Abstract: Soil organic carbon density (SOCD) is an important index to evaluate the carbon sink function of soil. Studying vegetation SOCD is of great significance for exploring the impacts of human activities on soil carbon cycle in urban areas. Based on the SOCD data of 206 sampling points (0-20 cm) and Sentinel-2 remote sensing images in the main urban area of Xi’an, we established the optimal SOCD estimation models of trees, shrubs, and grasslands by using multiple linear, quadratic fitting, and exponential models. The vegetation SOCD in the study area was estimated based on the optimal models, the spatial distribution of vegetation SOCD were further analyzed. Results showed that the SOCD of the urban vegetation ranged from 0 to 16.47 kg·m^-2, with an average of 3.24 kg·m^-2. Overall, the SOCD showed a spatial distribution characteristic of “high on the edge and low in the middle”, with high spatial heterogeneity. There were large differences in SOCD among different vegetation types. The average SOCD of the three vegetation types followed a pattern of trees (3.75 kg·m^-2) > shrubs (2.72 kg·m^-2) > grasslands (2.04 kg·m^-2). There was a negative correlation between urban vegetation SOCD and urbanization intensity (P<0.001), and the average SOCD was higher in areas with higher vegetation coverage. In general, urbanization had a negative impact on vegetation SOCD. Increasing urban vegetation coverage is one of the effective means to improve SOCD. The combination of remote sensing technology and field survey methods can quickly and effectively estimate SOCD of trees, shrubs and grasslands in the study area, which provides a reference for soil carbon management in future urban development.

Key words: soil organic carbon density, Sentinel-2, Xi’an City, remote sensing estimation, regression model