Atmospheric correction for China''s coastal water color remote sensing

INTRODUcrIONMeasurements from satellite have demonstrated that space sensor can be applied to measurethe sPectra of water-leaving radiance and proven that ocean color remote sensing is a powerfultcol for understanding oceanic, biological and physical processes. One sPecial ocean color satel-lite heaStar was successfully launched in August l997. It has brought to oceanOgraphers a wel-comed and improvd renewal of the ocean color of CZCS, VHRSR and arS. ffeaWiFS con-tributes much bette…     

我国海区SeaWiFS资料大气校正

利用光谱辐射传输理论，结合海上同步实测资料，开发出我国海区SeaWiFS资料大气校正模型。经卫星资料处理结果比对，本模型在一类水体，基本消除了412nm和443nm波段离水辐射率小于0的现象；在二类水体，利用临近一类水体的大气条件进行了有效的大气校正；同时建立了670nm,765nm,865nm波段的大气校正模型，这三个面适用于高浓度悬浮泥沙的信息提取。本模型用于处理我国海区的SeaWiFS资料比美国NASA模型更适合我国海区特定的大气和海洋环境，为SeaWiFS资料海洋水色信息提取和我国海洋一号（HY－1）及风云一号（FY－1C）卫星资料的大气校正研究提供了技术基础。     

Atmospheric Correction of SeaWiFS Data for the Retrieval of Suspended Sediment in East China Coastal Waters

The data of SeaWiFS (Sea-Viewing Wide Field-of-View Sensor), installed on SeaStar, has been used to generate SSC (suspended sediment concentration) of complex and turbid coastal waters in China. In view of the problems of the SeaDAS (SeaWiFS Data Analysis System) algorithm applied to China coastal waters, a new atmospheric correction algorithm is discussed, developed, and used for the SSC of East China coastal waters. The advantages of the new algorithm are described through the comparison of the restdts from different algorithms.     

Development of atmospheric correction algorithm for Geostationary Ocean Color Imager (GOCI)

This paper describes an atmospheric correction algorithm for Geostationary Ocean Color Imager (GOCI) and its early phase evaluation. This algorithm was implemented in GOCI Data Processing System (GDPS) version 1.1. The algorithm is based on the standard SeaWiFS method, which accounts for multiple scattering effects and partially updated in terms of turbid case-2 water correction, optimized aerosol models, and solar angle correction per slot. For turbid water correction, we used a regional empirical relationship between water reflectance at the red (660 nm) and near infrared bands (745 nm and 865 nm). The relationship was derived from turbid pixels in satellite images after atmospheric correction, and processed using aerosol properties derived for neighboring non-turbid waters. For validation of the GOCI atmospheric correction, we compared our results with in situ measurements of normalized water leaving radiance (nL _w ) spectra that were obtained during several cruises in 2011 around Korean peninsula. The match up showed an acceptable result with mean ratio of the GOCI to in situ nL _w (??), 1.17, 1.24, 1.26, 1.15, 0.86 and 0.99 at 412 nm, 443 nm, 490 nm, 555 nm, 660 nm and 680 nm, respectively. It is speculated that part of the deviation arose from a lack of vicarious calibration and uncertainties in the above water nLw measurements.     

Ocean color products retrieval and validation around China coast with MODIS

Waters along China coast are very turbid with high concentrations of suspended sediment nearly all the time,especially at the Hangzhou Bay,the Changjiang (Yangtze) River Estuary and the shoal along Jiangsu Province.In these turbid and optically complex waters,the standard MODIS ocean color products tend to have invalid values.Because the water-leaving radiances in the near-infrared (NIR) are significant resulting from the strong scattering of suspended particles,the standard MODIS atmospheric correction algorithm often gets no results or produces significant errors.And because of the complex water optical properties,the OC3 model used in the standard MODIS data processing tends to get extremely high chlorophyll-a (Chl-a) concentrations.In this paper,we present an atmospheric correction approach using MODIS short wave infrared (SWIR) bands based on the fact that water-leaving radiances are negligible in the SWIR region because of the extreme strong absorption of water even in turbid waters.A regional Chl-a concentration estimation model is also constructed for MODIS from in situ data.These algorithms are applied to MODIS Aqua data processing in the China coastal regions.In situ data collected in the Yellow Sea and the East China Sea in spring and autumn,2003 are used to validate the performance.Reasonably good results have been obtained.It is noted that water-leaving reflectance in the NIR bands are significant in waters along the China coast with high sediment loadings.The satellite derived and in-situ reflectance spectra can match in the turbid waters along China coast,and there is relatively good linear relationship between satellite derived and in-situ reflectance.The RMSE value of Rrs(λ) is 0.0031 sr ?1 for all the nine ocean color bands (412 to 869 nm).The satellite-derived Chl-a value is in the reasonable range and the root mean square percentage difference is 46.1%.     

Empirical ocean-color algorithms to retrieve chlorophyll-a, total suspended matter, and colored dissolved organic matter absorption coefficient in the Yellow and East China Seas

A bio-optical dataset collected during the 1998?C2007 period in the Yellow and East China Seas (YECS) was used to provide alternative empirical ocean-color algorithms in the retrieval of chlorophyll-a (Chl-a), total suspended matter (TSM), and colored dissolved organic matter (CDOM) absorption coefficients at 440 nm (ag₄₄₀). Assuming that remote-sensing reflectance (Rrs) could be retrieved accurately, empirical algorithms for T_Chl (regionally tuned Tassan??s Chl-a algorithm) in case-1 waters (T_Chl2i in case-2 waters), T_TSM (regionally tuned Tassan??s TSM algorithm), and T_ag440 or C_ag440 (regionally tuned Tassan??s or Carder??s ag₄₄₀ algorithm) were able to retrieve Chl-a, TSM, and ag₄₄₀ with uncertainties as high as 35, 46, and 35%, respectively. Applying the standard SeaWiFS Rrs, T_Chl was not viable in the eastern part of the YECS, which was associated with an inaccurate SeaWiFS Rrs retrieval because of improper atmospheric correction. T_Chl behaved better than other algorithms in the turbid case-2 waters, although overestimation was still observed. To retrieve more reliable Chl-a estimates with standard SeaWiFS Rrs in turbid water (a proxy for case-2 waters), we modified T_Chl for data with SeaWiFS normalized water-leaving radiance at 555 nm (nLw₅₅₅) > 2 mW cm^?2 ??m^?1 sr^?1 (T_Chl2s). Finally, with standard SeaWiFS Rrs, we recommend switching algorithms from T_Chl2s (for case-2 waters) to MOC_Chl (SeaWiFS-modified NASA OC4v4 standard algorithm for case-1 waters) for retrieving Chl-a, which resulted in uncertainties as high as 49%. To retrieve TSM and ag₄₄₀ using SeaWiFS Rrs, we recommend empirical algorithms for T_TSM (pre-SeaWiFS-modified form) and MT_ag440 or MC_ag440 (SeaWiFS Rrs-modified forms of T_ag440 or C_ag440). These could retrieve with uncertainties as high as 82 and 52%, respectively.     

Atmospheric correction scheme for GLI with absorptive aerosol correction

The present study proposes an atmospheric correction scheme for Advanced Earth Observation Satellite II (ADEOS-II)/Global Imager (GLI) ocean color retrieval that corrects for the atmospheric absorptive aerosol effect. Radiative transfer simulations were conducted assuming a non-absorptive model aerosol with a soot-type aerosol at various mixture ratios. The results indicate that while the spectral dependency of aerosol reflectance does not change in the longer (>550 nm) wavelength region, the reflectance at shorter wavelengths is highly variable and depends on the mixture ratio. The influence of aerosol absorption was also investigated using GLI data from ocean areas adjacent to Japan in the presence of absorptive Siberian fire smoke aerosol in the spring of 2003. The spectral curvature of the aerosol was estimated from the data obtained. An empirical, iterative scheme that detects and evaluates the influence of absorptive aerosols was developed by comparing 380 nm GLI-observed aerosol reflectance with predicted reflectances derived using an in-water optical model. To evaluate the performance of this scheme, satellite-derived normalized water-leaving radiances were compared with those measured from a ferry servicing Nagasaki and Fukue. The results of data acquired on March 20, 2003, indicate that this absorption correction scheme improved root mean square estimation error for normalized water-leaving radiance by approximately 40% in the 380, 400, and 412 nm bands. This atmospheric correction algorithm was used as a part of the second version of the GLI standard ocean color data process system at Japan Aerospace Exploration Agency (JAXA).     

Correction of atmospheric effect on ADEOS/OCTS ocean color data: Algorithm description and evaluation of its performance

This paper first describes the atmospheric correction algorithm for OCTS visible band data used at NASDA/EOC. Sharing a basic structure with Gordon and Wang’s Sea WiFS algorithm, it uses 10 candidate aerosol models including the “Asian dust model” introduced in consideration of the unique feature of aerosols over the east Asian waters. Based on the observations at 670 and 865 nm bands, the algorithm selects a pair of aerosol models that account best for the observed spectral reflectances, and synthesizes the aerosol reflectance used for the atmospheric correction. Two different schemes for determining the value of the parameter for the aerosol model selection are presented and their anticipated estimation error is analyzed in terms of retrieved water reflectance at 443 nm. The results of our numerical simulation show that the standard deviation of the estimation error of the “weighted average” scheme is mostly within the permissible level of ±0.002, reducing the error by 18% on average compared to the “simple average” scheme. The paper further discusses the expected error under the old CZCS-type atmospheric correction, which assumes constant aerosol optical properties throughout the given image. Although our algorithm has a better performance than the CZCS algorithm, further analysis shows that the error induced by the assumption taken in the algorithm that the water-leaving radiance at 670 nm band is negligibly small may be large in high pigment concentration waters, indicating the necessity for future improvements.     

二类水体的MODIS数据大气校正及在渤海表层水体叶绿素浓度监测反演中的应用研究

提出一种适用于中国近岸二类水体的大气校正方法,解决了水色遥感软件SEADAS由于无法计算近岸的二类水体离水辐射率而简单地将其设为0的问题。应用该方法计算出的离水辐射率反演了渤海表层水体的叶绿素浓度,并选取近岸同期监测点的实测值对反演结果进行验证。通过对比反演值与实测值,多数相对误差值均较小于10%,证明了反演结果较为准确,同时反演的叶绿素浓度的空间分布也符合冬季渤海的水团和环流状况。这说明应用改进后的大气校正方法能够很好地用于渤海表层水体的叶绿素浓度反演。     

Penetration of coastal waters into the Eastern Mediterranean Sea using the SeaWiFS data

Offward transport of surface coastal waters was studied in the Eastern Mediterranean from 31° to 37° N between 31° to 36° E using cluster analysis of water-leaving radiance spectra from SeaWiFS color scanner imagery for spring-summer period of 1998. Some types of spectra were found to be characteristic of products of mixing of coastal and offshore waters and served as a quasiconservative tracer of water transport. Three sectors of the eastern coastal zone supplied colored and suspended matter into the open sea areas as occasional pulse discharges of 3-5 week periodicity. The Haifa sector was the most powerful source of such matter producing traceable separate radiance inhomogeneities as far as 200 km west of Haifa. It appears that the discharges were triggered when a set of specific conditions took place and that evolution and life time of discharged waters depended on currents and horizontal mixing. Advantages of cluster approach to imagery analysis are briefly discussed.     

海水层辐射校准技术研究

为提高光学资料的利用率，确保光学遥感信息的精确度，必须对卫星获得的资料进行辐射校准，本文着重阐述我们通过在我国典型海区实测获取与卫星同步的海面归一化离水辐射率，对实用的大气模型进行选择及参数修正的研究，对卫星遥感资料获得的海面离水辐射率进行校准，论文给出了研究的一些主要结果。     

The black water around the Changjiang (Yangtze) Estuary in the spring of 2003

The Changjiang (Yangtze) Estuary is located in the East China Sea shelf with shallow water. Affected by the tide mixing and the runoff of the Changjiang River and the Qiantang River, the turbidity is very high. Generally, the water-leaving radiance is high in the turbid water because of the large particle scattering. Based on the in-situ data and ocean color remote sensing data of SeaWiFS, it was found that there was a black water region with the normalized water-leaving radiances less than 0.5 mW/(cm2·μm2·...     

2003年春季长江口海域黑水现象研究

长江口海域处于东海大陆架,水深较浅,受长江、钱塘江入海径流及潮汐搅拌的影响,水体浑浊度高,属于典型的二类水体。在浑浊二类水体由于水体中悬浮颗粒的散射作用,离水辐亮度通常较大洋清洁水体的大。通过分析2003年春季黄海、东海二类水体光学试验数据和同时段SeaWiFS卫星遥感数据,发现在长江口东南海域有一个归一化离水辐亮度极小的黑水区域,它在412~670nm波长的归一化离水辐亮度均小于0.5mW/(cm2·μm·sr)。利用现场实测的水体固有光学特性参数(色素吸收系数、非色素颗粒吸收系数、黄色物质吸收系数、颗粒散射系数、颗粒后向散射系数等)和水色要素浓度数据,分析了长江口海域黑水现象的机制,结果表明,长江口外黑水现象主要是由于水体后向散射系数值极小引起的,而造成颗粒后向散射系数值极小的原因主要有两种,一是颗粒物含量低,二是颗粒后向散射率的值小。黑水区浮游植物色素吸收所占的比例较高,大粒径的有机颗粒(浮游植物)导致黑水区域的颗粒后向散射率的值偏小。长江口海域黑水是在特定的颗粒物低含量及颗粒后向散射率极小的条件下出现的光学现象。     

Comparative Analysis of Chlorophyll-a Distribution from SeaWiFS,MODIS-Aqua,MODIS-Terra and MERIS in the Arabian Sea

The variability of Chlorophyll-a (Chl-a) distribution derived from MODIS (on Aqua and Terra platforms) and MERIS sensors have been compared with SeaWiFS data in the Arabian Sea. MODIS Aqua has overestimated the SeaWiFS Chl-a within 25–32% in the coastal turbid (eutrophic) waters and underestimated in open ocean waters with error within 20%. However, there is no significant bias (?0.1 on log-scale) observed as the slope is well within 0.97-1.1 (log transformed). MODIS-Terra has underestimated the Chl-a concentration in open ocean waters by about 29–31%, which is higher than MODIS-Aqua. MODIS-Terra is observed to be more accurate than MODIS-Aqua in the coastal waters. MERIS is overestimating the SeaWiFS Chl-a with log RMS error of ～0.15 and log bias of ～0.13–0.2. The differences in the Chl-a estimates between each sensor are possibly due to differences in the sensor design, bio-optical algorithms and also due to the time differences between the satellites over passes. We have examined that the MERIS is performing similar to SeaWiFS and the MODIS-Aqua (Terra) data are reliable in open ocean (coastal) waters. However, Chl-a retrieval algorithms need to be improved especially for coastal turbid waters to continue with SeaWiFS data for long-term studies.     

MODIS和GOCI卫星遥感反射率产品在浑浊海区交叉检验分析

对Geostationary Ocean Color Imager(GOCI)和Moderate Resolution Imaging Spectroradiometer(MODIS)传感器在中国渤海辽东湾海区的卫星大气校正算法开展评估工作。主要对比了GOCI和MODIS的412 nm,443 nm,488 nm,547 nm,678 nm波段的遥感反射率(Remote Sensing Reflectance:Rrs)。结果表明:GOCI的去云算法较严格,在卫星有效数据覆盖率方面差于MODIS;遥感反射率产品比对结果表明:GOCI和MODIS的遥感反射率产品有较好的线性相关,且GOCI反演值大于MODIS反演值;分区域的对比结果表明,MODIS和GOCI的遥感反射率差异随着水体的浑浊度增加而增大,GOCI需要开发适用于近岸水体的大气校正算法。     

Indication of bottom transport in shallow marine regions based on the data of satellite ocean color scanners

By example of a testing area in the northern Caspian Sea, a case study has been conducted to examine the feasibility of indication of near-bottom water transport from the data of ocean color scanners (OCS) for the basins whose bottom is shallower than the lower limit of the water-leaving radiance origination layer. The testing area has provided the desirable diversity of radiance origination conditions owing to an isolated underwater ridge. Based on the archived materials of the OCS SeaWiFS for 2000–2004, statistics of spectral normalized water-leaving radiances of the testing area have been computed. Hypothetically, the displacements of a radiance maximum in reference to the ridge were due to different mechanisms of interaction of water motions with bottom relief and bottom sediments. The correspondence between the patterns of radiance maximum displacements and the current notions of general water transport in the vicinity of the testing area has been established. Hopefully, the archived data of OCSs can be useful for revealing the patterns of seasonal and interannual variability of the near-bottom water transport at shallows with favorable bottom properties.     

Evaluation of GOCI sensitivity for At-Sensor radiance and GDPS-Retrieved chlorophyll-a products

The signal-to-noise ratio (SNR, or sensitivity) of an ocean color instrument is a critical parameter to determine the accuracy and precision of the data products. Yet published literature showed various formats in SNR specifications under different conditions, making a direct cross-sensor comparison difficult. Here, we compared the SNRs of GOCI spectral bands with those of SeaWiFS and MODIS/Aqua under the same radiance inputs. We also compared their ability to resolve small changes in the retrieved chlorophyll-a data products (Chl). While GOCI visible bands showed similar at-sensor SNRs to SeaWiFS, the near-infrared (NIR) bands showed significantly higher SNRs. Because the NIR bands were used for atmospheric correction, the increases in SNRs led to reduced noise in the retrieved Chl, as shown in the GOCI and SeaWiFS Chl products for Chl < 0.1 mg m^?3. The noise in the retrieved products also depends on the retrieval algorithms in addition to the sensor SNR. When a new band-subtraction algorithm (the Ocean Color Index or OCI algorithm) was applied to the same GOCI remotesensing reflectance data derived from the GDPS software package, significant noise reduction was found in the Chl product for low concentrations (< 0.25 mg m^?3), leading to product precision (??3% in Chl) comparable to those from MODIS/Aqua measurements. This is certainly a significant achievement, as GOCI spatial resolution is much higher than MODIS (500 m versus 1 km). In addition, artifacts across image mosaic edges over low-concentration waters have been removed nearly completely by the OCI algorithm. Data analyses also indicated that GOCI radiometric calibration requires further improvement.     

Satellite-measured temporal and spatial variability of the Tokachi River plume

River plumes have important effects on marine ecosystems. Variation in the extent and dispersal of river plumes is often associated with river discharge, wind characteristics and ocean circulation. The objectives of this study were to identify the Tokachi River plume by satellite, determine its relationship with river discharge and clarify its temporal and spatial dynamics. SeaWiFS multispectral satellite data (normalized water-leaving radiance: nL_w) with 1.1 km spatial resolution were used to determine the spatial and temporal variability of the plume during 1998–2002. Supervised maximum likelihood classification using six channels of nL_w at 412, 443, 490, 510, 555 and 670 nm with each band's spectral signature statistic was used to define classes of surface water and to estimate the plume area. Supervised maximum likelihood classification separated three to four classes of coastal water based on optical characteristics as a result of wind stress events. The satellite-observed plume area was correlated with the amount of river discharge from April to October. The plume distribution patterns were influenced by wind direction and magnitude, the occurrences of a near-shore eddy field and surface currents. Empirical orthogonal function (EOF) was used to express the spatial and temporal variability of the plume using anomalies of nL_w(555) monthly averaged images. The first mode (44% of variance) showed the turbid plume distribution resulting from re-suspension by strong wind mixing along the coast during winter. This mode also showed the plume was distributed along-shelf direction in spring to early autumn. The second mode (17% of variance) showed spring pattern across-shelf direction. EOF analysis also explained the interannual variability of the plume signature, which might have been affected by the flow of the Oyashio Current and the occurrence of a near-shore eddy field.     

Remote sensing and surface POC concentration in the South Atlantic

Several SeaWiFS products have been compared with shipboard data to assess the possibility of using remote sensing to estimate particulate organic carbon (POC) concentration in surface waters. Transmissometer data were collected during six South Atlantic Ventilation Experiment (SAVE) hydrographic expeditions conducted between November 1987 and March 1989 from R/V Knorr, and Melville. A total of 361 beam attenuation profiles were made with a SeaTech transmissometer interfaced with a CTD/rosette. In order to calculate the POC concentration from transmissometer profiles, a regression between beam attenuation and POC for open Atlantic Ocean waters derived from our research in the North Atlantic (North Atlantic Bloom Experiment, NABE) and enhanced by data from the Bermuda Atlantic Time Series Station (BATS) was applied. The profiles were processed and examined as vertical sections of the surface 250 m. The data were collected in two successive years, during the same season, which allowed us to compile a combined data set over the austral summer for examination. Beam attenuation/POC concentrations were integrated down to one attenuation depth with the intent of making comparisons with satellite optical data. No satellite optical data were available for 1987–1989, so the only option was to compare our integrated data with SeaWiFS-derived variables from later years averaged over the same season as SAVE data. Analysis of four SeaWiFS products acquired from 1997 to 2002 demonstrated very low variations from year to year for seasonally averaged data, suggesting that making comparisons of the beam attenuation/POC fields with averaged satellite optical products from later years is a valid (though not optimal) approach for this area. The highest correlation between beam attenuation/POC concentration and remotely derived products was found with normalized water-leaving radiance at 555 nm. Other SeaWiFS-derived variables—chlorophyll concentration, diffuse attenuation coefficient at 490 nm and integral chlorophyll (integrated over one attenuation depth)—were also compared but showed a slightly less satisfactory correlation.     

Validation of aerosol products derived from ocean colour in East African coastal waters

The aerosol products derived from the ocean colour missions SeaWiFS and MODIS (Aqua and Terra) were assessed with AERONET field measurements collected at sites in Mozambique (Inhaca) and Kenya (Malindi). The median of absolute relative differences between satellite and AERONET aerosol optical thickness τ_a at 443 nm varied between 12% and 22% for the different missions and sites. These differences tended to be higher at Malindi and for longer wavelengths. This analysis was supplemented by data collected offshore in the framework of the Maritime Aerosol Network. Results showed a general consistency between satellite missions and with global validation statistics. This suggests that the uncertainties associated with the atmospheric correction scheme and consequently with the distributions of ocean colour reflectance in the study area may be compared with those found in other regions. This result lends some confidence to the use of ocean colour data in the coastal waters of south-eastern Africa.