The Palmer Drought Severity Index (PDSI) is a popular meteorological drought index that is commonly used in the U.S. along with the Palmer Hydrological Drought Index (PHDI) and the Palmer Moisture Anomaly Index (Z-index). Using precipitation and air temperature as inputs, the Palmer indices estimate moisture supply and demand within a simple two-layered soil moisture model. The PDSI has some issues related to the lack of comparability between regions (Alley, 1984; Doesken and Garen, 1991; Hayes et al., 1999; Heim, 2002). To address this problem, Wells et al (2004) developed the self-calibrated (sc) Palmer Indices to automatically determine appropriate regional coefficients. This scPDSI makes the Palmer indices more spatially comparable. Another limitation of the Palmer indices is that they are calculated at a fixed timescale, which limits their ability to accurately monitor and quantify different types of drought (Vicente-Serrano et al., …show more content…
Proposed by (Vicente- Serrano et al., 2010), the SPEI is based on a monthly climate water balance (precipitation [P] minus reference evapotranspiration [ETo]), which is accumulated at different timescales and converted to a normal standard variable using a 3-parameter log-logistic distribution. Here the ETo was computed using the Hargreaves and Samani equation (Hargreaves and Samani, 1985), which is recommended by FAO for data scarce regions.
d) Standardized Palmer Drought Index (SPDI). Developed by Ma et al (2014), the SPDI is based on combining the methods of PDSI and SPI. This index shares the multi-scalar concept and the statistical nature of the SPI and SPEI (Vicente-Serrano et al., 2015) and the water balance defined by Palmer (1965). The SPDI is transformed to a standard normal variable using a generalized extreme value distribution.
The different drought indices were calculated from the mean climate series generated for each county. The multi-scalar indices (SPEI, SPI and SPDI) were calculated at timescales from 1 to 12-months. The monthly drought indices for each county were de-trended using the same method that was applied for de-trending the crop yield