The Role of Land Cover as a Driving Force for Regional Climate Change
Roger A. Pielke, Sr.
Colorado State University
Fort Collins, Colorado
The Role of Land Cover as a Driving Force for Regional Climate Change
Roger A. Pielke, Sr.
Colorado State University
Fort Collins, Colorado
There is general agreement on the importance of land surface characteristics for microclimate. When a forest patch is clear cut, for example, daytime temperature near the surface in the clear cut area increases relative to the same height within the forest canopy. Up to 80 percent of Earth's land surface has now been modified by humans. Pielke discussed how such land use change affects surface properties, atmospheric structure, and thus climate, at a variety of scales.
Pielke reports that irrigated areas are 10 degrees C cooler than non-irrigated areas in Colorado. Effects from such a change extend upwards into the atmosphere. For example, more energy is now available for thunderstorm production than was previously the case. There is evidence that this irrigation effect is convecting cooler air into Rocky Mountain National Park creating changes in run off and species ranges. Are these trends due to land use change? There is not much pre-disturbance data so it is difficult to prove.
One data set that does exist is for south Florida. Data from 1900 through 1973 coincide with widespread fragmentation and other dramatic changes in the landscape. Limited observational data indicate that rainfall has decreased in the interim during the rainy season. No trend is seen at Key West, far away from the highly disturbed area and Fort Lauderdale shows a slight increase. There appears to be a spatial pattern to the change. Model results are consistent with the observed changes and suggest that decreased rainfall since 1900 may be due to land use changes. The area of maximum rainfall is smaller and total amount of rain has decreased. This demonstrates that climate is sensitive to landscape changes on this scale.
Land use change affects surface properties, atmospheric structure, and thus climate, at a variety of scales.
The importance of landscape on mesoscale and regional scale weather and climate is also seldom questioned. For example, O'Brien (1995) has documented how deforestation in part of Chiapas, Mexico has resulted in an altered climate from what occurs in the undisturbed forested region. Pielke et al. (1997), as illustrated in Figure 1.30, demonstrate the very significant role that land use has in generating thunderstorms. In Figure 1.30, identical initial and lateral boundary condition meteorology were used; the only difference was that in the bottom simulation, a short grass prairie was assumed, while for the top simulation, the current heterogeneous landscape was prescribed. The use of the current landscape in the model is a necessary condition for a realistic simulation of thunderstorm activity in this region.
Human modification of the landscape has affected rainfall, generation of clouds, and the formation of tornadoes and thunderstorms.
In another example of the effects of landscape on weather and climate, Pielke points to a burn map in a boreal forest in Canada in which 16 percent of the landscape is in a recently burned environment. The burned areas are darker, resulting in preferential development of thunderstorms nearby which in turn cause more fires, generating a positive feedback with climatic effects.
In the Great Plains and the Texas Panhandle a correlation appears between heavily irrigated areas and tornadoes. In western Kansas, where center pivot irrigation is common, a mesoscale regional atmosphere modeling system has been used to simulate a day when a tornado formed (see Figure 1.30). A model run using the current landscape is compared to a model run using the natural landscape with the same meteorological conditions. Results indicate that the tornado formation is related to the current landscape. The implication of such research is that human modification of the landscape has affected rainfall, generation of clouds, and the formation of tornadoes and thunderstorms. Land use changes may also affect climate at larger scales, as the landscape has been fragmented and the type of vegetation has changed from tall grass prairie to crop land in states like Kansas.
At the continental scale, Pielke reports that model simulations at 60-km grid spacing were used to compare U. S. climatic patterns under the conditions of estimated natural vegetation with the those of the current landscape in which huge changes in vegetation distributions are apparent. When this model is run for one month, forced with current conditions, changes at the scale of the United States are apparent. Did landscape changes cause such changes as the observed warming in the Great Plains, and changes in wind speed and relative humidity? Pielke says that sensitivity experiments indicate that weather and climate are indeed sensitive to the landscape at the scale of the U.S.
Even on a global scale, regional landscape changes, in the tropics in particular, alter climate thousands of miles away in the mid- and high-latitude polar jet flow. In a 1996 study by Chase et al., long-wave tropospheric wind flow was shown to be substantially altered when current, as contrasted with potential, leaf area index (LAI) was specified as a lower boundary condition in the NCAR CCM2 general circulation model (GCM). Regional precipitation patterns were also changed in the model over southeast Asia as a result of the model simulated change in LAI. It was the change in the GCM-modeled thunderstorm activity, that resulted from the LAI change, that teleconnected to the higher latitudes and changed the polar jet flow. The conclusion from these and other studies is that land use plays a significant role in local, regional and global climate.
Interaction
Atmosphere-land cover two-way interactions also occur. These interactions can be on the diurnal scale, on the seasonal scale, and on the multi-year time scale. On the seasonal time scale, prescribed LAI significantly affects the meteorological model simulation of temperature and precipitation. Correspondingly, the prescription of temperature, and particularly precipitation, dramatically affects a biogeochemical model simulation of LAI and root density.
Vegetation dynamics interact with climate and weather through a coupled nonlinear interaction.
Since both the meteorological and ecological models are strongly influenced by what are dependent variables in the other models, this feedback must be considered in any climate simulation. The conclusion, therefore, is that vegetation dynamics interact with climate and weather through a coupled nonlinear interaction.
Predictability
In any nonlinear system, the time period of predictability is dependent on the degree of nonlinearity and the level of accuracy with which the feedbacks within the system can be represented. In the context of the climate system, the temporal limits on climate prediction are determined by (i) our understanding and ability to represent quantitatively the interactions between each important aspect of the Earth's climate system; and (ii) the degree of nonlinearity of these interaction. In the context of regional and global climate prediction, these limits have not been determined. Moreover, as discussed above, global and regional scale climate effects cannot be considered independently, but interaction across scale must be considered.
Conclusions
In terms of global climate predictability, Pielke asserts that accurate forecasts of future global and regional weather regimes beyond this time period are unattainable. He thus suggests that a vulnerability/susceptibility approach to climate change (with the inclusion of other environmental stresses) should be adopted. Such a procedure avoids the riskier approach of assuming we can forecast the future climate and its interaction with other environmental factors.
A second conclusion is that the use of land surface weather records to detect regional (and therefore, global) climate change/climate variability must include the influence of land use change over time on the record. Deforestation, and agricultural and grazing changes, for instance, must be included. The use of surface weather records in heterogeneous regions also requires that an assessment of the representation footprint of what the weather record is measuring be determined. This assessment must necessarily include microclimate and mesoclimate effects.
References
Chase, T. N., R. A. Pielke, T. G. F. Kittel, R. Nemani, and S. W. Running, 1996. The sensitivity of a general circulation model to global changes in leaf area index. J. Geophysical Res., 101:7393-7408.
O'Brien, K. L. 1995. Deforestation and Climate Change in the Selva Lacandona of Chiapas, Mexico. Unpublished Ph.D. Dissertation, The Pennsylvania State University.
Pielke, R. A., T. J. Lee, J. H. Copeland, J. L. Eastman, C. L. Ziegler, and C. A. Finley, 1997. Use of USGS-provided data to improve weather and climate simulations. Ecological Applications, 7:3-21.
In any nonlinear system, the time period of predictability is dependent on the degree of nonlinearity and the level of accuracy with which the feedbacks within the system can be represented.
Forward to Next Section//Back to Table of Contents//AGCI Homepage//Comments: agcimail@agci.org