An empirical standardized soil moisture index for agricultural drought assessment from remotely sensed data

We propose a simple, spatially invariant and probabilistic year-round Empirical Standardized Soil Moisture Index (ESSMI) that is designed to classify soil moisture anomalies from harmonized multi-satellite surface data into categories of agricultural drought intensity. The ESSMI is computed by fitting a nonparametric empirical probability density function (ePDF) to historical time-series of soil moisture observations and then transforming it into a normal distribution with a mean of zero and standard deviation of one. Negative standard normal values indicate dry soil conditions, whereas positive values indicate wet soil conditions. Drought intensity is defined as the number of negative standard deviations between the observed soil moisture value and the respective normal climatological conditions. To evaluate the performance of the ESSMI, we fitted the ePDF to the Essential Climate Variable Soil Moisture (ECV SM) v02.0 data values collected in the period between January 1981 and December 2010 at South–Central America, and compared the root-mean-square-errors (RMSE) of residuals with those of beta and normal probability density functions (bPDF and nPDF, respectively). Goodness-of-fit results attained with time-series of ECV SM values averaged at monthly, seasonal, half-yearly and yearly timescales suggest that the ePDF provides triggers of agricultural drought onset and intensity that are more accurate and precise than the bPDF and nPDF. Furthermore, by accurately mapping the occurrence of major drought events over the last three decades, the ESSMI proved to be spatio-temporal consistent and the ECV SM data to provide a well calibrated and homogenized soil moisture climatology for the region. Maize, soybean and wheat crop yields in the region are highly correlated (r > 0.82) with cumulative ESSMI values computed during the months of critical crop growing, indicating that the nonparametric index of soil moisture anomalies can be used for agricultural drought assessment.