Distribution of the suspended particulate matter (SPM) concentration is a key issue for analyzing the deposition and erosion variety of the estuary and evaluating the material fluxes from river to sea. Satellite remote sensing is a useful tool to investigate the spatial variation of SPM concentration in estuarial zones. However, algorithm developments and validations of the SPM concentrations in Yellow River Estuary (YRE) have been seldom performed before and therefore our knowledge on the quality of retrieval of SPM concentration is poor. In this study, we developed a new simple optical model to estimate SPM concentration in YRE by specifying the optimal wavelength ratios (600-710 nm)/ (530-590 nm) based on observations of 5 cruises during 2004 and 2011. The simple optical model was attentively calibrated and the optimal band ratios were selected for application to multiple sensors, 678/551 for the Moderate Resolution Imaging Spectroradiometer (MODIS), 705/560 for the Medium Resolution Imaging Spectrometer (MERIS) and 680/555 for the Geostationary Ocean Color Imager (GOCI). With the simple optical model, the relative percentage difference and the mean absolute error were 35.4% and 15.6 gm−3 respectively for MODIS, 42.2% and 16.3 gm−3 for MERIS, and 34.2% and 14.7 gm−3 for GOCI, based on an independent validation data set. Our results showed a good precision of estimation for SPM concentration using the new simple optical model, contrasting with the poor estimations derived from existing empirical models. Providing an available atmospheric correction scheme for satellite imagery, our simple model could be used for quantitative monitoring of SPM concentrations in YRE.
© 2013 Optical Society of America
Suspended particulate matter (SPM) has a great effect on the transparency, turbidity and water color of estuarine and coastal waters [1–3]. Knowledge of the loads, spatial distribution and physical properties of SPM is, therefore, essential to evaluate geomorphologic changes and to monitor water quality, since they relate total primary production to heavy metal and micro pollutants [4,5]. Furthermore, SPM in coastal and estuarine waters play an important role in biogeochemical cycles. The interaction between SPM and seawater constituents may strongly modify the nutrient concentration in estuarine systems. The fine-grained particles are an important carrier of various chemical compounds. Understanding the temporal and spatial dynamics of SPM in estuarine systems can thus allow for the estimation of the transport of terrestrial and anthropogenic materials to pelagic oceans. However, estuarial ecosystems typically exhibit SPM concentrations ([SPM]) with high temporal and spatial variability, which are often too difficult to be characterized by using traditional field sampling methods [2,4,6,7].
Fortunately, satellite imagery can be used to rapidly assess the [SPM] in coastal and estuarine environments at temporal and spatial scales difficult to attain with direct field measurements [2,3,7–9]. Considerable success in [SPM] estimation has been demonstrated with data from a variety of sensors with varying radiometric accuracy and sensitivity and various spatial and temporal resolutions, such as the Sea-viewing Wide Field-of view Sensor (SeaWiFS), the Moderate Resolution Imaging Spectroradiometer (MODIS), the Medium Resolution Imaging Spectrometer (MERIS) and the Geostationary Ocean Color Imager (GOCI) [2,7,9]. Two types of SPM algorithms are widely used for accurate [SPM] estimation: empirical models and semi-analytical models. The empirical methods are based on relationships between [SPM] and single-channel or multi-channel remote sensing reflectance (Rrs) [3,7,10–13]. The semi-analytical algorithm approach consists in connecting Rrs to [SPM] via a simplified optical model by which reflectance is expressed according to the inherent optical properties (IOPs) of absorption and backscattering [6,14–16]. The relationships used in empirical models are normally geographically specific and hardly be directly applied to other coastal areas. In addition, semi-analytical models depend on accurate information for the IOPs, which are difficult to be accurately measured especially in a turbid waters with high [SPM] variation [3,16]. Therefore, only “regional” SPM algorithms are implemented in areas with similar particle characteristics and where [SPM] products can be validated .
Despite considerable success in [SPM] estimation from satellite data, remote sensing of such a complex system in Yellow River Estuary (YRE) is quite challenging. Optical properties in YRE are quite complicated due to sediment resuspension and river discharging. [SPM] in YRE are strongly influence by a combination of hydrodynamic, physico-chemical, and biological processes . Previously, algorithm development and validation of [SPM] in YRE have seldom been performed , especially for the new ocean color sensor GOCI. The GOCI sensor, launched by the Korean Ocean Space Center in 2010, is the world’s first ocean color sensor in a geostationary orbit . This revolutionary design offers very significant new possibilities for remote sensing of sediment dynamics in tidal regions, because imagery is acquired every hour during daylight, up to a maximum of 8 images. It is possible to resolve high temporal variability and to obtain more days with usable data in periods of scattered clouds. GOCI has great potential to monitor [SPM] in an optically complex estuary such as YRE. However, to our knowledge, no applicable algorithms have been developed and validated to accurately derive [SPM] products in YRE by using GOCI data. Furthermore, no validation for the existing models (such as those listed in Table 3) with in situ observations in YRE has been published. The existing models might not be applicable to YRE, although they were successfully developed and used in other waters. Therefore, significant efforts on improving the accuracy of satellite-derived [SPM] are required in such areas.
In the present study a simple optical algorithm for estimating [SPM] is calibrated and validated based on in situ observations of 5 cruises during 2004 and 2011 in YRE. The specific calibration and validation for MODIS, MERIS and GOCI have been performed for the model's applications to multiple ocean color sensors. The existing empirical models are also validated by using the in situ observations.
2. Materials and Methods
2.1 Study area
The study area locates in the YRE and its adjacent waters (Fig. 1). The Yellow River is one of the most sediment-laden rivers in the world. The river is about 5460 km long with a drainage basin covering ca. 752,000 km2 . Historically, the river discharged an average of 574 × 108 m3 of water and 10 × 108 t of sediment annually, corresponding approximately 50–60% of the freshwater and more than 90% of the sediment received by the Bohai Sea [20,21]. The river outflow introduces SPM, dissolved and particulate organic matter into the Bohai Sea, potentially affecting the marine environment and representing an important flux of carbon. The riverine material strongly affects the optical properties of the coastal waters as seen in Fig. 1, making satellite sensors the most suitable tools available to map the river influence on the Bohai Sea. Previous studies  have shown that high spatio-temporal variability is observed in the distributions of SPM, which are closely related to hydrodynamic processes, such as winds, tides and currents. Both estuarine gravitational circulation and tidal asymmetry are important with the complex flow patterns largely determined by the interaction of Yellow River runoff, tidal currents, and varying topography, along with density differences and winds. The complicated hydrodynamics lead to very high SPM concentrations and high turbidity in the YRE.
2.2 Field data collection and processing
A total of 5 oceanographic surveys were conducted in YRE between June 2004 and December 2011 (Table 1). ). Sampling stations covered 119 °E-120 °E and 37.5 °N −38.5 °N (Fig. 1). Waters are optically complex with the values of the water constitutes varying in a wide range (in 2-3 orders). The concentrations of chlorophyll a (Chl) were in 0.01 mgm−3-62.9 mgm−3, and the concentrations of the SPM were in 1.69 gm−3-1896.5 gm−3. The [SPM], the Rrs and other environmental parameters were synchronously measured strictly following the NASA SIMBIOS ocean optic protocols .
The Rrs used in this study were measured by an ASD FieldSpec Dual VNIR, covering the spectral range of 350-1050 nm. The absolute radiance calibration of detectors was performed before each cruise. During the measurements, the tip of the optical fiber was kept ~1 m above the water surface by means of a 3 m long hand-handle pole. The zenith and azimuth angles viewing from the water surface are about 40° and 135°, respectively, which were determined by a hand-handle with adjusted-angle equipment. The integration time was chosen according to the intensity of radiance received by the ASD detector and the dark reading was obtained each time when the integration time was changed. The measurement location was selected with minimal shading, reflections from superstructure, ship wake and associated foam patches as well as whitecaps. Additionally, the measurement location was selected to point easily to a direction away from the sun glint.
The measured Rrs is calculated using Eq. (1):24] and wavelength-independent . The value of ρis always in the range of 0.022–0.05 [25,26]. When ρis out of the range of 0.022-0.05, it is determined from the sea conditions .
The [SPM] (units are gm−3), defined as the dry mass of particles per unit volume of water, was determined using a standard gravimetric technique . At each station, water samples were collected just below the sea surface with 10-liter Niskin bottles simultaneously with in situ optical measurements. The water sample was filtered with 0.45 m filter (Whatman GF/F filters) and vacuum filtration system. To remove salt, filters were washed with 250 mL of MilliQ water 3 times after filtration. Filters were dried for 24 h at 40°C and reweighed to obtain [SPM]. The dry-weight of the filter-pad was weighed by an electronic analytic scale. The blank filter and sampled filter-pad were weighed until the difference between two successive [SPM] calculated from the scale reading was within 0.01 gm−3.
2.3 Data analysis
After quality control we have the data set of 122 samples with paired Rrs and [SPM]. To calibrate and validate our SPM model, this data set was randomly divided into two groups, namely the calibration data set (n = 81) and the validation data set (n = 41).
Statistical analysis (mean value, linear and non-linear fitting) are performed with MATLAB software. The performances of the retrievals are evaluated by the correlation coefficient (r2), the relative percentage difference (RPD) and the mean absolute error (MAE). The calculation of RPD was the same as the previous study  and MAE was described as Eq. (2)
3.1 Variation of spectral characteristics and correlation with [SPM]
The Rrs spectra of the water mass with various [SPM] in the 2004 cruise in YRE are shown in Fig. 2.In general, the Rrs is highly variable over the visible and near-infrared spectral regions. In addition, the values of the Rrs increase with [SPM], especially in the red and near-infrared range (wavelength larger than 600 nm).
In the low turbidity water (with [SPM] <15 gm−3) [Fig. 2(a)], the values of the Rrs increase as a function of the wavelength from the blue to green band, reaching the peak at ca. 570 nm. The values decrease quickly in the range of 570-600 nm, and finally end with the minimum at the infrared band (larger than 700 nm). The downslope (from 570 to 600 nm) is generally steeper than the upslope (from the blue to green band).
In the modulate turbidity water (with 15 gm−3<[SPM]<150 gm−3) [Fig. 2(b)], the spectra shapes are similar to those in the low turbidity water in the blue and green bands. The values decrease slightly during 570- 700 nm and steep downslope is observed in the range of 720-740 nm. In the near-infrared range the values remain higher than ca. 0.012 Sr−1.
In the high turbidity water (with [SPM]>150 gm−3) [Fig. 2(c)], the Rrs spectra display an asymptotic approach to a broad peak ~570-700nm. In the near-infrared range (700-900 nm) the Rrs spectra vary widely with a second peak located at ca. 810 nm.
3.2 Estimation model ofSPM: calibration
The calibration data set contained 81 water samples, with [SPM] ranging from 1.9 to 1896.5 gm−3 with a mean value of 88.1 ± 256.8 gm−3 (mean ± standard deviation). To find the best wavelength band, or band ratios, by which to estimate [SPM] in YRE, the single band and the ratios of any two wavelengths from 400 to 900 nm were tested for correlation with [SPM] based on the exponential [Eq. (3)] and quadratic algorithms [Eq. (4)].
The quadratic algorithm using single band showed a reasonable correlation with [SPM] within wavelengths from 700 to 900 nm (Fig. 3). Two valleys near 740 nm and 810 nm were found in the Fig. 3(a). However, the relative percentage difference is high with the value larger than 80% at all wavelengths. The exponential algorithm using single band performed worse than the quadratic algorithm. The results showed that the two algorithms using single band could not be used to estimated accurately [SPM] in the calibration data set.
Algorithms using band ratios performed better than the algorithms using single band. For simplicity, we only present the RPD in Fig. 4 (while RPD < 60%). Both algorithms using band ratios showed a good precision with the band ratios (600-710 nm)/(530-590 nm), with the RPD<35%. In addition, low MAE and high r2 are also observed at the band ratios (data not shown). Therefore, the exponential algorithm using band ratios (600-710 nm)/ (530-590 nm) is recommended, considering the simplicity of the model.
To apply the new model presented in this study to the satellite data, the spectra recorded by ASD were aggregated using the spectral response functions of the satellite sensors MODIS, MERIS and GOCI. The model band ratios were determined for MODIS, MERIS and GOCI spectral bands. After tuning the band ratios are chosen as 678/551 for MODIS, 705/560 for MERIS and 680/555 for GOCI (Table 2). The models for all sensors give a good estimate of [SPM] for YRE (Fig. 5), with RPD 32.6%, 33.2% and 33.2%, and r2 0.95, 0.98 and 0.95 for MODIS, MERIS and GOCI, respectively. The model for MERIS (MAE = 16.6 gm−3) performs a little better than the MODIS (MAE = 24.5 gm−3) and GOCI (MAE = 26.2 gm−3).
3.3 Estimation model of SPM: validation
To further understand the applicability of the simple optical model to estimate [SPM], we evaluated its performance using a validation data set of 41 samples. The [SPM] in the validation data set varied from 1.69 to 393.6 gm−3 with a mean value of 51.8 ± 99.7 gm−3, which fell into the range of [SPM] used to calibrate the model.
Comparisons of the measured and estimated [SPM] from the calibrated simple optical models for MODIS, MERIS and GOCI showed that these values were in close agreement (Fig. 6), with a highly significant relationship with an r2 of 0.93, 0.93 and 0.94, the MAE of 15.6 gm−3, 16.3 gm−3 and 14.7 gm−3, and the corresponding RPD of 35.4%, 42.2% and 34.2%, respectively. The measured and estimated values of [SPM] were distributed along the 1:1 line (Fig. 6), indicating that the simple optical model could be used for the turbid waters of YRE.
3.4 Estimation model of SPM: sensitivity analysis
A sensitivity analysis was performed for the simple optical model by using the validation data set of 41 samples. Random errors were added 50 times into the in situ Rrs values. The random errors were drawn from the standard uniform distribution with the mean value 0 and the standard deviation 5%. Totally we have 2050 samples with <5% errors being randomly added into the Rrs values.
Comparisons of the measured and estimated [SPM] from the errors-added Rrs demonstrated that these values were in close agreement. The values of RPD of all three models varied in the range of <5%. Here for simplify we only presented the MODIS model in Fig. 7.Comparing Fig. 6(a) and Fig. 7 we can find that the distributions of estimated [SPM] are very similar in two Figs. The measured and estimated values of [SPM] were distributed along the 1:1 line. The value of RPD is 39.4% for the [SPM] estimated from the errors-added Rrs, which is 4% higher than that for the [SPM] estimated from the real in situ Rrs. The value of MAE is 16.2 gm−3 and 0.6 gm−3 higher accordingly. Therefore, the sensitivity analysis indicates that the simple model is robust to estimate [SPM] in the YRE, at least in the ranges of the data set.
4.1 Variation of the Rrs in YRE
In section 3.1 we introduced the variation of the Rrs over the visible and near-infrared spectral regions, which is similar to the results published in other turbid waters [3,14]. As shown in Fig. 2, the dramatic impact of [SPM] on volume reflectance is clear. The shapes of the Rrs are significantly different between various [SPM]. The [SPM] can substantially increase the volume reflectance in a manner that becomes more pronounced as the wavelength becomes longer.
The peak of the Rrs at ca. 570 nm is mainly due to high backscattering from SPM. The second peak of the Rrs at 810 nm in the turbid waters is the result of both high backscattering and a minimum in absorption by all optically active constituents including pure water. The Rrs at 570 nm is normally higher than that at 700 nm for [SPM] < 150 gm−3. This relation is reversed for [SPM] > 150 gm−3, because Rrs at 570 nm tends to saturate while [SPM] increases.
The variation of the Rrs relating to [SPM] is not only determined with [SPM], but also the different composition (refractive index, density) and size distribution. For example, Bowers and Binding  reported that the smaller sized sediments generally lead to a higher spectral reflectance. Shen et al.  also presented that particle size takes significant effect on Rrs with a dependency not only on spectral wavelength but also on [SPM] ranges.
4.2 Assessment and application of the simple optical model
In order to compare the estimation precision of our simple optical model with that of previous models, we firstly calibrated the coefficients for the same model expressions (Table 3) using the calibration data set, including Miller model , Doxaran model , Tassan model , and Zhang model . Then we assessed those models by using the same independent validation data set we used to validate our own model.
Significant differences were found between estimated and measured [SPM] values, with RPD of 61.6%, 125.5%, 73.31% and 76.49% for Miller model, Doxaran model, Tassan model and Zhang model, respectively (Table 3). The Miller model used reflectance at 645 nm as the indicator of [SPM]. Figure 3 have indicated that a single band model is not suitable for the data set.
The Doxaran model estimated [SPM] with the largest RPD among the four models, which also presented in Fig. 8.The reason is that the Doxaran model used the Rrs at 840 nm. Figure 2(a) displayed that the values of Rrs in the range of larger than 750 nm are small when [SPM] are lower than 15 gm−3. Therefore, The Doxaran model using ratio Rrs(840)/Rrs(545) is unable to characterize the Rrs differences in waters with low [SPM]. The estimated [SPM] are almost 10 gm−3 when the measured [SPM] varied in the range of 1- 15 gm−3[Fig. 8(b)], which denotes that the Doxaran model is failure in the clear or moderate turbid waters of YRE. Figure 4 have also indicated that the ratio Rrs(840)/Rrs(545) is not appropriate for estimating [SPM] in such a data set.
The Tassan model has attained a similar performance as the Zhang model. Both models used Rrs(555), Rrs(645) and Rrs(488)/Rrs(555). Although the ratio Rrs(488)/Rrs(555) can represent partly the variation of [SPM], Fig. 4 have displayed that the ratio is not the best choice for estimating [SPM]. Figure 8 denote that the estimated [SPM] scattering along the 1:1 line.
All the four models are successfully applied in some regions [3,12,13,16], they are not appropriate for YRE even after an attentive tuning (Fig. 8 and Table 3). One reason is that the relationship between reflectance and [SPM] are different in various regions. Shen et al.  have introduced that the interrelationships between Rrs and sediment characteristics are largely difference in the highly turbid waters of the Yangtze River estuary and the YRE. The second reason is that the variation of the Rrs relating to SPM is not only determined with [SPM], but also the other properties, such as size distribution. Shen et al.  presented that the effect of particle size of SPM on the observed Rrs is significant and depends on wavelengths and a [SPM] range. The SPM composition is also observed to affect the values of Rrs. Therefore, even though a regional algorithm was successfully developed in turbid water, the algorithm is difficult to directly apply in other regions, such as the turbid waters of YRE.
The design of the new approaches presented in this study is different from the four models, although they are all empirical algorithms. Our simple optical model used band ratios (600-710 nm)/ (530-590 nm). Figure 2 indicated that the first peak of Rrs values located in about 550-590 nm and the values varied significantly with the variation of [SPM] in the range 600-710 nm. Therefore, band ratios (600-710 nm)/ (530-590 nm) is capable to reflecting the variations of [SPM]. Our simple optical model is based on optical characteristics of Rrs in waters with various [SPM].
By comparison, our simple optical model produces a superior performance to all of the four models. Using of our simple optical model for the three sensors in estimating [SPM] in YRE decreases the RPD values of estimation by > 20%.
Our simple optical algorithm to estimate [SPM] is calibrated and validated using the in situ data set. A few caveats need to be considered when attempting to apply the model to satellite data. The successful applications of our simple model to satellite data depend heavily on the accuracy of the atmospheric corrections. Our model relies strongly on reflectance in red and NIR region (for example, reflectance at 705 nm used for MERIS). There are some specific hurdles that are to be expected. No operational atmospheric correction procedures have been proved universally robust in the NIR region across waters of varying geophysical features, especially in the turbid waters of YRE. In theory, the Wang's model seems a feasible option , which involves the use of shortwave infrared (SWIR) bands for aerosol model selection. However, some practical experiments with MODIS data has shown that even though ocean color products in turbid coastal waters can be improved using SWIR bands-based model, the extent of improvement is very limited due to the considerably lower sensor SNR values for the MODIS SWIR bands . Furthermore, no SWIR bands are available for GOCI data, which have shown great potentials in monitoring variations of SPM in turbid waters such as YRE. Nowadays great efforts are being made to improve the accuracy of atmospheric correction in turbid waters, especially in the waters in China seas, such as [3,7,8]. With the retrieval of accurate reflectance our simple models can be appropriately applied to estimate [SPM] in YRE.
4.3 SPM mapping based on the simple optical model
Providing an overall SPM model performance evaluation, Fig. 9 shows maps of [SPM] in YRE on July 16, 2005. Here for simplify we only present SPM mapping based on the simple optical model for MERIS. The [SPM] is derived from the atmospheric correction results of Schroeder's method , which has been recently validated by Cui et al. .
High [SPM] distributed along the coast areas (Fig. 9), especially in the mouth of YRE and the west coast of the Laizhou Bay. The concentrations along the coast are higher than in the outside part in the Bohai Sea. As a result, the apparent offshore decreases of the [SPM] indicate that waters along the offshore are more turbid than those of the Central Bohai Sea. A high [SPM] areas is obviously observed in the central Laizhou Bay, which is connected with the mouth of YRE. This pattern indicated that the SPM from the Yellow River affect mainly the areas in the Laizhou Bay. The distributions of [SPM] are consistent with the previous works [19, 20, 22], which indicated that our ratio algorithm is appropriate to estimate [SPM] in YRE.
Distribution of the SPM concentration is a key measure for analyzing the deposition and erosion variety of the estuary and evaluating the material fluxes from river to sea. Satellite remote sensing is a useful tool to investigate the spatial variation of SPM concentration in estuarial zones. In this study, we developed a new simple optical model to estimate [SPM] by specifying the optimal wavelength ratios (600-710 nm)/ (530-590 nm). The simple optical model was superior for application in YRE when compared to the published empirical models, using an independent validation data set. Thus, the simple optical model improved the [SPM] estimation precision in YRE.
Further study may include determining the temporal-spatial distribution of [SPM] using the satellite data based on the simple optical model calibrated and validated in this study. The new algorithms will be used to process large satellite data archives from MODIS, MERIS and GOCI in YRE. Inter-sensor comparisons will be made for [SPM], providing a quality check on the GOCI data, and GOCI data will be used to assess the temporal variability from the MODIS and MERIS archives. Furthermore, the satellite [SPM] data will be combined with available hydrodynamic and meteorological information on wind speed, tidal currents and river discharges. The integrated data set will be studied in order to interpret sediment transport processes in the region, characterizing SPM variability in terms of tide- and wind-driven advection and resuspension processes.
This study was jointly supported by the Public Science and Technology Research Funds Projects of Ocean (201005030), the National Natural Science Foundation of China (41276186), the Fund from NUIST (S8111005001)and a project funded by “the Priority Academic Program Development of Jiangsu Higher Education Institutions (PAPD)”. We are thankful to two anonymous reviewers who provided substantial comments and suggestions that led to the improvement of this manuscript.
References and links
1. D. Doxaran, P. Castaing, and S. J. Lavender, “Monitoring the maximum turbidity zone and detecting fine-scale turbidity features in the Gironde estuary using high spatial resolution satellite sensor (SPOT HRV, Landsat ETM+) data,” Int. J. Remote Sens. 27(11), 2303–2321 (2006). [CrossRef]
2. F. Shen, W. Verhoef, Y. Zhou, M. S. Salama, and X. Liu, “Satellite estimates of wide-range suspended sediment concentrations in Changjiang (Yangtze) estuary using MERIS data,” Estuaries Coasts 33(6), 1420–1429 (2010). [CrossRef]
3. M. Zhang, J. Tang, Q. Dong, Q. Song, and J. Ding, “Retrieval of total suspended matter concentration in the Yellow and East China Seas from MODIS imagery,” Remote Sens. Environ. 114(2), 392–403 (2010). [CrossRef]
4. M. P. Rainey, A. N. Tyler, D. J. Gilvear, R. G. Bryant, and P. McDonald, “Mapping intertidal estuarine sediment grain size distributions through airborne remote sensing,” Remote Sens. Environ. 86(4), 480–490 (2003). [CrossRef]
5. F. Shen, M. Suhyb Salama, Y. Zhou, J. Li, Z. Su, and D. Kuang, “Remote-sensing reflectance characteristics of highly turbid estuarine waters–a comparative experiment of the Yangtze River and the Yellow River,” Int. J. Remote Sens. 31, 2639–2654 (2010).
6. V. Volpe, S. Silvestri, and M. Marani, “Remote sensing retrieval of suspended sediment concentration in shallow waters,” Remote Sens. Environ. 115(1), 44–54 (2011). [CrossRef]
7. X. He, Y. Bai, D. Pan, N. Huang, X. Dong, J. Chen, C. A. Chen, and Q. Cui, “Using geostationary satellite ocean color data to map the diurnal dynamics of suspended particulate matter in coastal waters,” Remote Sens. Environ. 133, 225–239 (2013). [CrossRef]
8. Z. Mao, J. Chen, D. Pan, B. Tao, and Q. Zhu, “A regional remote sensing algorithm for total suspended matter in the East China Sea,” Remote Sens. Environ. 124, 819–831 (2012). [CrossRef]
9. R. Arnone, J. Acker, S. Ouillon, R. Gould, N. Rese, and N. Res, “Measuring marine suspended sediment concentrations from space: history and potential,” in 8th International Conference on Remote Sensing for Marine and Coastal Environments,Halifax, NS, Canada(2005).
10. Z. Chen, C. Hu, and F. Muller-Karger, “Monitoring turbidity in Tampa Bay using MODIS/Aqua 250-m imagery,” Remote Sens. Environ. 109(2), 207–220 (2007). [CrossRef]
11. Z. Han, Y. Q. Jin, and C. X. Yun, “Suspended sediment concentrations in the Yangtze River estuary retrieved from the CMODIS data,” Int. J. Remote Sens. 27(19), 4329–4336 (2006). [CrossRef]
12. R. L. Miller and B. A. McKee, “Using MODIS Terra 250 m imagery to map concentrations of total suspended matter in coastal waters,” Remote Sens. Environ. 93(1-2), 259–266 (2004). [CrossRef]
13. S. Tassan, “An improved in-water algorithm for the determination of chlorophyll and suspended sediment concentration from Thematic Mapper data in coastal waters,” Int. J. Remote Sens. 14(6), 1221–1229 (1993). [CrossRef]
14. B. Nechad, K. G. Ruddick, and Y. Park, “Calibration and validation of a generic multisensor algorithm for mapping of total suspended matter in turbid waters,” Remote Sens. Environ. 114(4), 854–866 (2010). [CrossRef]
15. J. A. Warrick, L. Mertes, D. A. Siegel, and C. Mackenzie, “Estimating suspended sediment concentrations in turbid coastal waters of the Santa Barbara Channel with SeaWiFS,” Int. J. Remote Sens. 25, 1995–2002 (2004). [CrossRef]
16. D. Doxaran, J. Froidefond, S. Lavender, and P. Castaing, “Spectral signature of highly turbid waters: application with SPOT data to quantify suspended particulate matter concentrations,” Remote Sens. Environ. 81(1), 149–161 (2002). [CrossRef]
17. M. Ren and Y. Shi, “Sediment discharge of the Yellow River (China) and its effect on the sedimentation of the Bohai and the Yellow Sea,” Cont. Shelf Res. 6(6), 785–810 (1986). [CrossRef]
18. J. Ryu, H. Han, S. Cho, Y. Park, and Y. Ahn, “Overview of geostationary ocean color imager (GOCI) and GOCI data processing system (GDPS),” Ocean Science Journal 47(3), 223–233 (2012). [CrossRef]
19. Y. Qian, Q. Ye, and W. Zhou, Fluctuation of water and sediment discharge and riverbed evolution in the main channel of the Huanghe River(China Building Material Industry Press, 1993).
20. H. Fan and H. Huang, “Response of coastal marine eco-environment to river fluxes into the sea: a case study of the Huanghe (Yellow) River mouth and adjacent waters,” Mar. Environ. Res. 65(5), 378–387 (2008). [CrossRef] [PubMed]
21. J. Z. Pang, M. X. Jiang, and F. L. Li, “Changes and development trend of runoff, sediment discharge and coastline of the Yellow River Estuary,” Transactions of Oceanology and Limnology 4, 1–6 (2000).
22. T. Cui, J. Zhang, Y. Ma, W. J. Zhao, and L. Sun, “The study on the distribution of suspended particulate matter in the Bohai Sea by remote sensing,” Acta Oceanol. Sin. 31, 10–17 (2009).
23. J. L. Mueller, G. S. Fargion, C. R. McClain, J. L. Mueller, S. W. Brown, D. K. Clark, B. C. Johnson, H. Yoon, K. R. Lykke, and S. J. Flora, “Ocean optics protocols for satellite ocean color sensor validation”, Revision 5, Volume VI:special topics in ocean optics protocols, Part 2, NASA Tech. Memo211621 (2003).
25. J. Tang, G. Tian, X. Wang, X. Wang, and Q. Song, “The methods of water spectra measurement and analysis I: above-water method,” Journal of Remote Sensing 8, 37–44 (2004).
26. Z. Lee, K. L. Carder, R. G. Steward, T. G. Peacock, C. O. Davis, and J. L. Mueller, “Protocols for measurement of remote-sensing reflectance from clear to turbid waters,” in SeaWiFS Workshop, Halifax(1996).
29. D. G. Bowers and C. E. Binding, “The optical properties of mineral suspended particles: a review and synthesis,” Estuar. Coast. Shelf Sci. 67(1-2), 219–230 (2006). [CrossRef]
30. M. Wang, S. Son, and W. Shi, “Evaluation of MODIS SWIR and NIR-SWIR atmospheric correction algorithms using SeaBASS data,” Remote Sens. Environ. 113(3), 635–644 (2009). [CrossRef]
31. P. J. Werdell, B. A. Franz, and S. W. Bailey, “Evaluation of shortwave infrared atmospheric correction for ocean color remote sensing of Chesapeake Bay,” Remote Sens. Environ. 114(10), 2238–2247 (2010). [CrossRef]
32. T. Schroeder, I. Behnert, M. Schaale, J. Fischer, and R. Doerffer, “Atmospheric correction algorithm for MERIS above case‐2 waters,” Int. J. Remote Sens. 28(7), 1469–1486 (2007). [CrossRef]
33. T. Cui, J. Zhang, S. Groom, L. Sun, T. Smyth, and S. Sathyendranath, “Validation of MERIS ocean-color products in the Bohai Sea: a case study for turbid coastal waters,” Remote Sens. Environ. 114(10), 2326–2336 (2010). [CrossRef]