Font Size: a A A

Discriminating Of Loess Landslide Based On GF-1 Satellite Images

Posted on:2017-03-17Degree:MasterType:Thesis
Country:ChinaCandidate:J J GaoFull Text:PDF
GTID:2283330485478651Subject:Soil and Water Conservation and Desertification Control
Abstract/Summary:PDF Full Text Request
After the implementation of the “Grain for Green” Project in the Loess Plateau,soil erosion on the sloping land have been significantly controlled. However, the frequency of loess landslides in the Loess Plateau increased due to the prolonged and heavy rainfall. The loess landslides occurred in the Yan’an Watershed on June, 2013 were study in this paper. We used the object-oriented technology combined with the "GF-1" image(2 m) and 1: 10,000 topographic map(5 m). We selected the spectral characteristics, spectral index, shape feature, texture, topography and other parameters to develop landslide feature extraction methods. Also the impact of landslides,topographic factors(elevation, slope, aspect, surface curvature) vegetation factors(woodland, grassland) were analyzed. Therefore, the rational allocation of measures to local conditions can provide an important scientific basis and reference for rational planning in the Loess Plateau disaster prevention and water conservation measures.The main conclusions are as follows:(1)Object-oriented multi-scale segmentation and combined techniques were used to build the high definition image segmentation and combination parameters based on recognition of loess landslides, when the multi-scale segmentation parameters was36.8, merge parameter was 85, it can yielded better results.(2)The spectrum, texture, shape and topography and other parameters were analyzed. Technical indicators and parameters of loess landslide which can identify the optimal combination were grouped to constructed loess landslide extraction algorithm: when Loess landslide =(0.13 < NDVI < 0.34) and(0.06 < blue band <0.10, 1.08 < elongation < 2.48 and 0.24 < roundness < 0.62) and(0.14 < near-infrared wavelengths < 0.25, 1100 <elevation < 1250) and(NDVI texture variance < 0.0025).(3)Based on the landslide areal density(LAP), the landslide point density(LND),we analyzed the loess landslides spatial distribution of Yangou Watershed, revealed the connections among loess landslides and terrain(elevation, slope, aspect,curvature) and the quantitative relationship with land use. Loess landslides occurred mainly in elevation between 1150 ~ 1200 m, LAP and LND were 1.41% and 3.33 ?km-2; gradient(5 m resolution) between 25° ~ 45°, accounting for 88.25%; Aspect South(S), Southeast(SE), southwest(SW), and a maximum of S direction, LAP andLND were 1.55% and 2.71 ? km-2; curvature of-4 to-1, landslides were easy to occur;especially in most woodland, LAP and LND were 2.25% and 5.5 ? km-2, respectively,the grass was less prone to landslide.
Keywords/Search Tags:Landslide, GF-1, Recognition algorithm
PDF Full Text Request
Related items