One task in BAHC is tO develop a physical understanding of the weather generator (WG). The core of the WG concept is scale interaction, represented by upscaling and downscaling, between the gridscale atmospheric fields and the subgrid-scale hydrological and ecosystem patterns. This paper considers scale interaction in terms of the convective (= latent plus sensible heat) and rain fluxes. Earth's surface and atmosphere are coupled through these fluxes. However, horizontal scale interaction in the free atmosphere is physically different from that at the surface. Upscaling atmospheric fluxes generates new sub-gridscale fluxes at the lower grid resolution, caused by the gridscale fluxes resolved at the higher grid resolution. Downscaling has to recover these secondary circulations, usually through mesoscale submodels. The same gridscale eddy effect is zero for surface fluxes because there is no cross-surface mass flux of air. The atmospheric fluxes on the various scales are quantified here with an atmospheric diagnostic model (DIAMOD) coupled to a surface flux model (SURFMOD). Two convectively active periods over Europe (one disturbed, one undisturbed) are considered. Upscaling from 100 to 1000 km horizontal resolution generates additional grid-scale eddy fluxes. However, these amount, in both cases, to just 10% of the diagnosed fluxes that are sub-gridscale at 100 km; this suggests that the additional gridscale eddy effect may be negligible in convective situations. We further apply a simple downscaling recipe for disturbed periods (scale the rain flux profiles with the observed surface rain) and for undisturbed periods (scale the convective flux profiles with the observed surface latent plus sensible heat flux). At the Earth's surface, we study the upscaling-downscaling mechanism, not with diagnosed, but with modeled latent and sensible heat fluxes. With a simplified version of SURFMOD, plus the probability density function of soil moisture, we reproduce the fatal impact of taking mean soil moisture for calculating mean evaporation: evaporation can be underestimated by 25% in dry situations and overestimated by 10% in moist situations. We demonstrate how to completely remove this bias. The technique presented here can be generalized to a wide class of deterministic and statistical models and offers a rational framework for the aggregation problem and the WG problem in general.
ASJC Scopus subject areas
- Water Science and Technology