Agricultural land use allocation involves assessment of land potential as well as land demand for various crops to identify the optimum land unit for each type. Current allocation models often focus on the perspective of increase in land demand. However, processes causing loss of arable land should be considered to correct the demand used in conventional allocation processes. In order to achieve that, a two stage model for crop allocation is presented which uses different models such as cellular automata, Markov chain, fuzzy rule-based system, goal programming and GIS raster analysis in a logical framework. In the presented model, at first by considering environmental (soil erosion and soil salinity) and economic (urban expansion) factors leading to arable land loss, agricultural use of each land unit is allocated. Next, crop types within the agricultural area are determined based on land suitability evaluation and demand assessment for each crop. This model is developed for rural land use planning on a regional scale and is applied to the Borkhar and Meymeh district in Isfahan province in Iran to predict land use map of 2015. The overall accuracy of 0.609 in the first step allowed us to draw conclusion that the inclusion of factors resulting farmland decline in allocation process gives more accurate results than the former allocation techniques which merely model factors causing increase in agricultural land areas. Moreover, according to the results for 2015, 27.82%, 21.64%, 7.27%, 5.85%, 7.36%, 6.36% and 1.74% of agricultural areas is allocated to wheat, wheat-maize, barley, barley-maize, maize, alfalfa and potato, respectively. The presented approach has advantageous to determine the crops with highest suitability for a given area according to the desired objectives.
