Using Time Series Optical and SAR Data to Assess the Impact of Historical Wetland Change on Current Wetland in Zhenlai County, Jilin Province, China

Abstract

Wetlands, as the most essential ecosystem, are degraded throughout the world. Wetlands in Zhenlai county, with the Momoge National Nature Reserve, which was included on the Ramsar list, have degraded by nearly 30%. Wetland degradation is a long-term continuous process with annual or interannual changes in water area, water level, or vegetation presence and growth. Therefore, it requires sufficiently frequent and high-spatial-resolution data to represent its dynamics. This study mapped yearly land-use maps with 30-m resolution from 1985 to 2018 using Landsat data in Google Earth Engine (GEE) to explore the wetland degradation process and mapped 12-day interval land-use maps with 15-m resolution using the Sentinel-1B and Sentinel-2 data in GEE and other assistant platforms to study the characteristics of wetland dynamics in 2018. Four sets of maps were generated using Sentinel-1B (S1), Sentinel-2 (S2), the combination of Sentinel-1B and Sentinel-2 (S12), and S12 with multitemporal remote sensing (S12’). All of the classifications were performed in the Random Forest Classification (RFC) method using remote sensing indicators. The results indicate that S12’ was the most accurate. Then, the impact of the historic land-use degradation process on current wetland change dynamics was discussed. Stable, degradation, and restoration periods were identified according to the annual changes in wetlands. The degraded, stable, restored, and vulnerable zones were assessed based on the transformation characteristics among wetlands and other land-use types. The impact of historical land-use trajectories on wetland change characteristics nowadays is diverse in land-use types and distributions, and the ecological environment quality is the comprehensive result of the effect of historical land-use trajectories and the amount of rainfall and receding water from paddy fields. This study offers a new method to map high-spatiotemporal-resolution land-use (S12’) and addresses the relationship between historic wetland change characteristics and its status quo. The findings are also applicable to wetland research in other regions. This study could provide more detailed scientific guidance for wetland managers by quickly detecting wetland changes at a finer spatiotemporal resolution.

Keywords

• wetland loss
• synthetic aperture radar
• optical remote sensing
• random forest classification
• land-use classification
• multitemporal remote sensing