Global observations of land-atmosphere interactions during flash drought

Abstract

Flash droughts, which intensify on subseasonal-to-seasonal (S2S) timescales (2 weeks–2 months), cause severe and sudden impacts on agriculture, ecosystems and economies. To evaluate and improve S2S forecasts of flash drought, we need to understand the land-atmosphere coupling processes that are critical to flash drought development, specifically the feedbacks between soil moisture and evapotranspiration. Previous investigations of flash droughts have either focused on specific regions or relied on global reanalysis datasets, which have known shortcomings in their representation of land-atmosphere interactions. Here, we use a variety of global long-term products of daily satellite observations to explore the evolution of the surface energy balance during flash droughts over the period 2000–2020. We investigate the differences between flash droughts with stronger and weaker land-atmosphere coupling, and assess feedbacks from the land surface to near-surface air temperatures during the events. Events with stronger evaporative stress are associated with perturbations in the surface energy budget for 4 months both before and after drought onset, indicating the importance of precursor land conditions for S2S predictability. For three semi-arid regions in Africa, we show that increased sensible heat flux feeds back to increase peak air temperatures during flash droughts. We also use Vegetation Optical Depth (VOD), a proxy for vegetation water content, to demonstrate that lower VOD 1–2 months before flash drought onset is linked to increased air temperatures during the peak of the drought in some regions. For example, in West African summer, 12 % of flash droughts with precursor VOD anomalies in the highest quartile experience a peak air temperature anomaly > 1.5σ, whereas this increases to 27 % for events with precursor VOD anomalies in the lowest quartile. This shows that globally-observable land surface conditions could provide useful information to S2S forecasts and motivates further assessment of land-atmosphere interactions in these forecasting models using observational datasets at the global scale.