An Algorithm Differentiating Sunlit and Shaded Leaves for Improving Canopy Conductance and vapotranspiration Estimates

Surface conductance (G(s)) is a key parameter in estimating land surface evapotranspiration (ET) and difficult to determine. Here we proposed an approach for determining G(s) according to the stomatal conductance of sunlit and shaded leaves that is estimated from their respective gross primary production (GPP) with the Ball-Berry model. Central to this approach, GPP is separately simulated for sunlit and shaded leaves with a revised two-leaf light use efficiency model. We tested the approach at 17 FLUXNET sites with seven different vegetation types. The revised two-leaf light use efficiency model outperforms its predecessor in estimating GPP at most sites. As to G(s) estimation, although our proposed algorithm has higher Akaike information criterion values than has the model estimating G(s) using vegetation indices, it was able to capture G(s) variations at all sites, while models estimating G(s) using leaf area index and vegetation indices performed poor at some sites. The proposed algorithm also improves ET estimation, indicated by lower Akaike information criterion, higher determination coefficient (R-2), and lower root mean square error of simulated daily ET for both calibration and validation data sets. This study demonstrates the usefulness of differentiating sunlit and shaded leaves in improving canopy conductance and ET estimates.