Land Cover Cloud Analytics: from Global Services to Regional Insights

Land Cover (LC) analyses and quantitiative as well as multi-thematic balances of (land use and) land cover are well established steps when identifying biogeoclimatic zones, estimating the potentials for human uses or habitat suitability, explore climate change impacts over time or dig deeper into the extent and co-location of specific categories like desert, forest, glaciers etc with determining factors in topography, climate or human impacts. While there are innumerable examples of land cover analysis in a range of projects at local scales covering catchments, smaller administrative districts or planning regions, a ‘big picture’ approach exploring national to global scales typically was constrained by the lack of easy access global data sets at high spatial resolution, and the resulting computation load hardly manageable on personal workstations. The recent availability of a variety of land cover services based on full and regular remote sensing coverage with automatic extraction of LC through deep learning approaches, in combination with geospatial cloud computing facilities enable researchers to leverage native (sensor) resolution analysis without the hassle of data download, preparation and local computational loads, as first implemented in Google Earth Engine. This paper supports this point by demonstrating LC analysis against topographic variables for the entire country of Kyrgyzstan. This kind of insights will lead to a better understanding of spatiotemporal LC dynamics and inform policy decisions from national to global levels.