|
RAP Seminar Series
Mesoscale land-surface/atmosphere interactions
by
Fei Chen
Wednesday, 7 March 2001
|
|
The coupled MM5/LSM modeling system has been used for realtime mesogamma-scale numerical weather forecast at several test ranges of the U.S. Army Test and Evaluation Command. Some of these test ranges feature complex terrain and surface variability in landuse and soil texture, which have important influence on the local weather. It is important to describe the surface variability at small scales in the coupled modeling system. We will discuss the design of this coupled system for mesogamma-scale applications and soil moisture and temperature initialization procedure. The MM5/LSM forecasts were compared to MM5 forecasts using a 5-layer slab model (SLAB). The forecasted near-surface variables in LSM are slightly better than that in SLAB. Nevertheless, the surface temperature in both LSM and SLAB falls too fast around or right after the sunset for about two or three hours before it recovers. It is then apparent that the morning and afternoon transition of stable PBL is not well resolved in MM5. This causes a rapid change of temperature (ranging from 2-4 C) during a 2-3 hour period after such a transition. In another example, this coupled was used to simulate the 1996 Buffalo Creek Flash Flood event. Numerous factors contributed to the mesoscale modulation of the large-scale, unstable, upslope flow thereby determining the specific location and intensity of this storm. Apparently, the MM5/LSM model, together with detailed specification of surface characteristics (including time-varying soil moisture, can enhance the predictability of this type of flash flood. As compared to the S-POL radar rainfall analysis, this coupled model well captured this flash flood event at the right time and in approximately the correct location, while an overly simplified land-surface physics seriously underestimated the rainfall amount. Also, we will present a study of the surface heterogeneity for the Walnut River watershed, Kansas. Based on data collected during the CAESE97 field experiment, a multi-scale (1, 5, and 10 km) gridded atmospheric forcing and surface data were constructed. These data were used to force three land-surface models to generate surface heat flux maps, which were validated against surface and aircraft measurements. Simulating the rapid greening process of grassland, dominant landuse type in the CASES97 domain, is challenging and requires an accurate description of underlying vegetation phenological development and its effect on canopy evapotranspiration in models. Across the CASES97 domain, the variability in surface heat fluxes typically ranges from 50-150 W/m2, induced by the variability in land use, soil moisture and surface radiation forcing. This variability, reasonably captured by land-surface models, tends to decrease when the grassland reaches its peak growing stage. |
|
|