地球信息科学学报 ›› 2023, Vol. 25 ›› Issue (10): 2093-2106.doi: 10.12082/dqxxkx.2023.220691

• 遥感科学与应用技术 • 上一篇    

基于GEE平台LandTrendr算法的海南岛森林扰动快速监测方法及分析

尹雄1,2,3(), 陈帮乾2,3, 古晓威2,3,6, 云挺4, 吴志祥2,3, 陈岳5, 赖虹燕2,3,6, 寇卫利1,*()   

  1. 1.西南林业大学大数据与智能工程学院,昆明 650224
    2.中国热带农业科学院橡胶研究所,海口 571101
    3.海南儋州热带农业生态系统国家野外科学观测研究站,儋州 571737
    4.南京林业大学林学院,南京 210037
    5.西南林业大学机械与交通学院,昆明 650224
    6.西南林业大学林学院,昆明 650224
  • 收稿日期:2022-09-14 修回日期:2022-11-26 出版日期:2023-10-25 发布日期:2023-09-22
  • 通讯作者: * 寇卫利(1979—),男,陕西西安人,博士,教授,主要从事林业信息化、遥感大数据、图像智能分析等研究。E-mail: kwl_eric@163.com
  • 作者简介:尹雄(1997—),男,云南昆明人,硕士生,主要从事林业遥感与信息技术研究。E-mail: yx1299556085@163.com
  • 基金资助:
    国家自然科学基金项目(42071418);国家自然科学基金项目(32260391);云南省教育厅科学研究项目(111722043/01117);海南省自然科学基金项目(422CXTD527);国家天然橡胶产业技术体系(CARS-33);江苏省自然科学基金项目(BK20221337)

Rapid Monitoring of Tropical Forest Disturbance in Hainan Island Based on GEE Platform and LandTrendr Algorithm

YIN Xiong1,2,3(), CHEN Bangqian2,3, GU Xiaowei2,3,6, YUN Ting4, WU Zhixiang2,3, CHEN Yue5, LAI Hongyan2,3,6, KOU Weili1,*()   

  1. 1. College of Big Data and Intelligence Engineering, Southwest Forestry University, Kunming 650224, China
    2. Rubber Research Institute, Chinese Academy of Tropical Agricultural Sciences, Haikou 571101, China
    3. Danzhou National Field Scientific Observation and Research Station for Tropical Agroecosystems, Danzhou 571737, China
    4. College of Forestry, Nanjing Forestry University, Nanjing 210037, China
    5. College of Mechanics and Transportation, Southwest Forestry University, Kunming 650224, China
    6. College of Forestry, Southwest Forestry University, Kunming 650224, China
  • Received:2022-09-14 Revised:2022-11-26 Online:2023-10-25 Published:2023-09-22
  • Contact: * KOU Weili, E-mail: kwl_eric@163.com
  • Supported by:
    National Natural Science Foundation of China(42071418);National Natural Science Foundation of China(32260391);Scientific Research Project of Yunnan Provincial Department of Education(111722043/01117);Natural Science Foundation of Hainan Province(422CXTD527);Earmarked Fund for Agricultural Basic Research System(CARS-33);Natural Science Foundation of Jiangsu Province(BK20221337)

摘要:

热带森林具有重要的经济价值和生态价值,快速准确地监测其扰动对促进热带森林的可持续发展有重要意义。为探究适用于热带森林扰动快速监测的方法,本研究基于1987年以来海南岛所有Landsat 5/7/8和Sentinel-2时间序列光学影像和野外调研数据,在谷歌地球引擎(Google Earth Engine, GEE)云平台利用LandTrendr算法快速监测海南岛1990—2020年的森林年度扰动时空分布特征,并结合7期天然橡胶种植历史分布图、林业政策和自然灾害等因素探讨森林扰动的驱动因子。研究表明:① 1990—2020年海南岛森林扰动总面积为2.53×103 km2(占2020年森林总面积的11.78%),集中分布在中部、北部和西北部地区,扰动面积最大的3个市县分别是儋州市、琼中县和白沙县;② 海拔300 m以下区域的森林扰动占83.40%,坡度25°以下的森林扰动占94.86%,高海拔地区森林保存完好,鲜有大面积的森林扰动发生;③ 2000—2010年森林扰动发生最频繁,其中2005年的扰动面积最大,2010年后森林扰动趋势明显减缓;④ 森林主要受橡胶种植、桉树发展和严重自然灾害(如台风与干旱)共同影响,其中因橡胶种植导致的森林扰动面积占全岛森林扰动总面积的43.48%。本研究建立的森林扰动快速监测方法和研制的海南岛热带森林扰动数据集可为森林监测研究及林业部门决策提供参考。

关键词: 海南岛, 森林扰动, 橡胶林, LandTrendr, 谷歌地球引擎(GEE), Landsat, Sentinel-2

Abstract:

Tropical forests have significant economic and ecological value, and timely and accurate monitoring of forest disturbance is critical to promoting their sustainable development. In this study, all Landsat 5/7/8 and Sentinel-2 optical images since 1987 and the LandTrendr algorithm were used to monitor tropical forest disturbance in Hainan Island. The dense time-series satellite images were quickly processed on the Google Earth Engine platform (GEE), and the spatiotemporal distribution characteristics of annual forest disturbances on Hainan Island over the past 30 years were identified. The main causes of forest disturbance were analyzed with the development of rubber plantations, changes in forestry policies, and several severe natural disasters. The results show that: 1) the total area of forest disturbance from 1990 to 2020 is 2.53×103 km2 (equivalent to 11.74% of the total forest area in 2020), and is mainly concentrated in the central, northern, and northwestern regions. The three regions with the largest forest disturbances are Danzhou City, Qiongzhong County, and Baisha County, respectively; 2) most forest disturbances occur at elevation below 300 m (83.40%) and slope less than 25° (94.86%), and forests at higher elevations are well preserved; 3) forest disturbance occurred more intensely between 2000 and 2010 and decreased significantly after 2010, with the largest affected area in 2005; 4) rapid development of rubber plantations (accounting for 43.48% of total forest disturbance area), changes in forestry policies (e.g., promotion of eucalyptus cultivation), and severe natural disasters (drought and hurricane) around 2005 are the main causes of forest disturbance. The rapid monitoring method of forest disturbance proposed in this study and long-term forest disturbance dataset provide a reference for forest monitoring research and forestry department decision making on Hainan Island.

Key words: Hainan Island, forest disturbance, rubber plantation, LandTrendr, Google Earth Engine, Landsat, Sentinel-2