Retrieving Sea Surface Salinity And Analyzing Its Temporal And Spatial Distribution Characteristics In The Gulf Of Mexico

Salinity is an important parameter characterizing marine physical and biogeochemical processes.It is the key factor to study the transportation of land-based material transport and the change of estuarine ecological environment.Optical satellite images have high resolution and can avoid radio frequency interference in near shore.This provide a feasible way to monitor sea surface salinity(SSS).Based on the ship-based SSS data and Moderate Resolution Imaging Spectroradiometer(MODIS)data,the machine learning algorithm is applied to build a high-precision SSS retrieval model in this paper.The model is used to obtain the SSS maps with high temporal and spatial resolution.Based on the predicted SSS,the paper studies the spatiotemporal heterogeneity of SSS in the Gulf of Mexico(GOM),the temporal and spatial heterogeneity of each factor with correlations coefficient,and the effects of wind and current on SSS.The contents of this paper are summarized as follows.(1)Based on the SSS retrieval mechanism and the correlation between the remote sensing reflectance(Rrs)and SSS,the Rrs data of 412 nm,443 nm,488 nm,555 nm and 667 nm are selected to build the SSS retrieval model.In addition,this paper analyzes the correlation between SSS and sea surface temperature(SST),and discusses the necessity of SST for the model.(2)The dataset,including Rrs of five bands(Rrs412,Rrs443,Rrs488,Rrs555,Rrs667),SST and ship-based SSS,are used to build multiple linear regression model(MLR),support vector machine model(SVM),BP neural network model(BP),and random forest model(RF).By evaluating the inversion accuracy of models,the RF with the highest accuracy is selected to the retrieve SSS,with RMSE=0.335 psu and R~2=0.93.(3)Using MODIS satellite data and RF,the SSS of the GOM has been retrieved.Based on the SSS maps of high temporal and spatial resolution,the characteristics of SSS in the GOM are analyzed.Combined with the data of wind and current,and taking typhoon as an example,we comprehensively analyze the variation mechanism of SSS in GOM.The spatial distribution pattern of SSS shows a ring-shaped inward value increase.The spatial distribution of sea surface salinity has seasonal characteristics,which is higher in winter and lower in summer.The spatial distribution is affected by river discharge,wind forcing and circulation and the interaction between wind and current is the dominant factor in the change of spatial distribution pattern of SSS.By analyzing the contribution of each factor,this paper provides a reference for the selection of modeling factors for SSS retrieval in complex nearshore environment.In addition,this paper also provides help for SSS retrieved studies by comparing several models.Furthermore,this paper provides ideas for studying the SSS spatial-temporal variation mechanism in nearshore by combining wind field,flow field,typhoon and other source data.