Coupling remote sensing retrieval with numerical simulation for SPM study—Taking Bohai Sea in China as a case

Abstract

Study of suspended particulate matter (SPM) transport in a coastal area, traditionally supported by field observation and numerical simulation, benefits now from numerous available remotely sensed ocean color images. Numerical simulation of SPM dynamics can be the supplement to remote sensing data which response the discrete states for different point-in-times. It would be helpful for monitoring coastal water environment in a real sense if remote sensing can be coupled with numerical simulation to get a promising water environmental dynamics in time and space domain. In this study, an approach coupling ocean color image retrieval with numerical simulation of SPM transport is proposed. Instead of traditionally specifying the spatial-constant SPM concentration as an initial condition of sediment transport models, the spatial-variant SPM concentration retrieved from MERIS image initializes the models. And the SPM resuspension and deposition can be interactively inspected and verified. Combining the simulated results with the remote sensing retrieved results, the parameters of hydrodynamic model and sediment transport model can be calibrated. The Bohai Sea in China is selected to test the approach. An optimized group of parameters are calibrated by using the SPM concentration retrieved from MERIS image in the Bohai Sea. Compared with the SPM initialization of spatial-constant, the spatial-variant SPM input retrieved results can quickly response to the changes of sediment concentration in real sense. The results show a well agreement with the in situ measurements. The results show that the approach is good at simulating SPM dynamics in clear waters with a SPM concentration lower than 20 mg/l and more effective in the area with less land source affected. The study suggests that the proposed approach is effective for monitoring coastal water environment roundly and three-dimensionally.

Introduction

Monitoring and modeling of the distribution of SPM is an important task, especially in coastal waters. SPM is one of the major parameters that regulates the penetration of light into waters and hence the primary production. In addition, SPM discharged from river is one of the nutrient sources which impact the primary production. As the nutrients absorbed by SPM can be conserved for several to tens of years (Moodley et al., 1998), influence of sediment on the marine ecological system is permanent and potential. Study on distribution and transport of SPM play an important role on marine ecological system.

The Bohai Sea has been heavily impacted by human activities related to fishery, harbors, oil, tourist and salt. The environment of Bohai Sea has deteriorated by contaminations dumping into the sea. The sediment run-off from land surface and four major rivers, the Yellow River, the Haihe River, the Luanhe River and the Liaohe River, is an important composition which carries plenty of nutrients flowing into the sea. Among of them, the Yellow River is the biggest contributor. Some studies have been focused on simulating the SPM transport in the Bohai Sea and the adjacent seas by using the numerical models since the late 1980s. Qin and Li (1982) discussed the effect of hydrodynamic condition to the distribution of SPM qualitatively. Graber et al. (1989) and Yanagi and Inoue (1995) used numerical methods to simulate the SPM transport or processes from different perspectives such as tidal currents and wind waves. Cao and Wang (1993) established numerical model for the transport of SPM based on dynamic procession of sand transport under wave and tidal current, and simulated the transport of SPM in Bohai Bay. Martin et al. (1993) investigated the transport of SPM from the Yellow River through the Bohai Sea to the western Pacific. Jiang and Sun, 2000, Jiang and Sun, 2001 applied 3D SPM model of particle tracking to simulate the concentration distribution of SPM which was closely related to meteorological condition, considering the force of meteorological element, tidal and wave. Li et al. (2006) and Wang et al. (2008) applied 3D SPM transport model to the Bohai Sea. The distribution and transport of suspended sediments off the Yellow River mouth and the nearby Bohai Sea was analyzed based on hydrological and sedimentological investigation (Qiao et al., 2009).

Remote sensing has been proved to be an effective approach to monitor the SPM in the coastal waters. How to retrieve SPM from remote sensing images has been focused by many researchers. Klemas et al. (1974) proposed a statistical model to quantitatively calculate SPM concentration using MSS data after Landsat-1 launching successfully in the early of 1970s. Retrieval approach of SPM concentration from more than 20 remotely sensors have been widely discussed in the past more than 30 years. Three kinds of models can be summarized, analysis model, Semi-analysis model and empirical model. The empirical method including lineal, logarithmic, minus exponential relation and other algorithms, has been widely discussed (Mertes et al., 1993, Tassan, 1994, Lira et al., 1997, Mishra, 2004, Van der Lee et al., 2009, Griet et al., 2009).

In fact, remote sensing and numerical simulation have been widely applied in studying water environment. Mohn and White (2007) selected 7 years (1998–2004) of remotely sensed data to create monthly climatological fields, and numerically simulated the spatio-temporal characteristics of chlorophyll-a in the waters. And then a relationship of monthly climatological fields and simulated chlorophyll-a was analyzed. SST observations of upwelling induced by tidal straining in the Rhine ROFI are studied by de Boer et al. (2009) by using NOAA/AHRRV SST images, supporting with a three-dimensional numerical modeling. Some studies tried to monitor SPM coupling satellite remote sensing retrieval with numerical simulation. Jensen et al. (1989) summarized studies of modeling SPM distribution using a coupled hydrodynamic and dispersion model and analysis of Landsat TM data. Atmospherically corrected chromaticity data derived from TM data were significantly correlated with modeled total SPM concentrations. Ouillon et al. (1998) summarized the benefits of coupling remote sensing and numerical modeling, and mentioned that the inverted SPM concentration map could be used to test different assumptions in the sediment transport model, and to improve the simulation results and understanding of sediment dynamics. Pleskachevsky et al. (2005) presented a quasi-3D and a 3D SPM transport model in synergy with ocean color (CZCS) data to simulate the vertical distribution of SPM concentration. Miller et al. (2005) used MODIS 250 red band reflectance data to calibrate and validate the output of sediment transport model (ECOMSED) through a comparison of the model predicted SPM concentrations with SPM images derived from cloud-free MODIS images in the Lake Pontchartrain, LA, USA. Kouts et al. (2007) combined satellite remote sensing with numerical modeling, consisting of a hydrodynamic model, a particle transport model and a benthic macro-algae growth model, to calculate distributions of SPM. Then comparison of SPM distributions from remote sensing images and numerical model results showed qualitatively similar patterns.

However, how to couple remote sensing with numerical simulation to monitor the coastal SPM distribution and transport is seldom reported. Mostly the derived SPM maps from remote sensing images were just used to test the spatial distribution of simulated SPM results, and did not integrate with numerical simulation to study the SPM pattern in time and space domains. As it just periodically records the instantaneous state for SPM in the coastal waters, the dynamics between the interval states cannot be explored. What's more, optical ocean color sensor is not available when it is cloudy. Successive distribution and transport of SPM can be simulated by hydrodynamic model coupled SPM resuspension and deposition models, but it is difficult to collect in situ measured initial conditions, and also consuming of computation resource and man power, which make it impossible for dynamic monitoring of SPM in a vast water area.

In this study, a method of coupling derived SPM image and numerical simulation is proposed to monitor the SPM dynamics in the Bohai Sea. The spatial-variant SPM concentration map retrieved from MERIS image is applied as the initial condition of the sedimentation resuspension and deposition model. The spatial-variant SPM concentrations are specified to the model instead of traditional spatial-constants. Parameters of the model are then calibrated according to the spatial distribution of SPM images on the premise that the derived results from remote sensing are satisfied. When a SPM image was set as an initial input to the model, it runs to simulate the SPM distribution and transport until a time when another SPM image and in situ observations are available. In this study, in situ measurements of SPM at 50 sites are used to verify the model. Comparison of simulated results using spatial-variant SPM concentration map and spatial-constant data as an initial condition, respectively, suggests that the modeling results are more reasonable when the spatial-variant SPM concentration map is initialized compared with the MERIS derived SPM results.