Response of changes in lake area to drought and land use change

• Published in: The Science of the total environment 2024-10, Vol.947, p.174638, Article 174638
• Main Authors: Luo, Yi, Chen, Rixiang, Yang, Kun, Zhou, Xiaolu, Jia, Tingfang, Shang, Chunxue, Pei, Xingfang, Wang, Qingqing, Li, Dingpu, Peng, Changqing, Guo, Hairui
• Format: Article
• Language: English
• Subjects: Deep learning; Drought; Lake area; Land use
• ISSN: 0048-9697; 1879-1026; 1879-1026
• DOI: 10.1016/j.scitotenv.2024.174638

The lake area is a crucial parameter that characterizes the state of a lake. Under the dual pressures of climate change and human activity, the magnitude and frequency of changes in lake areas become more pronounced. This process poses a serious threat to the local ecological environment. In this study, we constructed a lake water extraction model (LakeNet) based on a fully convolutional neural network. We extracted and analyzed the spatiotemporal characteristics of the area of nine major lakes from 1987 to 2022, as well as the driving factors behind these changes. Our results indicate that: 1) LakeNet exhibits high extraction accuracy and can remove some clouds. 2) The area of the nine major lakes shows a fluctuating downward trend (−8.11km2/10a), with drought and land use changes identified as significant driving forces behind the changes in lake boundaries, drought events caused the lake area to decrease, and the expansion of cropland further reduced the lake area. 3) Due to variations in lake area, the impact of drought on the area of the nine major lakes exhibits a lag effect, smaller lakes are likely to respond more quickly to drought. [Display omitted] •Proposed a lake boundary extraction algorithm based on fully convolutional neural networks.•Drought and changes in land use are the primary driving factors affecting the changes in lake boundaries.•The difference of lake area and geographical location leads to the difference of lake response to drought.