The tradeoffs between the spatial and temporal resolutions for the remote sensing instruments limit their capacity to monitor the eutrophic status of inland lakes.Spatiotemporal fusion(STF)provides a cost-effective wa...The tradeoffs between the spatial and temporal resolutions for the remote sensing instruments limit their capacity to monitor the eutrophic status of inland lakes.Spatiotemporal fusion(STF)provides a cost-effective way to obtain remote sensing data with both high spatial and temporal resolutions by blending multisensor observations.However,remote sensing reflectance(Rrs)over water surface with a relatively low signal-to-noise ratio is prone to be contaminated by large uncertainties in the fusion process.To present a comprehensive analysis on the influence of processing and modeling errors,we conducted an evaluation study to understand the potential,uncertainties,and limitations of using STF for monitoring chlorophyll a(Chla)concentration in an inland eutrophic water(Chaohu Lake,China).Specifically,comparative tests were conducted on the Sentinel-2 and Sentinel-3 image pairs.Three typical STF methods were selected for comparison,i.e.,Fit-FC,spatial and temporal nonlocal filter-based fusion model,and the flexible spatiotemporal data fusion.The results show as follows:(a)among the influencing factors,atmospheric correction uncertainties and geometric misregistration have larger impacts on the fusion results,compared with radiometric bias between the imaging sensors and STF modeling errors;and(b)the machine-learning-based Chla inversion accuracy of the fusion data[R^(2)=0.846 and root mean square error(RMSE)=17.835μg/l]is comparable with that of real Sentinel-2 data(R^(2)=0.856 and RMSE=16.601μg/l),and temporally dense Chla results can be produced with the integrated Sentinel-2 and fusion image datasets.These findings will help to provide guidelines to design STF framework for monitoring aquatic environment of inland waters with remote sensing data.展开更多
基金supported by the Science and Technology Major Project of Hubei Province,China(grant number 2023BCA003)in part by the Open Fund of Key Laboratory of Urban Land Resources Monitoring and Simulation,Ministry of Natural Resources(grant number KF-2023-08-23)the National Natural Science Foundation of China(grant number 42171375).
文摘The tradeoffs between the spatial and temporal resolutions for the remote sensing instruments limit their capacity to monitor the eutrophic status of inland lakes.Spatiotemporal fusion(STF)provides a cost-effective way to obtain remote sensing data with both high spatial and temporal resolutions by blending multisensor observations.However,remote sensing reflectance(Rrs)over water surface with a relatively low signal-to-noise ratio is prone to be contaminated by large uncertainties in the fusion process.To present a comprehensive analysis on the influence of processing and modeling errors,we conducted an evaluation study to understand the potential,uncertainties,and limitations of using STF for monitoring chlorophyll a(Chla)concentration in an inland eutrophic water(Chaohu Lake,China).Specifically,comparative tests were conducted on the Sentinel-2 and Sentinel-3 image pairs.Three typical STF methods were selected for comparison,i.e.,Fit-FC,spatial and temporal nonlocal filter-based fusion model,and the flexible spatiotemporal data fusion.The results show as follows:(a)among the influencing factors,atmospheric correction uncertainties and geometric misregistration have larger impacts on the fusion results,compared with radiometric bias between the imaging sensors and STF modeling errors;and(b)the machine-learning-based Chla inversion accuracy of the fusion data[R^(2)=0.846 and root mean square error(RMSE)=17.835μg/l]is comparable with that of real Sentinel-2 data(R^(2)=0.856 and RMSE=16.601μg/l),and temporally dense Chla results can be produced with the integrated Sentinel-2 and fusion image datasets.These findings will help to provide guidelines to design STF framework for monitoring aquatic environment of inland waters with remote sensing data.
