. The ratings are then introduced in a matrix to calculate the weights (Eastman, 1999). The traditional land suitability models use Boolean aggregation method of constructing suitability indexes that present a rigid binary choice of acceptance or rejection. The Weighted Linear Combination (WLC) method in MCE introduces a soft or ‘‘fuzzy’’ concept of suitability in standardising criteria. Instead of hard Boolean decision of assigning absolute suitability or unsuitability to a location for a given criteria, it is scaled to a particular common range where suitable and unsuitable areas are continuous measures. The aggregation method uses weighted linear combination, which retains the variability of continuous criteria and allows criteria to trade off with each other (Eastman, 1999). A low suitability defined by one criterion may be compensated by a high suitability score in another criterion. Trade-off between criteria will depend on weights assigned to them. As depicted in the decision
strategy space (see Fig. 4), this model is a significant improvement over the Boolean approach by avoiding its extreme risk aversion (binary rejection) and extreme risk taking (binary acceptance) nature. 3.3. Expert systems modelling e workshops To develop the LSA for this pilot study, each commodity was analysed using a series of expert systems model workshops involving agronomists, growers and soil scientists with expertise in the Gippsland region. The workshops had several objectives: To build awareness of current climate change science. To define the goal of the modelling exercise in the context of the project. To develop a model using Analytical Hierarchy Process e multi-criteria evaluation. To verify and fine-tune outputs following GIS modelling.
1286
A. Hood et al. / Environmental Modelling & Software 21 (2006) 1280e1289
Cool Climate Grape Heat Degree Days (Oct-Apr) Climate Change Scenario A1F1 (2050)
Victoria
Gippsland Region
Suitability Index Restricted 0 (Very low) 1 (Very low) 2 (Low) 3 (Low) 4 (Low) 5 (Moderate) 6 (Moderate) 7 (Moderate) This map is suitable for strategic planning purposes, rather than specific site investigation. Further detailed site analysis should be carried out prior to site-specific development proceeding.
8 (High)
10 (High)
Copyright © Department of Natural Resources and Environment (DNRE), 2002. Map prepared by the Strategic Resources Planning Unit, Agriculture Victoria, Department of Natural Resources and Environment. This data can not be guaranteed to be without flaw of any kind, and therefore AVS disclaims all liability of error or inappropriate use of the map data. Data source: Climate data - CSIRO-OzClim (Data Scale: 1:250,000); Soils and DEM data - DNRE, 2001/2002 (Data Scale: 1:25,000). Base data source: Corporate Geospatial Data Library, DNRE, 2001 (Data Scale: 1:25,000); Auslig Geodata Topo-250K.
10
0
10
20
30
40
50
Map Scale: 1:1,100,000
60
70
80 Kilometers
N
Gippsland CMA and LGA boundary Public Land State Border Principal Road
9 (High)
No Data
Legend
Lake River
Prepared for: Options for Agriculture in a "New Climate" Edition 3, June 2002, SRP Project V12CCa; Map Number M24CCb Model No. A1F1 2050 AMG Zone 55 (Transverse Mercator) Projection
Fig. 7. Heating Degree Days (OctobereApril; year 2050) for Cool Climate Grapes under the A1F1 extreme emissions scenario (A1F1; see Fig. 2).
The final model was developed in the form of a weighted hierarchy, as illustrated in Fig. 4.
3.4. Geographic Information Systems e Spatial Modelling In order to use the scenarios generated by OzClim as an input into the multi-criteria land suitability model; a suitable link between the two models was necessary. For the purposes of this pilot study, macros have been developed that take the scenarios generated by OzClim and convert them to a format readable by the land suitability models that have been developed using the ArcView Model Builder. The process involved: 1. Generation of all required scenarios using OzClim. 2. Exporting these scenarios in 16 bit IDRISI format. 3. Application of the interface macros, developed using IDRISI, to convert these IDRISI files to ArcView 32 bit Binary Raster format.
4. Importing the Binary Rasters into ArcView and converting them to ArcView grid layers. 5. Including these grid layers as inputs to the land suitability models. The selected OzClim climate change GCM was the CSIRO MK2 utilising both the B1 and A1F1 greenhouse gas emission scenarios (see Fig. 2). CSIRO MK2 B1 climate scenarios were created for the years 2020 and 2050. MK2 A1F1 scenario was only employed for the year 2050. The base climate (year 2000) was compiled from the average of the observational record between 1961 and 2000. 3.5. Comparison e other assessments of climate change impacts on agriculture Many modelling tools are appropriate for considering climate change impacts on agriculture and most variations in the approach to climate change occur around the choice of climate change scenario and the
1287
A. Hood et al. / Environmental Modelling & Software 21 (2006) 1280e1289
Cool Climate Grapes Climate Change
Climate 0.46
Soil 0.34
Slope 0.20 Degree 0.75
Aspect 0.25
Heat Degree Days (Oct-Apr) 0.47
Flower rainfall (Nov-Dec) 0.19
Rain Ripening (Mar-Apr) 0.09
Spring Frost (Sep-Nov) 0.20
Branas Index (Annual) 0.05
1350-1500 rating 1
<85mm Nov and Dec rating 1
<85mm Mar and Apr rating 1
<0-5 days rating 1
<4000 rating 1
1500-1650 rating 0.8
<85mm Nov >85mm Dec or alternate rating 0.8
<85m Mar& >85mm Apr or alternate rating 0.8
<5-10 days rating 0.8
4000-5000 rating 0.9
>85mm Nov and Dec rating 0.3
>85mm Mar and Apr rating 0.1
10-15 days rating 0.5
5000-6500 rating 0.7
900-1100 rating 0.1
15-25 days rating 0.2
6500-8000 rating 0.4
<900 rating 0
>25 days rating 0
>8000 rating 0.2
>1650 and 1100-1350 rating 0.4
Drainage 0.101
pH 0.06
Sodicity 0.19
Topsoil 0.75 Subsoil 0.25
Usable Depth 0.46
Texture 0.04
EC 0.14
Topsoil 0.75
Topsoil 0.75
Topsoil 0.75
Subsoil 0.25
Subsoil 0.25
Subsoil 0.25
Fig. 8. Example of the Analytical Hierarchy Process (AHP) hierarchy showing the criteria for growth in the model used for Cool Climate Grapes (numbers in brackets represent criteria weighting).
GCM used. Other significant differences are related to the spatial (varying scales) and non-spatial aspects of the agricultural model. Tao et al. (2003) map the implications for agricultural water cycle and agricultural production for the whole of China using crop specific water balance equations built from United Nations Food and Agriculture Organisation (UNFAO). This work finds that the key uncertainties lie in regionalisation of GCMs. Other studies have been conducted considering specific elements such as the impact of elevated CO2 or the changes in water supply (Fuhrer, 2003; Izaurralde et al., 2003). Interesting work has considered the spatial variations of vulnerability to climate change and variability across different socio-economic systems. The most poignant is a study of vulnerability of ranching on both side of US-Mexico border (Va´squez-Leo´n et al., 2003). All of these studies are dependent on
a modelling system that provides some insight to climate change over the long term and across space. The inputs to these assessments are complete with a range of substantial uncertainties. This study finds similar limiting issues with uncertainty. As a result there is a need to build uncertainty into a risk/vulnerability framework to provide a proper context in which these modelling and simulation exercises can operate.
4. Outputs of the pilot project 4.1. Regional commodity suitability mapping Three commodities were selected for analysis based on the commodities currently produced, new commodities that could be introduced into the region according
1288
A. Hood et al. / Environmental Modelling & Software 21 (2006) 1280e1289
Heat Degree Days Change 2000 - 2050 Climate Change Scenario B1 (2000) - A1F (2050)
Victoria
Gippsland Region
Legend Decreased Unchanged Increased Gippsland CMA and LGA boundary Public Land State Border Principal Road
This map is suitable for strategic planning purposes, rather than specific site investigation. Further detailed site analysis should be carried out prior to site-specific development proceeding.
Lake River
Copyright © Department of Natural Resources and Environment (DNRE), 2002. Map prepared by the Strategic Resources Planning Unit, Agriculture Victoria, Department of Natural Resources and Environment. This data can not be guaranteed to be without flaw of any kind, and therefore AVS disclaims all liability of error or inappropriate use of the map data. Data source: Climate data - CSIRO-OzClim (Data Scale: 1:250,000); Soils and DEM data - DNRE, 2001/2002 (Data Scale: 1:25,000). Base data source: Corporate Geospatial Data Library, DNRE, 2001 (Data Scale: 1:25,000); Auslig Geodata Topo-250K.
10
0
10
20
30
40
50
Map Scale: 1:1,100,000
60
70
80 Kilometers
N
Prepared for: Options for Agriculture in a "New Climate" Edition 3, June 2002, SRP Project V12CCa; Map Number M70CCa Model No. A1F1 2050 AMG Zone 55 (Transverse Mercator) Projection
Fig. 9. Change in Heating Degree Days (OctobereApril; year 2050 compared to 2000) for Cool Climate Grapes under the A1F1 extreme emissions scenario.
to potential market demand and advice from agronomists and other relevant experts. The three commodities selected were Cool Climate Grapes, high yield pasture and blue gum plantations. Initially, land suitability with base climate data (potential for the year 2000) was assessed and a series of maps prepared. Factors considered in this assessment were climate, landscape and soil. A panel of experts in a series of workshops assessed the accuracy of these maps. The base climate (year 2000) maps were verified at this stage. Land suitability outcomes were ranked in 12 categories covering high, moderate, low and very low suitability for commercial production and restrictive areas, which were deemed unsuitable for commercial production of the commodity. The process was repeated using LSA and a range of potentially ‘‘new climates’’ utilising the OzClim climate change scenarios, with a series of maps produced for the years 2020 and 2050. An example of these maps is shown in Fig. 5 depicting the future land suitability (year 2050) for Cool Climate Grapes under the extreme AIF1 emissions scenario (see Fig. 2).
4.2. Biophysical impact analysis change detection maps A further step in this pilot study was to develop change detection maps, using the land suitability maps created for the years 2000, 2020 and 2050. Fig. 6 shows the change in climate suitability (year 2050 compared to 2000) for Cool Climate Grapes under the A1F1 extreme emission scenario. These maps further enhance the information portrayed in the land suitability maps and clearly indicate areas where land suitability may increase, decrease or remain unchanged in the Gippsland region. Initial maps produced in the LSA were analysed using GIS, and consider temporal and spatial change in the Gippsland region. Change detection maps were produced for each of the commodities studied. The suitability of the Heating Degree Days (OctobereApril; year 2050) criterion for Cool Climate Grapes under the A1F1 extreme emissions scenario is shown in Fig. 7. This criterion is part of the model for Cool Climate Grapes shown in Fig. 8. The Heating Degree Days criterion has a weighting of 0.47 in
A. Hood et al. / Environmental Modelling & Software 21 (2006) 1280e1289
the AHP hierarchy of importance, and the weightings used for the factors climate, landscape/slop and soil were 0.46, 0.20 and 0.34, respectively. Considering the weightings, Heating Degree Days (OctobereApril) is the highest ranked criteria of importance in the Cool Climate Grapes model hierarchy (see Fig. 8). It is no surprise then that the change in Heating Degree Days (OctobereApril; year 2050 compared to 2000) for Cool Climate Grapes under the A1F1 extreme emissions scenario (shown in Fig. 9) displays a similar pattern to the change detection map for climate suitability for Cool Climate Grapes under climate forcing from the same emissions scenario (see Fig. 5).
Acknowledgments The Ecologically Sustainable Agriculture Initiative (EASI), Department of Primary Industries, Victoria and the Victorian Greenhouse Policy Unit, Department of Sustainability and Environment supported the pilot study. OzClim is a stand-alone climate change scenario generator that has been developed by CSIRO Atmospheric Research and the International Global Change Institute at the University of Waikato in New Zealand (http://www.dar.csiro.au/publications/ozclim.htm). References Burrough, P.A., MacMillan, R.A., van Deursen, W., 1992. Fuzzy classification methods for determining land suitability from soil profile observations and topography. Journal of Soil Science 43, 93e210. Conacher, A., Conacher, J., 2000. Environmental Planning and Management in Australia. Oxford University Press, South Melbourne, Victoria, Australia. Ceballos-Silva, A., Lopez-Blanco, J., 2003. Delineation of suitable areas for crops using a multi-criteria evaluation approach and land use/cover mapping: a case study in Central Mexico. Agricultural Systems 77 (2), 117e136. DNRE, 2002. Victorian Greenhouse Strategy. Department of Natural Resources and Environment, Melbourne, 110 pp. !http://www. greenhouse.vic.gov.auO (accessed 1.04.2005).
1289
Eastman, J.R., 1999. In: IDRISI 32: Guide to GIS and Image Processing, vol. 2. Clark Lab, Clark University, Worcestar, MA, USA. Fuhrer, J., 2003. Agroecosystem responses to combination of elevated CO2, ozone and global climate change. Agriculture, Ecosystems and Environment 97 (2003), 1e20. Ghafari, A., Cook, H.F., Lee, H.C., 2000. Integrating climate, soil and crop information: a land suitability study using GIS. In: Fourth International Conference on Integrating GIS and Environmental Modeling (GIS/EM4): Problems, Prospects and Research Needs. Banf, Alberta, Canada. IPCC, 2000. Special Report on Emission Scenarios e a Special Report of Working Group III of the Intergovernmental Panel on Climate Change. In: Nakicenovic, Nebojsa, Swart, Rob (Eds.). Cambridge University Press, 27 pp. !http://www.grida.no/climate/ipcc/ spmpdf/sres-e.pdfO (accessed 1.04.2005). IPCC, 2001. Climate Change 2000: the Science of Climate Change. Summary for Policymakers and Technical Summary of Working Group. Cambridge University Press, !http://www.grida.no/ climate/ipcc_tar/wg1/O (accessed 1.04.2005). Izaurralde, R.C., Rosenburg, N.J., Brown, R.A., Thomson, A.M., 2003. Integrated assessment of Hadley Center (HadCM2) climate-change impacts on agricultural productivity and irrigation water supply in the coterminous United States. Part II. Regional agricultural production in 2030 and 2095. Agricultural and Forest Meteorology 117 (2003), 97e122. Malczewski, J., 1999. GIS and Multicriteria Decision Analysis. John Wiley, New York. NLWRA, 2002. Australians and Natural Resource Management 2002. National Land and Water Resources Audit, Canberra, ACT. Page, C.M., Jones, R.N., 2001. OzClim: the development of a climate scenario generator for Australia. In: Ghassemi, F., McAleer, M., Oxley, L., Scoccimarro, M. (Eds.), Proceedings of the International Congress on Modelling and Simulation. Australian National University, Canberra, ACT, pp. 667e671. Saaty, T.L., 1980. The Analytic Hierarchy Process. McGraw-Hill, New York. Saaty, T.L., 1995. Decision Making for Leaders. RWS Publications, Pittsburgh. Tao, F., Yokozawa, M., Hayashi, Y., Lin, E., 2003. Future climate change, the agricultural water cycle, and agricultural production in China. Agriculture, Ecosystems and Environment 95 (2003), 203e215. Va´squez-Leo´n, M., West, C.T., Finan, T.J., 2003. A comparative assessment of climate vulnerability: agriculture and ranching on both sides of the USeMexico border. Global Environmental Change 13, 159e173. Whetton, P.H., Suppiah, R., McInnes, K.L., Hennessy, K.J., Jones, R.N., 2002. Climate Change in Victoria: High Resolution Regional Assessment of Climate Change Impacts. Victorian Department of Natural Resources and Environment, Melbourne.
Options for Victorian agriculture in a ‘‘new’’ climate: Pilot study linking climate change and land suitability modelling Adam Hood a,*, Bob Cechet b, Hemayet Hossain c, Kathryn Sheffield c a
Land and Catchments Division, Department of Sustainability and Environment, 2/8 Nicholson Street, East Melbourne, Victoria 3002, Australia b Risk Research Group, Geohazards Division, Geosciences Australia, Australia c Strategic Resources Planning Unit, Spatial Sciences, Department of Primary Industries (DPI), Victoria, Australia Received 16 April 2004; received in revised form 5 April 2005; accepted 5 April 2005 Available online 7 November 2005
Abstract This pilot project determines the potential for growing Cool Climate Grapes, high yield pasture and blue gum across Gippsland (Victoria, Australia), and the likely shift in that potential as projected climate change occurs in the region. It provides an indication of the information that can be generated combining Land Suitability Analysis and climate change scenarios. By considering the changing suitability of a few key agricultural enterprises in this region, the project offers an important direction for the agricultural elements of the Victorian Greenhouse Strategy. This pilot project utilises Land Suitability Analysis (LSA) developed by the former Victorian Department of Natural Resources and Environment (DNRE) under the National Land and Water Resources Audit of Australia and further enhanced with Local Government and Catchment Management Authorities across Victoria. This modelling technique allows for an assessment of impacts on a wide range of commodities beyond the historic focus of cropping and pasture models. Following a series of workshops with agronomists, soil scientists, climate scientists, strategic planners and growers, a series of maps/overlays were prepared for Gippsland showing potential suitability of blue gum plantations, Cool Climate Grapes and high yield pasture, as an initial illustration of the methodology. This potential suitability has since been adjusted/modified using a variety of climate change scenarios developed by CSIRO Climate Impacts Group, resulting in an assessment of the change in potential suitability for each enterprise at years 2020 and 2050. The resultant work can be analysed using Geographic Information Systems (GIS) to consider change over time and space in the region. Combined with changes in water availability, this may act as an innovative and powerful decision support tool for policy makers in government and industry. Ó 2005 Elsevier Ltd. All rights reserved. Keywords: Climate change; Expert systems; Modelling climate change; Land Suitability Analysis; Agricultural commodities; Victorian Greenhouse Strategy; Ecologically sustainable agriculture initiative; Climate impact scenarios
Software availability Name of software: OzClim Australian climate change scenario generator, version 2.3 Web page: http://www.dar.csiro.au/publications/ozclim. htm
* Corresponding author. Tel.: C61 3 96379021; fax: C61 3 96378485. E-mail address: [email protected] (A. Hood). 1364-8152/$ - see front matter Ó 2005 Elsevier Ltd. All rights reserved. doi:10.1016/j.envsoft.2005.04.022
Developer: CSIRO Atmospheric Research, PMB 1, Aspendale, Vic 3195, Australia (contact: rogerjones@ csiro.au or [email protected]) Year first available: 2001 Program language: Borland Delphi Operating system: Windows 2000/XP Program size: 40 MB approx (Program, GUI & 0.25 degree horizontal resolution data) Minimum hardware required: 500 MHz Pentium III processor, 128 MB RAM, 20 GB HDD, and CD-ROM
A. Hood et al. / Environmental Modelling & Software 21 (2006) 1280e1289
Hardware used: 1 GHz Pentium III processor, 1 GB RAM Availability and cost: Free (0.25 degree horizontal resolution data). Higher horizontal resolution data (range of geophysical fields available including percentile occurrence for daily and monthly data) available at a cost. Please contact developer (details above). Name of software: Land Suitability Analysis tool for agriculture (includes models for grapes, pasture and blue gum) Web page: Developer: Primary Industries Research, Department of Industries, Government of Victoria, Australia Year first available: 1992 Program language: Developed using Model Builder tools in ArcView GIS package and Macro language in IDRISI32 GIS package. Operating system: Windows 2000 Program size: 174 MB for ArcView and 95 MB for IDRISI32 Minimum hardware required: PC Pentium III, 512 MB RAM, 64 MB Graphics RAM, 60 GB HD Hardware used: Pentium 4 Xeon PC with 1 GB RAM, 128 MB Graphics RAM, 120 GB HD Availability and cost: Availability and Price on request
1. Introduction
1281
of the ‘‘Gippsland Model’’, a broader regional strategic planning model developed under the National Land and Water Resources Audit of Australia (NLWRA, 2002). The LSA is considered useful at this early phase of regional impact assessment for agriculture for several reasons: 1. It enables assessment at regional and strategic levels where data paucity and computing costs often limit more complex dynamic plant, catchment or farm models. 2. It is based on an ‘‘expert systems’’ approach and the associated communication effort delivers a range of peripheral outcomes for the project including raising awareness about climate change. 3. It is highly flexible and can be adjusted depending on the specific question being asked. 4. The tool is simple to manipulate and interrogate, providing a highly transparent modelling framework throughout the process. For this pilot study, LSA uses discrete spatial data sets of soil, landscape and climate attributes. The climate factors introduced into the model are developed using OzClim, a climate change scenario generator developed by the Australian Commonwealth Scientific and Industrial Research Organisation (CSIRO) Division of Atmospheric Research. Fig. 1 outlines the full extent of the study. For the purposes of this paper only the numbered boxes will be discussed in detail, while the dashed sections indicate future directions of the project.
1.1. Background The Greenhouse Strategy of the Australian State of Victoria (DNRE, 2002) highlighted the need for a range of response actions to deal with the goals of raising awareness of greenhouse issues, limiting greenhouse emissions, living in a carbon-constrained society, building capacity within industry and better understanding climate change impacts and the development of adaptation options. This paper examines part of a pilot project that sets out to consider some climate change issues facing agricultural production in parts of Victoria and some options for response. The modelling was seen as a starting point for opening dialogue between state and regional policy makers and specific agricultural industries. The project considers long-term averages and suitability of production. Full documentation of these projects and the methodologies used are available from the corresponding author. 1.2. Project design The pilot study utilises a specific Land Suitability Analysis (LSA) tool that was created as an integral part
2. Climate scenario tools Climate scenarios are generated using a combination of greenhouse gas emission scenarios and global climate general circulation models (GCMs).
2.1. SRES e emissions scenarios This pilot study employs the latest set of emissions scenarios issued by the Intergovernmental Panel on Climate Change (IPCC) (IPCC, 2000). The Special Report on Emission Scenarios (SRES) produced 40 future emission scenarios for greenhouse gases and sulfate aerosols. These scenarios cover a wide range of the main driving forces of future emissions, from demographic to technological and economic developments. Six of the 40 SRES emissions scenarios (A1B, A1T, A1F1, A2, B1 and B2 representing the families of technologyepopulatione economy futures) are shown in Fig. 2. Further details of these can be found at the SRES website (http://sres. ciesin.org).
1282
A. Hood et al. / Environmental Modelling & Software 21 (2006) 1280e1289
Climate Scenario Tools
Land Suitability Analysis Commodities for Analysis
IPCC Climate Change Evidence
• Rapid Assessment • GVAP Industry vs. Regional • Risk and Opportunity
• Recent Rapid Change - Industrial Society • Link to Greenhouse Gas Emissions • Relationship CO2 - Climate Change
Multi-criteria Evaluation 3.2
3.3
• AHP - IDRISI DSS Module • Weighted Growth Factors • Soil Attributes • Landform • Climate
General Circulation Models
2.1
SRES - Emissions Scenarios
2.2
Expert Systems Modelling • Workshops
OzClim - Climate Scenario Generator • Long-Term Averages
2.3
• Commodity Specific Attributes GIS - Spatial Modelling 3.4
• GIS Coverages to 2100 by Scenario and GCM
• ESRI Model Builder • Refine data coverage
Regional Commodity Suitability Mapping • 2020, 2050 •A1F1, B1 scenarios (Extreme, Moderate)
4.1
Biophysical Impact Analysis Change Detection Maps
4.2
Economic and Social Implications Industry and Regional Adaptation Options Fig. 1. Project design (relevant text sections indicated).
2.2. General circulation models The IPCC (2001) provides estimates of global-average warming based on 35 of the 40 SRES emission scenarios (Fig. 3). These estimates were determined using simple models tuned to match the global performance of
GCMs. The two inner lines represent the variation in results according to emission scenarios averaged across all models, while the two outer lines allow for model to model variation. The GCMs are the best available tools to study possible future climate and include complex equations
6
35 30 25 20
Temperature change (°C)
CO2 emissions (Gt/year)
A1F1 A2 A1B B2 A1T B1
15
IS92a
10 5 0 1980
2000
2020
2040
2060
2080
2100
Year Fig. 2. Six of the SRES emission scenarios (Giga-tonnes per year) for carbon dioxide (CO2), plus the IS92a mid-case scenario used by the IPCC in 1996 (IPCC, 2001).
5
Range due to emissions uncertainty and climate response uncertainty
4 Range due to emissions uncertainty
3 2 1 0 1980
2000
2020
2040
2060
2080
2100
Year Fig. 3. Projections for global-average warming relative to 1990 for an envelope of SRES greenhouse gas and sulfate aerosol emission scenarios (IPCC, 2001).
A. Hood et al. / Environmental Modelling & Software 21 (2006) 1280e1289
to represent processes that govern atmospheric and oceanic circulations, temperature, precipitation etc. Most GCMs satisfactorily simulate observed temperature, precipitation and mean sea-level pressure patterns over south-eastern Australia (Whetton et al., 2002).
2.3. OzClim climate scenario generator OzClim is a PC-based regional climate scenario generator for Australia that simplifies the process of calculating future climate change for application to impact models. It is chiefly used as a research tool for climate sensitivity studies, and it features a graphical user interface and point and click technology for ease of use, fast calculations and visualisation capabilities (Page and Jones, 2001). OzClim also allows comparison of output from a range of climate models and construction of future climate change scenarios for risk assessment. The details of the modelling science driving OzClim can be found at http://www.dar.csiro.au/impacts/ozclimscience. html. CSIRO Atmospheric Research has ‘‘tailored’’ a version of the OzClim scenario generator for the Victorian Government to describe how the Victorian climate may change during the next 100 years. This has been achieved by combining a range of projected global warming as given by IPCC (2001) (lowemiddleehigh climate sensitivity derived from Fig. 3) with projected regional climate changes (Victorian region) obtained from a range of atmosphereeocean general circulation models (AOGCMs) and also high horizontal resolution atmospheric general circulation models (AGCMs) forced by the output of the AOGCMs. OzClim is grid-based, and in this pilot study it employs an array of about 20,000 points (Victorian region), producing a grid spacing of about 5 km over Victoria. Climatic variables such as maximum and minimum temperature, and rainfall were augmented by evaporation, solar radiation and relative humidity. In addition, deciles for daily and monthly maximum and minimum temperature, and rainfall are available and can be modified according to selected emission scenarios. OzClim also has the capacity to ‘plug in’ a range of impact models to explore the sensitivity of impacts to different emissions scenarios and climatic ranges. Impact models for the Victorian region developed for this pilot study included simple water balance (precipitation minus potential evaporation), Heating Degree Days, Branas index (estimation of likely disease pressure on grapes caused by temperature and rain and is calculated by multiplying mean monthly temperature by mean monthly rainfall and summing the totals for the period October through March) and frost frequency. Models of effective rainfall, heat stress and chilling units are currently under development.
1283
3. Land Suitability Analysis 3.1. The land suitability framework The land suitability framework was developed by the Food and Agriculture Organisation of the United Nations (FAO). Unlike land capability assessment, which essentially focuses on the limitations to land use in an area, LSA assesses the suitability of land for a specific use or commodity (Conacher and Conacher, 2000). LSA can be used to assess potential climate change impacts by using spatial data sets and analysing the various elements based on a function of interaction between them and/or through a multi-criteria analysis method. Due to its probabilistic nature, LSA lends itself well to assessing the impacts of climate change as it can provide assessments of changing potential for production. 3.2. Multi-criteria evaluation tools and concepts This pilot study used LSA developed by the Victorian Government under the National Land and Water Resources Audit of Australia (NLWRA). The methodology incorporates a Multiple Criteria Evaluation (MCE) within a Geographic Information Systems (GIS) environment. MCE has been developed to improve spatial decision making when considering multiple objectives and conflicting preferences or priorities (Malczewski, 1999). MCE can be an effective decision-making tool for complex issues, using both qualitative and quantitative information. It combines the information from several criteria to form a single composite index of evaluation, and has been utilised around the world for Land Use Suitability modelling (Burrough et al., 1992; Ghafari et al., 2000; Ceballos-Silva and Lopez-Blanco, 2003). The MCE is implemented using the experts systems modelling approach of Analytical Hierarchy Process (AHP) developed by Saaty (1980). AHP provides a framework that allows hierarchical combination of criteria and incorporates experts participation in the decision-making process. Compared to empirical models based purely on the correlation amongst the data, the experts modelling system incorporates the knowledge of the experts that can address many of the other issues and understandings that may be important to decision making. The experts model is also better suited when access to good data is limited. The AHP orders critical factors into a hierarchy of importance. This improves the representativeness of the mapping because not all factors have an equal weighting of importance. It also allows criteria to trade-off with each other depending on the importance weights assigned to them. Furthermore, AHP can deal with criteria that are interdependent, both from the effect on land and in the interaction between spatial units.
1284
A. Hood et al. / Environmental Modelling & Software 21 (2006) 1280e1289
the quantitative aspects of human thought: the qualitative to define the problem and its hierarchy and the quantitative to express judgements and preferences concisely. The process itself is designed to integrate these dual properties.’’ (Saaty, 1995, p. 18). AHP also provides an effective structure for group decision making by imposing a discipline on the group thought’s processes. The necessity of assigning a numerical value (i.e., weight) to each variable of the problem helps decision makers to maintain cohesive thought patterns and to reach a conclusion. In addition, the consensual nature of group decision making improves the ‘‘consistency of judgements and enhances the reliability of the AHP as a decision-making tool’’ (Saaty, 1995, p. 5). Although a variety of techniques exist for the development of criterion weights, one of the most promising is that of pair-wise comparison in AHP. In the procedure for multi-criteria evaluation using Weighted Linear Combination (WLC), it is necessary that the weights sum to one. In AHP, these weights are calculated by taking the principal eigenvector of a square reciprocal matrix of pair-wise comparison between the criteria. All possible pairs of criteria are compared by experts
Weighted Linear Combination Model
Max
Trade off
Traditional Boolean Model
Min Risk averse (AND)
Risk
Risk-taking (OR)
Fig. 4. Decision strategy space (Eastman, 1999).
The Analytic Hierarchy Process breaks down a complex unstructured situation into its component parts. These parts, or variables, are arranged into a hierarchical order with numerical values assigned to the subjective judgements based on the relative importance of each variable. These judgements are synthesised to determine which variables have the highest priority and should be acted upon to influence the outcome of the situation. In this respect, ‘‘AHP incorporates both the qualitative and
Perennial Ryegrass/White Sub-Clover Pasture
Land Suitability Victoria
Climate Change Scenario A1F1 (2050) Gippsland Region
Suitability Index Restricted 0 (Very low) 1 (Very low) 2 (Low) 3 (Low) 4 (Low) 5 (Moderate)
This map is suitable for strategic planning purposes, rather than specific site investigation. Further detailed site analysis should be carried out prior to site-specific development proceeding.
Copyright © Department of Natural Resources and Environment (DNRE), 2002. Map prepared by the Strategic Resources Planning Unit, Agriculture Victoria, Department of Natural Resources and Environment. This data can not be guaranteed to be without flaw of any kind, and therefore AVS disclaims all liability of error or inappropriate use of the map data. Data source: Climate data - CSIRO-OzClim (Data Scale: 1:250,000); Soils and DEM data - DNRE, 2001/2002 (Data Scale: 1:25,000). Base data source: Corporate Geospatial Data Library, DNRE, 2001 (Data Scale: 1:25,000); Auslig Geodata Topo-250K.
10
0
10
20
30
40
50
Map Scale: 1:1,100,000
60
70
80 Kilometers
Legend
7 (Moderate)
Gippsland CMA and LGA boundary Public Land
8 (High)
State Border
9 (High)
Principal Road
10 (High)
Lake
No Data
River
6 (Moderate)
N
Prepared for: Options for Agriculture in a "New Climate" Edition 3, June 2002, SRP Project V12CCa; Map Number M24CCb Model No. A1F1 2050 land AMG Zone 55 (Transverse Mercator) Projection
Fig. 5. Future land suitability (year 2050) for Cool Climate Grapes under an extreme emissions scenario (A1F1; see Fig. 2).
1285
A. Hood et al. / Environmental Modelling & Software 21 (2006) 1280e1289
Climate Suitability Change 2000 - 2050 Cool Climate Grapes Climate Change Scenario B1 (2000) - A1F (2050)
Victoria
Gippsland Region
Legend Decreased Unchanged Increased Gippsland CMA and LGA boundary Public Land State Border
This map is suitable for strategic planning purposes, rather than specific site investigation. Further detailed site analysis should be carried out prior to site-specific development proceeding.
Copyright © Department of Natural Resources and Environment (DNRE), 2002. Map prepared by the Strategic Resources Planning Unit, Agriculture Victoria, Department of Natural Resources and Environment. This data can not be guaranteed to be without flaw of any kind, and therefore AVS disclaims all liability of error or inappropriate use of the map data. Data source: Climate data - CSIRO-OzClim (Data Scale: 1:250,000); Soils and DEM data - DNRE, 2001/2002 (Data Scale: 1:25,000). Base data source: Corporate Geospatial Data Library, DNRE, 2001 (Data Scale: 1:25,000); Auslig Geodata Topo-250K.
10
0
10
20
30
40
50
Principal Road Lake River
60
Map Scale: 1:1,100,000
70
80 Kilometers
N
Prepared for: Options for Agriculture in a "New Climate" January 29, 2003, SRP Project V12CCa Model No. B1 AMG Zone 55 (Transverse Mercator) Projection
Fig. 6. Change in climate suitability (year 2050 compared to 2000) for Cool Climate Grapes under the A1F1 extreme emissions scenario.
on a nine-point continuous scale for their relative importance in determining the suitability of the stated objective. The ratings are then introduced in a matrix to calculate the weights (Eastman, 1999). The traditional land suitability models use Boolean aggregation method of constructing suitability indexes that present a rigid binary choice of acceptance or rejection. The Weighted Linear Combination (WLC) method in MCE introduces a soft or ‘‘fuzzy’’ concept of suitability in standardising criteria. Instead of hard Boolean decision of assigning absolute suitability or unsuitability to a location for a given criteria, it is scaled to a particular common range where suitable and unsuitable areas are continuous measures. The aggregation method uses weighted linear combination, which retains the variability of continuous criteria and allows criteria to trade off with each other (Eastman, 1999). A low suitability defined by one criterion may be compensated by a high suitability score in another criterion. Trade-off between criteria will depend on weights assigned to them. As depicted in the decision
strategy space (see Fig. 4), this model is a significant improvement over the Boolean approach by avoiding its extreme risk aversion (binary rejection) and extreme risk taking (binary acceptance) nature. 3.3. Expert systems modelling e workshops To develop the LSA for this pilot study, each commodity was analysed using a series of expert systems model workshops involving agronomists, growers and soil scientists with expertise in the Gippsland region. The workshops had several objectives: To build awareness of current climate change science. To define the goal of the modelling exercise in the context of the project. To develop a model using Analytical Hierarchy Process e multi-criteria evaluation. To verify and fine-tune outputs following GIS modelling.
1286
A. Hood et al. / Environmental Modelling & Software 21 (2006) 1280e1289
Cool Climate Grape Heat Degree Days (Oct-Apr) Climate Change Scenario A1F1 (2050)
Victoria
Gippsland Region
Suitability Index Restricted 0 (Very low) 1 (Very low) 2 (Low) 3 (Low) 4 (Low) 5 (Moderate) 6 (Moderate) 7 (Moderate) This map is suitable for strategic planning purposes, rather than specific site investigation. Further detailed site analysis should be carried out prior to site-specific development proceeding.
8 (High)
10 (High)
Copyright © Department of Natural Resources and Environment (DNRE), 2002. Map prepared by the Strategic Resources Planning Unit, Agriculture Victoria, Department of Natural Resources and Environment. This data can not be guaranteed to be without flaw of any kind, and therefore AVS disclaims all liability of error or inappropriate use of the map data. Data source: Climate data - CSIRO-OzClim (Data Scale: 1:250,000); Soils and DEM data - DNRE, 2001/2002 (Data Scale: 1:25,000). Base data source: Corporate Geospatial Data Library, DNRE, 2001 (Data Scale: 1:25,000); Auslig Geodata Topo-250K.
10
0
10
20
30
40
50
Map Scale: 1:1,100,000
60
70
80 Kilometers
N
Gippsland CMA and LGA boundary Public Land State Border Principal Road
9 (High)
No Data
Legend
Lake River
Prepared for: Options for Agriculture in a "New Climate" Edition 3, June 2002, SRP Project V12CCa; Map Number M24CCb Model No. A1F1 2050 AMG Zone 55 (Transverse Mercator) Projection
Fig. 7. Heating Degree Days (OctobereApril; year 2050) for Cool Climate Grapes under the A1F1 extreme emissions scenario (A1F1; see Fig. 2).
The final model was developed in the form of a weighted hierarchy, as illustrated in Fig. 4.
3.4. Geographic Information Systems e Spatial Modelling In order to use the scenarios generated by OzClim as an input into the multi-criteria land suitability model; a suitable link between the two models was necessary. For the purposes of this pilot study, macros have been developed that take the scenarios generated by OzClim and convert them to a format readable by the land suitability models that have been developed using the ArcView Model Builder. The process involved: 1. Generation of all required scenarios using OzClim. 2. Exporting these scenarios in 16 bit IDRISI format. 3. Application of the interface macros, developed using IDRISI, to convert these IDRISI files to ArcView 32 bit Binary Raster format.
4. Importing the Binary Rasters into ArcView and converting them to ArcView grid layers. 5. Including these grid layers as inputs to the land suitability models. The selected OzClim climate change GCM was the CSIRO MK2 utilising both the B1 and A1F1 greenhouse gas emission scenarios (see Fig. 2). CSIRO MK2 B1 climate scenarios were created for the years 2020 and 2050. MK2 A1F1 scenario was only employed for the year 2050. The base climate (year 2000) was compiled from the average of the observational record between 1961 and 2000. 3.5. Comparison e other assessments of climate change impacts on agriculture Many modelling tools are appropriate for considering climate change impacts on agriculture and most variations in the approach to climate change occur around the choice of climate change scenario and the
1287
A. Hood et al. / Environmental Modelling & Software 21 (2006) 1280e1289
Cool Climate Grapes Climate Change
Climate 0.46
Soil 0.34
Slope 0.20 Degree 0.75
Aspect 0.25
Heat Degree Days (Oct-Apr) 0.47
Flower rainfall (Nov-Dec) 0.19
Rain Ripening (Mar-Apr) 0.09
Spring Frost (Sep-Nov) 0.20
Branas Index (Annual) 0.05
1350-1500 rating 1
<85mm Nov and Dec rating 1
<85mm Mar and Apr rating 1
<0-5 days rating 1
<4000 rating 1
1500-1650 rating 0.8
<85mm Nov >85mm Dec or alternate rating 0.8
<85m Mar& >85mm Apr or alternate rating 0.8
<5-10 days rating 0.8
4000-5000 rating 0.9
>85mm Nov and Dec rating 0.3
>85mm Mar and Apr rating 0.1
10-15 days rating 0.5
5000-6500 rating 0.7
900-1100 rating 0.1
15-25 days rating 0.2
6500-8000 rating 0.4
<900 rating 0
>25 days rating 0
>8000 rating 0.2
>1650 and 1100-1350 rating 0.4
Drainage 0.101
pH 0.06
Sodicity 0.19
Topsoil 0.75 Subsoil 0.25
Usable Depth 0.46
Texture 0.04
EC 0.14
Topsoil 0.75
Topsoil 0.75
Topsoil 0.75
Subsoil 0.25
Subsoil 0.25
Subsoil 0.25
Fig. 8. Example of the Analytical Hierarchy Process (AHP) hierarchy showing the criteria for growth in the model used for Cool Climate Grapes (numbers in brackets represent criteria weighting).
GCM used. Other significant differences are related to the spatial (varying scales) and non-spatial aspects of the agricultural model. Tao et al. (2003) map the implications for agricultural water cycle and agricultural production for the whole of China using crop specific water balance equations built from United Nations Food and Agriculture Organisation (UNFAO). This work finds that the key uncertainties lie in regionalisation of GCMs. Other studies have been conducted considering specific elements such as the impact of elevated CO2 or the changes in water supply (Fuhrer, 2003; Izaurralde et al., 2003). Interesting work has considered the spatial variations of vulnerability to climate change and variability across different socio-economic systems. The most poignant is a study of vulnerability of ranching on both side of US-Mexico border (Va´squez-Leo´n et al., 2003). All of these studies are dependent on
a modelling system that provides some insight to climate change over the long term and across space. The inputs to these assessments are complete with a range of substantial uncertainties. This study finds similar limiting issues with uncertainty. As a result there is a need to build uncertainty into a risk/vulnerability framework to provide a proper context in which these modelling and simulation exercises can operate.
4. Outputs of the pilot project 4.1. Regional commodity suitability mapping Three commodities were selected for analysis based on the commodities currently produced, new commodities that could be introduced into the region according
1288
A. Hood et al. / Environmental Modelling & Software 21 (2006) 1280e1289
Heat Degree Days Change 2000 - 2050 Climate Change Scenario B1 (2000) - A1F (2050)
Victoria
Gippsland Region
Legend Decreased Unchanged Increased Gippsland CMA and LGA boundary Public Land State Border Principal Road
This map is suitable for strategic planning purposes, rather than specific site investigation. Further detailed site analysis should be carried out prior to site-specific development proceeding.
Lake River
Copyright © Department of Natural Resources and Environment (DNRE), 2002. Map prepared by the Strategic Resources Planning Unit, Agriculture Victoria, Department of Natural Resources and Environment. This data can not be guaranteed to be without flaw of any kind, and therefore AVS disclaims all liability of error or inappropriate use of the map data. Data source: Climate data - CSIRO-OzClim (Data Scale: 1:250,000); Soils and DEM data - DNRE, 2001/2002 (Data Scale: 1:25,000). Base data source: Corporate Geospatial Data Library, DNRE, 2001 (Data Scale: 1:25,000); Auslig Geodata Topo-250K.
10
0
10
20
30
40
50
Map Scale: 1:1,100,000
60
70
80 Kilometers
N
Prepared for: Options for Agriculture in a "New Climate" Edition 3, June 2002, SRP Project V12CCa; Map Number M70CCa Model No. A1F1 2050 AMG Zone 55 (Transverse Mercator) Projection
Fig. 9. Change in Heating Degree Days (OctobereApril; year 2050 compared to 2000) for Cool Climate Grapes under the A1F1 extreme emissions scenario.
to potential market demand and advice from agronomists and other relevant experts. The three commodities selected were Cool Climate Grapes, high yield pasture and blue gum plantations. Initially, land suitability with base climate data (potential for the year 2000) was assessed and a series of maps prepared. Factors considered in this assessment were climate, landscape and soil. A panel of experts in a series of workshops assessed the accuracy of these maps. The base climate (year 2000) maps were verified at this stage. Land suitability outcomes were ranked in 12 categories covering high, moderate, low and very low suitability for commercial production and restrictive areas, which were deemed unsuitable for commercial production of the commodity. The process was repeated using LSA and a range of potentially ‘‘new climates’’ utilising the OzClim climate change scenarios, with a series of maps produced for the years 2020 and 2050. An example of these maps is shown in Fig. 5 depicting the future land suitability (year 2050) for Cool Climate Grapes under the extreme AIF1 emissions scenario (see Fig. 2).
4.2. Biophysical impact analysis change detection maps A further step in this pilot study was to develop change detection maps, using the land suitability maps created for the years 2000, 2020 and 2050. Fig. 6 shows the change in climate suitability (year 2050 compared to 2000) for Cool Climate Grapes under the A1F1 extreme emission scenario. These maps further enhance the information portrayed in the land suitability maps and clearly indicate areas where land suitability may increase, decrease or remain unchanged in the Gippsland region. Initial maps produced in the LSA were analysed using GIS, and consider temporal and spatial change in the Gippsland region. Change detection maps were produced for each of the commodities studied. The suitability of the Heating Degree Days (OctobereApril; year 2050) criterion for Cool Climate Grapes under the A1F1 extreme emissions scenario is shown in Fig. 7. This criterion is part of the model for Cool Climate Grapes shown in Fig. 8. The Heating Degree Days criterion has a weighting of 0.47 in
A. Hood et al. / Environmental Modelling & Software 21 (2006) 1280e1289
the AHP hierarchy of importance, and the weightings used for the factors climate, landscape/slop and soil were 0.46, 0.20 and 0.34, respectively. Considering the weightings, Heating Degree Days (OctobereApril) is the highest ranked criteria of importance in the Cool Climate Grapes model hierarchy (see Fig. 8). It is no surprise then that the change in Heating Degree Days (OctobereApril; year 2050 compared to 2000) for Cool Climate Grapes under the A1F1 extreme emissions scenario (shown in Fig. 9) displays a similar pattern to the change detection map for climate suitability for Cool Climate Grapes under climate forcing from the same emissions scenario (see Fig. 5).
Acknowledgments The Ecologically Sustainable Agriculture Initiative (EASI), Department of Primary Industries, Victoria and the Victorian Greenhouse Policy Unit, Department of Sustainability and Environment supported the pilot study. OzClim is a stand-alone climate change scenario generator that has been developed by CSIRO Atmospheric Research and the International Global Change Institute at the University of Waikato in New Zealand (http://www.dar.csiro.au/publications/ozclim.htm). References Burrough, P.A., MacMillan, R.A., van Deursen, W., 1992. Fuzzy classification methods for determining land suitability from soil profile observations and topography. Journal of Soil Science 43, 93e210. Conacher, A., Conacher, J., 2000. Environmental Planning and Management in Australia. Oxford University Press, South Melbourne, Victoria, Australia. Ceballos-Silva, A., Lopez-Blanco, J., 2003. Delineation of suitable areas for crops using a multi-criteria evaluation approach and land use/cover mapping: a case study in Central Mexico. Agricultural Systems 77 (2), 117e136. DNRE, 2002. Victorian Greenhouse Strategy. Department of Natural Resources and Environment, Melbourne, 110 pp. !http://www. greenhouse.vic.gov.auO (accessed 1.04.2005).
1289
Eastman, J.R., 1999. In: IDRISI 32: Guide to GIS and Image Processing, vol. 2. Clark Lab, Clark University, Worcestar, MA, USA. Fuhrer, J., 2003. Agroecosystem responses to combination of elevated CO2, ozone and global climate change. Agriculture, Ecosystems and Environment 97 (2003), 1e20. Ghafari, A., Cook, H.F., Lee, H.C., 2000. Integrating climate, soil and crop information: a land suitability study using GIS. In: Fourth International Conference on Integrating GIS and Environmental Modeling (GIS/EM4): Problems, Prospects and Research Needs. Banf, Alberta, Canada. IPCC, 2000. Special Report on Emission Scenarios e a Special Report of Working Group III of the Intergovernmental Panel on Climate Change. In: Nakicenovic, Nebojsa, Swart, Rob (Eds.). Cambridge University Press, 27 pp. !http://www.grida.no/climate/ipcc/ spmpdf/sres-e.pdfO (accessed 1.04.2005). IPCC, 2001. Climate Change 2000: the Science of Climate Change. Summary for Policymakers and Technical Summary of Working Group. Cambridge University Press, !http://www.grida.no/ climate/ipcc_tar/wg1/O (accessed 1.04.2005). Izaurralde, R.C., Rosenburg, N.J., Brown, R.A., Thomson, A.M., 2003. Integrated assessment of Hadley Center (HadCM2) climate-change impacts on agricultural productivity and irrigation water supply in the coterminous United States. Part II. Regional agricultural production in 2030 and 2095. Agricultural and Forest Meteorology 117 (2003), 97e122. Malczewski, J., 1999. GIS and Multicriteria Decision Analysis. John Wiley, New York. NLWRA, 2002. Australians and Natural Resource Management 2002. National Land and Water Resources Audit, Canberra, ACT. Page, C.M., Jones, R.N., 2001. OzClim: the development of a climate scenario generator for Australia. In: Ghassemi, F., McAleer, M., Oxley, L., Scoccimarro, M. (Eds.), Proceedings of the International Congress on Modelling and Simulation. Australian National University, Canberra, ACT, pp. 667e671. Saaty, T.L., 1980. The Analytic Hierarchy Process. McGraw-Hill, New York. Saaty, T.L., 1995. Decision Making for Leaders. RWS Publications, Pittsburgh. Tao, F., Yokozawa, M., Hayashi, Y., Lin, E., 2003. Future climate change, the agricultural water cycle, and agricultural production in China. Agriculture, Ecosystems and Environment 95 (2003), 203e215. Va´squez-Leo´n, M., West, C.T., Finan, T.J., 2003. A comparative assessment of climate vulnerability: agriculture and ranching on both sides of the USeMexico border. Global Environmental Change 13, 159e173. Whetton, P.H., Suppiah, R., McInnes, K.L., Hennessy, K.J., Jones, R.N., 2002. Climate Change in Victoria: High Resolution Regional Assessment of Climate Change Impacts. Victorian Department of Natural Resources and Environment, Melbourne.