A Spatio-temporal drought prediction for Sri Lanka using standard precipitation index

Abstract

This research intends to construct an efficient drought forecasting model employing Spatio-Temporal Interpolation with Standardized Precipitation Index (SPI). Notable deviations in rainfall are now observed across the globe due to the global increase in mean annual temperature causing recurrent drought events, and Sri Lanka is among the severely affected by these incidents bringing serious economic impacts, particularly to its agricultural sector. Hence, accurate draught predictions are imperative for early alerts to mitigate its consequences and execute readiness measures. This study thus scrutinizes geostatistical and deterministic descriptive Spatio-Temporal interpolation techniques to find an optimal method. Research findings based on cross validation suggest that Geostatistical Spatio-Temporal methods significantly outperform Deterministic Spatio- Temporal methods for drought forecasting in this region. The combined use of SPI-3 and SPI-6 enabled the formulation of reliable forecasts regarding the onset, intensity and offset of drought conditions. Consequently, this model was able to successfully predict drought hazards for the upcoming season. It is therefore recommended that the affected districts prepare accordingly with revised cropping and harvesting strategies suitable for YALA and MAHA seasons in Sri Lanka