Evaporation and Transpiration from Proximal Forests and Wetlands: Partitioning Water Fluxes to Improve Remote Sensing Algorithms and Hydrologic Theory

Evaporation (E) and transpiration (T) respond differently to ongoing global changes and are somewhat difficult to measure compared to their sum, evapotranspiration (ET). Remote sensing algorithms estimate E and T independently but have only been validated against ET, making it unclear if they correctly simulate the processes that control water flux between surface and atmosphere. There are promising new approaches to partition eddy covariance-measured ET into E and T at the ecosystem scale, which can be used to validate remote sensing products and understand how canopy conductance responds to environmental variability to control coupled carbon and water fluxes. Here, we use flux variance similarity (FVS) to partition E and T from eddy covariance ET measurements from 10 deciduous forests, 4 evergreen forests, and 3 wetlands/lakes in a 10 × 10 km area from the CHEESEHEAD19 experiment in northern Wisconsin, USA to gain insight into the processes that control these fluxes.

T decreased more than E as the growing season progressed in the forest ecosystems but not the wetlands, and the deciduous and coniferous forests showed similar E and T trajectories despite differences in vegetation phenology, with T/ET decreasing from 0.6 to 0.3, on average, from June to October. We applied the PT-JPL algorithm to half-hourly GOES-R observations and found that sub-daily E and T can be inferred accurately from some sites from space but the spatial heterogeneity of the CHEESEHEAD19 study domain was not fully captured by the ~1 km GOES-R pixels. A model based on the principle of maximum entropy production (MEP) showed good agreement with E and T observations at each site if the thermodynamic environment of the evaporating surfaces was accurately simulated, suggesting that their variability across ecosystems can be modeled with minimal inputs and assumptions. We introduce a model that couples MEP, stomatal optimality theory, and the evaporative capacitor model for E and apply it to our observations to demonstrate how simple models based on theory can be combined for a parsimonious description of ecosystem hydrology and carbon dynamics.