Evaluation of a plant physiological canopy conductance model in the ECMWF land surface scheme

A new representation of the canopy conductance in large scale atmospheric models, based on plant physiological theory. is evaluated. Calibration parameters for this canopy conductance model are considered to be more universal than the coefficients in the empirical relations used in many current large scale models. The new conductance model is compared with an empirical so-called Jarvis-Stewart approach in the ECMWF land surface scheme. A priori settings of model parameters were used for all vegetation types except for needle-leaf forests, for which model parameters have been tuned using field observations.
The behaviour under different environmental conditions is examined. It is concluded that in the new model synergetic relations exist between the environmental variables. which are not present in the empirical approach.
The new model is implemented in the ECMW land surface scheme. It is tested in offline mode with data from five surface field campaigns with different surface characteristics and vegetation types. In general the mean error and scatter have become smaller. The best results Were obtained for low vegetation surfaces.
Furthermore the new plant physiological approach is tested in the 3D Regional Atmospheric Climate Model RACMO where it is compared with the empirical Jarvis-Stewart approach.
One important aspect is that the evaporation has increased. Three subdomains were analysed separately, representing areas dominated by low vegetation, high vegetation, and needle-leat forests. Comparing the modelled 2m relative humidity and temperature with observations for these areas it is concluded that the bias has become larger, but the root mean square error has become smaller in the new approach. Again the best results were obtained for low vegetation surfaces.