The predictor-corrector method of computing screen temperatures, humidities and anemometer-height winds proposed by Hess et al. (1995) has been reconsidered in the light of a range of different methods of determining the fluxes and stresses in models of the land surface. We propose methods of computing effective roughness lengths and mean ‘surface’ values of temperature and moisture that are compatible with the computed stresses and fluxes at the scale of the grid box. These mean surface values can be used to generate appropriate grid-box values for screen temperatures, humidities and winds. Our method applies to the subgrid ‘tiles’ of individual surface types within a grid box (we include the possibility that a tile can be on bare or vegetated land, over the ocean or on sea-ice).
Hess et al. (Sun,) studied this question.