Historical Analysis and Future Projections of Meteorological Drought in Northeast Thailand Using Statistically Downscaled Climate Models

Abstract

Drought is a reoccurring meteorological hazard that seriously affects agricultural production, water resources and the livelihoods of rural people. The current study investigates the past, observed drought conditions as well as future climate change scenario in Northeast Thailand by combining historical meteorological observations, satellite-derived precipitation datasets, and statistically downscaled Global Climate Model (GCM) outputs. The SPI based on fitting the Gamma-distribution and through probability-of-normal transformation was employed in 12-month periods to describe drought severity and duration from 1994 to 2023. Future scenarios were projected using CanESM2 (CMIP5) and CanESM5 (CMIP6) under RCP4.5 and RCP8.5 emission pathways for mid-century (2041–2070) and late-century (2071–2100) periods. The accuracy of the climate projections was improved through bias correction via Quantile Mapping (QM) and empirical downscaling with Multiple Linear Regression (MLR), with calibration and validation of the models undertaken through a split sampling approach (70% calibration and 30% validation) over the baseline period (1994-2023). Climate Hazards Group InfraRed Precipitation with Station data (CHIRPS) rainfall datasets were incorporated primarily to fill gaps and extend the temporal continuity of station records. However, CHIRPS-based spatial maps were not produced, as the study focused on temporal drought patterns. The performance evaluation showed notable improvements, with R² values over 0.70 for precipitation and over 0.85 for temperature, along with RMSE reductions of 25 to 40% across representative stations. Historical analysis revealed an increase in the occurrence and intensity of events droughts at representative stations like Nongkai, Khon Kaen, and Ubon Ratchathani. In addition, RCP8.5 projections have depicted a growth in the number of extreme events where SPI values have fallen below -1.5 and their frequency is foreseen to double by 2100. The results demonstrate that satellite-derived precipitation datasets combined with statistically downscaled climate models can support regional drought assessment based on representative stations. Moreover, the results reflect a pressing need for water management, early warning systems, and climate-resilient crops. This work supports the National Adaptation Plan of Thailand and regional drought mitigation strategies with respect to evolving climatic conditions.