Chuqun Chen

Retrieval of oceanic chlorophyll concentration using support vector machines

This letter investigates the possibility of using a new universal approximator-support vector machines (SVMs)-as the nonlinear transfer function between oceanic chlorophyll concentration and marine reflectance. The SeaBAM dataset is used to evaluate the proposed approach. Experimental results show that the SVM performs as well as the optimal multilayer perceptron (MLP) and can be a promising alternative to the conventional MLPs for the retrieval of oceanic chlorophyll concentration from marine reflectance.

Retrieval of water optical properties for optically deep waters using genetic algorithms

Retrieval of water optical properties and concentrations can be identified as a nonlinear optimization problem. This problem may be difficult to solve by conventional optimization methods owing to its multimodel nonconvex nature. This letter explores the potential of genetic algorithms as the optimization scheme in such a problem. A remote sensing reflectance model for optically deep waters was used to illustrate the performance of the algorithms. The superiority of genetic algorithms over conventional optimization methods was demonstrated by experiments on a field dataset.

A Model of the 3D Circulation, Salinity Distribution, and Transport Pattern in the Pearl River Estuary, China

Abstract The three-dimensional circulation, salinity distribution, and overall transport pattern were simulated in the Pearl River Estuary, China, using a modified version of the Princeton Ocean Model (POM). The circulation in the estuary is mainly driven by tide, wind, and river runoff in a complex manner, where the relative importance of the different forcing factors varies over the year. Field data on currents from several locations in the estuary taken at different elevations through the water column were employed to validate the model together with measurements of the salinity distribution in the surface layer. Also, satellite images were utilized to qualitatively confirm the simulated overall transport pattern. Comparisons between measurements and calculations showed that the POM yielded satisfactory predictions without any particular calibration. However, for some events the coastal current outside the estuary induced by the large-scale circulation in the South China Sea affected the flow in the estuary, making it necessary to employ a more sophisticated boundary condition on the ocean side than what was initially implemented. The long-term objective of the numerical modeling of the flow and material transport in the Pearl River Estuary is to utilize the validated model for forecasting the circulation and water quality in the estuary in a comprehensive system where in-situ measurements and remote sensing are the other main components.

A local algorithm for estimation of yellow substance (gelbstoff) in coastal waters from SeaWiFS data: Pearl River estuary, China

A general three-component ocean colour model was used for simulation of water reflectance by inputting sea water component data measured in the Pearl River estuary of southern China. Based on the simulated reflectance data and Dissolved Organic Carbon (DOC) sea water component data a local algorithm for estimation of DOC concentration was developed. The application of the local algorithm shows that the estimated DOC is in close agreement in terms of concentration and distribution pattern with the sea water component data.

Absorption coefficient of yellow substance in the Pearl River estuary

The Pearl River system is mainly located in the Guangdong Province in southern China, with the length of 2214 km and total area of 453,690 km2. The Pearl River estuary is the largest estuary in the South China Sea (SCS), with a mean annual discharge of 326 billion m3, of which are about 30 million tons of dissolved matters annually discharged into the estuary. The high concentration of suspended sediments and dissolved matters makes the optical properties of the coastal waters very complex. The spectral absorption coefficient of yellow substance [Ay(λ)] is one of the inherent optical properties that influence the reflectance (or water-leaving radiance) of the water body. It is essential to measure Ay(λ) and to quantify its contributions to the total absorption of the water body. In this study, the Gelbstoff Optical Analyse Laboratory System (GOALS), with spectral range from 200 to 850 nm and with spectral resolution of 0.37 nm per pixel, was used to measure Ay(λ) in the Pearl River estuary and in the adjacent coastal waters in July 2002. Ay(400) was around 1.5 m-1 near the river mouth (zero salinity). It decreases with increasing salinity following an apparent non-linear mixing line. There is no apparent relationship between Ay(400) and dissolved organic carbon (DOC) concentration, indicating that the estuary is a complex, non-point source environment. This presents a great challenge to remote sensing study in this area.

Using in situ and Satellite Hyperspectral Data to Estimate the Surface Suspended Sediments Concentrations in the Pearl River Estuary

In situ remote sensing reflectance (Rrs) collected during 2004-2006 and the planetary reflectance (Rp) derived from EO-1/Hyperion image, are tested for estimating the surface total suspended matter (TSM), total inorganic particles (TIP) and water turbidity in the Pearl River Estuary (PRE). The in situ data show that the content of TIP and turbidity is proportional to the concentration of TSM which ranges from 6 mg/L to 140 mg/L. The band-subtraction of Rrs at 610 nm and 600 nm, [Rrs(610)-Rrs(600)], and the subtraction of the 26th and 25th Hyperion bands (609.97 nm and 599.80 nm), [Rp(B26)-Rp(B25)], are used in an exponential regression model to estimate the TSM concentrations, the mean relative errors between the estimated and measured TSM are 27.2% and 23.3%, respectively for Rrs and Rp, and the root mean square errors of estimation are 12.6 mg/L and 5.9 mg/L, respectively. This band-subtraction of two neighboring bands shows better performance than several popular single-band and band-combination models. This good performance may be mainly attributed to the band-subtraction of the two neighboring bands which improves the sensitivity of reflectance to suspended sediments by reducing the background impacts from water surface reflection and path radiance at the specific wavelengths. These methods and findings with the high spatial and high spectral resolution data may be used for the remote sensing of turbid estuary waters although further validation work with a wider range of TSM concentration may be necessary.

Estimation of Chlorophyll-a Concentrations in the Pearl River Estuary Using In Situ Hyperspectral Data: A Case Study

Taking Pearl River Estuary (PRE), China as an example, we explored the potential of in situ hyperspectral data in estimating chlorophyll-a concentrations of turbid waters. Two cruises were conducted on August 21, 2006 and May 18, 2004 to collect the data of water quality and remote sensing reflectance (Rrs). The field surveys showed that chlorophyll-a concentration ranged from 2.97 mu g/L to 49.97 mu g/L, and turbidity 13.6-128.9 NTU. The Rrs spectra were binned to 10 nm resolution, and then processed to be first-order and second-order derivatives. A linear algorithm is developed to estimate chlorophyll-a concentrations based on second order derivative at 670 nm; its mean relative errors of estimation is less than 58% and the root mean square error is 6.69 mu g/L, which is better than other popular algorithms for turbid waters, i.e., the ratio of Rrs at 700 nm and 670 nm. The Case-I algorithm of blue-green band ratio is also proved to be a failed application in PRE, and so does the algorithm of fluorescence line height (FLH), which is questionable for its application in waters with strong light scattering and absorption. All the above work was done without classification of cloud conditions. This suggests that the second-order derivative at 670 nm could be effective for estimation of chlorophyll-a concentrations in turbid waters especially in situ.

Using the normalized peak area of remote sensing reflectance in the near-infrared region to estimate total suspended matter

The normalized peak area (NPA) of remote-sensing reflectance (R rs) in the near-infrared region was used to estimate the concentration of total suspended matter (C TSM) in coastal waters. A linear regression model between C TSM and S NPA (R 2 = 0.83) was established, where S NPA is the area encompassed by the reflectance curve and the straight line between wavelengths 768 and 840 nm where there is a maximum of R rs near 715 nm. In the Pearl River estuary of South China, this NPA model performed better than other single-band and multi-band regression models, with a root mean square error (RMSE) of 4.07 mg l–1. This model may be widely applied to in situ measurements of TSM.

Retrieval of suspended sediment concentration in the Pearl River Estuary from MERIS using support vector machines

With the rapid industrialization and urbanization, more and more solid have been emitted into the Pearl River Estuary. The suspended sediment concentration is one of the most important water quality parameters. With in-situ optical data and suspended sediment data collected on four cruises from 2004 to 2006 in the Pearl River Estuary, analysis shows that with the increasing of the total suspended sediment (TSM) concentration, the intensive bands which have the best correlation relationship with the TSM concentration shift from Rrs(620) to Rrs(778). When the mean suspended concentration is 14.5 g.m−3, the Rrs(620) has best correlation with the suspended concentration. However, when the mean suspended concentration becomes more than 40g.m−3, the most correlated band shifts to 778nm. It seems that all of the Rrs(620), Rrs(665), Rrs(681), Rrs(708), Rrs(753), Rrs(760), Rrs(778) may be the most sensitive band for the different TSM concentration. This work investigates the possibility of using a new universal approximator-support vector machines (SVMs)-as the nonlinear transfer function between TSM concentration and remote sensing reflectance in the Pearl River Estuary. Experimental results show that the SVM performs better result than general empirical algorithms or the piecewise algorithm. The correlation coefficient between the in-situ and modeled TSM of the test dataset is 0.91 and the root mean squared error (RMSE) is 0.145. The algorithm based on the SVM is applied to MERIS satellite data in January 31, 2007. The distribution of TSM concentration was obtained and it shows that the algorithm could be a useful tool for the study of TSM distribution in Pearl River estuary.

Decadal variability of chlorophyll a in the South China Sea： a possible mechanism

Four climatologies on a monthly scale (January, April, May and November) of chlorophyll a within the South China Sea (SCS) were calculated using a Coastal Zone Color Scanner (CZCS) (1979-1983) and the Sea-viewing Wide Field-of-view Sensor (SeaWiFS) (1998-2002). We analyzed decadal variability of chlorophyll a by comparing the products of the two observation periods. The relationships of variability in chlorophyll a with sea surface wind speed (SSW), sea surface temperature (SST), wind stress (WS), and mixed layer depth (MLD) were determined. The results indicate that there is obvious chlorophyll a decadal variability in the SCS. The decadal chlorophyll a presents distinct seasonal variability in characteristics, which may be as a result of various different dynamic processes. The negative chlorophyll a concentration anomaly in January was associated with the warming of SST and a shallower MLD. Generally, there were higher chlorophyll a concentrations in spring during the SeaWiFS period compared with the CZCS period. However, the chlorophyll a concentration exhibits some regional differences during this season, leading to an explanation being diffi cult. The deepened MLD may have contributed to the positive chlorophyll a concentration anomalies from the northwestern Luzon Island to the northeastern region of Vietnam during April and May. The increases of chlorophyll a concentration in northwestern Borneo during May may be because the stronger SSW and higher WS produce a deeper mixed layer and convective mixing, leading to high levels of nutrient concentrations. The higher chlorophyll a off southeastern Vietnam may be associated with the advective transport of the colder water extending from the Karimata Strait to southeastern Vietnam.