Land cover-convective cloud interactions in the US Midwest

Warm season (June-August) land surface-atmosphere interactions for the period 1981-1995 in the Midwest USA are studied using satellite radiance information on vegetation activity (AVHRR-derived NDVI at 8km and 1km resolutions) and convective clouds from GOES satellite imagery (IR and VIS), and conventional land cover (USGS LULC digital maps). This period (1981-1995) was characterized by a wide range of summer climate conditions including extensive droughts (e.g., 1988) and major floods (1993). The Midwest Corn Belt is studied because of the relative lack of topographic influences on cloud development. Digital land cover maps and station meteorological data on temperature, precipitation, surface and atmospheric moisture, are used to calibrate the satellite retrievals and analysis methods.

Results obtained for periods having weak synoptic scale atmospheric forcing (no significant upper-level ageostrophic divergence or low-level baroclinicity) indicate the existence of mid-summer mesoscale land surface-convective cloud associations over a range of scales and for significant portions of the study area. Free convective cloud masses tend to be differentiated according to broad classes of land cover, particularly between deciduous forests and crop lands. The length scale dependency of the cloud-generating processes for cropped and natural growth regions is assessed using fractal analysis. The objective here is to test the stability of the land surface-convective cloud associations over a range of climate conditions. Results support the hypothesis that there exists an effective range of spatial scales within which the land surface climate signal (NDVI-convective cloud association) is expressed.