Spectral diffuse attenuation coefficient of down-welling irradiance (K_d(λ)) is a fundamental optical parameter for water color and of significance for heat budget calculation, photosynthesis modeling and carbon sequestration. Uncertainties exist in both in situ measured and satellite derived K_d(λ), therefore, evaluation of its data quality is essential. In addition, optical water classification of K_d(λ) is still limited. To address these issues, we proposed an optical water classification and quality control model (OC_QC model) based on a globally collected high-quality in situ K_d(λ) data. The model consists of the reference K_d(λ) spectra for 7 optical water types (OWT, ranging from clear open ocean water to highly turbid water) and a scoring system. An individual K_d(λ) spectrum can be assigned an OWT and quality control score (QC score) with this model. The performance of the OC_QC model is tested with independent in situ K_d(λ) spectra from the turbid coastal water (Bohai Sea) and clear oceanic water (Indian Ocean). Questionable or erroneous K_d(λ) spectra can be identified, suggesting the model’s robustness. The model was also evaluated with the Medium Resolution Imaging Spectrometer (MERIS) Level-2 and the Moderate Resolution Imaging Spectroradiometer (MODIS) Level-3 K_d(λ) data in the Bohai Sea, Yellow Sea, East China Sea and South China Sea. Satellite derived maps clearly showed the spatial and temporal variations of OWTs and QC scores, and the regions of low-quality K_d(λ) were reasonably extracted. Global maps of K_d(λ)-based optical water types and corresponding QC scores were also obtained with OC_QC model. With the quality controlled K_d(λ) data, the retrieval accuracy of Secchi disk depth was improved. The retrieval accuracies of Chlorophyll-a concentration and total seawater absorption coefficient were also improved with the K_d(λ) optical water classification based algorithms. The OC_QC model would be valuable for evaluating K_d(λ) data (especially the operational global K_d(λ) products), investigation of water optical properties, and development and merging of bio-optical algorithms.

下涌辐照度的光谱漫反射衰减系数（K _d (λ)）是水色的基本光学参数，对热收支计算、光合作用建模和碳封存具有重要意义。现场测量和卫星导出的 K _d (λ)都存在不确定性，因此，对其数据质量的评估至关重要。此外，K _d (λ) 的光学水分类仍然有限。为了解决这些问题，我们提出了一种基于全球收集的高质量原位K _d (λ) 数据的光学水分类和质量控制模型（OC_QC 模型）。该模型由参考 K _d(λ) 7 种光学水类型（OWT，范围从清澈的开阔海水到高度浑浊的水）的光谱和评分系统。使用此模型，可以为单个 K _d (λ) 光谱分配 OWT 和质量控制分数（QC 分数）。OC_QC 模型的性能通过来自浑浊的沿海水（渤海）和清澈的海洋水（印度洋）的独立原位K _{d (λ) 光谱进行测试。}可以识别有问题或错误的 K _{d (λ) 光谱，表明模型的稳健性。}该模型还使用中分辨率成像光谱仪 (MERIS) Level-2 和中分辨率成像光谱仪 (MODIS) Level-3 K _{d进行了评估}(λ) 渤海、黄海、东海和南海的数据。卫星衍生图清晰地显示了OWTs和QC分数的时空变化，合理提取了低质量K _d (λ)的区域。还使用 OC_QC 模型获得了基于K _{d (λ) 的光学水类型和相应的 QC 分数的全局图。}通过质量控制的 K _d (λ) 数据，提高了 Secchi 盘深度的检索精度。叶绿素a浓度和总海水吸收系数的反演精度也通过基于K _d (λ)光学水分类的算法得到提高。_{OC_QC 模型对于评估 K d}很有价值(λ) 数据（尤其是可操作的全球 K _d (λ) 产品），水光学特性的调查，以及生物光学算法的开发和合并。