Spatiotemporal variations in evapotranspiration partitioning within the North American monsoon region: Role of vegetation dynamics

A fundamental problem in land-atmosphere interactions is diagnosing the impacts of growing-season vegetation dynamics on evapotranspiration (ET) and its partitioning into soil evaporation (E), plant transpiration (T) and evaporation of intercepted water (I). In this study, a distributed hydrologic model and remote sensing observations are used jointly to simulate water and energy exchanges in ecosystems that dramatically respond to the North American monsoon. Remotely-sensed data from the Moderate Resolution Imaging Spectroradiometer (MODIS) are utilized to update model vegetation parameters related to radiation, interception and transpiration during four summer periods (2004, 2006-2008) exhibiting strong interannual differences in precipitation and vegetation response. Model simulations are compared against surface fluxes and hydrologic state variables measured at two eddy covariance tower sites in the study region in Sonora, Mexico and then applied more broadly over a mountain basin. Comparisons of simulations with static (leaf-on and leaf-off) and seasonally-varying vegetation reveal lower ET/P and a higher vegetation losses, (T+I)/ET, for green conditions. Vegetation fraction (vf) was found to be a primary control on ET partitioning, while interception parameters played a secondary role. As a result, spatial and temporal patterns in (T+I)/ET exhibited a strong signature of remotely-sensed vf. Simulations indicate that (T+I)/ET can dominate over large catchment areas during specific periods and exhibit ecosystem-dependent evolutions. In addition, interannual variations in precipitation and vegetation response lead to dramatic changes in ET partitioning across different summer periods. The fully-transient estimates of ET partitioning can help explain vegetation impacts on climate in drought-deciduous ecosystems of the North American monsoon region.