GLDAS Land Surface Models based Aridity Indices

Identification of dryland areas is crucial to guide policy aimed at intervening in water stressed areas and addressing its perennial livelihood or food insecurity. Aridity indices based on spatially relative soil moisture conditions such as NCEP aridity index allow cross comparison of dry conditions between sites. NCEP aridity index is based on the ratio of annual precipitation (supply) to annual potential evaporation (demand). Such an index ignores subannual scale competition between evaporation and drainage functions well as rainfall and temperature regimes. This determines partitioning of annual supply of precipitation into two competing (but met) evaporation and runoff demands. We here introduce aridity indices based on these additional considerations by using soil moisture time series for the past 3 decades from three Land Surface Models (LSM) models and compare it with NCEP index. We analyze global monthly soil moisture time series (385 months) at 1 x 1 degree spatial resolution as modeled by three GLDAS LSMs - VIC, MOSAIC and NOAH. The first eigen vector from Empirical Orthogonal Function (EOF) analysis, as it is the most dominant spatial template of global soil moisture conditions, is extracted. Frequency of nonexceedences of this dominant soil moisture mode for a location by other locations is calculated and is used as our proposed aridity index. An area is indexed drier (relative to other areas in the world) if its frequency of nonexceedence is lower. The EOF analysis reveals that their first eigen vector explains approximately 32%, 43% and 47% of variance explained by first 385 eigen vectors for VIC, MOSAIC and NOAH respectively. The temporal coefficients associated with it for all three LSMS show seasonality with a jump in trend around the year 1999 for NOAH and MOSAIC. The VIC aridity index displays a pattern most closely resembling that of NCEP though all LSM based indices isolate dominant dryland areas. However, all three LSMs identify some parts of south central Africa, southeast United States and eastern India as drier than NCEP classification. NOAH and MOSAIC indentify parts of Western Africa drier than the other two classifications, while NOAH and VIC indentify parts of central India as wetter than the other two classifications.