Construction of remote sensing inversion algorithm for water transparency of typical lakes and reservoirs in Henan Province
DENG Yuhao1,2, ZHU Chunhua1, QIU Shike3, DU Jun2, WANG Zheng2,4,5, ZHANG Yingzhuo6, WANG Chao2, SUN Tingting2, ZHAO Weixi7
1.School of Information Science and Engineering ，Henan University of Technology，Zhengzhou 450001，China；2.Institute of Geographical Sciences， Henan Academy of Sciences，Zhengzhou， 450052；3.Henan Academy of Sciences， Zhengzhou 450046，China；4.Collaborative Innovation Center of Geo-Information Technology for Smart Central Plains， Henan Province， Zhengzhou 450046，China；5.Key Laboratory of Spatiotemporal Perception and Intelligent processing， Ministry of Natural Resources， Zhengzhou 450046，China；6.School of Management and Economics， North China University of Water Resources and Electric Power，Zhengzhou 450003，China；7.Henan Province South-to-North Water Diversion Canal Head Ecological Environment Monitoring Emergency Center，Nanyang 474475，Henan，China
Abstract：Water transparency serves as a critical indicator for assessing water quality and the degree of eutrophication. Remote sensing technology offers unparalleled advantages in the inversion of water transparency. However， the optical characteristics of different water bodies exhibit significant variability， posing a substantial challenge in developing remote sensing algorithms applicable to various types of lakes and reservoirs. This study selected three water bodies in Henan Province：Danjiangkou Reservoir （Yangtze River Basin， Class I water quality）， Luhun Reservoir （Yellow River Basin， Class II water quality）， and Baisha Reservoir （Huai River Basin， Class IV water quality），which are situated in distinct river basins with varying water quality conditions， as research subjects. Utilizing an integrated monitoring approach combining spaceborne， airborne， and ground-based methods， this research employed measured data alongside Sentinel-2 MSI and Landsat 8/9 OLI satellite data to investigate remote sensing inversion algorithms for water transparency. Firstly， by integrating spectral normalization， first-order differentiation， and spectral equivalence analysis， we identified sensitive bands for water transparency （blue， green， and red bands） and established a unified empirical remote sensing inversion model， including single-band， ratio-band， and mixed-band methods. Through comparative analysis using satellite data synchronized with on-site transparency measurements， the accuracy of the two datasets was evaluated. The results are shown as follows. 1） Spectral equivalence combined with spectral normalization and first-order differentiation effectively identify sensitive bands for water transparency. 2） Model validation using reserved validation sets yielded satisfactory results， with R2 exceeding 0.9， RMSE approximately 0.5 m， and MRE below 20%； however， there is a notable disparity in inversion accuracy between the datasets. Specifically， the index model Y = 33.504e-1.379x based on the combination of （Green + Red）/Blue bands from Sentinel-2 MSI demonstrated the highest inversion accuracy （R2 = 0.9509， RMSE = 0.67 m）， followed by the algorithm model for Landsat-8 OLI （R2 = 0.7426， RMSE = 1.26 m）. 3） All three models based on Sentinel-2 MSI fitting achieved R2 greater than 0.9， while the highest R2 for models based on Landsat 8/9 OLI fitting was 0.742 6. Therefore， Sentinel-2 MSI data exhibits higher applicability and stability， making it more suitable as a data source for water transparency inversion. The study also explored why the most relevant sensitive band is not always the optimal model， noting that single-band models are susceptible to noise and outliers， whereas band ratio algorithms can better reflect the physical mechanisms of water body optical properties by incorporating multiple band information， thereby enhancing the robustness and interpretability of the model. These findings provide a reliable method and theoretical foundation for large-scale and efficient monitoring of water transparency in inland lakes and reservoirs， contributing significantly to water environmental protection and management practices.
邓玉浩,朱春华,邱士可,杜军,王正,张英卓,王超,孙婷婷,赵唯茜. 河南省典型湖库水体透明度遥感反演算法构建[J]. 华中师范大学学报(自然科学版), 2025, 59(6): 923-935.
DENG Yuhao,ZHU Chunhua,QIU Shike,DU Jun,WANG Zheng,ZHANG Yingzhuo,WANG Chao,SUN Tingting,ZHAO Weixi. Construction of remote sensing inversion algorithm for water transparency of typical lakes and reservoirs in Henan Province. journal1, 2025, 59(6): 923-935.
2025, Vol. 59 
