Modelling approach for estimation of rice yield reduction due to drought in Thailand

A rice simulation model was used to estimate yield reduction due to drought in rainfed lowlands using data from experiments conducted for three years at nine locations in Northern and Northeastern Thailand. Actual grain yield was well estimated by the simulations (r = 0.89, n =27). Irrigated conditions were simulated for each location and large differences in grain yield existed. These differences were attributed largely to differences in radiation-use efficiency, which reflects soil fertility, seeding date and seedling age at transplanting. Estimated yield loss due to drought after transplanting using 20 years rainfall data ranged between 11 and 58%, depending on location. The large yield reductions at some locations were attributed to low rainfall after transplanting, partly due to late seeding. Estimates of free water at these locations also indicated that a low water table at transplanting reduced yield. The simulation model is a useful tool, complementing experimental work, in rainfed lowland rice research.