Assessing transpiration partitioning of CMIP6 Earth System Models with ground and satellite observations

Abstract:

Earth System Models (ESMs) are found less skillful in simulating the response of terrestrial evapotranspiration (ET) to vegetation dynamics. Some of these models even reported a net increase in ET as a result of global deforestation. Much of the model error appears to arise from the unrealistic transpiration partitioning (T/ET), as the three components of ET (plant transpiration (Et), soil evaporation (Es) and canopy interception (Ei)) respond differently to changing environmental and/or vegetation conditions. The most recent studies show that, on average, global T/ET should be greater than 60%, yet most ESMs in the CMIP5 largely underestimate this ratio (around 40%)., resulting in irrational sensitivity of ET to vegetation changes. Given the rapidly changing land use and land vegetation composition, as well as the continuous increase of atmospheric CO₂ concentration and global warming, quantitative partitioning of ET by E and T will be very critical for understanding current and future trends in ET in the context of coupled vegetation-water-climate modelling. Currently, whether ESMs participating in CMIP6 have improved over those in CMIP5 in their ability to simulate ET partitioning has not been evaluated. In this study, we will comprehensively assess model outputs of ET fluxes and T/ET ratios in CMIP6 against numerous data streams of observations, including data from global FLUXNET observations, GRACE-based observations, MODIS products and other satellite data, as well as measurements based on isotopic discrimination, microlysimeters, and sap-flow sensors. We will also attempt to determine the key processes and parameters that affect ET partitioning in ESMs.