The use of satellite remote sensing to estimate coloured dissolved organic matter (CDOM) and identify its potential sources is important for monitoring lake water quality and implementing management strategies. In this study, taking Erhai Lake as an example and based on MODIS/Aqua satellite images and in-situ measured data, we used empirical and semi-empirical methods to develop algorithms for CDOM and the fluorescence index (FI₃₇₀) from remote sensing reflectance (Rrs(λ)). The temporal and spatial distributions of the CDOM concentration and FI₃₇₀ in Erhai Lake during 2013–2019 were retrieved. The results show the following. (1) The band ratio (Rrs(469)+Rrs(645))/Rrs(555) model could estimate the CDOM absorption coefficient at 412 nm (a_CDOM(412)) (R²=0.507), and it was relatively stable. Using the band ratio Rrs(645)/Rrs(469) combined with the chlorophyll-a (Chl-a) APProach by ELimination (APPEL) model, a semi-empirical inversion model of FI₃₇₀ performed with satisfactory accuracy (R²=0.550) and was more accurate than the empirical algorithm (R²=0.505). (2) During the period of 2013–2019, the CDOM concentration in Erhai Lake generally decreased from the northern to the central to the southern parts of the lake, and the CDOM concentration was higher in summer and autumn than in spring and winter. FI₃₇₀ was higher in the northern and western coastal waters and lower in the central, southern and eastern parts of the lake. FI₃₇₀ in autumn and winter was higher than that in spring and summer. CDOM was affected by both terrestrial and internal sources, and their relative contributions were not the same in different seasons. (3) For different Chl-a concentrations, different CDOM concentration models had better retrieval effects, i.e., Rrs(645))/Rrs(555) and (Rrs(469)+Rrs(645))/Rrs(555) had the best performance when Chl-a<10 μg/L and Chl-a>10 μg/L, respectively. The inversion models established in this study offer improved quantifications of the CDOM concentration and the FI₃₇₀ in Erhai Lake, providing important support for monitoring water quality and implementing efficient management strategies.

使用卫星遥感来估计有色溶解有机物 (CDOM) 并确定其潜在来源对于监测湖泊水质和实施管理策略非常重要。本研究以洱海为例，基于 MODIS/Aqua 卫星图像和原位测量数据，我们采用经验和半经验方法开发了 CDOM 和遥感反射率荧光指数（FI ₃₇₀）的算法。(Rrs(λ))。检索2013- ₂₀₁₉年洱海CDOM浓度和FI ₃₇₀的时空分布。结果如下。(1) 带比(Rrs(469)+Rrs(645))/Rrs(555)模型可以估计412 nm处的CDOM吸收系数(a _CDOM (412)) (R² =0.507），相对稳定。使用带比 Rrs(645)/Rrs(469) 结合叶绿素-a (Chl-a) 消除法 (APPEL) 模型，FI ₃₇₀的半经验反演模型具有令人满意的精度 (R ² =0.550 ) 并且比经验算法更准确 (R ² =0.505)。(2) 2013-2019年，洱海CDOM浓度总体呈从北向湖中向南递减的趋势，夏秋季CDOM浓度高于春季和冬季。FI ₃₇₀在北部和西部沿海水域较高，在湖的中部、南部和东部较低。FI ₃₇₀秋冬季高于春夏季。CDOM受陆源和内源双重影响，不同季节的相对贡献不同。(3) 对于不同的 Chl-a 浓度，不同的 CDOM 浓度模型具有更好的反演效果，即 Rrs(645))/Rrs(555) 和 (Rrs(469)+Rrs(645))/Rrs(555)当 Chl-a<10 μg/L 和 Chl-a>10 μg/L 时，性能最佳。本研究建立的反演模型改进了对洱海CDOM 浓度和 FI _{370 的}量化，为监测水质和实施有效管理策略提供了重要支持。