Deriving long-term continous 30-m resolution surface albedo over Greenland based on Landsat data and preliminary accuracy evaluation

Surface albedo is one of the important climate variables that plays a key role in regulating the absorption of solar radiation available for melting over the Greenland Ice Sheet (GrIS). There is an increasing need for albedo data to be available for use in applications that require a medium to fine spatial resolution, such as GrIS melting changes. The different sensor characteristics and the active lifespan of the associated satellite result in the inherent limitations of the existing satellite albedo products, namely, coarse spatial resolution or short effective data time span. In earlier studies, the direct estimation method was applied to Landsat data for albedo estimations, which has been widely demonstrated to have high accuracy. However, due to the data collection frequency (16-day), cloud coverage (about 57% missing data caused by cloud coverage), and instrument failure (approximately 22% of the normal scene area missing in an SLC-off image), continuous data gaps often occur in Landsat observations. This paper proposed a spatiotemporal shape matching method to generate spatiotemporally continuous surface albedo data at a spatial resolution of 30m for snow surfaces. Validation results show that this algorithm can effectively fill the data gaps, in addition, it can be extended to the pre-MODIS era. Almost all the missing data resulting from cloud cover and instrument failure can be filled. And the temporal resolution of the fused Landsat albedo is 8-day. The fused Landsat albedo was demonstrated good accuracy with an R2 of greater than 0.8 and a RMSE of less than 0.075 against ground observations, which is comparable to the accuracy of the MODIS albedo product. By analyzing long-term albedo changes on the pixel level over the automatic weather stations operated by the Programme for Monitoring of the Greenland Ice Sheet (PROMICE), we found that the albedo of most sites was abnormally low in 1995, 1998, 2010, 2012, and 2019, indicating that large-scale GrIS melting occurred in these years. In addition, the variations in surface albedo over PROMICE stations exhibit geographic variability.