Multilayer canopy model outperforms big-leaf model for evapotranspiration predictions under high water and heat stress conditions

Accurate estimation of evapotranspiration (ET) is critical for understanding land–atmosphere interactions, assessing climate impacts on ecosystem function, and informing water resource management. Most land surface models (LSMs) represent vegetation using a simplified big-leaf formulation, in which the canopy is treated as a single, uniform layer. Multilayer canopy (MLCAN) models offer a more mechanistic alternative by explicitly resolving vertical gradients in radiative transfer, photosynthesis, aerodynamics, and leaf temperature, yet it remains unclear when this added complexity leads to improved ET estimates. Here, we compared ET estimates from multilayer and big-leaf models constrained by eddy covariance observations and GEDI-derived canopy structure across sites in diverse ecoclimate regions. We show that MLCAN outperforms the big-leaf model under high-stress environments, particularly during periods of low soil moisture (SM) and high vapor pressure deficit (VPD), under clear-sky conditions, at drier sites, and in canopies with dense and vertically symmetric leaf area distributions. In contrast, MLCAN amplifies model errors under low-stress or cloudy conditions, when within-canopy gradients are muted. Process-level diagnostics reveal that the improved performance of MLCAN stems from its ability to represent within-canopy microclimate, resulting in less negative leaf water potentials and cooler leaf temperatures during high-VPD and low-SM conditions. Together, these findings reveal the climate and vegetation configurations where multilayer canopy complexity enhances ET predictions, which becomes increasingly relevant under increasing climate stress.