A new land-surface treatment for HIRLAM - comparisons with NOPEX measurements

In order to improve the accuracy of forecasting near-surface atmospheric va riables over a heterogeneous landscape, a framework of subgrid surface type s and the ISBA parameterisation scheme for land surfaces have been tested i n the operational weather forecast model HIRLAM, using a 5.5 km grid resolu tion. Surface energy fluxes measured during a single summer day at six fixe d sites in the NOPEX area, representing agricultural fields, boreal forests and lakes, were used for verification. Both, in-situ field measurements and the HIRLAM simulation indicated that t he Bowen ratio over forests was about twice as large as that of adjacent ag ricultural fields. This difference could be explained by the more effective turbulent mixing and larger surface resistance associated with the forest, thus making the sensible heat flux relatively large there. The use of init ial soil moisture from a routine hydrological model gave improved agreement with measured surface fluxes and radiosonde temperature and humidity profi les compared to initialising from routine HIRLAM surface data. The differences in heat fluxes between the various surface types were also demonstrated by airborne flux measurements flown along a track at a height of ca. 100 m above the terrain. Modelled heat fluxes along the flight track were considerably smoothed due to the grid resolution used, e.g. the effec t of a lake in reducing grid-averaged sensible heat flux could only be weak ly detected, because the lake surface represented only 10% of the grid area . When the proportion of a contrasting surface type (lake) was altered from 10 to 100%, the surface fluxes calculated for the lake surface were almost unchanged; the results of the comparison did not provide evidence that mor e complex aggregation schemes for heat fluxes than straightforward area-wei ghted averaging would be required. The hourly variation of the modelled and simulated heat fluxes during the d ay studied could not be directly compared, because the simulated cloudiness did not exactly match that observed at the field sites. When the simulated net radiation was replaced with direct measurements, the model-based estim ates of sensible and latent heat fluxes were closer to the corresponding fi eld measurements. The divergence of sensible heat flux with height, as infe rred from the tower measurements made over the forest, were supported by th e aircraft measurements and the HIRLAM simulations. (C) 1999 Elsevier Scien ce B.V. All rights reserved.