The transient behavior of whole-canopy fluxes during dynamic light conditions for midlatitude and tropical forests

Cloud shadows lead to alternating light and dark periods at the surface. Understanding how clouds affect whole-canopy fluxes suffer from two knowledge gaps that limit scaling from leaf to canopy scales, an effort currently done by assertion alone. First, there is a lack a clear quantitative definition of the incident light time series that occur on specific types of cloudy days. Second, the characteristic time scales for leaves to respond to for stomatal opening and closing is 1-10 minutes, a period too short to allow accurate eddy fluxes. We help to close the first gap by linking the durations of alternating light and dark periods statistically to conventional meteorological sky types at a midlatitude mixed deciduous forest (Harvard Forest, MA, USA: 42.53N, 72.17W) and in a tropical rain forest (Tapajós National Forest, Brazil; 2.86S, 54.96W). The second gap is narrowed by measuring the dynamic response whole canopy exchanges in the flux footprint at intervals of only a few seconds using the classical ensemble average method, keying on step changes in light intensity. Combining light and shadow periods of different lengths we estimate ensemble fluxes sensible heat (H), net ecosystem exchange (NEE), and latent heat (LE) fluxes initiated by abrupt radiation changes at intervals of 30 s over 20 minutes. We present composite results of the transient behavior of whole-canopy fluxes at each forest, showing distinct features of each forest type. Observed time constants and transient flux parameterizations are then used to force a simple model to yield NEE, LE, WUE, and Bowen ratio extrema under periodic shadow-light conditions and given cloud amount. We offer the hypothesis that, at least on certain types of cloudy days, the well-known correlation between diffuse light and WUE does not represent a causal connection at the canopy scale.