Spatial optimal allocation of land and water resources using GIS and genetic algorithm

This research focuses on spatial optimal allocation of land and water resources for crop production in agricultural watersheds. The process of optimal allocation is complex due to spatial and temporal variation in supply and demand parameters. In this study methodology that integrates the system simulation models (hydrological and crop growth), economic analysis model, and resource allocation model (using genetic algorithm evolutionary optimisation technique) within GIS is developed to build a spatial decision support system (SDSS) for spatial and optimal allocation of resources. This study investigated different ways of integrating simulation models with GIS (loose coupling, tight coupling and full coupling). The study revealed that the full coupling method is superior to other two methods of integration. The full coupling (integrated) approach is used to develop the SDSS. The hydrological processes such as rainfall, interception, infiltration, runoff, channel routing, deep percolation, evaporation, crop evapotranspiration, irrigation and crop growth are considered for the development of distributed hydrological simulation model in this study. The outputs of this model are runoff, net benefits, crop yields and water use pattern for the specified landuse plan. The resource allocation (optimisation) model developed for optimal spatial allocation of land and water resources in the watershed uses the hydrological simulation model as external evaluation function for GA optimisation technique. The optimisation model is designed to handle various objective functions (to maximise cropped area, crop yields and net benefit; to minimise runoff). The GA generates initial population (landuse plans). These landuse plans are evaluated by the hydrological simulation model and are then ranked according to their fitness. The best performing landuse plans are used to reproduce new landuse plans using crossover and mutation operators of GA. The newly generated landuse plans are evaluated and are competed with the initial set of population to get included in the next generation. The next generation is reranked according to their fitness and the process is repeated till the optimal solution is obtained. The optimal set of population contains land and water resources allocation plans performing on par. The developed SDSS was applied to the Pimpalgaon Ujjaini watershed, a case study watershed from Ahmednagar District, Maharashtra, India. The satellite remote sensing images of the study area were used to develop the landuse and other thematic maps. These maps were used to generate the initial population. The application of the model resulted in spatial optimal land and water resource allocation plans. These plans enable the decision makers to investigate on what has to be changed and where the changes have to be made for sustainable development. The SDSS gives the decision maker a powerful tool to study the effect of changes in watershed.