| This research summarized forest fuel loads in the domestic and foreign, took Ta He forestregion, in HeiLongjiang Province, as the study object for its largely forest fuel loads and pro-firereason. Using the Landsat7ETM~+ remote sensing image in 2002, we selected the linear ridgeestimation method to research that forest type which had been classified. Through analyzingdifferent grey level of wave band, the grey level ratio of wave band and the Judeich factor in thedifferent forest type, we wanted to know that they could affect fuel loads or not. Finally, we hadestablished the dynamic linear model to have carried on the forecast for fuel loads of different foresttype in the Ta He area. Linear model had achieved the good forecast effect, and we had carried onthe analysis to major effect factor of the fuel loads.(1)In geometry adjustment process of remote sensing image, the gradation heavy samplingused the bilinear interpolation. 250 ground control points were selected to carry on proofreading theimage, and we had guaranteed the precision in 0.5 element. We had carded on the synthesis usingthe relevant small wave band, which improved the image quality. By computing it, we finallyobtained 4-5-3 wave band as the best wave band combination.(2)This research classified TaHe area into 7 kinds by the inspector general classification,which were foliage forest, coniferous forest, farmland, waters, mark and so on. In the classifiedprocess, we found that the confusion degree was quite serious in farmland, coniferous forest, coniferand broadleaf mixed forest.(3)The overall classification precision was 85.61％, and the overall classification effect wasapplied to our request, in which foliage forest charting precision was 90.38％, and the user precisionwas 89.23％; The coniferous forest charting precision was 88.89％, and the user precision was90.25％. The classification precision of the foliage forest and the coniferous forest was all high,which could be satisfied the need of the research.(4)The precision of fuel loads estimation model for the foliage forest and the coniferous forestwas so high, which had achieved 90.33％, 94.40％respectively, but there were so many independentvariables in the model, which led to its complexity.(5)The result of the analysis on the independent variables of the foliage forest and theconiferous forest principal components showed that: the normalized vegetation index and the ratiovegetation index load value are all higher, therefore they are playing the major effect role to the fuelloads; Load values of TM3, TM4 and TM5 are also relative higher, which reflect that the chlorophyllabsorbed red light spectral in 0.63～0.69 um and infrared light spectral in 1.55～1.75 um that iscorrelative with Plant's water content, and has the close relation near-infrared light spectral in0.76～0.90 um with the biomass determination. Therefore the forest fuel loads, the chlorophyll absorption sensitive area and the reflected plant water content sensitive area have the closerelationship; In the third, the forth, the fifth principal components, load values of the slope position,the sunny slope and the shade density are very high, which is also playing the vital role to the foliageforest fuel load.(6) In the Ta He area, the fuel loads of the foliage forest is 94.3904 t/hm~2, and the fuel loads ofthe coniferous forest is 112.0601 t/hm~2, which are smaller 2.9037 t/hm~2, 4.7971 t/hm~2than theactual value. This possible related with the local fire, forestry production and the way of forestmanagement. That simultaneously also related with the error which was engendered in the test.
|
PDF Full Text Request