Climate Land Interaction Project (CLIP) - East Africa
| Home | Research | People | Publications | Login



overview | systems | climate | land use | land cover    Matrix of Experiments

Climate Modeling

The regional climate model that we are using is based on the Regional Atmospheric Modeling System (RAMS, Pielke et al. 1992). Lofgren (2000) has refined the version used here under the name Coupled Hydrosphere-Atmosphere Research Model (CHARM) in order to make it more useful for simulations over long time periods. Some additional adjustment may need to be made to make it better suited to the needs of the current project. In particular, the convective precipitation scheme may benefit from some adjustment in order to work more satisfactorily in a tropical setting. The input data that are required to drive CHARM include lower boundary conditions and lateral boundary conditions. The lower boundary conditions that are required are specification of land and water fractions within each grid square, plus land cover type. Land surface parameterizations can be adapted to accommodate additional land cover types, multiple land cover types within one grid square, and seasonal variation of vegetation characteristics

As a proof of concept for using a regional climate model as a part of the overall study, a regional climate model based on RAMS was run for the first two months of 1993 for two experimental cases, with additional simulations currently pending. The two cases that have been run represent one case with an approximation of present-day land cover, albeit at a coarse resolution (vegetated case), and a case of extreme desertification, with all vegetation removed and only bare soil remaining (unvegetated case). The model was parameterized with 40-km grid spacing, while the vegetation in the vegetated case was specified on a 1-degree grid (approximately 110 km), but we plan to improve the resolution of the vegetation input. The bare soil prevents any withdrawal of moisture from the soil other than immediately at the surface, resulting in little evapotranspiration.

Although the direct effect is on the evapotranspiration, a more striking transformation occurs in the simulation sensible heat flux. The vegetated case during February 1993 shows a pattern in which the sensible heat flux (Fig. 1a) is high (much heat is transferred from the surface to the atmosphere) where the surface is specified as grassland, especially toward the eastern edge (upwind in prevailing conditions) of those zones, and lower in the savanna regions. This is complemented by the rates of evapotranspiration or latent heat flux (not shown), which are higher in the savanna regions than in the grasslands. The unvegetated case shows a very different pattern (Fig. 1b), with a land breeze cell developing near the Indian Ocean coast and dissipating the clouds there. This results in high insolation near the coast and, combined with the lack of evapotranspiration from the land in this case, values of sensible heat flux are far higher than any simulated in the vegetated case.

Under a more reasonable land use/land cover transformation, the results could not be expected to be as dramatic as this, but localized alterations of energy fluxes are very likely to be a result. Changes in large-scale circulation cloud also result within the study area. The model domain could also be expanded, using multiple nesting at varying resolutions, to include most of Africa and the Indian Ocean, to enable such scenarios as deforestation of the Congo. Finer-scale grids can also be nested into the 40-km grid over limited regions.




RAMS simulation for current vegetation.

RAMS simulation for no vegetation.

RAMS simulations for current vegetation (Fig 1a)
and no vegetation (Fig 1b). The two month simulation
represented 1993 conditions.