As solar farms supply an increasing share of electricity, comprehending their overall impact on climateis becoming ever more critical. Solar farms typically alter the surface reflectance, as their reflectiveproperties differ from those of the original ground where they are installed. However, in life cycleassessments, the climatic effects of surface albedo changes resulting from the installation of solar farmsare not considered. Therefore, this research aims to determine the albedo-climate effect of solar farmsglobally. Since earlier studies did not account for solar farm degradation or the effective albedo, thesefactors will be included in this analysis.

In the first part of this research a model was developed to determine the surface albedo. The model relieson Sentinel-2 satellite imagery to obtain information on surface reflectance. The albedo is calculatedusing a linear combination of the available spectral bands. To validate this albedo estimation approach,radiative flux data from SURFRAD measurement station in the USA were employed. The results showthat the model produces albedo estimates with a RMSE of 0.032. A bias correction was developed thatdepends on the solar zenith angle at the time the Sentinel-2 image was acquired. The bias correctionwas theoretically derived from the data and the anticipated error introduced by the assumption that thesurface reflection is Lambertian. With this correction applied, the albedo estimation RMSE was reducedto 0.021. The findings highlighted how the Lambertian assumption affects the outcomes and confirmedthat applying a bias correction is appropriate for improving the accuracy of the albedo estimates.

The second model developed in this study quantifies how albedo is altered by the installation of solarfarms. The location of 500 solar farms were sourced from the Solar Asset Mapper dataset developedby Transition Zero. For the albedo change, the difference between the albedo of the PV area andthat of the surrounding area is used. Google Earth Engine was employed to simplify satellite imageprocessing and enhance computational capacity. The mean albedo difference observed was−0.0198.Seasonal variation in the PV region’s albedo was detected, indicating that the solar farm boundariesare incorrectly defined, causing the surrounding area’s seasonality to be introduced into the PV albedo. In addition, the results revealed problems associated with glare.

In the third model, the climatic impact of the albedo changes of solar farms was assessed. To remove low-quality measurements, for example those distorted by glare, the dataset was cleaned before conductingfurther analysis, resulting in a total of 157 farms. Here, the effective albedo of the PV area wascalculated, and both the absolute and effective changes in albedo were applied throughout the remainderof the model. On average an absolute albedo change of−0.0299 and effective albedo change of 0.0496was found for these 157 farms. The radiative forcing caused by the albedo change was calculated withthe use of a radiative kernel dataset. The radiative forcing associated with the solar farm’s avoidedemissions was estimated based on the local electricity carbon intensity. The net radiative forcing wasthen obtained by the sum of these contributions together with the radiative forcing from embodiedemissions. Based on this net radiative forcing, the carbon break-even time was determined for eachfarm. On average, a carbon break-even time of 7.79 yr was found when considering the absolute albedochange, whereas a much shorter average carbon break-even time of 0.41 yr was obtained for the effectivealbedo change. These results indicate that the effective albedo exerts a cooling, rather than warming,effect on the climate, thereby reducing the overall climate impact of solar farms.