Atmospheric Correction Technology of Sentinel-2/MSI Data Aiming for Small Water Body:A Case of Daguangba Reservoir
Binbin Jiang,Xiaoyu Zhang,Ran Bi,Yong Du,Chengqi Huang
DOI: https://doi.org/10.1109/agro-geoinformatics262780.2024.10660993
2024-01-01
Abstract:Water quality holds most important in aquatic environments, particularly within the context of reservoirs located in Hainan Island, where its availability is limited yet its benefits are profoundly significant. The assessment of water quality in such regions is imperative for comprehending ecosystem health and ensuring the sustainable management of water resources. High-resolution satellites equipped with Multispectral Instruments (MSI), such as Sentinel-2, offer a promising avenue for monitoring water quality through atmospheric correction and water quality modeling techniques. In the domain of water quality assessment, the selection of atmospheric correction (AC) and water quality parameter (WQP) models plays a pivotal role in determining the accuracy of results. This variability underscores the importance of choosing appropriate models to precisely estimate parameters such as suspended particle matter (SPM) and chlorophyll a (Chl a). To delve deeper into this aspect, Sentinel-2/MSI images underwent processing utilizing three different AC models: POLYMER, ACOLITE, and C2RCC. Subsequently, two different WQP models were applied: OCX for ChI a and Nechad for SPM. Upon analysis, the results revealed notable discrepancies based on the combinations of AC and WQP models employed. The Polymer-Nechad demonstrated lower errors across four metrics, with an RMSE of 19.14, RMSLE of 0.50, MAPE of $72.5 \%$, Bias of 1.21, and MAE of 2.54. Specifically, the POLYMER-OCX and C2RCC-OCX combination exhibited superior performance in estimating Chl a concentration, while the POLYMER-Nechad combination yielded the most accurate estimations for SPM. Moreover, this study contributes to the advancement of long-time series remote sensing techniques for reservoir (Daguangba) monitoring in Hainan Island. The study also presents a detailed analysis of the annual changes in Chl a and SPM concentrations in the Daguangba Reservoir from 2016 to 2023, highlighting the potential of Sentinel-2/MSI data for long-term environmental monitoring. The year-to-year analysis from 2016 to 2023 depicted a fluctuating trend in both parameters, with the highest values recorded in 2017. Following this peak, a general decreasing trend in Chl a was observed, with a temporary rise in 2020, possibly due to specific environmental events or changes in the ecosystem. The implications of these findings extend beyond academic research, with practical applications in environmental management and resource conservation. In conclusion, this study highlights the importance of remote sensing technology in assessing water quality parameters and emphasizes the potential benefits of employing Sentinel-2/MSI data for effective environmental monitoring and management in reservoir located in Hainan Island. The integration of advanced remote sensing techniques holds promise for enhancing our ability to monitor and safeguard aquatic ecosystems for future generations.