Fusing existing land cover products to produce an improved spatio-temporal record

Abstract:

The natural and anthropogenic cover on the surface of the Earth, more commonly referred to as land cover (LC), is a critical variable driving many environmental processes, including biogeochemical cycling, hydrologic processes, soil erosion, and ecosystems productivity. Thus, the assessment, monitoring, and characterization of LC is essential for global-change research. The previous tasks have become primary applications of remote sensing, resulting in increased availability of LC products. Although those products typically are spatially continuous, they rarely provide the temporal frequency and/or total duration required for environmental studies. Limited research has focused on the assimilation and synthesis of existing LC datasets using statistical interpolation techniques. In this study, we will integrate the two lines of research, one on space-time interpolation of a single LC product and one on fusing multiple existing LC products, to create a methodology for fusing multiple existing LC products over space and time to produce a single high-frequency LC record over a longer time period and a large spatial domain. LC maps of various spatial and temporal resolutions will be constructed, which can be used in a variety of applications. The proposed methodology is demonstrated by application to a region of the conterminous U.S. that comprises the Rocky Mountains (RM). The RM region is one of the most important mountain regions in the world because it has a considerably high ratio of water demand with respect to water availability, has a high ecosystem diversity, and is a global sink of atmospheric carbon. The LC products available for this case study and that are detailed enough spatially to be applied in environmental studies at a regional scale are fused using the proposed methodology. The fused products include the NLCD, Collection 5 MODIS Global LC Type, Historical LC Data Sets of the USGS, and LANDFIRE Existing Vegetation Cover.