Flocculation Characteristics Of Fine Sediment In The Surface Water Of The Yangtze Estuary And Its Impact On Spectral Reflectance

The Yangtze Estuary,where fresh water from the river basin meets the salt water from the sea,is frequent of water and sediment exchange,including runoff sedimentation,local sediment resuspension and wave lift sand.The suspended sediment concentration in the water is high,which suspended particles less than 0.032mm are the majority,coupled with salinity,water flow and so on,make flocculation of fine particles sediment frequent and important.Flocculation changes the propagation of light into the water,resulting in changing the radiation characteristics of the water.The core problem of remote sensing method is to establish the quantitative relationship between water components and radiation,so we need to consider flocculation and its solution which affect water radiation characteristics.Based on in situ measurement at three sites,corresponding to the the upper part,the turbidity maximum zone and the external part in the Yangtze Estuary in the winter season in year 2014,the paper wants to discuss the flocculation characteristics of different estuarine parts,analyze the influence of flocs on spectral reflectance,and then establish appropriate different inversion models.According to the variation process of floc size in different parts,the influence of the parameters on the particle size of flocs is analyzed by comparing the change of flow velocity,turbidity and salinity in the same time.The principal component analysis shows,the floc average diameter is controlled by the flow velocity and turbidity in the upper part of the estuary,in the turbidity maximum zone of the estuary is controlled by the flow velocity of neap tide and by salinity of spring tide,and in the external part is controlled by salinity.The PSA(Projected Surface Area)is a measure of the suspended sediment concentration and the floc average diameter,which is proportional to the suspended sediment concentration and is inversely proportional to the floc average diameter.The composition,floe settling velocity,particle size,effective density and the PSA of flocs vary in different parts.The distribution of floe disperse particle size is in the form of"small-large-small" in three different parts,the floe average diameter is "large-small-large" pattern,floc settling velocity is "large-small-large" pattern,the PSA is"small-large-small" pattern.In addition to the suspended sediment concentration,the turbidity,the floc average diameter,the PSA of the floc and the spectral reflectance have a good correlation,the multi-layer perceptron model is used to analyze the contribution rate of each parameter to the remote sensing factor.Then establish the traditional single factor regression model,consider the flocculation of the multi-factors regression model and neural network model.The results shows that the Gordon model of the suspended sediment concentration in the single factor regression model is better,R2 is 0.9091,however the fitting accuracy of the neural network model is 0.9688,which is close to the real situation and the best performance.