Multi‐Model Intercomparison of the Complementary Relationship of Evaporation Across Global Environmental Settings

Abstract The Complementary Relationship of evaporation (CR) theory has gained attention in recent years, in part because it relies solely on standard weather data for estimating evaporation, eliminating the need for land surface moisture and resistance information. CR models show varied skill across diverse landscapes and climates. However, inconsistent parameterization, calibration strategies, station data sets, and geographic coverage have hindered consistent and systematic intercomparison of model performance. This study intercompares four widely used CR models, uncalibrated and calibrated, with eddy covariance data worldwide, and assesses how environmental conditions affect model skill and calibration. Eddy covariance data from 227 FLUXNET and AmeriFLUX stations were quality assured and controlled, filtered, and energy balance closure corrected using open‐source software and benchmarking procedures for reproducibility, after which 82 sites remained. Systematic intercomparison showed that, overall, the Rescaled Power function (Szilagyi et al., 2022, https://doi.org/10.1029/2022wr033095 ) had the highest skill followed by Sigmoid (Han & Tian, 2018, https://doi.org/10.1029/2017wr021755 ), while Polynomial model (Brutsaert, 2015, https://doi.org/10.1002/2015wr017720 ) generally biased high, and Advection‐Aridity (Brutsaert & Stricker, 1979, https://doi.org/10.1029/wr015i002p00443 ) often produced unrealistic negative values when observed evaporation was low. Two‐parameter calibration greatly improved model skill, with single‐parameter PT‐α calibration performing comparably, suggesting potential to reduce model non‐uniqueness. Calibration and model skill rankings were consistent using energy balance closure corrected and uncorrected data sets. Results indicate that environmental conditions such as aridity index, relative humidity, vapor pressure deficit, and the ratio between radiative and apparent potential evaporation can explain spatial patterns of CR model parameters, revealing opportunities for future development of improved, calibration‐free CR evaporation modeling.