Study On Models For Estimating Concentration Of Tsm In Meiliang Bay Based On Field Spectral Data

Based on the data of Meiliang Bay, located in the north of Taihu Lake, our paper divided several datasets of spectrum reflectance into two categories according to the apparent optical properties and water quality parameters-one dominated by phytoplankton, and the other dominated by non-algae particles. On this basis, we evaluated the accuracy and applicability of the typical inversion models of suspended matter using these two datasets, and then reconstructed the model with its sensitive bands. Finally, suspended sediment inversion model of Meiliang Bay was constructed after summarizing the previous research. The main conclusions of this study are listed as follows:(1) Optical properties of the water in Meiliang bayThe spectral characteristics of the water in Meiliang Bay show similar characteristics with that in typical inland water. However, the apparent optical properties vary greatly because of the difference of sampling time. The optical properties of datasets collected in September2010, March2011, April2011, and July2011showed inland water optical properties of phytoplankton. The optical properties of datasets in October2010and May2011were dominated by non-algal particles. Although the bleeding of the algae was largely related with the seasons, there was some difference for each month. Our study found that chlorophyll concentration was low in May which was donated by non-algal particulate particles. Therefore, dividing the inversion models by the seasonal difference had the defect.(2) The division method of the water categories of Meiliang BayThis study analyzed the apparent optical properties of different datasets of Meiliang Bay. According to concentration of chlorophyll and suspended solids, as well as the surface characteristics of the spectral data, we divided the datasets into two categories-Class I water dominated by algae and Class II water dominated by non-algae particles. Actually, this separation was also the division of the water category. This division method could be confirmed through the ratio of Chlorophyll concentration and total suspended matter, and the ratio of organic suspended solids and inorganic suspended matter concentration. The significance of the water body category was to test the adaptability of the model to different categories of data.(3) Validation of typical TSM estimating models The TSM models were validated by different datasets, and it was found that the different measuring and computing modes of apparent spectrum make the model invalidated. Simple models, such as linear model, power function model, could always achieve high precision; whereas the complicated models were opposite. In Meiliang Bay, validation of single band linear model, power function model and band ratio model was good.(4) Sensitive band analysis and model Establishment of Meiliang BayThe two categories of spectrum reflectance in the Meiliang Bay were used to analyze the sensitive band. It was found that there was a very high correlation between near-infrared remote sensing reflectance and TSM, making it the preferred band for the inversion of the TSM. The sensitivity of the entire wavelength in the planktonic algae-led Class I water was lower than that of the non-algal particulate matter led class II water. This paper established that R824linear model, R720power function model and the ratio of R824and R555model could achieve better results than the other models.