| Coastal zone is the important area contacting the ocean and the continent, which isthe transition zone where matters and energy interact and exchange with each other.Natural resources is rich in there, and many people live in there, where is closely relatedto people’s work and live. Suspended sediment concentration(SSC) is one of the criticalparameters in water quality and environment evaluation, which not only determines thewater optical properties, such as transparency, turbidity, but also influence the marineenvironment, coast construction and so on. As one of the bay in Bohai Sea, Bohai Baynot only occupies an important place about nature resources, but also play a huge part insocial economy of our country. In recent years, a large number of projects work in there.Therefore, monitoring the SSC levels in Bohai Bay offshore area is significance.Remote sensing technology is an effective method to monitor marine ecologicalenvironment in large scale. Compared to traditional method using vessels to measure insome points, the great advantage of remote sensing technology is that it can monitorocean water in real time, continuous and large scale. Using remote sensing data toretrieve SSC is a good way, what’s more, the result of retrieval can guide real workabout ocean environment and construction.The key of retrieval suspended sediment concentration using remote sensing is toestablish a quantitative model between the information of remote sensing and SSC.However, no generally retrieval model have been developed for case Ⅱ waters,because the complexity of the waters’ components make waters optical propertiescomplex, and that cause regionalism and instantaneity of model. Therefore, if we wantto established a retrieval model with high precision and a certain generalization, wemust study the spectral characteristics of the study area and other related factors.The major objective of this study is to develop and analyze some retrieval modelsto calculate the SSC levels in Bohai Bay offshore area based on measure of SSC,particle size and spectral data in situ and analysis of water spectral characteristics in thestudy area. First, set up connection between SSC and RS data using statistical regressionalgorithm(SRA-based model), principal component analysis(PCA-based model) andneural network algorithm(NN-based model). However, RS data which is not reasonablewhen there is a big difference among the particle size of the study area, this paperbrought out two parameters models which used both the RS data and particle size of suspended sediment. In addition, this paper developed a semi-analysis model based oninherent optical coefficient which calculated from apparent optical coefficient byquasi-analytical algorithm. Then on this basis, all models are contrasted and analyzed tofind merit and demerit. Finally, using the best retrieval model and LANDSAT5-TM data,SSC levels of the entire region were estimated, and the characteristics of the SSC levelswere analyzed. Combined with previous studies, this paper pointed out that the factorsinfluence SSC levels are sediment carrying by rivers and human activities.Main research contents in this paper are the following:⑴Researched water spectral characteristics in Bohai bay⑵Established one parameter models based on direct relationship between SSCand RS reflectance, concluding SRA-based model, PCA-based model and NN-basedmodel.⑶Established two parameters models which considering particle size ofsuspended sediment to be one factor, concluding SRA-based model, PCA-based modeland NN-based model too.⑷Estimated inherent optical quantity by quasi-analytical algorithm.⑸Established a semi-analysis model to retrieve the SSC levels.⑹Comparison and application all of the retrieval models.⑺Analysis the SSC levels in the study area and the influence factors.Main results contents in this paper are the following:⑴The spectral reflection curve of Bohai Bay water express the characteristics ofcase Ⅱ waters representatively. It has two obvious peaks, the first peak at the band of550-600nm, and the second peak at the band about800nm.⑵In all of one parameter models, the precision of PCA-based model is highest.The absolute error is3.46mgL-1, the relative error is26.54%, and the root mean squareerror is4.44mgL-1. However, NN-based model is not satisfactory.⑶In addition to the model used statistical regression, two parameters modelsconsidering particle size of suspended sediment are better than one parameter models.In all of two parameters models, NN-based model is best. The absolute error is2.98mgL-1, the relative error is24.23%, and the root mean square error is4.16mgL-1.SRA-based model is not good.⑷Semi-analysis model based on QAA is the best one in all of models, the precision is very high, and the results of retrieval is satisfactory. The absolute error is2.49mgL-1, the relative error is16.07%, and the root mean square error is4.30mgL-1.⑸In all of models, semi-analysis model and two parameters NN-based model arebest, precision of models is high, and the stability of two parameters NN-based model ishigh. PCA-based models are better, and PCA-based models are modest. One parameterNN-based model is not so good.⑹The levels of SSC in this study area are in the middle. The levels of SSCmainly distributed in the range20-60mgL-1. SSC distribution in the study area presentstwo characteristics: SSC levels in the southern sea is greater than these in the northernsea. Secondly, SSC levels from coastal waters to deep sea show a trend of decrease.⑺The factors influencing the distribution of SSC are sediment carrying by riversand human activities. |
PDF Full Text Request