Flash Drought Monitoring in Pakistan Using Machine Learning Techniques and Multivariate Drought Indices

Abstract

The rising global temperature owing to global warming has led to a disruption in the annual rainfall pattern and an increase in drought severity. Various drought indices have been developed over the period out of which the Standardized Precipitation Index (SPI)and Standardized Precipitation and Evaporation Index (SPEI) were found to be the best performing under the Pakistan climate that comprises mostly arid and semi-arid regions. Precipitation, Temperature, and evapotranspiration are among the most notable hydrological parameters to identify drought. The grid-based satellite data of resolution 0.25°x0.25 has been used for Precipitation and Temperature and Evapotranspiration has been calculated by the Hargraves method.  During the drought monitoring assessments, a severe drought period from 1999-2002 was witnessed in Pakistan with a negative value of SPI and SPEI ranging in between 1.6-2.0 signifying severe drought conditions over that period Main reason for such drought events was a disruption in the rainfall patterns. In addition, it was observed that an increased maximum temperature mainly caused drought in some regions in 2012. The most severely affected areas were Tharparkar in Sindh, parts of Baluchistan province, and some districts in Punjab. Such mapping and monitoring of flash droughts are important to plan our water resources well.