Inland Water Color Parameters Retrieval And Algal Bloom Monitoring Using Multi-source Remote Sensing

Water is the source of life,which is one of the most important elements of the natural ecological environment.Water resources are closely related to the human survival and development.Earth's freshwater resources,as a very scarce resource,account for less than one percent of the total amount of water resources.Therefore,inland water resources monitoring has a very important role.With the development of society,more and more inland waters have been polluted,causing a great threat to human life and social production.In this paper,the hyperspectral remote sensing data and multi-source satellite remote sensing data are used to retrieve the water color parameters of inland water bodies,and the cyanobacteria blooms of the inland water is monitored by using temporal spectral data.The main contents and results are summarized as follows:1.The differences between the measured spectrum and the water quality parameters of the different zones of the Nansi Lake,especially for the different types of water plants and the common water bodies,were analyzed and compared.Then,the influence of the main aquatic plants and their growth coverage on the water quality and remote sensing retrieval were explored.For the water body which is cover by aquatic plants,this paper proposed a practical technique to use different kinds of water plants and their growth status to indicate the water quality of the water body,and then indirectly accomplished the water color retrieval.For the water body which is not cover by aquatic plants,remote sensing models for retrieving water color parameters were established by using the hyperspectral data;2.Based on the unique phenological feature,the remote sensing vegetation index temporal spectral image(i.e.temporal spectral cubic image)was successfully constructed.Then on this basis,the traditional hyperspectral analysis technique was used to realize the fine partition of the Nansi Lake basin,which is classified as the farmland,wetland,aquatic water body and ordinary water,etc.These results laid an important foundation for the remote sensing inversion of Nansi Lake water color parameters;3.In addition,the water color parameters of Nansi Lake were retrieved by multi-source remote sensing data,and the accuracies of these models were satisfied.Meanwhile,it explored the influence of spatial resolution,temporal resolution and spectral resolution on the inversion of water color parameters of inland water bodies,which can provide a reference for the design of the inland water satellite sensor;4.This study selected MODIS NDVI data of the Lake Taihu from 2001 to 2013 and constructed temporal spectral data for each year.Then,the feature temporal spectra of severe cyanobacteria bloom,moderate cyanobacteria bloom,slight cyanobacteria bloom and aquatic plants were extracted.Based on SVM(Support Vector Machine)algorithm,these four kinds of objects were separated,obtaining their spatial distribution and area.The results were compared with the traditional threshold segmentation method,showing that the method proposed in this paper can classify the cyanobacteria bloom intensity of Taihu Lake more accurately and comprehensively.The rich phenological information provided by NDVI temporal spectra can accurately distinguish the cyanobacteria and aquatic plants,which is expected to provide effective means to accurately monitor and predict the algal bloom trends and strength.