Abstract:

Objective: Changsha, Dalian, Nanchang and Shenzhen were selected to detect the spatio-temporal patterns of greenspace changes and main impact factors.This study aims to provide the new spatial data for the improvement of urban greenspace theories and models, and for the optimization of greenspace network. Method: Based on the time-series Landsat composites imagess, NDVI-thresholding technique was applied to reclassify greenspace into four categories and landscape indices were used to measure the characteristics of greenspace patterns. Temporal dynamics of greenspace was assessed using the Mann-Kendall trend test. Normalized distance index (NDI) was calculated to measure the urban-to-rural gradients. Dynamic time warping (DTW) was applied to detect the differences of long-term trends trajectories of landscape indices in four cities. Finally, maximum information-based non-parametric exploration method (MINE) was used to detect the pairwise relationships between the changes in greenspace patterns and main impact factors. Result: From 1985 to 2011, the percentage of dense vegetation in Changsha, Dalian, and Shenzhen had an average increase of 0.26% and the mean patch size of dense vegetation increased 0.11 hm². The percentage of medium-dense vegetation in these three cities had an average decrease of 0.40%, while the mean patch size in Nanchang and Shenzhen increased 0.15, 0.04 hm² respectively, however, Changsha and Dalian decreased 0.01, 0.08 hm² respectively. The percentage and mean patch size of sparse vegetation in Changsha and Nanchang showed an increase trend. The percentage increased by 0.12% and 0.23%, the mean patch size increased by 0.04 and 0.01 hm², respectively. To the contrary, the percentage of sparse vegetation in Dalian and Shenzhen decreased 0.22% and 0.63% while the mean patch size decreased 0.04 and 0.06 hm².The urban-to-rural gradients of dense vegetation in Changsha and Nanchang had a ∩-shaped trajectory. Sparse vegetation in Changsha and Shenzhen had a monotonic decreasing trajectory, while Dalian, and Nanchang had a ∪-shaped trajectory. The changes in greenspace patterns had strong non-linear relationships with income, gross domestic product per person, proportion of urban population and proportion of city area. Conclusion: From 1985 to 2011, the dense vegetation in Changsha, Dalian and Shenzhen increased and the patches became more connected, contrasting to medium-dense vegetations. In the four selected cities, sparse vegetation increased in inland cities and the patterns became less fragmented, contrasting to the coastal cities. There are three trajectories of urban-to-rural gradients of greenspace, the first one is parabola-shaped trajectory ("∩or∪"), second is constant trajectory ("-"), third is monotonic increasing or decreasing trajectory. The greenspaces tended to have large changes in the city centers and suburb areas.

Key words: urban greenspace, landscape patterns, urban-to-rural gradient, spatio-temporal dynamics