【摘要】：三江源地區(qū)地處青藏高原腹地,由于其獨特的地理位置,形成了我國特有的高寒草地生態(tài)系統(tǒng),在水源涵養(yǎng)方面具有極其重要的生態(tài)地位,是我國水資源供給的核心發(fā)源地,具有以草地生境為主的豐富的動植物多樣性。由于長期發(fā)生在這一地區(qū)的土壤侵蝕現(xiàn)象的不斷演變惡化,導(dǎo)致牧草質(zhì)量下降、草地退化,以及土地荒漠化和“黑土灘”化等一系列問題,因此,對該地區(qū)土壤侵蝕監(jiān)測與評估方法的研究具有十分重要的意義。本研究以GIS為技術(shù)平臺,RUSEL模型為參考,結(jié)合多種機器學(xué)習(xí)方法。依據(jù)137Cs侵蝕模數(shù)測算法建立自定義土壤侵蝕模型,并根據(jù)其精度進行比對分析,在研究區(qū)取得相對精確的侵蝕估算量的同時,評價分析了相應(yīng)的參數(shù)篩選、建模方法和RUSLE模型結(jié)果,最后依據(jù)侵蝕結(jié)果提出了相應(yīng)的防治對策。主要取得了以下研究結(jié)果:1)應(yīng)用多種基于機器學(xué)習(xí)的參數(shù)篩選和建模方法,獲取了精度較高的三江源侵蝕模數(shù)空間分布圖和相應(yīng)的侵蝕量,精度最高的自定義模型是模擬退火參數(shù)篩選法和Cubist建模方法結(jié)合的模型。2)利用RUSLE和模擬退火-Cubist模型分別估算的三江源多年平均侵蝕量分別為3.1*109 t*a-1和2.3*109 t*a-1,且模擬退火-Cubist模型精度顯著大于RUSLE模型的精度。說明結(jié)合了實測侵蝕數(shù)據(jù)和GIS遙感大面積應(yīng)用優(yōu)勢的自定義模型估算精度上明顯優(yōu)于RUSLE模型。3)通過侵蝕強度分級的重心轉(zhuǎn)移分析,兩種模型結(jié)果的侵蝕強度等級在空間上都符合由東南向西北逐步升高的分布規(guī)律�；诳h界分區(qū)的侵蝕程度比較表明,侵蝕程度最高的三個地區(qū)為格爾木市唐古拉鄉(xiāng)、治多縣和興�？h。4)三江源區(qū)不同植被群系合并類型下的土壤侵蝕程度由高到低依次為:農(nóng)耕地稀疏植被/沼澤/荒漠草原林地草甸灌叢。除草原類侵蝕強度顯著偏高,大部分類型基本符合其他研究對各類下墊面土壤侵蝕難易程度的比較結(jié)果。結(jié)合人文因素分析,草原類植被群系中多以適口性良好的牧草為優(yōu)勢種,故過度放牧是引起草原類下墊面侵蝕現(xiàn)象嚴重的主要原因。5)通過參數(shù)篩選方法對參數(shù)重要性的評價,表明風(fēng)蝕和凍融侵蝕是研究高寒草地生態(tài)系統(tǒng)土壤侵蝕時是不可忽略的兩種侵蝕動力,是有別于其他下墊面土壤侵蝕的主要特征,而RUSLE模型參數(shù)中多年平均風(fēng)速和土壤凍融數(shù)據(jù)的應(yīng)用缺失,是本研究自定義模型整體優(yōu)于RUSLE模型的主要因素。6)三江源區(qū)的土壤侵蝕程度總體偏高。由土壤流失量結(jié)合市場價值法,估算研究區(qū)多年平均有機質(zhì)經(jīng)濟損失價值,模擬退火-Cubist模型估算的年均經(jīng)濟損失為299億元,并且低于用于參考的RUSLE模型估算結(jié)果。由此可知,整個研究區(qū)受土壤侵蝕現(xiàn)象所產(chǎn)生的經(jīng)濟損失數(shù)額巨大。最后,依據(jù)三江源侵蝕模數(shù)的空間分布的特征,提出了一些土壤侵蝕防治措施及政策建議。
[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.
【學(xué)位授予單位】：蘭州大學(xué)
【學(xué)位級別】：博士
【學(xué)位授予年份】：2016
【分類號】：S812.2

【相似文獻】

相關(guān)期刊論文 前10條

1 陳永宗;我國土壤侵蝕研究工作的新進展[J];中國水土保持;1989年09期

2 張翼;黃土高原丘陵溝壑區(qū)土壤侵蝕研究進展[J];水土保持學(xué)報;2001年02期

3 李立青,劉普靈,楊明義;放射性核素~7Be在土壤侵蝕研究中的應(yīng)用現(xiàn)狀及前景[J];核農(nóng)學(xué)報;2003年05期

4 張利華,薛重生;“3S”技術(shù)在土壤侵蝕研究中的應(yīng)用——以湖北省東北地區(qū)為例[J];長江流域資源與環(huán)境;2004年05期

5 胡文生,蔡強國,陳浩;攝影測量技術(shù)在土壤侵蝕研究中的應(yīng)用[J];水土保持研究;2004年04期

6 牛振華,王占禮,黃新會;我國黃土地區(qū)土壤侵蝕研究歷史與進展[J];水土保持研究;2004年04期

7 周江紅;雷廷武;;流域土壤侵蝕研究方法與預(yù)報模型的發(fā)展[J];東北農(nóng)業(yè)大學(xué)學(xué)報;2006年01期

8 呂明輝;王紅亞;蔡運龍;;基于湖泊(水庫)沉積物分析的土壤侵蝕研究[J];水土保持通報;2007年03期

9 朱連奇;馮文峰;朱文博;;土壤侵蝕研究中的“3S”技術(shù)應(yīng)用進展[J];地理科學(xué)進展;2008年06期

10 衛(wèi)亞星;王莉雯;劉闖;;基于遙感技術(shù)的土壤侵蝕研究現(xiàn)狀及實例分析[J];干旱區(qū)地理;2010年01期

相關(guān)會議論文 前7條

1 董元杰;史衍璽;;磁測技術(shù)在土壤侵蝕研究中的應(yīng)用[A];“全國水土流失與江河泥沙災(zāi)害及其防治對策”學(xué)術(shù)研討會會議文摘[C];2003年

2 李旭;吳麗萍;馮學(xué)武;王學(xué)東;;內(nèi)蒙古內(nèi)陸河流域農(nóng)牧交錯區(qū)旱作農(nóng)田土壤侵蝕研究[A];“全國水土流失與江河泥沙災(zāi)害及其防治對策”學(xué)術(shù)研討會會議文摘[C];2003年

3 曹忠杰;林素蘭;賈天會;;遼東山區(qū)柞蠶場土壤侵蝕研究[A];海峽兩岸減輕災(zāi)害與可持續(xù)發(fā)展論文專輯[C];2000年

4 魏彥昌;歐陽志云;王效科;;~(137)Cs法土壤侵蝕研究在熱帶地區(qū)的應(yīng)用[A];生態(tài)學(xué)與全面·協(xié)調(diào)·可持續(xù)發(fā)展——中國生態(tài)學(xué)會第七屆全國會員代表大會論文摘要薈萃[C];2004年

5 琚彤軍;劉普靈;;中子活化REE示蹤技術(shù)在土壤侵蝕研究中的應(yīng)用[A];中國原子能農(nóng)學(xué)會第七次代表大會暨學(xué)術(shù)研討會論文集[C];2004年

6 林洪;于泳;宋立旺;;曹娥江大閘樞紐工程建設(shè)期土壤侵蝕研究[A];中國原水論壇專輯[C];2010年

7 張保華;王喜;劉子亭;何毓蓉;;貢嘎山天然林和鹽亭人工林土壤侵蝕研究[A];《中國水土保持科學(xué)》第1屆4次編委會會議論文集[C];2006年

相關(guān)博士學(xué)位論文 前3條

1 王雪璐;青藏高原三江源高寒草地生態(tài)系統(tǒng)土壤侵蝕研究[D];蘭州大學(xué);2016年

2 溫熙勝;三峽庫區(qū)坡耕地土壤侵蝕研究[D];北京林業(yè)大學(xué);2007年

3 伍鐵牛;更新世黃土高原中南部土壤侵蝕研究[D];中國科學(xué)院研究生院（教育部水土保持與生態(tài)環(huán)境研究中心）;2010年

相關(guān)碩士學(xué)位論文 前8條

1 楊廣斌;基于網(wǎng)格數(shù)據(jù)的喀斯特山區(qū)土壤侵蝕研究[D];貴州師范大學(xué);2004年

2 閆夏嬌;降雨過程中堿性坡耕地土壤侵蝕研究[D];太原理工大學(xué);2012年

3 辛偉;四川鹽亭縣城鎮(zhèn)土壤侵蝕研究[D];西南大學(xué);2008年

4 蔡欣宇;砒砂巖地區(qū)高速公路邊坡土壤侵蝕研究[D];長安大學(xué);2008年

5 肖繼兵;遼西地區(qū)坡耕地土壤侵蝕研究[D];中國農(nóng)業(yè)科學(xué)院;2011年

6 崔文靜;江西紅壤區(qū)雙核素示蹤土壤侵蝕研究[D];南京師范大學(xué);2012年

7 陶炳德;基于~(137)Cs和~(210)Pb_(ex)的共和盆地塔拉灘土壤侵蝕研究[D];青海師范大學(xué);2013年

8 姜洪濤;基于~（137）Cs技術(shù)的中國北方農(nóng)牧交錯帶土壤侵蝕研究[D];內(nèi)蒙古師范大學(xué);2010年

，

本文編號：2348780