Evaluation and combination of monthly evapotranspiration equations for the conterminous U.S. from 1895
Abstract
Evapotranspiration (ET) is the largest component of the water budget, accounting for 70% of precipitation nationwide. Accurately estimating ET is therefore required for a variety of water resource assessment and modeling applications. Many equations have been developed to predict ET, ranging from purely empirical to process-motivated formulations, calibrated in a range of settings, are based on differing types and numbers of explanatory variables, and have varying degrees of complexity. The wide spectrum of formulations in use to estimate ET leads to uncertainties in the selection process and in the resulting predictions for ET. Here we undertake a national-scale evaluation of 21 commonly used formulations for ET, comparing overall performance, trends in performance with time, and relative performance in different settings (e.g., climate region). The national-scale data we use for evaluation are (1) water balance data from a set of ~1800 gaged watersheds; (2) flux tower data from 170 flux towers; and (3) remotely sensed ET from the MODIS-based SSEBop product. We then combine the estimates to produce ensemble, monthly, km-scale maps of ET for the CONUS from 1895-2018. The ensemble estimation makes use of a modified Bayesian model averaging approach, where the contribution of each formulation is weighted by its relative performance against the data. Finally, we present evaluations of the accuracy of our ensemble predictions, and comparisons with other available estimates of historical monthly ET.
- Publication:
-
AGU Fall Meeting Abstracts
- Pub Date:
- December 2020
- Bibcode:
- 2020AGUFMH111.0003R
- Keywords:
-
- 1805 Computational hydrology;
- HYDROLOGY;
- 1816 Estimation and forecasting;
- HYDROLOGY;
- 1839 Hydrologic scaling;
- HYDROLOGY;
- 1847 Modeling;
- HYDROLOGY