Causes of tea land dynamics in Sri Lanka between 1995 and 2030

REGIONAL ENVIRONMENTAL CHANGE(2023)

引用 0|浏览2
暂无评分
摘要
It is vital to investigate changes in tea land use patterns in key tea-growing regions, which are among the most vulnerable to climate change. The objective of this research was to assess the spatiotemporal dynamics of tea land in the main tea-growing areas in Sri Lanka such as Badulla, Nuwara Eliya, and Ratnapura over a 25-year period from 1995 to 2020 using Landsat satellite images. We estimated tea land changes in five time windows using change detection maps, assessed major driving forces’ impact (temperature, rainfall, elevation, slope, population) on tea occurrences using a random forest (RF) algorithm coupled with a model-agnostic approach, and predicted tea land cover for 2030 using cellular automata and artificial neural networks (CA-ANNs) algorithms and independent variables. The results revealed an overall accuracy of over 83% from 1995 to 2020. Tea-growing areas increased by 13.1% in Nuwara Eliya and 19.3% in Ratnapura, while decreasing by 16.6% in Badulla between 1995 and 2020. The Badulla, Nuwara Eliya, and Ratnapura districts exhibited the greatest gains between 2010 and 2015, with gains of 36.7%, 36%, and 19.4%, respectively. As per the RF analyses, elevation (0.4) was the most important variable, followed by rainfall (0.32), and population (0.24). The model suggests that the selected variables impact tea land use and land cover differently. Tea lands are projected to shrink further by 2030, covering approximately 105 km 2 in the study area. The findings provide valuable insights to stakeholders and policymakers, ensuring a sustainable tea sector in Sri Lanka in the future.
更多
查看译文
关键词
CA-ANN,GIS,Land use/land cover,Random forest,Remote sensing,Tea
AI 理解论文
溯源树
样例
生成溯源树,研究论文发展脉络
Chat Paper
正在生成论文摘要