Study On The Prediction Of Hyperspectral Characteristics And Vegetation Cover Of Several Typical Sandy Plants

The paper selects 11 species of sand plants Artemisia ordosica,Salix psammophila,Elaeagnus angustifolia,Peganum harmala,Kalidium foliatum,and small fruit white in Toke County.Nitraria sibirica,Bassia dasyphylla,Tamarix chinensis,Caragana korshinskii,Artemisia blepharolepis,and Sabina vulgaris are used as research objects.The HR-1024 ground feature spectrometer performs spectral measurements on the desert plants shrubs,extracts the spectral characteristics parameters of typical desert plants,analyzes the similarities and differences of the spectral parameters of different desert plants,and uses the vegetation of different life forms Spectral characteristics are distinguished,and a hyperspectral inversion model is established for Salix salsa vegetation coverage.The research results can provide theoretical basis and data support for hyperspectral remote sensing inversion and vegetation monitoring in semi-arid areas.The main research conclusions are as follows:(1)The spectral curve shapes for the 11 species of psammmophytes are mainly in the visible-light blue wave band(350-500nm),550nm,670nm,680-760nm bands,according to the red edge parameters,the bassia dasyphylla,nitraria sibirica,tamarix chinensis,salix psammophila are well differentiated out;the sabina vulgaris,nitraria sibirica and tamarix.chinensis are identified out by use of the characteristic parameters of spectral absorption(2)The three(trees,shrubs and herbs)different life types of vegetation spectral curve shapes have basically shown the same,which are distinguished from the differences between the wave valley and peak positions,the more obvious difference bands are mainly concentrated in the bands of 400-700nm and 700-900nm;The derivative transformation for the three different life types of vegetation spectral curves is carried out,the trees,shrubs and herbs are identified according to the differences of the red edge parameter,blue edge parameter,yellow edge parameter,absorption area and absorption depth.(3)With the increasing of vegetation coverage,the characteristics represented by the spectral curve of salix psammophila are also more and more obvious,in 400-760 nm wave band,the characteristic representation of the "two valley and one peak" is stronger and stronger,the "twin peaks" phenomenon near the red edge position is more and more obvious,the "red edge" area of spectrum curves appears the obvious red shift phenomenon to the right,the red edge area is also constantly increasing;while the absorption depth increases,the absorption intensity also becomes very strong.(4)The two bands normalized difference vegetation index(NDVI)has showed the best form,followed by the difference and ratio values;the best correlation wave bands with the vegetation coverage are respectively concentrated in DI(1390,960),RI(690,670)and NI(210,1520).(5)After the wavelet transformation for the original spectral reflectance of salix psammophila,the spectral characteristic representation is more obvious,which can highlight the more wave band spectral information of original spectrum,the blue shift phenomenon is happened at the 400 nm,760 nm,1000 nm,1220 nm,1370 nm and 1870 nm wave valley location,the red shift phenomenon has been happened at the 500 nm,800 nm,920 nm,1190 nm wave peak position.(6)The correlation among the original spectrum wavelet coefficient,first-order differential wavelet coefficients and vegetation coverage is higher than the correlation between the second order differential wavelet coefficients and vegetation coverage,the correlation is analyzed by choosing the wavelet coefficients in the different transformation and different scales and the vegetation coverage,the sensitive wave bands and characteristics of spectral wave bands can effectively been extracted out,it is beneficial to further study the spectral information of vegetation.(7)By the comparison and analysis,in the inversion models of vegetation coverage established by the five methods,the partial least squares regression model established with the wavelet coefficient as the input variable is the most stable,and has the highest precision,and the estimation effect of this method on the vegetation coverage is very obvious.