Article ID | Journal | Published Year | Pages | File Type |
---|---|---|---|---|
4539617 | Estuarine, Coastal and Shelf Science | 2015 | 10 Pages |
•A neural network model is developed for TSM estimation in East China Seas.•Neural network model produces ∼41% uncertainty in TSM estimation from MODIS data.•Variations of Kd(490) in East China Seas are mainly controlled by TSM.
The objectives of this study are to evaluate the applicability of three existing retrieval models of total suspended matter (TSM) in the Bohai, Yellow, and East China Seas (BYES), and to propose a multilayer back propagation neural network (MBPNN) model. Based on the comparison of TSM predicted by these models with in situ measurements taken in the BYES, it was found that the MBPNN model produces a superior performance to the three existing models selected. Near-infrared band-based and shortwave infrared band-based combined model was used to remove the atmospheric effects from the moderate resolution imaging spectroradiometer (MODIS) data, and the TSM concentration was quantified from the MODIS data after atmospheric correction using the MBPNN model. It was found that the MBPNN model produces 41.4% uncertainty in deriving TSM concentration from the MODIS data. The MBPNN model-based climatological seasonal mean TSM concentration indicated that the highest TSM concentrations were found around the river estuaries, while the lowest values were found in the far offshore regions of the BYES. The MBPNN model-based monthly mean TSM concentration revealed that the highest TSM concentrations occurred in February, and the lowest in July. The diffuse attenuation coefficient at 490 nm (Kd(490)) depended heavily on TSM concentration in the BYES. The significant relationship (R2 = 0.72, p < 0.05) between Kd(490) and TSM concentration indicated that the BYES waters are TSM dominated types.
Graphical abstractFigure optionsDownload full-size imageDownload high-quality image (318 K)Download as PowerPoint slide