Attribution of Flux Partitioning Variations between Land Surface Models over the Continental United States

Recent studies have noted large disagreement in the relative contributions of soil evaporation (E), plant transpiration (T), and evaporation of intercepted water (I) to the total flux of terrestrial evaporation (ET). This disagreement exists both between land surface models (LSMs), but is also reflected in model-based retrievals. The noted disagreement in the ratio of T to total ET is a significant contributor of uncertainty in the global water balance estimates. Even if two different models have the exact same T fraction at each model grid-cell, they can yield different fractions in their global comparison if they don’t agree on the distribution of total ET between T dominated regions versus E dominated regions. It is important to distinguish between these two alternative explanations (local partitioning vs regional distribution of total ET) so that efforts to improve the models can be directed effectively.

In this study, we explore for the continental US the level of agreement in ET partitioning for the LSMs that are implemented in the North American Land Data Assimilation System (NLDAS). These models also show large disagreement in T fraction of the global ET. For these models, we examine a regional breakdown to determine whether the disagreement stems from flux partitioning or regional variations in the estimates of total latent flux. Specifically, the study examines the sources of disagreements in the bulk partitioning. In addition, the ensemble model outputs are used to estimate the uncertainty in the flux partitioning and how the uncertainty varies spatially across the CONUS domain.