A Study On The Rhinopithecus Roxellana Habitat Evaluation Based On SVM And Multi-source Data

Based on the Support Vector Machine(SVM) and Geographic Information Technology(GIS) the habitat suitability of Rhinopithecus roxellana in Shennongjia Natural Reserve Hubei Province was evaluated and analyzed in the paper. It was the first time that the domestic GF-1 satellite images was used in the Rhinopithecus roxellana habitat assessment. And the feasibility of GF-1 satellite image replacing foreign satellite images in the habitat assessment of wild animal was tested. The results showed that:1, The vegetation classification of the protected areas was accomplished by satellite remote sensing technology. Based on the GF-1 remote sensing images in October 2013, forest vegetation types in Shennongjia Natural Reserve were interpreted by using multi-source data and support vector machine technology. The forest vegetation was classified into 5 types: needle forest, broadleaf forest, mixed coniferous, shrubbery and farmland. It was showed that and the 95.92% area was covered by forests in the reserve, broadleaf forests and mixed coniferous forests were the main vegetation types, accounting for about 80.7% of the reserve area. Based on the GF-1 satellite images the vegetation index and texture feature were extracted to classify of forest vegetation types, the overall classification accuracy is 76.46%, Kappa=0.705, which meet the application requirements of habitat evaluation.2, Based on Geographic Information Technology, the Rhinopithecus roxellana habitat database of Shennongjia Natural Reserve has been established, including altitude, slope, aspect, vegetation type, water system, road and residential area. The analysis of the layer stack by using spatial analysis technology showed that: without considering the influence of human activities, the optimal habitat and suitable habitat were mainly distributed in the core area of Shennongjia Natural Reserve. The optimal habitat area was 56.876 km2, accounting for 7.77% of the total area. The suitable habitat area was 305.156 km2, accounting for 41.69% of the reserve area. Due to the impact of residential and road, Rhinopithecus roxellana actual suitable habitat in the reserve became smaller, The optimal habitat area was 46.935 km2, accounting for 6.41% of the total area, the suitable habitat area was 214.617 km2, accounting for 29.32% of the total; the barely suitable habitat area was 302.385 km2, accounting for 41.31% of the total area; the unsuitable area for Rhinopithecus roxellana survival was 167.9757 km2, accounting for 22.95% of the total area. Spatial pattern analysis showed that, most of the eastern areas of the west Shennongjia were available for the Rhinopithecus roxellana habitat, and fewer were in the south, northwest and southwest Shennongjia available for Rhinopithecus roxellana habitat. There were more areas in central part of the reserve available for habitat, but due to the development of ecological tourism, the main road construction limiting the scope of activities of the Rhinopithecus roxellana, and wild Rhinopithecus roxellana has not been found in central part of the reserve area in actual investigation. Suggestions were made including strengthening the protection of the existing suitable habitat and establishing habitat corridor to increase the continuity of suitable habitat and ensure migration and movement of the Rhinopithecus roxellana. The evaluation result was very consistent with the Rhinopithecus roxellana current distribution, which showed that the habitat evaluation method which based on support vector machine classification of vegetation and the use of geographic information system technology was feasible and can be used in establishment of a commonly used golden monkey habitat evaluation system.3, In this study it was the first time that the high resolution remote sensing image data obtained by the Chinese newly launched GF-1 satellite was applied, and it was verified the feasibility of the application of domestic high resolution satellite images in assessment of wildlife habitat. The GF-1 satellite images can replace Landsat satellite ones effectively in the study of Rhinopithecus roxellana habitat evaluation, which broke the dependence of foreign data in the field of wild animal habitat research. And with the characteristics of high spatial resolution, high time resolution, low cost and reliable data source the GF-1 satellite image undoubtedly became a new data source of the remote sensing technology application for wild animal habitat studies in China.