Spectral Characteristics Analysis And Identification Of Dominant Vegetation Clusters In Caohai Wetland,Weining,Guizhou

Wetland vegetation,as the primary producer of wetland ecosystems,plays an important role in the maintenance and functioning of this ecosystem.The application of remote sensing technology to the extraction and classification of wetland vegetation has the advantages of rapid,large-area and real-time updating,which can provide a reasonable scientific basis for the investigation of wetland vegetation resources and ecological restoration.In this paper,we use the dominant wetland clusters of wetland(Mus.Artemisia,sorghum,Zoysia,Goosegrass,and clover),which are widely distributed in Weihai Caohai National Nature Reserve,Guizhou,to study the advantages of wetland based on field measurements.Hyperspectral information of vegetation,through spectral feature analysis(original spectrum,derivative spectrum,logarithm of reciprocal and envelope removal),statistics and contrasts of spectral characteristics of wetland vegetation and their spectral characteristics,and then spectral standard deviation threshold method Extracting the spectral characteristic bands between wetland vegetations effectively,and simulating different vegetation sensors(HJ1A-CCD,HJ1A-HSI)based on field measured vegetation spectra to construct vegetation indices commonly used for wetland vegetation extraction,such as normalized vegetation index NDVI,environmental vegetation.Index HJVI,Soil Regulated Vegetation Index SAVI,Difference Vegetation Index DVI,Modified Normalized Water Index MNDWI,Ratio Vegetation Index RVI,and explored the ability of simulated vegetation index to identify five dominant wetland vegetations and screened out better evaluation results.Wetland vegetation and vegetation index;finally based on the above analysis results,respectively Identification and classification of vegetation maximum Caohai protected wetland Likelihood decision tree is complete.The relevant research results are as follows:(1)The canopy spectral reflectance of the five dominant wetland vegetation clusters in Caohai Nature Reserve has obvious differences,and the recognition and classification under the canopy scale can obtain higher recognition accuracy;(2)For the characteristic difference bands of the five dominant wetland vegetations,the distribution statistics and sensitivity detection are carried out according to the blue,green,red,red edge and near-infrared band intervals,and the characteristic bands in the green and red light ranges account for a large proportion.The spectral difference bands of the characteristic difference bands in the red and red edges are higher,and 622-659 nm,665-679 nm and 687-701 nm are the best spectral characteristic bands of the five dominant wetland vegetation spectra;(3)From the field measured vegetation hyperspectral data to simulate the vegetation index of different sensors(HJ1A-CCD,HJ1A-HSI)to identify the five dominant wetland vegetation,the DVI and SAVI vegetation index have no obvious effect on the grassland dominant wetland vegetation.NDVI,HJVI,MNDWI,and RVI have different degrees of vegetation discrimination and have better effects in HJ1 ACCD simulation recognition.(4)Maximum likelihood method The classification accuracy of the five dominant wetland vegetation clusters in Caohai Nature Reserve is 51.43,kappa coefficient is 0.45,and the overall classification accuracy of decision tree is 67.7,kappa coefficient is 0.53;the classification effect is significantly better than the maximum Law.