Abstract: Aiming at the situation that rapid urbanization had weakened river network connectivity, taking Pudong New Area as an example, and applying gridding method of ArcGIS software, this paper analyzes the temporal and special variations of land use and river network connectivity from 1989 to 2010, and detects the indexes of urban landscape pattern that affected river network connectivity using spatial autoregression model. The results are: ① From 1989 to 2010, agricultural land and water area of Pudong New Area were mainly transformed to builtup land. The average values of β dropped from 0.80 to 0.72 while that of γ declined from 0.30 to 0.25. And there existed notable spatial differences between the northeastern and western regions both in land use and river network connectivity. ② The model shows that number of patches (NP) for public building land and mean of patches area (AREA_MN) for water area are the most significant factors that influence river network connectivity. Besides, mean of patches area (AREA_MN) for road and traffic land, largest patch index (LPI) for industrial land, and interspersion juxtaposition index (IJI) of industrial land, residential land and water area also affect the change of river network connectivity.

Key words: river network connectivity, urbanization, landscape pattern change, spatial auto regression, Pudong New Area