Analysis Of The Plants Spectral Characteristics In The Ranium Ore District

Hyperspectral remote sensing can provide large amounts of data to vegetationmonitoring technology, conducive to the mild and moderate pollution monitoring of theplant, and to achieve quantitative remote sensing technology. The technology has security,economic, fast features, can be earlier, more timely discover the ecological changes in theenvironment.This thesis focuses on the uranium mining areas with practice and relevant literature,using the SVC HR-768portable spectroradiometer to collection spectral data of thedominant plants (Vitex negundo, Dicranopteris pedata and Phytolacca) in mine, andconducted indepth discussion. Using wavelet thresholding analysis method denoising,through the continuous removal method and the first derivative to pretreatment deal withthe spectral curve, and analysis the obtain spectral characteristic parameters.The results show that there are more serious noise in the spectral data in thewild-collected plants, the selected wavelet denoising method can achieve better noiseremoval effect in the test, and can better retain the spectral curve of the basiccharacteristics; But to obtained denoising results in wavelength range of1800~2000nmand2400~2500nm, mainly due to the impact of water vapor, it can’t achieve a comparableeffect of indoor spectral curve test, which indicates that the wavelet denoising feasible, butthere is further room for improvement; Wild-collected plants spectral curve shows thetypical "peak and valley" spectral curve characteristics, transform processing to plantspectral data of350~850nm, we can see, compared to non-uranium mine, the foliarreflectance spectrum of Vitex negundo, Dicranopteris pedata and Phytolacca changessignificantly in the uranium mine, the overall reflectance increases, and a system red shiftphenomenon, the maximum absorption depth of Red Valley shallow, absorption area andabsorption peak symmetry for smaller, plants in uranium ore all have "red edge blue shift"phenomenon, the red edge peak are increasing trend; Vegetation indices(NDVI、SEI、AI1、AI2、AI3、PSDN_A、PSDN_B、PSDN_C)are small, there is a certain correlation between these parameters and uranium poison, indicating the uranium directly or indirectly affectthe spectral reflectance characteristics of the plant.In short, transformation and analysis of plants’ hyperspectral data in the uraniummining area, provide a basis for better understanding the spectral characteristics of theuranium ore plant, the uranium plant prediction model, combined with hyperspectralremote sensing image interpretation, analysis and inversion, may be in order to achievelarge-scale vegetation change monitoring and plant looking for uranium.