Multi-source Remote Sensing Data Fusion For Chlorophyll Concentration Products

Chlorophyll concentration in nearshore water is an important indicator of marine eutrophication,red tide and carbon sink,which is of great significance for remote sensing monitoring.However,the lack of high spatial resolution and high remote sensing reflectance accuracy at the present stage is difficult to meet the needs of nearshore water monitoring.Data fusion can comprehensively utilize the advantages of multi-source remote sensing data to obtain remote sensing images with high spatial resolution and high Rrs accuracy,which is of great significance to the monitoring of chlorophyll concentration in coastal water and ecological environment.In this paper,Sentinel-2 MSI（Multi Spectral Instrument）images with high spatial resolution and Sentinel-3 OLCI（Ocean Land Colour Instrument）images with high Rrs accuracy were used as data sources.Based on this data source,three data fusion experiments（WTBF（wavelet transform based fusion）,BOBF（bio-optical based fusion）,and IUBF（improved unmixing based fusion））are carried out.The accuracy of fusion data and regional applicability of each method are evaluated with the in-situ Rrs data.After the optimal fusion algorithm was evaluated,the effect of chlorophyll concentration inversion from image generated by the optimal fusion algorithm was evaluated.Then a pair of MSI and OLCI images were selected monthly in Wenzhou coastal waters for fusion,and OC4Me was used to retrieve chlorophyll and perform quarterly fusion.The results show that:（1）the Rrs accuracy of OLCI is higher than that of MSI.The MRE and RMSE of MSI at 443 nm,560 nm and 665 nm are higher than those of OLCI,indicating that the Rrs accuracy of MSI is lower than that of OLCI.The visual effect and mean gradient showed that the clarity of MSI was higher than OLCI.（2）BOBF is the best fusion algorithm among the three methods.Combined with MRE,RMSE and bias,the Rrs accuracy of fusion images generated by BOBF and WTBF was higher than that of MSI.The MRE,RMSE and bias of fusion images generated by IUBF were higher than that of WTBF and BOBF,the accuracy of IUBF was worse.The clarity of images generated by WTBF and BOBF was similar to that of MSI evaluated by visual evaluation and mean gradient.The clarity of images generated by BOBF was higher than that of WTBF,while the clarity of images generated by IUBF was higher than that of OLCI but lower than that of WTBF and BOBF.（3）In the coastal area of Yantai with lower Rrs,the clarity of fusion image generated by BOBF is equal to that of MSI and the Rrs accuracy is higher than that of MSI,showing good applicability.（4）Compared with OLCI,the clarity of chlorophyll concentration inversion based on BOBF fusion image was improved,and the correlation coefficient r between the inversion and OLCI was higher and the RMSE was lower.（5）Chlorophyll distribution in coastal waters of Wenzhou has spatial difference.The chlorophyll concentration was higher in the coastal area,decreased as the coastal area extended to the open sea.The distribution of chlorophyll concentration had obvious seasonal characteristics.The average concentration of chlorophyll in Wenzhou coastal waters in 2020 was 4.7098 mg/m~3 in spring,2.5614 mg/m~3 in summer,4.1979 mg/m~3 in autumn and 7.3196 mg/m~3 in winter,with the highest concentration in winter and the lowest concentration in summer.