青藏高原三江源高寒草地生态系统土壤侵蚀研究

发布时间：2018-11-22 08:58

【摘要】：三江源地区地处青藏高原腹地,由于其独特的地理位置,形成了我国特有的高寒草地生态系统,在水源涵养方面具有极其重要的生态地位,是我国水资源供给的核心发源地,具有以草地生境为主的丰富的动植物多样性。由于长期发生在这一地区的土壤侵蚀现象的不断演变恶化,导致牧草质量下降、草地退化,以及土地荒漠化和“黑土滩”化等一系列问题,因此,对该地区土壤侵蚀监测与评估方法的研究具有十分重要的意义。本研究以GIS为技术平台,RUSEL模型为参考,结合多种机器学习方法。依据137Cs侵蚀模数测算法建立自定义土壤侵蚀模型,并根据其精度进行比对分析,在研究区取得相对精确的侵蚀估算量的同时,评价分析了相应的参数筛选、建模方法和RUSLE模型结果,最后依据侵蚀结果提出了相应的防治对策。主要取得了以下研究结果:1)应用多种基于机器学习的参数筛选和建模方法,获取了精度较高的三江源侵蚀模数空间分布图和相应的侵蚀量,精度最高的自定义模型是模拟退火参数筛选法和Cubist建模方法结合的模型。2)利用RUSLE和模拟退火-Cubist模型分别估算的三江源多年平均侵蚀量分别为3.1*109 t*a-1和2.3*109 t*a-1,且模拟退火-Cubist模型精度显著大于RUSLE模型的精度。说明结合了实测侵蚀数据和GIS遥感大面积应用优势的自定义模型估算精度上明显优于RUSLE模型。3)通过侵蚀强度分级的重心转移分析,两种模型结果的侵蚀强度等级在空间上都符合由东南向西北逐步升高的分布规律。基于县界分区的侵蚀程度比较表明,侵蚀程度最高的三个地区为格尔木市唐古拉乡、治多县和兴海县。4)三江源区不同植被群系合并类型下的土壤侵蚀程度由高到低依次为:农耕地稀疏植被/沼泽/荒漠草原林地草甸灌丛。除草原类侵蚀强度显著偏高,大部分类型基本符合其他研究对各类下垫面土壤侵蚀难易程度的比较结果。结合人文因素分析,草原类植被群系中多以适口性良好的牧草为优势种,故过度放牧是引起草原类下垫面侵蚀现象严重的主要原因。5)通过参数筛选方法对参数重要性的评价,表明风蚀和冻融侵蚀是研究高寒草地生态系统土壤侵蚀时是不可忽略的两种侵蚀动力,是有别于其他下垫面土壤侵蚀的主要特征,而RUSLE模型参数中多年平均风速和土壤冻融数据的应用缺失,是本研究自定义模型整体优于RUSLE模型的主要因素。6)三江源区的土壤侵蚀程度总体偏高。由土壤流失量结合市场价值法,估算研究区多年平均有机质经济损失价值,模拟退火-Cubist模型估算的年均经济损失为299亿元,并且低于用于参考的RUSLE模型估算结果。由此可知,整个研究区受土壤侵蚀现象所产生的经济损失数额巨大。最后,依据三江源侵蚀模数的空间分布的特征,提出了一些土壤侵蚀防治措施及政策建议。
[Abstract]:The Sanjiangyuan region is located in the hinterland of the Qinghai-Xizang Plateau. Because of its unique geographical position, it has formed a unique alpine grassland ecosystem in China and has an extremely important ecological status in water conservation. It is the core source of water resources supply in China. It has abundant animal and plant diversity, which is dominated by grassland habitat. As a result of the continuous evolution and deterioration of soil erosion, which has been occurring in this area for a long time, resulting in a series of problems, such as the decline in the quality of forage, the degradation of grassland, the desertification of land and the transformation of "black soil beaches", and so on, It is of great significance to study the methods of soil erosion monitoring and assessment in this area. This research takes GIS as the technical platform, RUSEL model as the reference, and combines various machine learning methods. According to the calculation method of 137Cs erosion modulus, the self-defined soil erosion model was established, and the precision of soil erosion model was compared and analyzed according to its precision. At the same time, the relative accurate erosion estimation was obtained in the study area, and the corresponding parameter selection was evaluated and analyzed. Modeling method and RUSLE model result. Finally, the corresponding control measures are put forward according to the erosion results. The main results are as follows: 1) by using a variety of parameter selection and modeling methods based on machine learning, the erosion modulus spatial distribution map and the corresponding erosion amount of the three rivers source are obtained. The highest precision of the custom model is the combination of simulated annealing parameter screening method and Cubist modeling method. 2) the annual average erosion amount estimated by RUSLE and simulated annealing Cubist model is 3.1 109t, respectively. * a-1 and 2.3 109t / a, The precision of simulated annealing-Cubist model is significantly higher than that of RUSLE model. It shows that the estimation accuracy of the self-defined model which combines the measured erosion data and the advantages of GIS remote sensing in large area application is obviously superior to that of the RUSLE model. 3) through the gravity center transfer analysis of erosion intensity classification, The erosion intensity grades of the two models are consistent with the distribution law of increasing gradually from southeast to northwest in space. Based on the comparison of erosion degree between county boundaries and subzones, the three areas with the highest erosion degree are Tanggula Township of Golmud City. 4) the degree of soil erosion from high to low under the combination of different vegetation formations in the source region of the three rivers is: sparse vegetation / swamp / desert steppe forest meadow shrub. The soil erosion intensity of steppe was significantly higher than that of steppe, and most of the types of soil erosion were basically consistent with the results of other studies on the soil erosion difficulty and ease of various underlying surfaces. Based on the analysis of human factors, the herbage with good palatability is the dominant species in grassland vegetation formation. Therefore, overgrazing is the main reason for the serious erosion of steppe subsurface. 5) the importance of parameters is evaluated by parameter selection method. The results show that wind erosion and freeze-thaw erosion are two kinds of erosion power which can not be ignored in the study of soil erosion in alpine grassland ecosystem, and are the main characteristics of soil erosion different from those of other underlying surfaces. However, the lack of annual mean wind speed and soil freeze-thaw data in the parameters of RUSLE model is the main factor that the self-defined model is better than the RUSLE model. 6) the soil erosion degree in the source region of the three Rivers is relatively high. Based on the soil loss and market value method, the average economic loss of organic matter in the study area is estimated. The average annual economic loss estimated by simulated annealing-Cubist model is 29.9 billion yuan per year, which is lower than that estimated by the RUSLE model used for reference. Therefore, the economic loss caused by soil erosion in the whole study area is huge. Finally, according to the spatial distribution characteristics of erosion modulus of the source of three rivers, some measures of soil erosion prevention and control and policy suggestions are put forward.
【学位授予单位】：兰州大学
【学位级别】：博士
【学位授予年份】：2016
【分类号】：S812.2

【相似文献】