Yin et al. (2025) Hybrid wetland city map: Improved wetland characterization through the synergy of global land cover products

Identification

• Journal: International Journal of Applied Earth Observation and Geoinformation
• Year: 2025
• Date: 2025-12-06
• Authors: Xiaogan Yin, Weiguo Jiang, Zhe Yang, Ziyan Ling, Nandin-Erdene Tsendbazar, Peng Hou, Yue Deng, Xiaoya Wang, Zhijie Xiao, Xiao Li, Miaolong Lin
• DOI: 10.1016/j.jag.2025.104994

Research Groups

• Key Laboratory of Land Use, Ministry of Natural Resources, Faculty of Geographical Science, Beijing Normal University, Beijing, China
• State Key Laboratory of Remote Sensing and Digital Earth, Faculty of Geographical Science, Beijing Normal University, Beijing, China
• Laboratory of Geo-information Science and Remote Sensing, Wageningen University & Research, Wageningen, the Netherlands
• School of Information and Control Engineering, Qingdao University of Technology, Qingdao, China
• Satellite Environment Centre, Ministry of Ecology and Environment, Beijing, China
• School of Architecture and Civil Engineering, Chengdu University, Chengdu, China
• College of Remote Sensing and Geomatics Engineering, Nanjing University of Information Science and Technology, Nanjing, China

Short Summary

This study developed a hybrid Wetland City Map (WCM) by fusing three global 10-meter resolution land cover products using a Weighted Voting-Knowledge based Decision Rule method. The WCM significantly improved wetland classification accuracy within 43 Ramsar Wetland Cities, providing crucial high-resolution data for urban wetland monitoring and management.

Objective

• To develop a hybrid Wetland City Map (WCM) by fusing three global 10-meter resolution land cover products (Dynamic World, ESA WorldCover, and Esri Land Cover) using a novel Weighted Voting-Knowledge based Decision Rule (WV-KDR) method.
• To overcome the limitations of existing global land cover datasets in accurately identifying and characterizing wetland types within urban areas, particularly for Ramsar Wetland Cities.
• To provide high-resolution fusion maps to support the Wetland City Accreditation process, national wetland reporting, and the implementation of the Ramsar Strategic Plan and Sustainable Development Goals.

Study Configuration

• Spatial Scale: 43 Ramsar Wetland Cities across 17 countries in Asia, Europe, Africa, and North America. The output map has a spatial resolution of 10 meters.
• Temporal Scale: Annual land cover maps for 2020 and 2021, with analysis of wetland changes between these two years.

Methodology and Data

• Models used:
  • Weighted Voting-Knowledge based Decision Rule (WV-KDR) method for data fusion.
  • Input global land cover products (Dynamic World, Esri Land Cover) were developed using deep learning models (Fully Convolutional Neural Network, Convolutional Neural Network-UNet).
  • ESA WorldCover used a Gradient Boosting Decision Tree (CatBoost) algorithm.
• Data sources:
  • Global Land Cover (GLC) products: Dynamic World (DW), ESA WorldCover (ESA), Esri Land Cover (ESRI).
  • Satellite imagery: Sentinel-1 and Sentinel-2 (used by input products and for visual interpretation of reference data).
  • Reference data: Visually interpreted points generated using Sentinel-2 imagery in Google Earth Engine (GEE) and high-resolution Google Earth imagery.
  • Ancillary data: ETOPO 2022 (bathymetry data), 2024 OSM global coastline dataset, GWL_FCS30D dataset (for validation of herbaceous wetlands, tidal flats, and mangroves).

Main Results

• The WCM achieved an average overall accuracy (OA) of 86.93 % and an average Kappa coefficient of 0.825 across all 43 wetland cities for 2020 and 2021.
• WCM's accuracy surpassed the three input land cover products by an average of 8 % in OA and 0.12 in Kappa coefficient. Individual city improvements ranged from 2 % to 26 %.
• Coastal wetland cities showed greater accuracy improvements (average 10 % OA) compared to inland cities (average 4 % OA).
• F1 scores for wetland categories in WCM were: 90.33 % (water), 64.09 % (marsh), 71.67 % (tidal flat/flooded flat), and 92.17 % (mangrove), significantly outperforming input datasets.
• Wetland coverage varied widely among cities in 2020, from 0.24 % (Ifrane, Morocco) to 65.83 % (Al Chibayish, Iraq), with Asian cities generally having higher coverage.
• Water bodies constituted the dominant wetland type, averaging 76 % of the total wetland area across cities.
• Wetland area changes between 2020 and 2021 ranged from a maximum increase of 1.3 % (Jining, +174 km²) to a maximum decrease of 6.5 % (Al Chibayish, -162 km²). The WCM more accurately captured these changes than individual products.

Contributions

• Developed a novel, wetland-specific data fusion method (WV-KDR) that integrates class confidence and spatial weights with knowledge-based decision rules, specifically tailored for urban wetland characterization.
• Generated the first unified global wetland urban land cover map at 10-meter resolution, providing unprecedented detail for urban wetland monitoring.
• Significantly improved the accuracy and spatial detail of wetland classification, particularly for diverse wetland types (water, marsh, tidal flat/flooded flat, mangroves) within urban environments.
• Optimized sample selection strategy by allocating more reference samples to areas of disagreement among input datasets, maximizing the incorporation of complementary information.
• Incorporated ecological and spatial logic, such as intra-annual water inundation frequency and coastal extension zones, into the decision rules to enhance wetland delineation and reduce misclassification.
• Provides a robust spatial information foundation for the Ramsar Wetland City Accreditation process, supporting national authorities in evaluating city nominations and monitoring conservation outcomes.
• Contributes directly to the implementation of the Ramsar Strategic Plan and relevant Sustainable Development Goals (SDG 6 and SDG 11) by enabling high-precision monitoring of urban wetlands.

Funding

• National Key Research and Development Program of China (Grant No. 2024YFF1306105, No.2023YFF0807204)
• National Natural Science Foundation of China (Grant No. U21A2022)
• Open Fund of the State Key Laboratory of Remote Sensing Science and Digital Earth, Beijing Engineering Research Center for Global Land Remote Sensing Products (Grant No. OF202414)
• Key Laboratory of Land Use, Ministry of Natural Resources

Citation

@article{Yin2025Hybrid,
  author = {Yin, Xiaogan and Jiang, Weiguo and Yang, Zhe and Ling, Ziyan and Tsendbazar, Nandin-Erdene and Hou, Peng and Deng, Yue and Wang, Xiaoya and Xiao, Zhijie and Li, Xiao and Lin, Miaolong},
  title = {Hybrid wetland city map: Improved wetland characterization through the synergy of global land cover products},
  journal = {International Journal of Applied Earth Observation and Geoinformation},
  year = {2025},
  doi = {10.1016/j.jag.2025.104994},
  url = {https://doi.org/10.1016/j.jag.2025.104994}
}