Study On Remote Sensing Quantitative Retrieval Of Non-optically Active Water Quality Parameters Based On MODIS In The Chinese Bohai Sea

The Chinese Bohai Sea,which located in the north of China,suffers from the development of the Bohai Economic Circle of China.Terrigenous input of pollutants induces marine ecological environment deterioration,serious eutrophication,red tides and other severe environment problems.A new monitoring technology is essential to gain accurate and cyclical seawater quality monitoring data.The current in situ techniques for measuring and monitoring seawater quality are time-consuming with a low input and output ratio.In situ measureent suffers from severe natural environment and bad weather conditions,and do not provide a synoptic view of a water body across the landscape.Fortunately,remote sensing provides the potential of gaining spatial and temporal coverage needed for seawater quality monitoring with the advantages of wide survey and freq uent monitoring.It provides huge amounts data of multi-format with various resolutions.The marine remote sensing have developed rapidly,and formed a complete system of marine monitoring till now.A lot of achievements have been gained in the research area of marine physical remote sensing and ocean color remote sensing.Remote sensing had been generally used to monitor seawater quality parameters,like temperature,chlorophyll-a,total suspended solids and colored dissolved organic matter(CDOM),etc.However,most of the remote sensing retrievals of water quality have focused on the indicators which have clearly characteristic wavelengths.Few studies have focused on non-optically active water quality parameters whose characteristic wavelengths remain controversial.This study attempted to carry out an experiment and test whether nutrients,carbon fractions and salinity can be potentially estimated by using remotely sensed data in the Chinese Bohai Sea.The objectives of this study were as follows:(1)to explore the retrieval mechanism of non-optically active water quality parameters based on remote sensing;(2)to develop inverse models to estimate non-optically active water quality parameters with sensitive ? remotely sensed variables;(3)to map the spatial distributions of non-optically active water quality parameters in the Chinese Bohai Sea with MODIS images,and(4)to evaluate the robustness of the developed retrieval models.The main research content of this article was as follows:(1)to develop remote sensing inversion data set for non-optically active water quality parameters.The concentrations of nutrients,carbon fractions and colored dissolved organic matter(CDOM),were determined by laboratory analysis.The downloaded MODIS images were preprocessed by ENVI to seawater remote sensing reflectance of the sampling plots;(2)to explore the quantitative remote sensing inversion mechanisms for the non-optically active water quality parameters.The utility of direct method was evaluated based on the correlations between the MODIS reflectance of the sampling plots and the non-optically active water quality parameters concentrations.The utility of indirect method was evaluated based on the correlations between the nutrition concentrations and turbility,the carbon fraction concentrations and CDOM and the salinity and CDOM.Reasonable modeling scheme were selected based on the different inversion mechanisms for the three types of non-optically active water quality parameters;(3)to develop and validate the inversion models for the three types of non-optically active water quality parameters,and to evaluate the robustness of the developed models.The equivalence ratio variables were developed based on the equivalence relation between the remotely sensed data and the measured reflectance of the sampling plots.The inversion models of nutritions and carbon fractions were gained by multiple stepwise regression.However,the inversion models of salinity was gained by single variable regression.All the developed models were validated based on different independent data sets and the robustness of these models were also evaluated.The yielded results showed that:(1)the correlations between different nutrient concentrations and MODIS reflectances were obviously different.The band ratios,which were developed based on the equivalent relation between the remotely sensed reflectance and the measured reflectance,meet the absence of the measured reflectance.The band ratios highlighted the reflectiviy difference of different bands,and improved the modeling accuracies;(2)the whole modeling inversion and the partition modeling inversion accuracy was roughly the same.The inversion accuracies of nitrate nitrogen and total dissolved inorganic nitrogen were higher than nitrate nitrogen,ammonium nitrogen and phosphate;(3)some equivalent band ratios were highly related to carbon fraction concentrations in the Laizhou Bay.The carbon fractions retrieval models were developed and validated,and the yielded results were reasonable;(4)the measured salinity was significantly related to CDOM.The salinity retrieval model was developed by using CDOM as proxy,and the yielded results were in accord with the fact.The developed retrieval model was credited as robustness and could be extended to map salinity during 2010 – 2014.Characteristics and innovations of this paper were as follows:(1)several remote sensing equivalent band ratios were developed based on the the equivalent relation between the remotely sensed reflectance and the measured reflectance;(2)the inversion mechanism of three types of non-optically active water quality parameters were explored and remote sensing quantitative retrieval of three types of non-optically active water quality parameters were confirmed.In conclusion,the study analyzed the inversion mechanism of the gived non-optically active water quality parameters based on the analysis of data acquisition and processing.Reasonable modeling schemes were operated in modelling,and the feasibility of non-optically active water quality parameters retrievals was confirmed.The yielded results provided scientific basis for remote sensing quantitative retrieval of non-optically active seawater quality parameters.In addition,the research ideas and methods provide support for future study in this area.