Characterization of land surface water flux under vegetation greening introduced by changes in climate and land-use

Detailed changes in land surface water fluxes under vegetation greening is unclear especially in different patterns of climate change and land-use change. A typical vegetation greening region located in Loess Plateau was selected as a studying area. Because of high spatial heterogeneous site conditions, similar spectral reflectance and shapes of different vegetation types, there is a low accuracy of land cover mapping in mixed regions of multiple vegetation types, thereby leading to a pronounced underestimate of land use change, such as grain for green project. Besides spectrum, topography, and some usually used features, a novel land cover mapping framework is constructed with evapotranspiration which vary dramatically in vegetation types. Generally, the classification accuracy of all kinds of land cover is above 90%, and improved by 5.4-15.3%, 0-15.7%, 3.0-20.4%, of cropland, forest and grassland, respectively. Then water fluxes including precipitation, evapotranspiration, soil moisture, and runoff were analyzed in nine different vegetation types, considering the three types of vegetation found in cropland, forest and grassland along with respective stable, loss, and gain states. The result indicated that the cropland returning and afforestation has successfully facilitated a positive regional water cycle. This finding is useful for supporting ecological restoration and future water resources management, and enhancing the carrying capacity and resilience of the region.