戈壁表面多尺度砾石特征参数估算及其空间分布规律研究

发布时间：2018-04-13 18:51

本文选题：戈壁地貌 + 砾石粒径　；参考：《中国林业科学研究院》2017年博士论文

【摘要】：戈壁是干旱环境中的一种独特地貌景观。戈壁表面砾石形貌参数研究对于研究戈壁形成演化过程和风沙活动影响,以及开展戈壁地区生态保护和经济开发都具有重要意义。目前在戈壁区开展的大范围、高精度的砾石形貌特征参数研究较少,究其原因主要是地面测量精度高,但受人工测量效率的限制难以获得大量的样本;遥感估算覆盖面积大但是由于遥感影像空间分辨率较低,难以直接测量砾石形貌特征。随着近地面遥感技术的发展,在戈壁表面获取较大覆盖范围、超高地面分辨率的数字图像越来越便利。因此,利用计算机图像模式识别领域先进的图像分析技术发展针对高空间分辨率图像的图像分析技术具有重要的科学意义和应用前景。然而,迄今为止,基于图像自动计算砾石形貌参数的方法在戈壁的形成演化研究中尚未见到报道。因此,本研究提出一种基于图像,利用机器学习算法快速、准确测量戈壁表面砾石覆盖度、粒径、形状比率和方位角等形貌参数的新方法,并利用地面实测和人工矢量化地面摄影数字图像验证了新方法的精度。应用该方法以中国新疆维吾尔自治区哈密市天山南麓洪积扇(93°0′0″E,43°10′0″N)为研究区域,利用野外地面摄影、近地面无人机遥感和卫星影像三种不同尺度数据源,实现了天空地一体化的戈壁表面砾石形貌特征和空间分布规律研究。主要研究成果如下:(1)提出了一套基于决策树图像分类算法、分水岭图像分割算法,快速、准确测定戈壁表面砾石覆盖度、粒径、形状比率和方位角等形貌特征的新技术(Morphological characteristics gain effectively technique,McGET)。通过与地面测量和人工矢量化测量的形貌特征参数进行比较,评估了McGET方法的精度。结果显示:与地面测量结果比较,McGET计算的大砾石(粒径不小于对应样方地面测量粒径最小值)粒径均值与地面测量结果非常接近(y=1.00x+2.74,R2=0.89,P0.001)。与手动勾绘结果比较,McGET计算的砾石覆盖度精度范围在92.6%~99.9%;在戈壁表面不存在大量叠置的砾石的情况下,McGET计算的粒径均值精度大于69%;McGET计算的砾石形状比率均值精度在88%~99%;在戈壁表面不存在大量近圆形砾石的情况下,McGET获得的砾石方位角分布与手动勾绘一致。更重要的是,McGET大大缩短了野外地面测量及室内图像处理花费的时间。因此,McGET可以同时处理大量戈壁样方图片,快速、准确获取多类型的砾石形貌参数数据。(2)McGET技术在样方(1m×1m)、景观(3 ha)和区域(300 km2)尺度上的适用范围有所差异。在样方尺度上,McGET可以提取到单个砾石的所有形貌特征,包含砾石覆盖度、粒径、形状、方位角等,只不过受图像地面分辨率和覆盖范围的限制,样方尺度上提取的砾石粒级范围多在4mm到256mm左右,比较适合粒级较小的戈壁表面砾石研究。在样区尺度上,McGET也可以提取到单个砾石,只不过受图像地面分辨率限制,提取的砾石粒径在32mm以上,比较适合粒级较大的戈壁表面砾石研究。在区域尺度上,McGET可以获取像元内砾石特征的均值,适合通过遥感反演进行大尺度上砾石粒径的分级研究。(3)应用McGET技术结合运动恢复结构和遥感反演技术,分别以地面数字图像、无人机高分辨率图像和卫星影像为数据源,在样方、景观和区域尺度上上对戈壁表面的砾石形貌特征进行了估算。样方尺度上,砾石覆盖度均值约为75%;砾石呈现单峰左偏态分布,砾石粒径均值约为15mm;砾石形状比率均值约为1.57。景观尺度上,自扇心到靠近扇缘,三个样区的砾石覆盖度分别为34.22%,26.85%,21.88%;砾石粒径均值分别为130,95,78 mm。区域尺度上,以Landsat 8 OLI地表反射率产品为数据源,结合地面样方调查,建立了戈壁表面砾石粒径遥感估算模型,获得了较好精度(R2=0.549)。(4)本研究通过综合样方、景观及区域尺度上砾石特征的测定结果,分析了砾石特征随海拔、坡度等因素的变化,得到了景观和区域尺度上戈壁表面砾石的空间分异规律。在景观尺度上,沿海拔变化方向,砾石覆盖度和砾石粒径变化与海拔关系不大,可能受到植被分布和局部地形的影响。以地貌单元为单位分析,在靠近扇心区域,平地砾石覆盖度和粒径都明显高于坡地,扇中区域二者差异不大,靠近扇缘区域平地上砾石覆盖度和粒径都略低于坡地上。自扇心到扇缘,平地上砾石覆盖度和粒径的下降速度由快变缓,坡地上砾石覆盖度和粒径的下降速度相对平地较为均匀。在进行景观尺度的戈壁表面砾石特征空间分布研究时,局部微地形的影响不可忽略。在区域尺度上,从洪积扇扇心到扇缘,戈壁表面的砾石覆盖度和砾石粒径都逐渐减小,而砾石形状比率略微增大。线性回归分析显示,海拔每升高100m,砾石覆盖度显著增加了2.2%,砾石粒径均值显著增加了0.5mm左右,而砾石形状比率下降了0.004左右。沿着海拔变化方向,砾石粒径均值逐渐减小且在扇心和扇缘区域下降速度较快,随水平距离增加分别为1.83mm/km和0.31mm/km,在扇中区域下降速度较慢,为0.15mm/km。同一海拔梯度上,在接近洪积扇中轴线的位置砾石粒径较高,而两侧粒径较小;其中扇心区域,二者差别最大,大于3mm,扇中区域二者差别最小,接近2mm。在区域尺度上,砾石粒径变化受宏观地形下的坡度影响较大。
[Abstract]:Gobi is a unique landscape in arid environment. Study on morphology parameters of Gobi gravel surface influence the formation of the evolutionary process and the aeolian activities for the study of Gobi and Gobi area, carry out ecological protection and economic development are of great significance. At present in the Gobi region of large range, gravel topography of the high precision. The main reason is the high measurement accuracy, but by the manual measurement efficiency limit is difficult to obtain a large number of samples; remote sensing coverage area is large due to the low spatial resolution remote sensing image, is difficult to directly measure the gravel morphology. With the development of remote sensing technology in the near ground, Gobi acquired a large surface coverage, digital image and super the high ground resolution more convenient. Therefore, the use of advanced computer image analysis image pattern recognition technology development for high spatial resolution The rate of image image analysis technology has important scientific significance and application prospect. However, so far, the automatic calculation method based on morphology parameters in the formation of gravel image has not seen the evolution of Gobi. Therefore, this study proposes a new method based on image, using machine learning algorithm is fast, accurate measurement of the surface of the gravel coverage in Gobi a new method, particle size, shape ratio and azimuth angle shape parameters, and the use of ground truth and artificial vector ground photography digital image to verify the accuracy of the new method. The application of this method to Chinese Hami City, south of the Tianshan Mountains in the Xinjiang Uygur Autonomous Region alluvial fan (93 degrees 0 '0 "E, 43 degrees 10' 0" for N) the study area, using field near ground photography, UAV remote sensing and satellite images of three different scales of data source, realize the integration of the distribution of sky surface morphology and spatial characteristics of Gobi gravel Study on the law. The main results are as follows: (1) proposed a decision tree based image classification algorithm, watershed segmentation algorithm, fast and accurate determination of Gobi surface gravel coverage, particle size, new technology of shape ratio and azimuth angle morphology (Morphological characteristics gain effectively technique, McGET) through the comparison. The morphology parameters and the ground measurement and artificial vector measurement, evaluate the accuracy of the McGET method. The results show that: compared with ground measurements, McGET calculation of boulders (diameter not less than the minimum particle size measurement of ground plots) mean particle diameter and the ground are very close to the measured results (y=1.00x+2.74, R2=0.89, P0.001). Compared with the results of manual drawing, McGET calculation accuracy of gravel coverage in the range of 92.6%~99.9%; in Gobi a lot of gravel doesn't exist on the surface of the laminated case, McGET calculation The mean size accuracy is higher than 69%; McGET calculation of gravel shape ratio accuracy in the 88%~99% mean; in Gobi there is a large surface nearly round gravel under the condition that the angular distribution is consistent with the manual drawing range gravel McGET obtained. More importantly, McGET greatly reduces the field measurement and indoor image processing time. Therefore, McGET can handle a large number of Gobi sample pictures, fast, accurate data acquisition of multi parameter gravel morphology types. (2) McGET in the plot (1m * 1m), landscape (3 HA) and region (300 km2) on the scale of scope vary. In scale, McGET can be extracted to all appearance characteristics of single gravel, including gravel coverage, particle size, shape, azimuth angle, but the image ground resolution and coverage restrictions, extraction scale of gravel particle size range in 4mm to about 256mm, than Gobi is suitable for the size of the gravel surface. In the area of small scale, McGET can also be extracted into a single gravel, but by the image ground resolution limit, the extraction of gravel size above 32mm, more suitable for the study of Gobi large size gravel surface. In the regional scale, the mean McGET can obtain the characteristics of gravel in a pixel, suitable for large scale gravel particle size classification by remote sensing. (3) the application of McGET technology combined with the structure and motion recovery of remote sensing technology, respectively, to ground digital image, UAV high-resolution images and satellite images as data sources in the sample, the landscape and regional scale on the gravel morphology the surface of Gobi were estimated. Quadrat scale, gravel coverage average is about 75%; gravel unimodal left skewed distribution, the mean grain size of gravel is about 15mm; the gravel shape ratio average is about 1.5 7. on the scale of landscape, from the heart to close to the edge of the fan fan, three area gravel coverage were 34.22%, 26.85%, 21.88%; the mean grain size of gravel were 130,95,78 mm. on a regional scale, using Landsat 8 OLI surface reflectance products as the data source, combined with ground investigation, established the Gobi gravel surface size estimation models, obtain good precision (R2=0.549). (4) the sample determination results of gravel characteristics landscape and regional scale, analyzes the characteristics of gravel with altitude, slope changes and other factors, the variation of landscape and regional scale space in Gobi. In the landscape of gravel surface on the scale, along the elevation direction, gravel covered little and gravel size and altitude change, may be affected by the distribution of vegetation and the effects of local topography. The landform unit analysis, near the fan area, the ground covered with pebbles The degree and particle size were significantly higher than that in the two fan slope, regional difference, close to the edge of the fan area on the ground gravel coverage and particle size are slightly lower than the slope. From the heart to the edge of the fan fan, flat gravel coverage and particle size decrease speed is slow, the slope and gravel coverage the particle size decreased the relative speed of the ground is more uniform. In landscape scale spatial distribution characteristics of Gobi gravel surface, the effects of local micro topography can not be neglected. In the regional scale, from the heart to the fan fan, Gobi surface gravel coverage and gravel particle size gradually decreases, and the ratio of gravel shape increased slightly. The linear regression analysis showed that the altitude of every 100m increase of gravel coverage increased 2.2%, the mean grain size of gravel increased about 0.5mm, while the gravel shape ratio decreased by about 0.004. Along the elevation change direction, size of rock fragments were Decreases in the fan and heart and edge of the fan area decreased faster, with increasing horizontal distance were 1.83mm/km and 0.31mm/km, in the area of decline was slower in the fan, 0.15mm/km. at the same altitude, near the alluvial fan in axis size of rock fragments is high, and the two side size is smaller; the fan area two, the difference between the maximum, more than 3mm, the difference between the two smallest fan in the area, close to 2mm. in the regional scale, gravel size change by slope under the influence of macro topography is large.

【学位授予单位】：中国林业科学研究院
【学位级别】：博士
【学位授予年份】：2017
【分类号】：P931.3

【相似文献】