Monitoring Ice Phenology in Lake Wetlands Based on Optical Satellite Data: A Case Study of Wuliangsu Lake

Abstract

It is challenging to obtain the ice phenology for a lake covered with a vast area of aquatic (shallow lake wetlands) using optical satellite data because possible clouds above the lake could contaminate the result. We developed a new method to tackle this challenge. Our target was Wuliangsu Lake, a large (330 km2) and shallow (1.6 m average depth) lake wetland in the Inner Mongolia Plateau. We used Landsat and Sentinel-2 imageries to extract the lake water boundary. The MOD09GQ/MYD09GQ dataset, having the highest spatial resolution among MODIS reflectivity products, was first selected to differentiate water and ice pixels. Then, we used the reflectivity state parameters containing cloud information in the dataset to filter out the cloud pixels. The ice phenology characteristics, such as freeze-up, break-up dates, and ice cover duration (ICD) between 2013 and 2022 were obtained. We further applied the air temperature correction technique to remove the outliers. The average of ICD in Wuliangsu Lake was about 127 ± 6 days. The freeze-up start and break-up end occurred on 17 November ± 5 days and 25 March ± 4 days, respectively. The remote sensing results agree well with the field observation, with a mean absolute error of 2 days. The algorithm can effectively remove the influence of aquatic plants and clouds on lake ice identification, thereby satisfying the needs of daily monitoring and ice phenology research in the lake wetlands.