Understanding the Impact of Forcing Data Uncertainty in Evaporation Modeling

Understanding the role of data uncertainty in evaporation estimation is an important issue in hydrological modeling. Little is known on the relative sensitivity of evaporation models to variables and parameters, and insights into the acceptable error in required data forcing are required for both terrestrial and spaceborne applications. This study focuses on evaluating the impact of uncertainties in input data for a number of commonly used evaporation models, including SEBS, PT-JPL and PM-Mu. High-quality records from a set of FLUXNET towers over various biomes were used to provide a semi-synthetic dataset of consistent meteorological and flux observations, accompanied with MODIS observations of vegetation phenology. The input meteorological data to the models were contaminated with a defined error structure and used to force the evaporation models. Subsequent variations in the simulated evaporation were observed by comparison against tower data. Diagnosis of the impact of forcing data accuracy was aided by tracing back the range in the perturbed input errors for a given prediction error in evaporation. Frequency analysis on the range of data errors from the previous step for a large ensemble of modeling records provided insights on the sensitivity of the models to data and the needed accuracy of the input data required for accurate evaporation modeling.