Highest Quality Remote Sensing Reflectance Database Compiled from 20+ years of MODIS-Aqua measurements

数据说明文档 1. 数据简介 本数据集提供经严格筛选的 MODIS Aqua 传感器最高质量遥感反射比（Rrs）光谱（412 nm、443 nm、488 nm、531 nm、547 nm、667 nm、678 nm）数据，空间分辨率 4km，时间 尺度包含 8天合成与月合成产品。数据采用最高质量 Rrs 筛选标准(CHQR)三重筛选方案进 行处理，确保其同时满足以下特性： ⚫ 最优观测条件（排除云层、耀斑等干扰） ⚫ 符合水体真实光谱形态 ⚫ 残余误差最小化 该数据集适用于海洋水色遥感反演、水质监测及相关算法开发、长期趋势等研究。 2. 数据筛选标准 图 1 CHQR：最高质量 Rrs 的筛选标准示意图。 最高质量 Rrs 数据筛选标准示意图如图 1所示，具体包括以下三个方面： 1）基于观测条件或数据处理失败进行过滤。排除以下 Level-2 质量控制标记（l2-flags） 对应的 Rrs 产品：ATMFAIL (1)、LAND (2)、HIGLINT (4)、HILT (5)、HISATZEN (6)、 STRAYLIGHT (9)、CLDICE (10)、COCCOLITH (11)、HISOLZEN (13)、LOWLW (15)、CHLFAIL (16)、NAVWARN (17)、MAXAERITER (20)、CHLWARN (22)、ATMWARN (23)、NAVFAIL (26)。 2）光谱形状一致性。利用质量评价系统（QA system）对目标 Rrs 光谱进行筛选，仅保 留 QA Score = 1 的数据，以确保所保留的 Rrs 能反映真实的水体状态（Wei et al., 2016）。 3）最小或无残余误差。基于吸收系数约束方法（CBAC）进一步筛除残余误差较大的 Rrs 数据。我们结合两种算法估算 443nm 的系数系数( a(443))：对大气校正误差敏感的 Rrs 波段比值法（Lee et al., 1998）； 对大气校正误差较不敏感的波段差值法（Lee et al., 2023）。 通过二者的对比，若二者近似一致，则认为得到了残余误差极小的 Rrs 数据。 算法细节说明： 比值算法（Band-Ratio for a(443)，记为 a(443)BR）：： 𝑎(443)𝐵𝑅 =10𝛼0+𝛼1×𝜌25+𝛼2×𝜌25 2+𝛼3×𝜌35+𝛼4×𝜌35 2 (1a) 𝜌25 = 𝑙𝑜𝑔10[𝑅𝑟𝑠(443) 𝑅𝑟𝑠(547)],𝜌35 = 𝑙𝑜𝑔10[𝑅𝑟𝑠(488) 𝑅𝑟𝑠(547)] (1b) 拟合系数𝛼0= −0.883, 𝛼1= −0.955, 𝛼2= −0.010, 𝛼3= −0.450, 𝛼4= 0.345。 差值算法（Band-Difference for a(443)，记为 a(443)BD）： 𝑀𝐵𝐷𝑅𝑟𝑠(443)= 𝑅𝑟𝑠(547)−[𝑅𝑟𝑠(443)+547−443 667−443(𝑅𝑟𝑠(667)−𝑅𝑟𝑠(443))] (2a) 𝑎(443)𝐵𝐷 =10𝛽0+𝛽1 exp(𝛽2 × 𝑀𝐵𝐷𝑅𝑟𝑠(443)) (2b) 拟合系数𝛽0= −2.12, 𝛽1= 0.91, 𝛽2=247.38。 吸收系数一致性检验（CBAC） 参考 Hu et al. (2013) 提出的叶绿素约束（CBCC）策略，定义 a(443)相对差异为： δ𝑎(443) = |𝑎(443)𝐵𝐷−𝑎(443)𝐵𝑅| 𝑎(443)𝐵𝐷 (3) 设置阈值 δ𝑎(443)为0.15，满足 δ𝑎(443)≤0.15 为最高质量 Rrs 的条件之一。 3. 数据结构 3.1 文件目录 TEXT ├── 8_day_Rrs/ # 8 天合成产品 │ └── HQ_flag/ # 质量标记(1=最高质量) ├── monthly_Rrs/ # 月合成产品（结构同上） └── readme.docx 3.2 数据读取 MATLAB 示例： MATLAB % 读取质量标记 HQ_flag=ncread('AQUA_MODIS.20020704_20020711.L3m.8D.RRS.HQ _flag.4km.nc','HQ_flag'); % 读取 Rrs 数据（已自动应用质量掩膜） Rrs_443=ncread('AQUA_MODIS.20020704_20020711.L3m.8D.RRS.44 3nm.4km.nc','Rrs_443'); 3.3 变量说明 变量名 类型 描述 有效范围 HQ_flag int8 1=最高质量，0=其他 0/1 Rrs_*nm float 各波段遥感反射率 (sr⁻¹) ≥0 1. Data Overview This database provides rigorously screened, highest-quality remote sensing reflectance (Rrs), spectra from the MODIS Aqua sensor at 4 km spatial resolution, covering seven wavelengths (412 nm, 443 nm, 488 nm, 531 nm, 547 nm, 667 nm, and 678 nm). It includes both 8-day and monthly composite products. The data are screened using CHQR (Criteria of the Highest-Quality Rrs) to have the selected Rrs spectrum meet the following conditions simultaneously: ⚫ Optimal observation conditions (excluding clouds, sun glints, etc.) ⚫ Spectral consistency with true water optical properties ⚫ Minimal residual uncertainties This dataset is suitable for research in ocean color remote sensing inversion, water quality monitoring, algorithm development, and long-term trends. 2. Data Screening Method Fig. 1. Criteria for the highest-quality Rrs (CHQR). Criteria of the Highest-Quality Rrs are shown in Fig. 1, which specifically includes the following three aspects: 1) Filtering based on observation condition or failure in data processing. Any Rrs product having the following level-2 quality-control flags (l2-flags) are excluded: ATMFAIL (1), LAND (2), HIGLINT (4), HILT (5), HISATZEN (6), STRAYLIGHT (9), CLDICE (10), COCCOLITH (11), HISOLZEN (13), LOWLW (15), CHLFAIL (16), NAVWARN (17), MAXAERITER (20), CHLWARN (22), ATMWARN (23), and NAVFAIL (26); 2) Spectral shape consistency. The QA system will be applied to a target Rrs spectrum and only QA Score = 1 will be kept. This helps ensure the data product represents realistic water conditions (Wei et al. 2016). 3) Minimum to no residual errors. We can implement a constraint based on the absorption coefficient (CBAC) to screen “error-free” Rrs. We estimated the absorption coefficient at 443 nm (a(443)) using two algorithms: the Rrs band ratio method (Lee et al., 1998), which is sensitive to atmospheric correction errors, and the band-difference method (Lee et al., 2023), which is less sensitive to such errors. By comparing the results from both methods, Rrs data were considered to have minimal residual errors when the two estimates of a(443) were approximately consistent. Algorithm details: The band-ratio algorithm for a(443) (a(443)BR) 𝑎(443)𝐵𝑅 =10𝛼0+𝛼1×𝜌25+𝛼2×𝜌25 2+𝛼3×𝜌35+𝛼4×𝜌35 2 (1a) 𝜌25 = 𝑙𝑜𝑔10[𝑅𝑟𝑠(443) 𝑅𝑟𝑠(547)],𝜌35 = 𝑙𝑜𝑔10[𝑅𝑟𝑠(488) 𝑅𝑟𝑠(547)] (1b) where 𝛼0−4 are the fitting coefficients (Lee et al. 1998), 𝛼0= −0.883 , 𝛼1= −0.955 , 𝛼2= −0.010, 𝛼3= −0.450, and 𝛼4= 0.345. On the other hand, the band-difference algorithm for a(443) (a(443)BD) is: 𝑀𝐵𝐷𝑅𝑟𝑠(443)= 𝑅𝑟𝑠(547)−[𝑅𝑟𝑠(443)+547−443 667−443(𝑅𝑟𝑠(667)−𝑅𝑟𝑠(443))] (2a) 𝑎(443)𝐵𝐷 =10𝛽0+𝛽1 exp(𝛽2 × 𝑀𝐵𝐷𝑅𝑟𝑠(443)) (2b) with 𝛽0−2 the fitting coefficients (Lee et al. 2023), 𝛽0= −2.12, 𝛽1= 0.91, and 𝛽2=247.38. Constraint Based on Absorption Coefficient (CBAC): Following the constraint scheme based on chlorophyll concentration (CBCC) (Hu et al. 2013), the relative difference between a(443)BD and a(443)BR is calculated as: δ𝑎(443) = |𝑎(443)𝐵𝐷−𝑎(443)𝐵𝑅| 𝑎(443)𝐵𝐷 (3) The δa(443) threshold is 0.15, and the highest quality Rrs spectra having δa(443) ≤ 0.15. 3. Data Structure 3.1 Directory Organization TEXT ├── 8_day_Rrs/ # 8-day composite products │ └── HQ_flag/ # Quality flags (1=highest-quality) ├── monthly_Rrs/ # Monthly composites (same structure) └── readme.docx 3.2 Data Access Here is an example of how to read the data in MATLAB: MATLAB % Read quality flag HQ_flag=ncread('AQUA_MODIS.20020704_20020711.L3m.8D.RRS.HQ_fla g.4km.nc','HQ_flag'); % Read masked Rrs data Rrs_443=ncread('AQUA_MODIS.20020704_20020711.L3m.8D.RRS.443nm.4 km.nc','Rrs_443'); 3.3 Variable Description Variable Type Description Valid Range HQ_flag int8 1=highest-quality, 0=others 0/1 Rrs_*nm float Rrs at specified band (sr⁻¹) ≥0 Reference: Hu, C., Feng, L., & Lee, Z. (2013). Uncertainties of SeaWiFS and MODIS remote sensing reflectance: Implications from clear water measurements. Remote Sensing of Environment, 133, 168-182 Lee, Z., Zhao, L., Hu, C., Wang, D., Lin, J., & Shang, S. (2023). Absorption Coefficient and Chlorophyll Concentration of Oceanic Waters Estimated from Band Difference of Satellite-Measured Remote Sensing Reflectance. Journal of Remote Sensing, 3, 0063 Lee, Z.P., Carder, K.L., Steward, R.G., Peacock, T.G., Davis, C.O., & Patch, J.S. (1998). An empirical algorithm for light absorption by ocean water based on color. Journal of Geophysical Research-Oceans, 103, 27967-27978 Wei, J.W., Lee, Z.P., & Shang, S.L. (2016). A system to measure the data quality of spectral remote-sensing reflectance of aquatic environments. Journal of Geophysical Research-Oceans, 121, 8189-8207