| Green tide disaster has endangered the ecological environment and damaged the economic benefits of China coastal areas.Remote sensing plays an important role in green tides monitoring for prevention and control of green tide.Satellite images are valuable data source for extracting a large area of floating green macroalgae on the sea surface.However,there are large errors in green macroalgae coverage derived on mixed pixels.This paper mainly studies on the remote sensing monitoring of Ulva pertusa.In this paper,using the collected spectral data of Ulva pertusa,the obtained synchronous unmanned aerial vehicle(UAV)and Landsat images,retrieval models were established between sub-pixel coverage of green macroalgae and reflectance of characteristic bands and vegetation indexes by analyzing spectral characteristics of green macroalgae in the study area,based on the results of green macroalgae coverage derived from UAV image.The main conclusions are listed as follows:(1)The hyperspectral data of Ulva pertusa collected under different water depths were processed and analyzed.It is found that its reflectivity increased first and then decreased in the near-infrared band,and with the shallower water depth,the turning point moved to the right of wavelength.By fitting the relationship between the resampled spectral reflectance and water depth,it is found that there is inactive linear relationship between water depth and reflectance in blue band,and there is no linear relationship between water depth and reflectance in green and red bands,and there is a positive linear relationship between water depth and reflectance in near-infrared band,RVI,NDVI,EVI and VB-FAH.(2)Monitoring the floating green macroalgae on the sea surface with the high-resolution UAV image of RGB.By analyzing the spectral characteristics of green macroalgae and sea water and comparing the monitoring accuracy of different indexes NGRDI,NGBDI,RGRI,Ex G and red band DN value,the dynamic threshold method of red band DN value is determined to have the highest accuracy in extracting green macroalgae from this UAV image.Finally,this method is used to monitoring green macroalgae in the whole study area.(3)Based on the green macroalgae extraction results of the UAV image,by analyzing the green macroalgae spectral curve of the Landsat image,the retrieval model between sub-pixel coverage of green macroalgae and multi vegetation indices and reflectance of multiple characteristic bands is established.Firstly,by analyzing the spectral curve of typical green macroalgae and sea water pixels in the Landsat image,the characteristic bands of blue,green,red and near-infrared and vegetation indexes of NGRDI,NDVI and VB-FAH are selected to fitting model formulas with sub-pixel coverage of green macroalgae in the sample area.The results show excellent linear relationships between the reflectance of blue,green,red bands and the sub-pixel coverage of green macroalgae,and the reflectance decrease monotonically with increasing sub-pixel coverage.Then,these three models were verified in the verification area,and the results show that the model established by the reflectance of green band is more accuracy than other indexes or indices combination,with the coefficient of determination(R~2),root mean square error(RMSE),and mean relative error(MRE)values of 0.92,0.07,and 10.85%,respectively.This formula can be used for the inversion of green macroalgae coverage in this study area.Hence,we provide a model in this paper that could estimate the sub-pixel coverage of green macroalgae,and realize the precise monitoring of the coverage of green macroalgae extracted from Landsat images. |
PDF Full Text Request