Emergent Simplicity of Continental Evapotranspiration Across Multi-Scale Observations and Climate Models

Abstract:

Evapotranspiration (ET) is extraordinarily difficult to model because it depends, in part, on heterogeneous land surface features -- such as soil moisture, land cover type and plant physiology -- resulting in substantial disagreement between models. This poses a major challenge for projecting changes in the continental water cycle. By leveraging the fact that the land and atmosphere are tightly coupled in many continental environments, we show that the evaporative fraction (ET as a proportion of available energy at the surface) can be estimated across a broad range of water- and energy-limited conditions using a simple, closed-form equation with no free parameters and no land surface inputs; only air temperature and specific humidity observations are required. The method is based on a recent theory of land-atmosphere coupling (‘surface flux equilibrium’ or SFE), which we here evaluate using multi-scale observations and climate models. The equation performs well when compared to daily eddy covariance measurements at 76 inland continental sites, with overall prediction errors indistinguishable from errors in the measurements themselves, despite substantial variability in surface conditions across sites. When evaluated against multiyear mean annual ET obtained from catchment water balance estimates from 221 catchments across the United States, the resulting error statistics are comparable to those in the catchment water balance estimates themselves. The theory is also reproduced well within a broad selection of CMIP6 climate models. In total, these results reveal an emergent simplicity to continental ET, in which land-atmosphere coupling efficiently embeds land surface information in the near-surface atmospheric state on daily to monthly time scales. This result allows ET to be studied globally using weather data, which have much greater spatial and temporal coverage compared to current ET observation networks. Since lack of observations has been a major limitation in studying ET, this work opens up new opportunities for constraining projected changes to the continental water cycle.