资源科学 ›› 2021, Vol. 43 ›› Issue (12): 2403-2415.doi: 10.18402/resci.2021.12.04

• “澜沧江—湄公河流域农业资源与环境”专栏 • 上一篇    下一篇

2000—2019年缅甸南部橡胶林时空演变

李贺1(), 何志杰1,2, 黄翀1(), 刘庆生1, 刘高焕1, 张晨晨1   

  1. 1.中国科学院地理科学与资源研究所,资源与环境信息系统国家重点实验室,北京100101
    2.中国地质大学(北京),地球科学与资源学院,北京100083
  • 收稿日期:2021-05-19 修回日期:2021-11-02 出版日期:2021-12-25 发布日期:2022-02-25
  • 通讯作者: 黄翀,男,安徽六安人,博士,副研究员,主要从事生态遥感研究。E-mail: huangch@lreis.ac.cn
    黄翀,男,安徽六安人,博士,副研究员,主要从事生态遥感研究。E-mail: huangch@lreis.ac.cn
  • 作者简介:李贺,男,江苏徐州人,博士,副研究员,主要从事资源环境遥感研究。E-mail: lih@lreis.ac.cn
  • 基金资助:
    澜沧江—湄公河合作专项;国家自然科学基金项目(41801353);国家自然科学基金项目(41890854)

Spatiotemporal evolution of rubber forests in southern Myanmar during 2000-2019

LI He1(), HE Zhijie1,2, HUANG Chong1(), LIU Qingsheng1, LIU Gaohuan1, ZHANG Chenchen1   

  1. 1. State Key Laboratory of Resources and Environmental Information System, Institute of Geographic Sciences and Natural Resources Research, CAS, Beijing 100101, China
    2. School of Earth Sciences and Resources, China University of Geosciences, Beijing 100083, China
  • Received:2021-05-19 Revised:2021-11-02 Online:2021-12-25 Published:2022-02-25

摘要:

橡胶林是东南亚地区重要的经济林种和战略物资,及时、准确掌握橡胶林时空动态信息对于可持续的森林资源监测、管理和维护生态环境健康稳定具有重要意义。为深入探索缅甸孟邦橡胶林的时空动态过程,本文选取2000—2019年Landsat长时间序列遥感影像,利用连续土地覆被变化检测与分类(CCDC)算法,通过变化检测,得到研究区的各连续变化像元;然后,利用随机森林(RF)算法对变化像元进行分类得到橡胶林及相关地物分类结果;最后,在精度验证的基础上,探索分析橡胶林时空演变和造成的其他土地覆被类型变化。研究结果表明:①利用Landsat时序遥感数据,结合CCDC算法可以准确提取橡胶林及相关地物的时空分布,总体分类精度优于85%,F1分数大于0.80,其中橡胶林的分类精度优于80%。②2000—2019年缅甸孟邦橡胶林分布格局总体呈逐年扩张趋势,至2019年橡胶林面积由7.25万hm2增加至19.72万hm2,面积增加1.72倍。③从土地利用转换角度得出,橡胶林的扩张主要由天然林和耕地转换而来,20年来转换总面积为12.47万hm2;其中,天然林减少面积最大,为10.52万hm2,占总土地变化面积的84.36%。橡胶林扩张受社会经济因素的价格影响,20年来,橡胶林扩张造成的天然林和耕地变化均呈现先增大后减少趋势,与橡胶年平均出口价格波动速率基本吻合。相关结果可以为当地橡胶林持续监测与生态环境可持续发展提供决策支持。

关键词: 橡胶林, 缅甸, Landsat, 连续土地覆被变化检测与分类(CCDC), 随机森林(RF), 时空演变

Abstract:

Rubber forests are important economic forest species and strategic materials in Southeast Asia. Timely and accurate detection of the spatial and temporal dynamics of rubber forests is important for the scientific formulation of rubber forest plantation planning, promoting local economic development, and maintaining a healthy and stable ecological environment. For an in-depth exploration of the spatiotemporal dynamics of rubber forests in Mon State, Myanmar, this study selected Landsat long time series remote sensing images from 2000 to 2019 and used the continuous land cover change detection and classification (CCDC) algorithm to obtain each continuous change image pixel through change detection. Then, using the random forest (RF) classification algorithm, the classification results of rubber forests and related land cover types were obtained. At last, on the basis of accuracy verification, an analysis of the spatiotemporal expansion of rubber forests and their encroachment on other land cover types were conducted. The results show that: (1) The spatial and temporal distribution of rubber forests and related land cover types can be extracted accurately by the CCDC algorithm using time series Landsat data, with the overall classification accuracy higher than 85% and an F1 score greater than 80%, and the classification accuracy of rubber forests is higher than 80%. (2) Rubber forests in Mon State was expanding year by year, with the area of rubber forests increasing from 7.25×10 4 hm2 to 19.72×104 hm2 between 2000 and 2019, an increase of 1.72 times. (3) From the perspective of land cover, the expansion of rubber forests was mainly achieved by encroaching on natural forest and cropland. The encroachment area was 12.47×104 hm2 over the past 20 years; among this, the encroachment area of natural forest was the largest, at 10.52×104 hm2, accounting for 84.36% of the total encroached land area. The expansion of rubber forests is affected by social and economic factors such as price. In the past 20 years, the encroachment of rubber forests on both natural forest and cropland showed a trend of first increasing and then decreasing, which is basically consistent with the fluctuation rate of the average annual export price of rubber. These results may provide a decision support for rational rubber production planning and sustainable development of the ecological environment in Mon State, Myanmar.

Key words: rubber forests, Myanmar, Landsat, continuous land cover change detection and classification, random forest algorithm, spatiotemporal evolution