Estimating dynamic non‐water‐limited canopy resistance over the globe: Changes, contributors and implications

Abstract Non‐water‐limited canopy resistance ( r cs , also known as the bulk stomatal resistance or surface resistance) is a critical variable in estimating potential evapotranspiration (PET), which is widely used in ecohydrology related fields. However, quantifying r cs is a challenging work. Here we develop an approach for estimating r cs over the globe. Comparing results over the globe and across 10 ET data sets (used as inputs), which are based on diverse mechanisms and algorithms, we find that the approach can capture canopy resistance well (mean correlation of 0.84 ± 0.04, mean relative Root Mean Squared Error of 4.4% ± 1.0%, and mean relative Mean Absolute Error of 5.8% ± 1.4%), and the estimated r cs are very close to those estimated using another method ( R 2 = 0.92), which is based on a quite different hypothesis that is only suitable for saturated regions. Based on these, we find that the r cs shows an overall increasing trend (0.43 ± 0.13 s m −1 year −1 ) over the globe (at 77.6% ± 3.9% of the land grid cells) during 1982–2014, and the air temperature dominates the variabilities of r cs in regions with decreasing r cs (mean relative contribution of 57.9% ± 11.4%), while air CO 2 concentration controls the changes in r cs in regions with increasing r cs (mean relative contribution of 47.3% ± 8.0%). Moreover, we also find that the traditional PET estimator explicitly overestimates the increasing trends in PET, and tends to overestimate (underestimate) the increasing (decreasing) trends in regions with increasing (decreasing) PET. These findings can improve our knowledge on the complex water‐vegetation‐environment interactions and would be helpful for developing more accurate models for quantifying the impacts of global change on water resources.