Development of Nonstationary Standardized Precipitation-Evapotranspiration Drought Index (nSPEI) using Climate Indices as Covariates

Abstract:

There is need to develop multiscale, multivariable drought indices for use in nonstationary climate change scenarios. They aid in (i) monitoring and/or forecasting drought events over various time scales and (ii) comparison of drought conditions across locations, and (iii) account for non-stationarities in precipitation and several other meteorological variables (e.g., evapotranspiration, temperatures, and relative humidity) which affect drought in arid and semi-arid regions. Recent contributions made in this connection make unrealistic assumption of monotonic trend for location, scale and shape parameters (in frequency analysis), considering time as their covariate. To address this issue, we suggest using large-scale climate indices as covariates in lieu of time. A new nonstationary Standardized Precipitation-Evapotranspiration drought Index (nSPEI) is developed. It is estimated using time-varying probability distribution function (tv-PDF) whose parameters are considered as functions of the suggested covariates. The nSPEI therefore appears more robust and reliable to characterize droughts in climate change scenarios. This is demonstrated through a case study on Karnataka state of India. Covariates corresponding to different locations (0.5 degree resolution grids) in the study area were identified from ten global and regional scale climate indices. The covariates were subsequently used to arrive at parameters of tv-PDF for nSPEI estimation corresponding to 0 to 12 months lead times. In the analysis, precipitation and evapotranspiration were considered for the period 1951-2005 from Climatic Research Unit (CRU) reanalysis data. Equatorial Indian Ocean Oscillation (EQUINOO) and Multivariate ENSO Index followed by Arctic Oscillation and Atlantic Multi-decadal Oscillation were selected as covariates for most locations. The nSPEI was found to be effective and consistent with SPEI in identifying historical drought events. There were, however, differences in intensity, duration, onset and termination of droughts discerned using the two indices. The nSPEI offers scope to determine future projections of droughts corresponding to different lead times based on prediction of climate covariates identified for different parts of the study area.