Research On The Forest Health Evaluation Of Qinling Huoditang Based On The Hyperspcctral Data

The environmental protection and the ecology construction have been the attentionfocuses of the whole world due to the growing demand in the aspects of environment andecology. The forestry is the emphasis in the environmental protection and the ecology field.With the increasing of the forest coverage, improving the quality of the forest has been thepriority field of forest resource management.Nowadays, the research methods about forest health elevation are confined to thetraditionally ones of investigating typical sample plots. Those methods can not respond to theimperceptible changes timely and effectively for the long evaluation period and complicatedwork. With the development of the spectral imager and the hyperspectral image technology,the spectral information that can be achieved is more and more exhaustive. The spectralsignificances are disparate between the healthy vegetation and the ill-healthy vegetation. Thisarticle discussed the forest health evaluation method based on the vegetation indexesextracted from the hyperspectral image. Then the reliability of the model was verified.Eventually, the relationship between forest stand factors and forest health was analyzed.Conclusions were obtained as follows:(1) The indicator system was established to estimate the health status of the forest onthree levels (the vegetation distribution, the photosynthetic intensity of the trees and the stressof damage by disease and insect or the moisture the trees faced) according to the spectralcharacteristics of the vegetation. The evaluation indexes which can reflect the forest healthlevel sensitively in the research areas were extracted by means of sensitivity analysis.Eventually, the vegetation indexes chosen were ARI2, PRI, CRI1, ARI1,CRI2, SG, EVI,NDVI705, RGRI, NDVI、RVI, base on which, an indicator system for the estimation of foresthealth was built.(2) The weights of the indexes were worked out depend on the analytic hierarchy processand the following evaluation model was established.(3) The precision of the model on the stand level and the sample plot level was varified.The result demonstrated that there was significant correlation between the results we obtained from the model and those from the traditional investigation method both on the two levels.The determination coefficients reached0.707and0.655respectively. The FHI of the modelwas below which of the traditional investigation method. Nevertheless, the trend wascoincident. On both the stand level and the sample plot level, data from the model and thosefrom traditional investigation method were compared. The comparison results demonstratedthat there was significant correlation between the two sets of data, with the determinationcoefficients reaching0.707and0.655on the stand level and the sample plot level respectively.Though the FHI from the model were smaller than those from traditional investigationmethod, their trends were coincident. The FHI were reclassified into four levels (ill-health,sub-health, health, good-health) based on the natural discontinuities classification method.Ultimately the distribution map of the forest health level was worked out.(4) The relationship between the forest health index and the stand description factors wasanalyzed depend on the forest resource check data. The correlation analysis results showedthat there was dramatically positive correlation between FHI and the factors, being the canopydensity, the age-class, the average DBH, with the slope (P<0.01). The degree of density hadobviously positive correlation with the FHI (P<0.05). However, negative correlation onlyexisted between the FHI and the elevation (P<0.01). The divergence analysis demonstratedthat significant differences existed only between the FHI and the elevation, the aspect, theslope, the vegetation types, the age-class, the DBH, the degree of the density, the soil texture,the agrotype, the canopy density，the average tree height, the accumulation of per hectare.Then relation model was established which is between the FHI and the relevant factors. Thereis Quadratic polynomial relationship between the FHI and the elevation (R~2=0.884, P<0.01),the aspect (R~2=0.597, P<0.01), the age-class(R~2=0.847, P<0.01), the degree of the density(R~2=0.951,P<0.01), the average tree height (R~2=0.751, P<0.01), the accumulation of perhectare(R~2=0.597, P<0.01). There is the linear relation between the FHI and the canopydensity (R~2=0.597, P<0.01) and the DBH (R~2=0.750, P<0.01).