Difference and consistency of multi-sources land cover products in global large river estuaries

Abstract

Many land cover products with different resolutions and types have been produced by researchers, which provide important information on changes in the terrestrial surface. However, multi-source land cover products around the world differ greatly at the coastal regional scale. Information on coastal land cover remains very limited for the large river estuaries around the world, which are one of the important ecosystems on Earth. There is an urgent need to review land cover products to provide a reference for understanding the exquisite evolutionary process and resource management of the delta ecosystems. Here, we evaluated the consistency and accuracy of the current mainstream land cover products through area deviation, area correlation, pixel superposition, and error matrix. We found that 1) the comprehensive land cover products had basically the same land cover type composition and distribution features in estuaries, but the area of different types in vary products vary to a degree; 2) the correlation coefficient of the comprehensive land cover products was 0.64–0.99, with the highest correlation being between GLC_FCS30 and GlobeLand30 and the lowest correlation being between Esri_Land_Cover and GlobeLand30; 3) among the comprehensive products, land cover types are inconsistent in their spatial distribution and wetlands were the most inconsistent, followed by bare land, vegetation, impervious area, cropland, and water body. Among the special products, tidal flats were the most inconsistent, followed by impervious areas and water bodies; 4) the special land cover products were highly consistent at the pixel-level in the delta regions, and the water body of special products were more consistent than impervious area and tidal flats; 5) the wetland classification accuracy of the comprehensive products in the delta areas must be improved and the overall accuracy of the Esri_Land_Cover was the highest. GLC_FCS30, the Global Surface Water dataset, and the Global Tidal Wetland Change dataset had higher accuracy than the special types, including for impervious areas, water bodies, and tidal flats. Our findings can help users choose the appropriate land cover data for their purpose, and supports the sustainable management of estuaries.