Simulating agricultural decision making to project future land use

Abstract: This paper presents an approach to predicting the impact of future agricultural policies, socio-economic change and climate change on agriculture land use. The approach is based on individual farmers’ decision making as they attempt to maximise their long-term profit, given the physical attributes of their land. Soil type, climate and slope determine the limits for mechanised arable cropping, and whether and how well a crop will grow and mature. Crop simulation models predict the average crop yield and its variability. These soil water models do not include disease, so a further factor for this variability is derived from statistics on yields. The models tend to be validated on old (eg 1970s) experimental data and under-predict current crop yields. Further, break crops (eg peas/beans) tend to be under-represented in the simulated crops, so adjustments to the simulated crops are used to represent all crops. The same adjustments are then applied to future simulations. A linear programming model simulates a collection of farmers determining the cropping which maximises profit margin over labour and machinery costs given soil workability. Because of variability in yields and prices, each farmer given the same information, processes it differently due to perceptions, experiences and atittudes to risk. The model simulates this by randomly selecting yields and prices based on their variability. The outcome is thus a regional prediction of the cropping. Examples of the model are given for two contrasting regions of England. The model results are compared with observed distributions of agricultural land use. The model is very good at representing land use aggregated at the regional level and general spatial trends in land use patterns. Results for the future show that socio-economic change causes larger effects then climate change