[Dynamic change and prediction of vegetation cover in Shenzhen, China from 2000 to 2018]

With landsat-series multi-temporal image data, percentage of vegetation cover (PVC) was estimated by pixel dichotomy. The linear regression analysis and center of gravity migration methods were used to explore the characteristics of the spatiotemporal changes of vegetation cover in Shenzhen from 2000 to 2018. The CA-Markov model was combined to predict future land cover in Shenzhen. The results showed that the PVC in Shenzhen demonstrated obvious regional differentiation characteristics from 2000 to 2018. The eastern region occupied larger proportion than the wes-tern part, while the southern region was larger than the north part. This feature exhibited good consistency with regional topographic effect. The spatial migration characteristic of the center of gravity of PVC was from northwest to southeast, and then from southeast to northwest, with a migration rate of 551.2 m·a^-1. This process was closely related to urbanization in Shenzhen. The PVC in Shen-zhen tended to be generally improved from 2000-2018, with a improvement rate of 0.005·a^-1. The percentage of significantly improved and degraded PVC area was 30.8% and 12.8%, respectively. The CA-Markov method was used to predict the land cover/use pattern of Shenzhen in 2024 under two scenarios, theoretical scenario and natural scenario. There was no significant difference in proportion of the area of the land cover/use patterns obtained by the two kinds of prediction method, with the difference threshold being 0-1.2%. Compared with the data before 2018, the proportion of arbor forests and arable land converted into construction land in Shenzhen would be significantly reduced in 2024, whereas the contradiction between supply and demand would be still tense.