A probabilistic approach to assess the Standardized Precipitation Index in a region of southern Italy

Drought is a normal, recurrent feature of climate that may occur everywhere, even though its characteristics and impacts vary significantly from region to region. Drought indices are useful for monitoring and assessing drought; however, these indices are calculated at sites and it is generally required to estimate the spatial distribution of drought in the forms of maps with proper spatial accuracy. Geostatistical methods allow the interpolation of spatially referenced data. In this study a probabilistic approach, based on the acceptance of uncertainty and a finite probability of making errors in SPI estimation, is presented as an alternative to the traditional mapping methodology. The approach utilises geostatistical techniques of stochastic simulation to identify and model the spatial continuity of SPI and to develop alternate plausible simulations of SPI values in the Calabria Region. First, the SPI values have been evaluated using a daily homogeneous and gap-filled precipitation dataset. Then, five hundred simulations of SPI values were produced, using the sequential Gaussian simulation method. As a result, probabilistic summaries of the simulations provided tools for (1) estimating the range of plausible estimates of SPI at unsampled locations, (2) identifying the locations of the boundaries between significant portions of the Calabria Region with different SPI values, (3) estimating the uncertainty in SPI values and evaluate the consequences of overestimation and underestimation in decision making. Results suggest that the probabilistic prediction of SPI is a necessary pre-requisite for a rational management of environment.