• 
• 
• 

Integrated Whole-Farm Modelling — An Application for Policy Analysis of Climate Change Adaptation

Research Papers 2012.6

Paper presented at the 2012 annual conference of the Australian Agricultural and Research Economics Society

7–10 February, Fremantle, Western Australia

March 2012

Authors

Rukman Wimalasuriya
Senior Quantitative Economist
Economics and Policy Research Branch
Department of Primary Industries, Victoria

Chris Chan
Principal Scientist, Economic and Policy Modelling
Economics and Policy Research Branch
Department of Primary Industries, Victoria

Oscar Cacho
Professor, Agricultural and Resource Economics
School of Business, Economics and Public Policy
University of New England, Armidale, New South Wales

Acknowledgments

The authors would like to thank Dr Craig Beverly of the Victorian Department of Primary Industries and Dr Jason Crean of NSW Department of Trade and Investment, for providing helpful comments on a draft of this paper.

Acknowledgement also goes to Dr Richard Eckard, Dr Brendan Cullen and Dr Kithsiri Dassanayake of the School of Land and Environment, University of Melbourne, as well as to Dr Garry O’Leary, Esther Liu and Prof. Bill Malcolm of the Victorian Department of Primary Industries, who all rendered valuable assistance to the research underpinning this paper.

Special thanks to Brendan Christy of the Victorian Department of Primary Industries for providing GIS data used in the model.
The authors gratefully appreciate the encouragement and support rendered by Dr Deborah Peterson, Ms Deirdre Rose and Mr Gavan Dwyer of the Policy and Strategy Group within the Victorian Department of Primary Industries.

Any errors remaining in the paper are the authors’ responsibility. Moreover, the views expressed in this paper are those of the authors and do not necessarily reflect the views of their affiliated organisations.

Abstract

This paper describes an integrated modelling approach suitable for analysing a wide range of policy, economic and environmental issues where spatial heterogeneity is important. The approach involves overlaying whole-farm models onto GIS map layers of land use, soil, climate and topographic information. The integration is implemented through a Matlab^® platform connected to a suite of Excel^® based farm models. This approach enables flexible, efficient data processing and scenario analysis. The integrated modelling platform can be used to assess Victoria-wide and regional farm-level impacts of climate and weather changes, policies, market developments and new technologies. It supports comparative assessments of the ‘before’ and ‘after’ scenarios, but not the identification of farm adjustment paths over time. To demonstrate its functionality, we present an application that evaluates possible consequences of technological adaptation under a common climate change scenario in affecting farming systems and land-use patterns in Victoria.

Keywords: Whole-farm modelling, spatial land use analysis, GIS, mathematical optimisation, climate change

Contents

Acknowledgement

Abstract

1. Introduction

2. Whole-farm models of farming systems in Victoria

• 2.1 Cropping component
• 2.2 Livestock component

3. Integrating EPRB whole-farm models

• 3.1 Integrated model structure
• 3.2 Solution process

4. An illustrative application

• 4.1 Impacts of climate change on regional agriculture
• 4.2 Potential benefits of adaptation options
• 4.3 Results and discussion

5. Conclusions

References

Appendix 1

Appendix 2

1. Introduction

The Victorian Department of Primary Industries (DPI), through its Economics and Policy Research Branch (EPRB), maintains a suite of whole-farm models representing major farming systems in the State. The models have been developed by different staff at different times over the past two decades. As such, they do not have a uniform structure and are based on separate databases. For further background on the development of approaches to farm modelling, see Appendix 1.

This paper reports a methodology to developing an integrated whole-farm modelling platform that can manage spatial heterogeneity. GIS layers are used to link whole-farm models to the regions that the models represent. This type of integration is helpful in analysing policy issues that affect farming practices across the State, such as climate change adaptation and land use change. The paper also includes an application of the integrated modelling platform to test its functionality, on a climate change research issue.

The rest of the paper is structured as follows. Section 2 provides a general description of the whole-farm models maintained by the DPI. The methodology adopted to integrate them is explained in section 3. An application of the integrated framework to a climate change scenario is reported in section 4. Section 5 concludes the paper with a discussion of model limitations and proposals for further model development. History and different approaches to modelling farm systems, and a complete inventory of the regional whole-farm models are included in Appendices.

2. Whole-farm models of farming systems in Victoria

Dryland dairy and broadacre (grain, sheep and beef) farming accounts for roughly 50 per cent of the total annual output value of agriculture in Victoria (ABS, 2006). Reflecting the economic significance of these activities, the EPRB whole-farm models are primarily concerned with broadacre and dairy farming in different dryland regions. Table 1 presents the list of whole-farm models maintained by the EPRB. Two types of cropping model, PRISM and EMAR, exist for capturing regional differences in crop farming. CAB model, is used to represent livestock production. An inventory of these models, including their full names, regions represented and other details, is given in Appendix 2.

EPRB also maintains a separate model, namely the Water Policy Model (see Eigenraam et al. 2003), to represent agricultural activities and water use across irrigation districts of the Northern Victorian Irrigation Region. This model has not been included in the integrated modelling platform.

Whole-farm model solutions are based on the maximisation of whole-farm gross margin, generating estimates of the optimal activity mix in terms of crop rotation, livestock stocking rate and feed mix. Optimal activity mix can be estimated for specific rainfall, price and technology conditions. These models are applicable for evaluating farming practice changes associated with technology adoption, changing crop and pasture cultivars, management practice improvement and policy change.

Data for the models come from various sources. Prices of farm inputs and outputs are sourced from the Australian Bureau of Agricultural and Resource Economics and Sciences (ABARES) and DPI. All commodity price data are averaged over five-years. Yield of crops, pasture and livestock products, and rates of farm input use are calibrated using local industry intelligence from DPI researchers and extension staff. Yield and input use estimates are region-specific and are assumed to remain stable over time.

The linear programming (LP) framework for the EPRB whole-farm models comprises an objective function for maximising whole-farm gross margin and a set of constraints on farming activity and resource availability such as land areas for particular farm systems. An exogenous change to the farming system is effected through ‘shocking’ specific model parameters. The impacts of a system change are measured by comparing model variables between the model runs with and without the parameter shock.

Table 1: EPRB whole farm models

Name of the model	Farming system represented	Region represented
PRISM–NC	Mixed grain/sheep	North-central
PRISM–NE	Mixed grain/sheep	North East
EMAR–Mallee	Mixed grain/sheep	Mallee
EMAR–Wimmera	Grains	Wimmera
Dairy CAB–Gippsland	Dairy cattle	Gippsland
Dairy CAB–SW	Dairy cattle	South West
Beef CAB–Gippsland	Beef cattle	Gippsland
Beef CAB–SW	Beef cattle	South West
Beef CAB–NE/GV	Beef cattle	North East/Goulburn Valley
Sheep CAB–Wimmera	Sheep	Wimmera
Sheep CAB–SW	Sheep	South West
Sheep CAB–NE/GV	Sheep	North East/Goulburn Valley

All the EPRB whole-farm models are coded in Excel® spreadsheets, with the LP matrices optimised using the add-in software ‘What’s Best®’. Model validation is based on comparing model estimates with actual farm data on crop rotation practices, crop yields, stocking rates and crop areas. Below is a brief account of the cropping and livestock components in these models.

2.1 Cropping component

For broadacre enterprises, the crop and crop–pasture rotation sequences are based on a combination of annual pasture, cereal crops (wheat, barley, oats and triticale), pulse crops (lupins, field peas, chickpeas and lentils) and oilseed crops (canola). The set of crops included for rotation varies across regions.

The key to determining optimal crop rotation is the impact on crop yields. For specific crops, the yield rate depends on the paddock history of the two previous years. In addition, the crop yield is dependent on the history of weed, disease, nitrogen and moisture status of soil. The dependence of yield on rainfall is modelled following the ‘water use efficiency’ (WUE) approach suggested by French and Schultz (1984) and Fischer (1979).

For the annual models to represent inter-seasonal rotation, it is assumed that all crops of the rotation are simultaneously grown on certain proportions of the farm land area and the crop on each proportion of the land is rotated from year to year. This modelling setup is similar to that of MIDAS (Kingwell and Pannell 1987).

2.2 Livestock component

Sheep enterprises are self-replacing Merino ewes, Dorset over Merino ewes (or first-cross lamb), first-cross ewes (or second-cross lamb) and Merino wethers. Beef cattle enterprises are beef vealer production, store weaner production, store steer production, yearling steer production, bullock production and a backgrounding operation. Dairy cattle production is operated by winter and autumn calving enterprises with the ability to choose seasonal or split calving in any month.

Modelling livestock production is based on monthly feed budgets. Optimal stocking rates are determined under the criterion of least-cost feed mix. Under a demand-driven approach, feed requirements per animal are fixed at levels dependent on the production per animal and the monthly body weight pattern.

The energy requirements of animals are calculated in relation to the monthly flock/herd structure, body weights, weight changes and physiological status of animals. Energy demand is assumed to be first met by metabolisable energy available from the pasture, either within the month or after ‘carrying-over’ to the following month while allowing for certain decline in pasture quantity and quality. For the summer period when there is no active pasture growth, energy available from stubble is used to feed livestock in the mixed farming systems. Hay and feed grain may also be used to fill any feed gaps as necessary.

During this period, the feed intake of sheep is restricted to their dry matter intake capacity to account for low energy concentrations in hay and stubble.

Subject to matching the energy demand patterns of the sheep enterprises with the energy suppliy from pasture and stubble on a monthly basis, modelling solutions determine the most profitable animal enterprise and the number of animals that the farm system can carry.

Potential annual dry matter production from pasture is estimated as a function of the growing season rainfall (French 1992). This potential production is further adjusted to reflect local system conditions to arrive at region-specific pasture production levels. Estimates on percentage monthly distributions of pasture production in the region and monthly energy contents are used to derive the monthly availability of dry matter and energy. Estimates on average pasture utilisation are used to convert the amount of energy and nutrients from pasture into animal intake.

3. Integrating EPRB whole-farm models

We integrated all 12 EPRB regional whole-farm models into a Victoria-wide framework using the Matlab® software (The MathWorks 2000), and linked this framework to Geographic Information System (GIS) map layers. The integration structure and the solution process are as follows.

3.1 Integrated model structure

The design of model integration seeks to preserve transparency of the whole-farm models. To this end, we keep the original Excel files containing the LP matrices and all the model parameters (Figure 1a), and create a Matlab interface to read these matrices and perform model optimisation.

This integration procedure is largely automated, requiring minimal intervention by the user. Yet it provides the clarity and transparency necessary to facilitate communication between modellers and regional staff when updating data, developing scenarios, calibrating shocks and interpreting results.

Under the integrated framework, the LP matrix of each whole-farm model remains to be solved independently. For each region, the resultant optimised whole-farm gross margins are aggregated using the relevant GIS raster map layers (Figure 1b). This aggregation procedure can then produce a farm profit map of all agricultural regions in Victoria (Figure 1c).

The mapping of farm system changes to Victoria-wide production impacts is the most significant improvement to the modelling procedure, which greatly enhances the capability of whole-farm modelling to undertake policy analysis in a spatial context and to visualise modelling results for improved communication with stakeholders. The GIS maps can be used to provide detailed modelling results at sub-region level, capturing spatial variability within and across regions.

Figure 1: Illustration of the integrated model: a set of whole-farm models on spreadsheets allows communication with scientists and the creation of scenarios and shocks (a); farm (LP) models are combined with map layers (b); and solution of the LP models produces farm profit maps (c).

3.2 Solution process

The solution process is centred on the Matlab code, which is used to (i) read the existing GIS files; (ii) solve the LP matrices; and (iii) assign results to individual regions (Figure 2).

Specifically, the GIS raster map layers are read into Matlab as matrices and saved in a binary ‘*.mat’ file. At the moment, only the map layers of land use and regional boundaries are stored and used. Other map layers such as rainfall, soil type or topography can be added in the future.

An Excel file is created to contain model names that are read into the Matlab workspace. Any subset of the 12 whole-farm models can be integrated in this way, with the name list directing Matlab to read the relevant LP matrices, vector of constraints and vector of objective function coefficients from the corresponding whole-farm models.

Read list of excel file names to read from

↓

Load map files

↓

Read LP matrices of whole-farm models from excel

↓

Solve LP models in Matlab

↓

Assign results to regions, on maps

↓

Calculate total profit for regions

↓

Save LP matrix data, solutions and profits

Figure 2: Representation of the solution process

When solving the LP models, the relevant ‘*.mat’ files are loaded. Results from individual models are then aggregated by region, or by sector, or across the whole state. For each farming system within a particular region, the relative contribution (i.e. the weighting factor for aggregation) is derived using crop and pasture area statistics, livestock numbers and production value estimates published by the ABS (2006) — the most recent agricultural census of which small area data have been released. For each region, the total gross margin is calculated as a product of the per hectare gross margin for a specific farming system, the number of map grids in the region, and the standard grid size (i.e. area in hectares).

The final step is to save the LP data, LP solutions and profit maps into a binary ‘*.mat’ file. Accordingly, a single file is created to store all the relevant modelling information for each scenario run. Between different scenarios runs, the binary files will have exactly the same structure — which facilitates the extraction, manipulation and batch processing of the stored information in these binary files.

Model shocks can be designed in numerous flexible ways. One way is to adjust activity budgets (also called ‘objective values’) to represent cost and price changes. Another way is to adjust resource constraints to represent regulatory impacts on land use or availability of other natural resources. Yet another way is to adjust matrix coefficients to represent technological and environmental impacts on farm system operations. It is also possible to adjust model parameters representing biophysical aspects of the farming system, such as when assessing the impact of introducing a high-yielding crop variety.

4. An illustrative application

The integrated platform is applied to climate change adaptation. This was done in two steps. First, we estimated the impact of climate change on dryland farming systems in Victoria. This was to represent a scenario where no adaptation response occurs. Second, we assessed the benefit of adapted farming systems in a scenario where farmers take up certain adaptation options recommended by DPI scientists. This was to represent the outcome of DPI research and extension services promoting technological opportunities to mitigate possible adverse impacts of climate change on Victoria’s agriculture

4.1 Impacts of climate change on regional agriculture

Some scientists projected that the temperature in regional Victoria would likely increase as a result of climate change (Soste et al. 2011). Moreover, rainfall was projected to decrease and the seasonality of rainfall and temperature to shift in some areas of the state. If such projections materialise, the temperature rise and rainfall reduction could hamper crop growth and yield.

On the other hand, the elevation of atmospheric carbon dioxide (CO2) concentration along with a temperature rise could stimulate photosynthesis and improve nutrient and water use efficiency in growing crops and pasture (Tostovrsnik et al. 2010). Accordingly, climate change could bring positive effects on vegetative growth that offset the negative impacts of lower rainfall in the crop growing season. However, if the climate is to become increasingly dry and hot, the net efficiency of plant growth would eventually reach a threshold and decline thereafter (O’Leary et al. 2011).

A number of studies produced scientific evidence about the potential impacts of climate change on plant growth. According to the study by Sheehy et al. 2005, rice yields would increase by 0.5 t/ha (ton per hectare) for every 75 ppm (parts per million) increase in CO2 concentration, and decrease by 0.6 t/ha for every one Celsius degree increase in temperature. The study by Howden et al. (2008) suggested that, in Australia, a 10 per cent rainfall reduction would cancel out the CO2-induced fertilisation benefits in livestock regions subject to modest net drying over time.

We primarily used the Intergovernmental Panel on Climate Change (IPCC 2001) ‘2050 High’ emissions (also called ‘A1Fi’) scenario for modelling. However, compatible regional climate impact estimates were not all available for this scenario. Therefore, as suggested by Cullen (2011, pers. com.), we also drew on the ‘2070 Medium’ scenario (also called ‘A1B’) when sourcing data on climate impacts for a few regions under study.

Estimates of the potential impacts associated with these climate change scenarios on crops and pasture in each region were collected from various sources as listed in Table 2. These estimates were mostly based on simulation modelling. While they show differential regional impacts, they do not capture climate variability within and between seasons. Due to this data limitation, the present study did not consider seasonality shift and the associated potential effect of lowering the quality of grain and livestock products.

Table 2: Climate change impacts on crop and pasture yields in Victorian regions

Region	Grain crops	Sheep/Beef pasture	Dairy pasture
Mallee	-20% (GO)	-20% (Assumed)	(not applicable)
Wimmera	+10% (GO)	-6% (BC1)	(not applicable)
North-central	+10% (GO)	-6% (BC1)	(not applicable)
North East	+10% (GO)	-6% (BC1)	(not applicable)
South West	(not applicable)	+4% (BC2)	-1% (BC2)
Gippsland	(not applicable)	-5% (Assumed)	-3% (BC3)

(a) (GO): O’Leary et al. (2011)

(b) (BC1): Cullen, 2011, pers. Com., unpublished data

(c) (BC2): Cullen et al. (2008)

(d) (BC3): Cullen et al. (2009)

Specifically, yield of grain crops in the Mallee region was projected to decline while grain yield in other regions was projected to increase. Pasture yield was projected to decline in all regions except for sheep/beef pasture in South West Victoria.

Using these data, the integrated whole-farm models were solved for scenarios with and without the projected changes in crop and pasture yields. The ‘without changes’ scenario represents the base case reflecting current climate conditions and yields. The ‘with changes’ scenario reflects the incidence of climate change. The impact of climate change was then derived as the difference between those two scenarios. The impact of climate change adaptation was derived as described below.

4.2 Potential benefits of adaptation options

An objective of our whole-farm modelling is to build up the capability to, among other things, identify and evaluate adaptation options that can help mitigate the potential adverse impacts of climate change. This capability is essential for informing policies in relation to research and development and structural adjustment. For example, Scott et al. (2011) used APSIM and enterprise budgeting to assess different adaptation options in sorghum systems in NSW.

A number of scientists had suggested adaptation strategies for Victorian agriculture over different timeframes — namely: develop and adopt new crop varieties and alter management practices in the medium term; and shift some farming systems towards south in the longer term (Howden et al. 2007; Howden and Stokes, 2010; O’Leary et al. 2011; Tostovrsnik et al. 2010). The medium-term strategies focus on achieving ‘incremental’ changes to farming systems to bring about adaptation benefits. However, these benefits might not be sustainable if climate change becomes more extreme. Consequently, the long-term strategies could be necessary to achieve ‘transformational’ adaptations through land use and system changes (Howden et al. 2010).

While the above adaptation strategies aim to reduce losses in crop, pasture and livestock production, there are also strategies being investigated for identifying and capitalising on opportunities to actually gain from climate change. This type of research remains at an early stage, but has already generated innovative ideas such as adapting farming systems to changed seasonality through the introduction of summer crops into winter-grain cropping systems in northern Victoria (Dassanayake 2011).

For our study, the modelling was concerned with medium-term adaptation options — specifically, the introduction of new crop varieties. Other modelling capabilities, including the incorporation of transformational adaptation options, can be developed using the integrated modelling framework in the future.

We modelled the outcomes of adopting new varieties of grain crops that are tolerant to the changed climate. This adaptation is relevant only to Mallee where grain crop yields were projected to be lower by 20% under the assumed climate change scenario. Grain crop yields in the other regions were projected to increase under climate change, hence not requiring intervention. As such, the modelling of climate change adaptation was confined to the Mallee region.

Two adaptation scenarios were considered: namely, (Adaptation 1) new varieties of grain crops are developed to halve the projected yield loss under climate change (Table 2); and (Adaptation 2) new crop varieties are developed to prevent the loss totally. It was assumed that the introduction of new crop varieties does not require additional farm inputs.

4.3 Results and discussion

A main outcome of our research is the computationally efficient methodology for integrating EPRB whole-farm LP models into a flexible framework that can be applied at Victoria-wide spatial scale. The case study of climate change adaptation options confirmed a particular strength of the integrated modelling approach — that is, enabling the use of GIS layers to aggregate modelling results by different sectoral and geographical scales.

The impacts of climate change were found to vary between farm systems and between regions (Table 3). Modelling results suggest that climate change could lead to increased profitability of mixed farming systems in north east (by 14%) and north central Victoria (by 9%), grain farming systems in Wimmera (by 16%), and beef cattle and sheep farming systems in south west Victoria (both by 5%). For the other regional farming systems, however, climate change could reduce profitability. Potential decline in profitability ranges from 1% for dairy farming systems in the south west to 26% for mixed farming systems in Mallee.

Using the land area utilised by each regional farm type as the gross-up factor, we converted the farm-level results into regional aggregates (Figure 3). Without adaptation, the impact of climate change on dryland farming in Victoria was estimated to be a reduction in total farm profit by $80 million (or 2%) relative to the base case. As a point of comparison, the study by Gunasekara et al. (2007) suggested a climate change impact of 4% reduction in gross state product of Victoria. Our estimate is lower because it was derived with the inclusion of carbon fertilisation benefits whereas the fertilisation effect was ignored in the study by Gunasekara et al. (2007).

Table 3: Summary of modelling results for different farm types

Region	Farm system	Farm gross margin ($/ha)		Change
		Base	CC	($/ha)	(%)
North-central	Mixed	381.14	415.43	+34.29	+9.0
North East	Mixed	271.63	309.58	+37.95	+14.0
Mallee	Mixed	198.87	147.98	-50.89	-25.6
Wimmera	Grains	321.75	371.68	+49.93	+15.5
Gippsland	Dairy cattle	1,223.45	1,180.61	-42.85	-3.5
South West	Dairy cattle	778.26	768.77	-9.49	-1.2
Gippsland	Beef cattle	369.54	349.35	-20.19	-5.5
South West	Beef cattle	148.56	155.87	+7.31	+4.9
North East	Beef cattle	193.25	179.60	-13.65	-7.1
Wimmera	Sheep	146.32	135.48	-10.83	-7.4
South West	Sheep	189.72	198.68	+8.96	+4.7
North East	Sheep	133.66	123.59	-10.07	-7.5

The impact of climate change was estimated to vary across regions, with the estimated impact ranging from a gain of $42 million (or 7% relative to the base case) for Wimmera to a reduction of $128 million (a decline by 26%) for Mallee (Table 4). Climate change was projected to increase grain crop yields in mid-north Victoria encompassing Wimmera and North East. Such projected yield increases underpin the estimated positive impact of climate change on Wimmera, but were offset by the projected pasture yield reductions in North East (Figure 3). For Mallee, grain crop and pasture yields were projected to decline under climate change, hence contributing to a significant negative regional impact.

A small but positive regional impact of climate change ($17 million or 2% relative to the base case) was estimated for the South West region. This result is consistent with the projection of higher grain and beef/sheep pasture yields under climate change, notwithstanding the projected marginal decline in dairy pasture yields.

Figure 3: Impact of the 2050 High Climate Change scenario on farm profits

Table 3: Summary of modelling results for different farm types

Region	Impact of CC $m change	Impact of CC % change	Value of Adaptation 1 ($m)	Value of Adaptation 2 ($m)
Mallee	-128.3	-26	63	126
Wimmera	+41.5	+7	(not applicable)	(not applicable)
North-central	+40.1	+9	(not applicable)	(not applicable)
North East	-19.2	-4	(not applicable)	(not applicable)
South West	+16.6	+2	(not applicable)	(not applicable)
Gippsland	-30.3	-5	(not applicable)	(not applicable)

For the South West region, the modelling did not consider a widely projected change in land use from livestock to grain cropping under climate change. The study by Tostovrsnik et al. 2010 raised the likelihood of grain cropping activities spreading into the high-rainfall southern regions of Victoria under climate change. This projection was attributed to the potential reduction in the frequency of water logging and flooding due to the reduced rainfall, which would make the region more suitable for grain crops. For future research, this type of spatial redistribution of farm types should be considered.

A small negative regional impact of climate change (minus $30 million or negative 5% relative to the base case) was estimated for the Gippsland region. As this region is dominated by livestock farming activities, the result largely reflects the projection of reduced sheep/beef and dairy pasture yields (Table 2).

The economic benefit of introducing new crop varieties to Mallee was estimated to be $63 million for Adaptation 1 and $126 million for Adaptation 2 (Table 4).

5. Conclusions

This paper describes an integrated approach to connecting regional whole-farm models and GIS layers. The aim has been to revitalise a set of existing LP models representing key farming systems in Victoria, hence expanding the capability of the EPRB in using these models for policy analysis. These models have been proven useful through time, but they are not suited to the type of spatial analysis that is increasingly required today. Different regional sectors and communities will be affected in different ways and to different degrees by climate change, water availability, government policy and other market developments. Some regions will benefit and others will suffer. The knowledge of differential regional impacts would be useful to policy makers.

Here we have reported the first step in a process that should eventually lead to a complete integration within a true spatially explicit model. The spatial features of the current model are aggregated at the regional level. For each region we know the total area of each farm type, so we can calculate regional impacts by solving all 12 whole-farm models and aggregating the results according to regional composition of farm types. The results are then presented on maps. This level of spatial resolution is sufficient for many problems, but for others we may want to know the actual location of the farms so we can relate farm performance to land features such as soil type or some productivity index. The capability to do this is present in our Matlab model, which can relate any set of maps to any set of farm models. The main challenge, however, is the extension of the whole-farm models so they can be adjusted according to some land productivity measure that may vary in space.

Even without the extensions discussed above, the current model has the advantage that it is able to package and process large amounts information without requiring users to copy and paste data into spreadsheets. The illustration we present regarding a climate change scenario shows the potential of the model, but additional work will be required to exploit the richness of the results. For example, we could perform analysis of shadow prices for any number of scenarios by reading and processing a set of results files. The results of these scenarios could have been obtained over a period of days or months or in different computers simultaneously, but each scenario run is associated to a binary file containing a full set of results. To access these results and analyse them requires Matlab code to be written for specific problems. Eventually the model could be wrapped within a friendly user interface to undertake the more common analyses. In the meantime there are many important questions that can be explored with the integrated system in its current form.

Further applications of integrated whole-farm modelling, as we have developed for Victorian dryland farming systems, are possible in the following areas:

• Carbon pricing — Objective coefficients (or activity budgets) can be adjusted to represent scenarios with and without a carbon price for quantifying the impact on each region, agricultural industry and farming system.
• Emissions reduction — Cost to farmers of mitigating requirements can be estimated by including new constraints reflecting allowable limits on emissions for different farming systems.
• Carbon offset — Value of a carbon offset activity can be estimated by including offset options, their costs and returns in whole-farm models.
• Environmental management — Impact of preventing or reducing nutrient run-off from a farming system can be estimated by including new constraints on the related environmental outcomes in the LP models.
• Drought — Impact of drought on farming systems in different regions can be quantified by linking crop and pasture yields to growing-season rainfall.

References and Appendices

View References and Appendices

---

If you would like to receive this information/publication in an accessible format (such as large print or audio) please call the Customer Service Centre on 136 186, TTY 1800 122 969, or email customer.service@dpi.vic.gov.au.

Published by the Department of Primary Industries
Economics and Policy Research Branch, March 2012

© The State of Victoria 2012.

This publication is copyright. No part may be reproduced by any process except in accordance with the provisions of the Copyright Act 1968.

Authorised by the Department of Primary Industries
1 Spring Street, Melbourne 3000.

Wimalasuriya, R., Chan, C. and Cacho, O. 2012. Integrated whole-farm modelling - an application for policy analysis of climate change adaptation. Research Paper 2012.6. Policy and Strategy Group, Department of Primary Industries (Victoria), Melbourne.

ISBN 978-1-74326-124-8 (online)

Disclaimer

This publication may be of assistance to you but the State of Victoria and its employees do not guarantee that the publication is without flaw of any kind or is wholly appropriate for your particular purposes and therefore disclaims all liability for any error, loss or other consequence which may arise from you relying on any information in this publication.

For more information about DPI go to www.dpi.vic.gov.au or phone the Customer Service Centre on 136 186.