Water Savings with Irrigation Water Management at Multi-week Lead Time Using Extended Range Predictions

Abstract

The Extended Range Predictions (ERP) provide Sub-seasonal to Seasonal (S2S) prediction of meteorological variables at a lead time of multiple weeks. Despite the considerable improvements in the ERP system, their applications in real-time irrigation water management are limited. Such limited use of ERP is due to uncertainty in predictions and a spatial mismatch in scales between the model grids and farms. Here, we develop a chance-constrained optimization framework for farm-scale irrigation water management planning for multiple weeks in advance. The degrading model performance with time and the related uncertainty are addressed by generating large ensembles using weather generators. Working with grape farmers in India has revealed that such an advance planning could save water by 3–25 % at a farm-scale without losing yield, compared with the existing approach requiring near real-time decisions. Moreover, the co-development of such irrigation tools with farmers increases the use and usability of S2S predictions.