Deng Ying, Shao Zhenfeng, Dang Chaoya, Huang Xiao, Wu Wenfu, Zhuang Qingwei, Ding Qing
State Key Laboratory of Information Engineering in Surveying, Mapping and Remote Sensing, Wuhan University, Wuhan 430079, China.
State Key Laboratory of Information Engineering in Surveying, Mapping and Remote Sensing, Wuhan University, Wuhan 430079, China.
Sci Total Environ. 2023 Nov 25;901:165777. doi: 10.1016/j.scitotenv.2023.165777. Epub 2023 Jul 29.
Urban wetlands play a crucial role in sustainable social development. However, current research mainly focuses on specific wetland types, and fine extraction of urban wetlands remains a challenge. This study proposes a fine extraction framework based on hierarchical decision trees and shape features for urban wetlands, using Sentinel-2 remote sensing data to obtain detailed wetland data of Wuhan and Nanchang from 2016 to 2022. Our framework applies random forests to classify land cover, extracts urban fine wetlands by hierarchical decision trees and shape features, and assesses the dynamics of wetlands in the two cities. We also analyzed and discussed the characteristics of urban wetlands in the two cities. The results show that wetland accuracies of Wuhan and Nanchang are greater than 84.5 % and 82.9 %, respectively. The wetland areas of Wuhan in 2016, 2019, and 2022 are 1969.4 km, 1713.8 km, and 1681.1 km, while those in Nanchang are 1405.9 km, 1361.6 km, and 766.9 km. Inland wetlands are the main wetland types in both regions, with lake wetlands accounting for the highest proportion (over 40 %). The urban wetlands in the two cities exhibit different spatial and temporal evolution patterns, with varying change trends of wetland area and the structural proportions of fine wetlands. Besides, Wuhan's urban wetlands are primarily located in the south, while Nanchang's urban wetlands are concentrated in the east, exhibiting higher spatial and temporal dynamics. Analysis suggests that the reduced urban wetlands from 2016 to 2022 are related to fluctuating decreasing precipitation, growing population, and gross domestic product (GDP). Our study provides support for the conservation of urban wetland resources in Wuhan and Nanchang and highlights the need for targeted management strategies.
城市湿地在社会可持续发展中发挥着至关重要的作用。然而，目前的研究主要集中在特定的湿地类型上，城市湿地的精细提取仍然是一项挑战。本研究提出了一种基于层次决策树和形状特征的城市湿地精细提取框架，利用哨兵 - 2 遥感数据获取了 2016 年至 2022 年武汉和南昌的详细湿地数据。我们的框架应用随机森林对土地覆盖进行分类，通过层次决策树和形状特征提取城市精细湿地，并评估了两个城市湿地的动态变化。我们还分析和讨论了两个城市城市湿地的特征。结果表明，武汉和南昌的湿地精度分别大于 84.5%和 82.9%。武汉 2016 年、2019 年和 2022 年的湿地面积分别为 1969.4 平方千米、1713.8 平方千米和 1681.1 平方千米，而南昌的湿地面积分别为 1405.9 平方千米、1361.6 平方千米和 766.9 平方千米。内陆湿地是两个地区的主要湿地类型，其中湖泊湿地占比最高（超过 40%）。两个城市的城市湿地呈现出不同的时空演变模式，湿地面积和精细湿地的结构比例变化趋势各异。此外，武汉的城市湿地主要位于南部，而南昌的城市湿地集中在东部，呈现出较高的时空动态性。分析表明，2016 年至 2022 年城市湿地减少与降水波动减少、人口增长和国内生产总值（GDP）增长有关。我们的研究为武汉和南昌城市湿地资源的保护提供了支持，并强调了制定针对性管理策略的必要性。
