Abstract:  The coastal beach has abundant resources and diverse ecological functions, and plays a crucial role in regulating climate, purifying water quality and protecting biodiversity.In recent years, frequent human activities have continued to affect the wetland ecosystem, which has been causing the wetland area to shrink gradually and destroying the living environment. As a result, the ecological function of the wetland is limited. Phragmites australis and Spartina alterniflora are typical vegetation types in coastal beaches of southeast China. Phragmites australisisa native plant which plays important ecological roles in coastal wetlands.Spartina alterniflora is an invasive species, which hinders the growth of local plants and changes the spatial distribution of original vegetation. It has been spreading continuously in recent decades and posing a serious threat to the local ecosystem and biodiversity. Traditional vegetation monitoring methods mainly rely on manual sampling, which takes a lot of time and effort and may damage the wetland. It has become a hot topic to use remote sensing technology to monitor Spartina alterniflora. However, it is very similar to Phragmites australis in leaf shape, color and other aspects. And the traditional remote sensing information is difficult to distinguish them. In the present study,Phragmites australis and Spartina alterniflorainthe coastal beach of Jiangsu Yancheng Wetland National Nature Reserve Rare Birds were selected as the research objects. The original hyperspectral reflectance curves, the reflectance spectral curves after continuum removed, and the first derivative spectral curves were utilized to accurately identify Phragmites australis and Spartina alterniflora in different growing periods, in order to realize fast and accurate identification of different vegetation in wetland and to provide references for dynamic monitoring of the wetland vegetation and protection of the wetland ecosystem. Continuum removal could effectively highlight the absorption and reflection characteristics of spectral curves, which could be used to extract characteristic bands for classification and recognition. It was shown that the original spectral curves of Phragmites australis and Spartina alterniflorawere basically the same insummer. Both of them showed a similar reflection peak near 550nm in the green band and a similar absorption valley near 670nm in the red band. Analyzing the spectral curves after continuum removed in the visible bands could effectively distinguish Phragmites australis from Spartina alterniflora in summer and the reflectance of Phragmites australis was lower than that of Spartina alterniflora(P<0.05). However, the original spectral curves in autumn showed that the reflection peak and absorption valley of Phragmites australis were higher than those of Spartina alterniflora. The analysis of the original spectral curves in the whole range of 350-1000nm made it easier to distinguish Phragmites australis from Spartina alterniflora in autumn, and Phragmites australis had higher reflectance values than Spartina alterniflora(P<0.05). The analysis of the original spectral curves of Phragmites australis in the range of 350-540nm and near-infrared bands could easily distinguish its growing periods and the spectral reflectance of Phragmites australis in summer was higher than that in autumn(P<0.05). Meanwhile, the continuum removal method was easier to distinguish them in the range of 520-700nm, and the spectral reflectance of Phragmites australis in autumn was higher than that in summer(P<0.05). In the range of 350-1000nm, the analysis of the original spectral curves of Spartina alternifloracould intuitively distinguish different growing periods and the spectral reflectance of Spartina alterniflora in summer was significantly higher than that in autumn(P<0.05). The first derivative spectra curve could not directly distinguish Phragmites australis from Spartina alterniflora, but their different growth periods could be identified according to the changes of their red edge and spectral characteristic parameters. The research could provide a basis for the identification of vegetation types and provide effective information for the protection and planning of the wetland ecosystem.

Key words:  Phragmites australis, Spartina alterniflora, Hyperspectral remote sensing, Spectral signature, Continuum removal