Canopy conductance subject to minute-scale variations in insolation: validation of canopy conductance parameterizations using innovative observations

The canopy conductance (g_c) is a key-variable for the energy and carbon exchange in atmospheric models, i.e. in their land-surface schemes that represent land-atmosphere interactions. For meteorological purposes, commonly used parameterizations are the fully empirical Jarvis-Stewart (JS) and the A-gs parameterizations. The latter, in part based on plant-physiological insights, relates the net assimilation rate (A_n) to g_s, the stomatal conductance. These parameterizations were originally developed for typical time scales of 30-60 minutes. However, nowadays atmospheric models run on increasingly finer spatial and temporal resolutions. At the same time, the skill of the g_c parameterizations at these short time scales remains unknown.

This study aims to validate both parameterizations for g_c at short time scales (∼ minutes). To gain more insight in the A-gs parameterization, its sub models for the internal CO₂ concentration (C_i) and the net assimilation rate (A_n) are examined as well. Data from an experiment over growing winter-wheat have been used for validation purposes. Turbulent fluxes with 1-min time resolution were obtained with an innovative method that combines scintillometer measurements with scalar turbulence measurements. Hence, we can focus on non-stationary conditions, mainly caused by rapid changes in incoming short wave radiation.

We found that the JS- and the A-gs parameterizations had difficulty in simulating the observed responses of g_c to rapid variations in insolation. Particularly in the morning hours, before 9 UTC (which is 2 hours before local noon), the observed g_c increased with cloud cover, in contrast to the decrease predicted by both parameterizations. Concerning the sub models of the A-gs parameterization, we found that the variability of A_n was reproduced well, but the predicted magnitude of the variations was too small. C_i was not reproduced realistically. Moreover, our results suggest that the assumption of C_i being a function of only VPD and the CO₂ concentration at the leaf surface does not hold for insolation lower than 500 W m^âˆ'2. We found that the representation of C_i is crucial in explaining the limited skill of the A-gs parameterization under rapidly varying insolation.