There is an excellent article on the role of landscape within the climate system. It is
Allard, J. and Carleton, A.M. (2010) Mesoscale Associations between Midwest Land Surface Properties and Convective Cloud Development in the Warm Season. Physical Geography, Vol. 31, 107-136.
The abstract reads [highlight added]
“The study of land surface–atmosphere interactions is vital to understanding climatic variations in the Earth’s planetary boundary layer, particularly given continual land cover modifications by humans on local to regional scales. An agriculturally important region ideal for the study of land surface–atmosphere interactions is the Midwest United States “Corn Belt.” To evaluate the mesoscale relationships between Corn Belt land surface properties and a key indicator of the surface–atmosphere feedback in humid climates—warm-season convective cloud development—conventional meteorological data, digital maps of land surface properties, and satellite data were examined in a GIS framework for the May–September periods of 1991 through 1999. The results indicate associations between the surface and atmospheric moisture content and the formation of convective clouds: cumulus clouds initiate first and persist longer over a dry (moist) surface with an initially dry (moist) atmosphere. These relationships are evident when forcing from the free atmosphere is either extremely weak (i.e., when fronts and other disturbances are absent and 500 mb winds are 18 ms–1). However, the association between convective cloud development and other land surface properties (e.g., land use–land cover [LULC], soil order, elevation, and slope) is not consistent spatially. We find that a surface moisture–convective cloud relationship dominates Corn Belt land surface–atmosphere interactions across a range of barotropic synoptic conditions under different atmospheric moisture contents. The study results can help lead to improved prediction of convective cloud formation, and more realistic modeling of land surface–atmospheric interactions for weather and climate forecasting.”
The conclusion includes the text
“The results of this study ultimately could help lead to more realistic modeling of land surface–atmosphere interactions. Knowledge of the locations of climatically important land surface boundaries, including the relative proportions of cropland to forest at mesoscales, could help improve forecasts of deep convection for the Corn Belt during the warm season using mesoscale models. These improvements conceivably could come about through emphasizing the mesoscale land surface–atmosphere interaction important for deep convection, and would complement forecasting using model-predicted, synoptic-scale atmospheric dynamic and thermodynamic fields. Moreover, our findings are significant because they provide further observation-based knowledge of the feedbacks between the Earth’s surface and atmosphere in middle-latitude locations that have seen significant human modifications to the landscape (e.g., deforestation for intensive agriculture). This type of information is crucial as climate continues to change on regional and global scales: land-cover modifications may effect regional and local climate changes comparable to those driven by increased global emissions of greenhouse gases (Pielke et al., 2002). Future empirical and modeling research should continue to examine multiple synoptic flow types for other subregions within the Midwest at synoptic and mesoscales to afford further insights into the mechanisms of warm-season climate variations for the Midwest United States Corn Belt.”