Extraction of long time series wetland information based on Google Earth Engine and random forest algorithm for a plateau lake basin – A case study of Dianchi Lake, Yunnan Province, China. (February 2023)

• Record Type: Journal Article
• Authors: Zhao, Fei; Feng, Siwen; Xie, Fei; Zhu, Sijin; Zhang, Sujin
• Is Part Of: Ecological indicators. Volume 146(2023)
• Publication Date: 2023-02
• Subjects: DT Decision tree -- EVI Enhanced vegetation index -- GEE Google Earth Engine -- GLCM grey level co-occurrence matrix -- JPL Jet Propulsion Laboratory -- mNDWI Modified normalized difference water index -- NDBI Normalized build-up index -- NDVI Normalized difference vegetation index -- OP Overall accuracy -- PA Producer accuracy -- QA Quality assessment -- RF Random forest -- SRTM Shuttle radar topography mission -- SVM Support vector machine -- UA User accuracy -- QADI Quantitative and allocation disagreement indexq; Long time series -- Change detection -- Google Earth Engine -- Random forest -- Wetland information extraction; Environmental monitoring -- Periodicals; Environmental management -- Periodicals; Environmental impact analysis -- Periodicals; Environmental risk assessment -- Periodicals; Sustainable development -- Periodicals; 333.71405
• DOI: 10.1016/j.ecolind.2022.109813
• Languages: English
• ISSNs: 1470-160X

Abstract

Highlights: Landsat images were classified using the Google Earth Engine platform. Wetland classification result accuracy was high compared to control data. Dianchi Basin wetlands areas increased overall from 1988 to 2020. The study methodology can be used to fill data gaps where historical data are missing. Abstract: Wetlands are transitional zones between terrestrial and aquatic ecosystems. They have the potential to continuously provide human beings with food, raw materials, and other substances. Also, wetland landscape pattern changes have profound impacts on the climate of plateau areas and accelerate the rate of climate change, so it is crucial to extract long time series of plateau wetland information. Concurrent with large-scale urbanization, industrial development and construction, and rapid population increases, the lakeside wetlands in the Dianchi Lake Basin are changing rapidly. However, scholars have not yet extracted long time series wetland information for this area, so the long-term evolution of these wetlands cannot be clarified. In this study, we used the Google Earth Engine (GEE) platform to extract wetland information for the Dianchi Basin from 1988 to 2020 from Landsat data and generated 33 wetland maps by applying a random forest classification model to identify trends and a confusion matrix to assess accuracy. The results showed that the overall trend of wetland area from 1988 to 2020 showed an increasing area, and the total wetland area increased by 31.11 km 2 (+9.67 %). Furthermore, the overall accuracy of most years exceeded 80 %, the QADI was no higher than 0.2, the user accuracy and producer accuracy were higher than 80 % and 70 %, respectively, for swamps and water bodies, and the remaining non-wetland categories also achieved robust classification accuracy. Thus, the results of this study can compensate for the lack of information on wetland changes in the region over the past 33 years and provide data support and a scientific basis for sustainable wetland development in the plateau.