Flash Drought Climatology Over Southeastern South America: Sensitivity to Index and Reanalysis Selection, and Potential Causes

ABSTRACT Flash droughts (FD) develop rapidly, impacting agricultural production and natural ecosystems. Though studies have used various indices and datasets to study their characteristics, there is limited understanding of the regional uncertainty among FD indices and the possible causes. Therefore, this study uses three indices (FDII, SESR and SPEI) and two reanalysis datasets (ERA‐5 Land and MERRA‐2) to assess the agreement and uncertainties in FD climatology and trends over southeastern South America. In addition, the possible role of land–atmosphere coupling strength and the Aridity Index as possible explanations are explored. Although FD patterns vary depending on the index and dataset used, most FD events occur during the austral warm season, with other seasons also standing out. The FDII shows the highest frequency mainly over transitional and humid climates, coinciding with SESR in humid regions of Brazil. The SPEI‐1 shows FD maxima over humid and arid regions. A comparison of each FD index between ERA‐5 Land and MERRA‐2 reveals both notable similarities and important differences. On one hand, both SPEI‐1 and SESR show similar spatial patterns. On the other hand, high uncertainties are observed in the soil moisture‐based FDII. Moreover, while all three indices indicate an increase in the spatial occurrence of FD in recent years for both datasets, FDII did not exhibit significant temporal changes. The 2017–2018 FD event in the Argentine Pampas revealed that the variables used inside the FD indices agree to represent the severe drought conditions and their further impacts on vegetation, but strong uncertainties in the temporal and spatial coincidence among the indices are evident. While differences in land–atmosphere coupling help explain some regional similarities and differences among FD indices, discrepancies largely stem from their varying sensitivities and threshold definitions. These findings underscore the need to refine FD indices and adopt multi‐index dataset approaches to improve drought characterisation.