Comparison of two modelling approaches for an integrated crop economic model

Abstract

Integrated crop models with biophysical and economic component models were developed in order to support decisions on agricultural land productivity assessments in northern Thailand. Over the past few decades, efforts to produce higher crop yield focussed on extending crop land to increase crop yield per capita. This process included considerations of land and water quality, improved land and water use efficiency and greater involvement of farmer communities in the planning process. In order to support the planning process, decision makers needed a tool to assist in assessing the complex production system, focusing on the beginning of the cropping production process to leaving the farm. Such a tool should also support the dynamic assessment of economic land suitability for the 19 major economic crop types used in northern Thailand. The Land Development Department (LDD) in Thailand framework provides a guide to the suitability of different crop types to a range of land quality attributes. Two modelling approaches were used to develop an integrated crop-economic model based on the framework of the LDD. The first model type is a mechanistic model, which estimates crop yields using soil and climate information and estimates economic returns. To introduce uncertainty into the model, fuzzy sets were used. The second approach used a Bayesian network model. The Bayesian network was used to estimate the probability of achieving a crop yield given climate and soil input data. Economic returns in the Bayesian network were estimated using utilities. The LDD framework is used in the models to estimate crop yield and economic returns using available biophysical and economic information. This paper will introduce both models and assess the constraints that influenced the construction of the models. A set of criteria will be used to evaluate the models in order to examine their usefulness, representativeness and robustness. The comparison of models will focus on the: data type, technique, and model outcomes. The results of this comparison will help in evaluating the strengths and weaknesses of each of the modelling approaches, and based on these outcomes, recommendations on methods for building cropeconomic models will be made.