The Utilization Of High Resolution Remote Sensing Data In Forest Harvesting Monitoring

The forest cutting quota verification obains the actual situation of forest harv esting in a certain time mainly by field survey to assess the quality manageme nt of local harvesting of forest resources.While the traditional forest cutting qu ota verification is difficult to meet the need for monitor deforestation and fores t sustainable development as strength and long.There are many advantages in r emote sensing technology, such as the macro,dynamic, convenient, reproducible and low cost, With the gradual improvement of the resolution remote sensing i mages, high resolution remote sensing images has been widely used in Remot Sensing in forest resources.This paper is for the purpose of meeting the need for monitor deforestation, and selects the TM5 and SPOT5 two image data sources, forest of timber extraction as the research object, from concept and micro levels, formed 2 gra de resolution image sequence Pyramid, provides earth observation data sources from coarse to fine. Information extraction and the change detection analysis to cutting area and studied the vegetation change within the region by different in formation extracted methods.The results are as follows:(1)For the first time, by using DeltaCue as module, and TM images as the d ata source, then pre-classificating the 2 image change information of test area, we got 6878 formed of 1163 positive changed pixel and 5715 negative change d pixel. By this means, we've got the range of information preliminary then c omposited the spectral features and the vegetation index to extracted the small forest harvesting directly. The results showed that use TM to extract the cuttin g polygon by taking advantage of its multi-image and good spectral characteris tics, weighted composite correct determine rate for polygon number was 78.9%, and weighted composite correct determine rate for polygon area was 96.3%, Minimal extraction of logging on patch is 3 pixel (0.27hm2)(2) Proposes integrated optimal segmentation scale is 20 of SPOT5 by used multiscale segmentation,Image segmentation-a rule-based classification-the classification based on segmentation, and proposed two kinds of classification s chemes:classification of vegetation and non-vegetation and vegetation of the fi ne classification to extrat SPOT5 integration of data(2.5m spatial resolution) by the method that object-oriented image segmentation and object information.Co mposite optimal segmentation scale, classified two image with the nearest neigh bor classifier, and classification accuracy was all above 85%. For the vegetatio n to class on the early and late non-vegetation land class changed pixel analys is, nearest neighbor classification using images found on the post-NDVI (after) It 0.2, Brightness gt 75, Texture-Correlation-band4 It-148.2 conditions. The pi xel is extracted. after additional interpretation, weighted composite correct deter mine rate for polygon number was 90.8% and weighted composite correct dete mine rate for polygon area was 96.9%, Minimal extraction of logging on patc h is 3 pixel (0.03hm2).The result showed that: object-oriented feature extractio n method considerably improve the classification accuracy of SPOT5 image, ex tracted the small forest harvesting, on-site verification.(3) The locations of cutting area could be accurately monitor by using TM a nd SPOT images data. TM data extracted cutting area from the level of view, the correct determined rate of spot number was 78.9%, and of spot area w as 96.3% after additional interpretation. Multi-spectral data of SPOT5 was bett er than that of TM in identifying the number of cutting area, size and accurac y. In extracting the data of clear-cutting timber, the correct determined rate of spot number used additional interpretation of TM image timber information ex traction result was 65.2%, which was much lower than that of SPOT5 which was 94%.(4)It is difficult to identify the non-clear-cutting area automatically by this m ethod, and the effect is related to the consistency of phase image of two satell ite and the harvesting intensity. From the predict data of remote sensing could be found that the proportion of non-clear-cutting polygons in total number of p re-interpretation polygons was very low. The number of non-clear-cutting polyg ons extracted from TM was 2,8% of the total; the number of non-clear-cuttin g polygons extracted from SPOT5 was 4,6% of the total.Research on the utilization of high-resolution remote sensing of forest reso urces harvesting quota method, the accuracy presented in this paper for the pro duction of cutting quota monitoring and verification of information provided in the application of remote sensing information.