Combining remote sensing-simulation modeling and genetic algorithm optimization to explore water management options in irrigated agriculture

We present an innovative approach to explore water management options in irrigated agriculture considering the constraints of water availability and the heterogeneity of irrigation system properties. The method is two-folds: (i) system characterization using a stochastic data assimilation procedure where the irrigation system properties and operational management practices are estimated using remote sensing (RS) data; and (ii) water management optimization where we explored water management options under various levels of water availability. We set up a soil–water–atmosphere–plant model (SWAP) in a deterministic–stochastic mode for regional modeling. The distributed data, e.g. sowing dates, irrigation practices, soil properties, depth to groundwater and water quality, required as inputs for the regional modeling were estimated by minimizing the residuals between the distributions of field-scale evapotranspiration (ET) simulated by the regional application of SWAP, and by surface energy balance algorithm for land (SEBAL) using two Landsat7 ETM+ images. The derived distributed data were used as inputs in exploring water management options. Genetic algorithm was used in data assimilation and water management optimizations. The case study was conducted in Bata minor (lateral canal), Kaithal, Haryana, India during 2000–2001 rabi (dry) season. Our results showed that under limited water condition, regional wheat yield could improve further if water and crop management practices are considered simultaneously and not independently. Adjusting sowing dates and their distribution in the irrigated area could improve the regional yield, which also complements the practice of deficit irrigation when water availability is largely a constraint. This result was also found in agreement with the scenario that water is non-limited with the exception that the farmers have more degrees of freedom in their agricultural activities. An improvement of the regional yield to 8.5% is expected under the current scenario.