High resolution monitoring and probabilistic prediction of meteorological drought in a Mediterranean environment

Abstract

Drought remains a costly natural disaster, with strong socio-economic and environmental impacts. Skilful seasonal drought forecasts can help to make early decisions. Here we assess the quality of a prototype seasonal forecasting system for a Mediterranean region (peninsular Spain + Balearic Islands) to predict meteorological drought as measured by the standardised precipitation index (SPI). We first show that there is a high agreement between the official data provided by the Spanish Meteorological Agency and the state-of-the art ERA5 reanalysis, building confidence in using these datasets. Thus, since the ERA5 data are provided in near-real time as it is updated on a monthly basis, it can be used to monitor drought evolution. Then, we demonstrate that it is possible to obtain skilful and reliable seasonal drought predictions several months in advance by applying an ensemble-based streamflow prediction system (ESP, an ensemble based on the rearrangement of historical data) using ERA5 data as initial conditions. The results indicate that a statistical persistence-based model could lead to an actionable seasonal drought forecasting skill thus providing the basis for a cheap and fast prototype for drought early warning.