Abstract:

Urban waterlogging posed a serious threat to social security and development, and identifying the key influencing factors of urban waterlogging was the basis of studying waterlogging disasters. Global regression model (OLS), geographically weighted regression model (GWR) and multi-scale geographically weighted regression model (MGWR) were used to analyze and evaluate the correlation between waterlogging degree and land use type, topography, river density and other influencing factors in Wuhan in 2016. The results showed that after OLS screening, the influencing factors selected for GWR and MGWR analysis were cultivated land area, grassland area and impervious area. The comparison of model performance showed that MGWR was better than GWR and OLS. MGWR showed that the correlation between the influencing factors and the degree of waterlogging was spatially non-stationary, and the influence of different factors had spatial scale differences. Impervious area had the smallest spatial scale, and cultivated land area had the relatively small spatial scale. The grassland area and constant term had nearly global scale. Impermeable area positively affect the degree of waterlogging, while cultivated land area and grassland area negatively affect the degree of waterlogging. The impervious area was the most important factor affecting the degree of waterlogging, and the mean regression coefficient was 0.934. The regression coefficient of impervious area in the middle of Wuchang District and Hongshan District was the largest.

Key words: urban waterlogging, multi-scale geographically weighted regression analysis, impact factor, type of land use