A revised drought index based on precipitation and pan evaporation

Drought episode identification and its severity characterization are crucial to evaluate and reduce the risk of drought disaster. The Standardized Precipitation Evapotranspiration Index (SPEI) with two different potential evapotranspiration formulations, i.e. the Thornthwaite and Penman-Monteith equations, (denoted as SPEI-Th and SPEI-Pm, respectively) are two popular drought indices used today. These two SPEI versions have the limitations of overestimation of drought (as a response to global warming) and high data requirements, respectively. A new SPEI variant was developed with precipitation and pan evaporation (E-pan) as its inputs (denoted as SPEI-Pe), which overcomes the shortcomings of SPEI-Pm and SPEI-Th. Climatic data at ground stations in the Yunnan Province (YP) of China were collected to validate this new SPEI variant with SPEI-Pm as the standard measure. Month-by-month comparison of SPEIs suggested that SPEI-Pe had a better agreement with the SPEI-Pm standard than SPEI-Th, with the criterion measures of mean absolute error and refined index of agreement (d(r)). The sensitivity analysis showed that two energy variables, net and global radiation are the most influential factors for evapotranspiration, and for SPEI-Pm and differences between SPEI-Pm and other two variants too. The validity of SPEI-Pe was also tested against the historical drought records in YP. Results demonstrated that SPEI-Pe generally agreed with historical drought events at different durations with only several exceptions, and thus it had a comparable performance to identify drought episodes with the other two SPEI variants in YP. The results suggest that SPEI-Pe has a better agreement with SPEI-Pm than SPEI-Th, and that it provides a good replacement for the standard measure when underlying climatic data (especially the energy variables) are not available.