Increases in extreme ET leading to a higher risk of flash droughts

Evapotranspiration (ET) is a crucial process liking the surface energy balance, the hydrological and the carbon cycles. However, ET often remains underexplored due to climate model limitations as well as sparse and poor observational coverage.

While mean ET projections of CMIP6 models are highly uncertain, we explore whether climate models are in clearer agreement in terms of extreme ET, similar to what has been shown for mean versus extreme precipitation. We first define extreme ET (ETxx) as the annual 7-day ET maximum and investigate the physical drivers behind such events in a mid-latitude region (Central Europe). Typically, extreme ET events are characterized by high temperatures and incoming surface radiation, characteristic of a heat wave.  

We find an increase in extreme ET during the recent historical period and throughout scenario SSP5-8.5 in most CMIP6 models, together with a shift of these extremes from summer towards spring. We also find a higher degree of climate model agreement in the ET extremes, partially due to constraints in the boundary conditions of such an event, meaning that the drivers behind an extreme ET event are better constrained than the drivers of annual mean ET. This is a somewhat expected result due to the increase in vapor pressure deficit with higher temperature. The agreement also extends to all considered observational products, which agree on an increase in extreme ET, however the magnitude of this increase remains uncertain across observations-based products. We find that the observed trends lie outside the likely range of trends found in unforced climate simulations, indicating that the recent shift in observed extreme ET is attributable to climate change. We further find that records in extreme ET have been disproportionally set in more recent years, compared to what would be expected in a stationary climate in both observations and CMIP6 models.

Overall, mean ET projections and trends are complex and notoriously uncertain. Here we show that extreme ET events are better constrained than mean ET projections, making them a natural target for more robust inference from observations, attribution studies and emergent constraints. Our findings indicate an elevated risk for flash drought due to higher evaporative demand. The fact that future changes in peak water demand are less uncertain than changes in the mean demand is a highly relevant information for decision-makers and for the design of future water supply infrastructure (such as irrigation systems).