Simple Models Outperform More Complex Big‐Leaf Models of Daily Transpiration in Forested Biomes

Transpiration makes up the bulk of total evaporation in forested environments yet remains challenging to predict at landscape‐to‐global scales. We harnessed independent estimates of daily transpiration derived from co‐located sap flow and eddy‐covariance measurement systems and applied the triple collocation technique to evaluate predictions from big leaf models requiring no calibration. In total, four models in 608 unique configurations were evaluated at 21 forested sites spanning a wide diversity of biophysical attributes and environmental backgrounds. We found that simpler models that neither explicitly represented aerodynamic forcing nor canopy conductance achieved higher accuracy and signal‐to‐noise levels when optimally configured (rRMSE = 20%; R2 = 0.89). Irrespective of model type, optimal configurations were those making use of key plant functional type dependent parameters, daily LAI, and constraints based on atmospheric moisture demand over soil moisture supply. Our findings have implications for more informed water resource management based on hydrological modeling and remote sensing. Plain Language Summary Forests comprise the largest share of Earth's vegetated surface area and play an integral role in its hydrological cycle. Forests transfer moisture from below the surface to the atmosphere via transpiration, affecting surface moisture budgets and weather patterns at local‐to‐regional scales. Our ability to accurately predict transpiration in forests is thus critical to reliable weather prediction and more informed water resource management. The most accurate predictions stem from process‐oriented models with detailed representations of plant hydraulic architecture and leaf stomata regulation. These models, however, rely on inputs that are not widely available and thus are not well‐suited for predictions across broader spatial scales. Here, we sought to identify models that could be readily applied using conventional input data streams to predict daily transpiration across a wide diversity of forested ecosystems and over large spatial scales. This was carried out by evaluating predictions emanating from four models of varying complexity against two independent estimates of daily transpiration. We found the most parsimonious models to be those requiring few meterological variables and one forest structural variable as input, achieving an accuracy 33% higher and explaining 16% greater variance than the most complex models requiring additional m