Quantifying and interpreting annual and daily slopes between actual and potential evapotranspiration by land cover type

In the water cycle, the lion’s share of precipitation transforms into evapotranspiration (ET), which is the water converted to water vapor through evaporation and plant transpiration. Potential evapotranspiration (PET) characterizes the atmospheric demand for water. Both ET and PET are driven to first order by radiative forcing, but the manner and extent to which ET and PET affect each other is unclear. Many hydrologic models assume that a rise in PET induces a rise in ET, whereas in complementary theory a rise in PET is caused by a decrease in ET. Here we use data from the conterminous U.S. to test the competing conceptual frameworks for the relationship between ET and PET by partially controlling for the influence of radiative forcing. We observe negative correlations between annual ET and PET, consistent with the feedback described by complementary theory. At the daily timescale, we observe the sign and magnitude of the slope between ET and PET to vary with time of year and landscape type, with values that are not well explained by either theory. The varying slopes are consistent with observed differences in vegetation response to atmospheric demand, and may indicate that plants have adapted to their typical water availability conditions with varying transpiration responses to PET. The results indicate an opportunity to improve the accuracy of current methods of estimating ET from PET by including these dynamics.