A comparison of the Diel cycle of modeled and measured latent heat flux during the warm season in a Colorado subalpine forest

Precipitation changes the physiological characteristics of an ecosystem. Because land-surface models are often used to project changes in the hydrological cycle, modeling the effect of precipitation on the latent heat flux lambda E is an important aspect of land-surface models. Here we contrast conditionally sampled diel composites of the eddy-covariance fluxes from the Niwot Ridge Subalpine Forest AmeriFlux tower with the Community Land Model (CLM, version 4.5). With respect to measured lambda E during the warm season: for the day following above-average precipitation, lambda E was enhanced at midday by approximate to 40 W m(-2) (relative to dry conditions), and nocturnal lambda E increased from approximate to 10 W m(-2) in dry conditions to over 20 W m(-2) in wet conditions. With default settings, CLM4.5 did not successfully model these changes. By increasing the amount of time that rainwater was retained by the canopy/needles, CLM was able to match the observed midday increase in lambda E on a dry day following a wet day. Stable nighttime conditions were problematic for CLM4.5. Nocturnal CLM lambda E had only a small (approximate to 3 W m(-2)) increase during wet conditions, CLM nocturnal friction velocity u() was smaller than observed u(), and CLM canopy air temperature was 2 degrees C less than those measured at the site. Using observed u() as input to CLM increased lambda E; however, this caused CLM lambda E to be increased during both wet and dry periods. We suggest that sloped topography and the ever-present drainage flow enhanced nocturnal u() and lambda E. Such phenomena would not be properly captured by topographically blind land-surface models, such as CLM.