The radiative effect of changing human land-use patterns on the climate of the past 300 years is discussed through analysis of a series of equilibrium and transient climate simulations using the UVic Earth System Climate Model. Land-surface changes are prescribed through varying land cover type, representing the replacement of natural vegetation by human agricultural systems from 1700 to 1992. All land cover simulations show a cooling in the range of 0.09 to 0.22°C with larger regional changes caused by local positive feedbacks.
If you can't find a tool you're looking for, please click the link at the top of the page to "Go to old article view". Alternatively, view our Knowledge Base articles for additional help. Your feedback is important to us, so please let us know if you have comments or ideas for improvement.
 In the study of anthropogenic climate change, there is a growing interest in the ways that human activities interact to affect the climate system. In particular, human land-use over the last several centuries, through large scale changes to the global vegetation cover, has had a potentially significant effect on global climate. These land cover changes have recently been quantified in global datasets that provide estimates of historical land cover change over the past three centuries [Klein Goldewijk, 2001; Ramankutty and Foley, 1999]. These datasets now allow for historically realistic analyses of land cover change using general circulation climate models.
 Land cover change can influence global climate in a number of ways, by altering parameters such as leaf area index, root distribution, soil water holding capacity, surface resistance to evaporation, fractional vegetation cover, surface albedo, and fluxes and storage of carbon and other nutrients [see for example Betts, 2001; Bolin et al., 2000]. As many of these changes have quite complex effects, this paper focuses on the effects of two key processes, whose mechanisms are conceptually simple. First, as natural vegetation is replaced by cropland, surface albedo generally increases, as trees are replaced by agricultural crops. Second, the energy available for evaporation is affected and the plant transpiration pathway is altered as a result of a modified vegetation cover.
 This paper seeks to quantify the global temperature change resulting from the radiative forcing associated with historical land cover change by incorporating a global croplands dataset into an intermediate complexity climate model. Results from a series of equilibrium and transient climate runs are presented so as to compare the climate response to land cover change over the past 300 years with the responses to greenhouse gas and solar orbital forcing. This work uses the UVic Earth System Model [Weaver et al., 2001], which in contrast to previous modelling studies, includes a full ocean general circulation model (GCM) and a thermodynamic/dynamic sea-ice model. As such, this study allows for an estimate of the radiative impact of historical land cover change on global climate, including feedbacks associated with ocean circulation and changing sea-ice distribution.
 The UVic Earth System Model [Weaver et al., 2001] is an intermediate complexity climate model comprising a full ocean GCM (version 2.2 of the GFDL Modular Ocean Model) coupled to a vertically integrated energy/moisture balance atmospheric model and a dynamic/thermodynamic sea-ice model. Surface wind stress as well as vertically-integrated atmospheric winds used for advection of moisture are specified from NCEP reanalysis data. A dynamic wind feedback parameterisation allows for wind perturbations to be applied when simulating past climates. The coupled ocean-atmosphere model has a resolution of 3.6° in longitude and 1.8° in latitude, and conserves both energy and water without the use of flux adjustments [Weaver et al., 2001].
 In this work, the UVic Model is further coupled to a simple land surface model. The model used is a version of Manabe's  simple bucket model, modified to allow for on-line calculation of the aerodynamic resistance to evaporation; the Dalton number for evaporation over land is calculated from a spatially variable surface roughness length, following the methodology of [Brutsaert, 1982]. A single dominant vegetation type is specified at each grid cell, using seven vegetation types based on the vegetation field of DeFries and Townsend . The specified vegetation type determines the surface albedo, as well as the roughness length used in the calculation of the surface resistance to evaporation. These values (shown in Table 1) are obtained by spatially averaging the surface albedo and roughness length fields provided in Sellers et al.  according to the chosen vegetation type categories. The presence of snow or sea-ice increases the surface albedo locally by 0.18.
 Our model results are from three 2000 year equilibrium runs and four transient model runs covering the period from 1700 to 1992. The croplands dataset used is that of Ramankutty and Foley , which provides a potential or natural vegetation field (representing vegetation distributions in the absence of human disturbance) and fractional cropland areas at each grid cell yearly from 1700 to 1992. As Ramankutty and Foley  consider only historical changes in cropland area, and not other land cover changes such as pasture and extensively logged areas [see also Klein Goldewijk, 2001], this dataset could be expected to generate a lower bound on the estimate of the radiative effect of land cover changes.
 The three equilibrium runs presented here are differentiated by the specified vegetation fields. Equilibrium (1) uses present day vegetation as determined from satellite data [DeFries and Townsend, 1994]. Equilibria (2) and (3) use the fractional cropland areas of Ramankutty and Foley  corresponding to years 1992 and 1700, respectively (as shown in Figure 1), superimposed onto the potential vegetation field. All three equilibria use present day CO2 concentration (365 ppm) and orbital parameters. As the croplands dataset used here does not include all land cover changes that have occurred since 1700, it would be expected that equilibrium 1 (present day vegetation) will show a stronger land cover signal than equilibrium 2 (1992 land cover). Differences in the definition and placement of vegetation types between the two vegetation fields does have the potential to have a confounding influence on the land cover signal in equilibrium (1), but this is nevertheless a useful experiment to estimate the effect of land cover changes not captured in the Ramankutty and Foley  dataset.
 Transient climate simulations begin from a control equilibrium climate defined by year 1700 land cover (as in equilibrium 3 above), pre-industrial CO2 (280 ppm) and year 1700 orbital parameters. This control equilibrium is spun up for 1700 years, and transient simulations are run from year 1700 to present. The three transient runs are forced by: (1) changing croplands (land cover) only; (2) changing CO2 (as represented by the observed increase to present-day levels beginning in 1850); and (3) changing land cover and atmospheric CO2. A control experiment (4) is also conducted with changing orbital parameters, but not other radiative forcings. In runs 1 and 3, croplands are specified yearly from 1700 to 1992, and surface albedos and roughness lengths are changed according to the fractional cropland area in each grid cell.
3.1. Equilibrium Results
 Results of the equilibrium model runs are presented as differences: (a) present day vegetation equilibrium minus year 1700 land cover equilibrium; and (b) year 1992 land cover equilibrium minus year 1700 land cover equilibrium. Negative differences thus represent a decrease resulting from changes in land cover between the year 1700 and the present day.
Figure 2 shows zonally averaged temperature, precipitation and evaporation differences for comparisons (a) and (b). Both comparisons show a cooling and a decrease in precipitation at all latitudes. Comparison (a) shows a globally averaged cooling of 0.22°C and an average precipitation decrease of 14.2 mm/year. While the spatial pattern is similar, the magnitude of the change shown for comparison (b) is less: a cooling of 0.10°C and precipitation decrease of 3.3 mm/year. As can be seen in the shape of the temperature plot, cooling is amplified in the Northern Hemisphere, consistent with the location of the majority of the specified land cover changes (see Figure 1). There is also an amplification at around 70°S, resulting from local sea-ice albedo feedbacks.
 Much of the decrease in precipitation can be attributed to atmospheric temperature change. However, the large decreases seen in the tropics and sub-tropics can be further explained by examining the roughness lengths listed in Table 1. According to the parameterisation of evaporation in the bucket model, a larger roughness length results in a smaller aerodymanic resistance to evaporation and hence a larger Dalton number and (given equal moisture availability) more evaporation. By changing roughness length alone, a much larger decrease in evaporation would be expected as a result of a conversion from forest to grassland in the tropics than by a similar change in the temperate/boreal regions. As such, the greatest evaporation (and hence precipitation) differences are seen in the tropics where roughness length changes are largest, with a less pronounced decrease seen in the northern mid-latitudes.
 Spatially-varying temperature differences are shown in Figure 3. While the majority of the temperature change can be seen in the Northern Hemisphere, the largest local changes are the result of local snow and sea-ice albedo feedbacks. Particularly in the case of comparison (a), these positive feedbacks serve to amplify the cooling signal initiated by land cover change. Comparison (b) shows less cooling, but as in the case of the zonally averaged results, the spatial pattern of cooling is similar to (a). As the model used here does not contain atmospheric variability [Weaver et al., 2001], the differences shown here represent significant changes in the climate mean state.
3.2. Transient Results
 Transient runs were chosen so as to compare the importance of land cover forcing of climate to greenhouse gas forcing. The results of these runs are shown in Figure 4. The transient effect of land cover change is −0.09°C, quite close to that found in comparison (b) in the equilibrium results (−0.10°C). This indicates that there is little evidence of an oceanic cooling commitment associated with land cover change, as the majority of the temperature change seen in the equilibrium comparison is also seen in the transient run. While the land cover effect on global temperature is smaller than that of CO2 (+0.76°C), it is of significant magnitude, and has a noticeable dampening effect on the total warming when both forcings are combined (+0.68°C). It is also clear that land cover change is much larger than any model variability or orbitally induced response seen in the control run (+0.003°C).
 Transient changes in global temperature, evaporation, precipitation, and short-wave radiation for the four runs (as well as a fifth run described below) are summarized in Table 2. The globally averaged short-wave radiative forcing resulting from land cover change is −0.15 W/m2, slightly less than, but within the error bounds of that found by Hansen et al. .
Table 2. Total Changes in Temperature, Evaporation, Precipitation and Downward Short-Wave Radiation Absorbed at the Surface
Transient Model Run
1. Land Cover (LC) only
2. CO2 only
3. LC + CO2
5. LC (crop albedo = 0.2)
3.3. Model Sensitivity
 As shown in Table 1, the albedo value assigned to croplands in this model is 0.17. As a globally and temporally averaged albedo value for grassland, savanna and cropland vegetation types, this value reproduces well the satellite albedo data provided by Sellers et al. . Previous studies of vegetation albedo, however, have placed the albedo for agricultural land in the range of 0.12 to 0.25, depending on the crop type, with only rice having a crop albedo of less than 0.17 [Wilson and Henderson-Sellers, 1985]. It is not clear that 0.17 is a good representative albedo for a pure cropland vegetation type. In order to address this concern, a second land cover transient run was performed using a cropland albedo value of 0.2 for all fractional cropland areas applied onto the background vegetation field. This change almost doubles the albedo forcing associated with land cover change, as not only do forest grid cells now see a 50 to 75% larger albedo increase, but the albedo of grassland, savanna and shrubland grid cells is also increased from 0.17 to 0.2 in those areas where croplands are applied.
 As can be seen in the last line of Table 2, a higher specified cropland albedo value has a large impact on the total temperature change for the transient run. Instead of a cooling of 0.09°C as in the case of the original run, this second land cover transient run shows a much larger cooling of 0.17°C. Based on this result, it is clear that the specified cropland albedo is of primary importance for determining the magnitude of the global temperature change forced by land cover change.
4. Conclusions and Future Directions
 This study demonstrates that historical land cover change has had a significant radiative effect on global climate over the past three centuries. All model runs show a cooling associated with land cover change, both in global averages, and as amplified locally by positive feedbacks. In the equilibrium runs, global cooling was in the range of 0.10°C to 0.22°C, depending on which vegetation field was used to represent “present day” vegetation. In the transient case, the cooling was in the range of 0.09°C to 0.17°C, depending on the specified albedo value for cropland. From this analysis, it is clear that uncertainties in surface albedo values can have a large impact on the magnitude of the modelled land cover effect. It is also evident from comparing the land cover only transient run (0.09°C cooling) with equilibrium runs (2) and (3) (0.10°C cooling) that the effect of land cover change does not carry with it a large oceanic cooling commitment. This is in contrast to the effect of greenhouse gas forcing where the large thermal memory of the ocean has resulted in differences as large as 0.61°C in the UVic model between transient and equilibrium simulations of the present day climate [Weaver et al., 2001].
 This work, and others in this area of research, raises important questions about the future of human-induced climate change over the next several centuries. In addition to the physical changes addressed in this paper, land cover change has a large impact on the climate system through the release of stored carbon reserves to the atmosphere [Bolin et al., 2000]. In the transient climate runs presented here, the terrestrial carbon cycle was not treated as an interactive component of the climate system, and it was assumed that releases of greenhouse gases due to land cover change were captured by the specified atmospheric CO2. Much research remains to be done to determine the net effect of land cover change on climate. In a simulation of Northern Hemisphere afforestation, for example, [Betts, 2000] found that in some areas of boreal forest, the warming effect of albedo change was equivalent to the cooling effect of carbon sequestration, resulting in a neutral total forcing. This question has yet to be addressed using realistic scenarios of transient land cover change.
 A related question is that of the effect of land cover change on the future of the terrestrial carbon sink. In many instances, land-use change reduces the total ability of the earth's terrestrial ecosystems to draw carbon from the atmosphere [Bolin et al., 2000]. How will continued land cover change affect the capacity of the terrestrial carbon sink to respond to and mitigate future climate change? Questions such as these must remain a focus of research if we are to continue to improve the accuracy of climate predictions for the next century.
 The authors wish to thank N. Ramankutty and J. Foley for making available the historical land cover dataset used in this study, as well as M. Claussen, V. Brovkin and two anonymous reviewers for useful comments and suggestions. Financial support for this research from NSERC and CFCAS is gratefully acknowledged.