Journal Article

Gap filling for satellite-derived products of lake aquatic environment using historical big data

Wei, Chengguo, Wang, Haoran, Jiao, Yaqin, Luo, Juhua, Zhang, Fukang, Xiao, Qitao, Ge, Yilin, Qiu, Yinguo

2025

Overview

Effective monitoring of lake aquatic environments is crucial for assessing lake health, identifying issues, and developing emergency plans. Satellite-based remote sensing has been recognized as an effective method for timely and comprehensive monitoring of these environments. However, satellite-derived products often lack complete spatial coverage due to invalid pixels resulting from factors such as cloud cover, high sun glint contamination, and high satellite-viewing angles. To address this issue, we propose a novel gap filling method for satellite-derived products of lake aquatic environments, utilizing historical big data. We initially developed a machine-learning-based model for similarity matching across various dates. This model was based on 10 factors, selected from water quality and meteorological conditions that have a significant correlation with the lake aquatic environment. This model allows for the assignment of values to invalid pixels in a specific satellite-derived product, derived from the corresponding pixels in the products of historical dates. The proposed method has been applied to the satellite-derived Chl-a products of Lake Chaohu. The experimental findings demonstrate that the computed mean value of the peak signal-to-noise ratio (PSNR) stands at 35.75, as derived from the experimental data. This substantiates the precision of the gap filling method applied to satellite-derived products. This study underscores the significant value of the proposed method in gap filling for satellite-derived products, as well as in predicting the aquatic environment of lakes.

• Publisher: Taylor & Francis Group
• Subject: historical big data / Lake aquatic environment / machine learning / missing data reconstruction / satellite-derived products