EGI OpenIR
玛纳斯河流域积雪参数反演及径流模拟研究
Alternative TitleThe Retrieval of Snow parameters and Runoff Simulation in Manas River Basin
任伟伟
Subtype博士
Thesis Advisor杨涛
2020-06-30
Degree Grantor中国科学院大学
Place of Conferral北京
Degree Discipline理学博士
KeywordMODIS 积雪去云 综合概率 雪水当量 TRMM 降水校正 SPHY 水文模型 MODIS snow remove cloud Comprehensive probability Snow water equivalent TRMM precipitation correction SPHY hydrological model
Abstract积雪是西北干旱区水资源重要的组成部分,然而高质量积雪产品的缺乏严重制约了流域积雪灾害的实时监测以及融雪径流模拟研究。积雪对气候变化的影响十分敏感,研制出高精度的积雪和雪水当量产品对于深入了解山区积雪水资源的变化、形成过程、影响因素和未来变化趋势以及对流域水资源管理具有十分重要的意义。本论文以天山北坡玛纳斯河流域为研究区域, 利用遥感反演和模型反演的方法提高积雪产品精度, 并研究积融雪的时空变化规律, 进一步利用积雪产品提高水文模型的模拟精度。主要取得如下认识:(1) 利用 MODIS 积雪产品和 DEM 数据,提出了时空综合概率的 MODIS积雪去云新方法,结果表明,该方法可以显著提高去云精度,平均精度达 96.8%,可以获得高质量的无云积雪产品。然后,利用该积雪产品分析了玛纳斯河流域积雪的时空变化规律。 结果表明, 一年当中,流域最大积雪面积出现在 2 月初,流域最小积雪面积出现在七月底八月初。积雪面积从 2003 年-2015 年有增加的趋势,但增加趋势不显著。年均积雪天数大于 300 天的区域主要分布于流域的西侧和东侧中间区域,以及流域南部的部分区域。而流域内积雪天数小于 50 天的区域主要分布在山谷。在冬季,平均积雪天数大于 70 天的地区,不仅在高山区,在低海拔区也存在。平均积雪天数在海拔 900~2300m 之间呈明显的下降趋势,在 2300~3200m 之间,积雪天数达到最小值,平均积雪天数为 19 天,并且几乎无变化;然而,在 3200m 以上的区域,积雪天数随着海拔的升高而明显升高,最终平均积雪天数达到约 80 天。(2) 利用 AMSR-E 被动微波数据,实测雪压数据和土地利用数据构建了混合像元雪水当量反演模型。通过对比 ERA-interim 雪水当量产品和 ESA 雪水当量产品发现,构建的雪水当量反演模型具有较高的反演精度, R2 达 0.53,同时,ERA-interim 雪水当量产品和 ESA 雪水当量产品相对的 R2 分别是 0.32 和 0.38,能有效地降低反演的误差,提高反演精度。利用反演的雪水当量产品分析了玛纳斯河流域累计雪水当量的变化趋势, 结果表明, 年累计雪水当量较高的区域在分布在流域的南部的冰川区和流域中西部的冰川区,年累计雪水当量达 300mm 以上,在流域中部的河谷地区,年累计雪水当量却是最少的区域,最低处不到100mm,在流域出山口附近,年累计雪水当量到 200mm 以上。整个流域内降雪量最多的时间是冬末夏初。流域内雪水当量增加最明显的区域位于流域东南角的高海拔山区,而雪水当量减少最明显的区域在流域下游平原区。(3) 基于实测降水数据, TRMM 3B43 数据, 植被指数数据,再分析数据,利用贝叶斯神经网络(BNN) 模型和数据融合技术,构建了一个 TRMM 降水数据空间降尺度-校正模型。该模型不仅可以提高 TRMM 降水数据的空间分辨率,而且还提高了降水产品精度,空间分布更合理。 然后利用降水校正模型校正后的1998-2014 年新的 TRMM 月降水数据,获得了流域内的年、季、 月降水随高程的变化规律。 结果发现年降水随高程变化规律比较明显,降水中心主要分布于冰川区和高山区,年降水随高程的变化规律是“增-减-增”。春季和秋季的降水主要发生在中山带,而夏季降水会随着海拔的升高而一直增多,冬季基本上相反,降水随着海拔的增高而降低。流域内存在第二大降水带的可能原因有三:低温,高海拔和冰川的吸水效应。(4)融雪径流是玛纳斯河流域重要的径流组成,准确的模拟积雪径流对于融雪洪水预报和水资源管理具有重要的意义,但是仅用径流校正水文模型存在较大的误差,而用多变量约束模型可以提高模拟精度和稳定性。设置了三组实验: ①仅用日径流数据校正 SPHY 水文模型;②同时积雪面积百分比(fSCA) 和日径流数据校正 SPHY 模型;③同时日均雪水当量(SWE) 和日径流校正 SPHY 水文模型。结果表明综合利用积雪面积百分比和径流数据校正水文模型可以显著的提高模型模拟精度,不仅表现在径流过程上而且还表现在空间分布上。最后分析了玛纳斯河径流组成,在 2003-2012 年,玛纳斯河以融冰径流为主,占了总径流的 32.61%,其次是融雪径流为 26.35%,然后降雨径流和基流分别占 19.91%,21.13%。
Other AbstractSnow cover plays an important role in the northwest arid region for the inlandrivers. However, lacking high-precision snow product (i.e., snow cover data and snowwater equivalent data) seriously restricts the real-time monitoring of snow disaster,snow dynamic and snowmelt simulation in basins. Climate change has dramaticallyaltered snow areas, snow amount and snow duration because snow can be easilychanged between solid and liquid states in response to relatively minor changes intemperature. Therefore, the development of high-precision snow cover and snow waterequivalent (SWE) products is of prime importance for the better understanding of thetotal amount, formation process, influencing factors and future trends of snow waterresources in mountainous areas, as well as the basin water resources management. Here,we take the Manas river basin, on the north slope of Tianshan Mountains, as theresearch area. This study reproduced a cloud free MODIS snow cover product based ona new removal cloud algorithm named as Spatio-Temporal Comprehensive Probabilitymethod. Besides, this study also rebuilt snow water equivalent using mixed pixel snowwater equivalent retrieval algorithm. And then the effects of snow cover and snow waterequivalent product on runoff calibration were verified. Main findings were stated asfollows:(1) Using the original MODIS snow product (i.e., MOD10A1 and MYD10A1) andDEM data, a new MODIS cloud removal method, namely Spatio-TemporalComprehensive probability cloud removal method, is proposed. The results show thatthis method can significantly improve the cloud removal accuracy, with an averageaccuracy of 96.8%. And high-precision cloud-free snow cover product (i.e., new snowcover product) is reproduced by this method. Following that, the new snow coverproduct is used to analyze the temporal and spatial variation of snow in the Manas RiverBasin. Results show that the largest snow area in the basin appeared in early February,and the smallest snow area in the basin appears at the end of July and early August. The snow cover area increases from 2003 to 2015, but the increase is not significant.Regions over 300 snow days are mainly distributed in the southwest side and mid-eastside, as well as parts of the southern side. The region with snow days less than 50 ismainly distributed in valleys. In winter, not only high mountain areas but also lowaltitude areas have at least 70 snow-coverd days. The average snow-covered days showa significant decrease trend between 900-2300m. Between 2300-3200m, the snowcovered days reaches a minimum value. The average snow-covered days is 19 dayswith almost no change. However, above 3200m, the number of snow-covered daysincrease significantly with altitude, and finally the average snow-covered days reaches80 days.(2) Using AMSR-E passive microwave data, measured snow pressure data andland use data, a mixed pixel snow water equivalent retrieval model is constructed. Thennew SWE product is produced by using this model. By comparing the new SWEproduct with ERA-interim SWE product, ESA SWE product, it is found that the newSWE product has a higher average accuracy which R2 is 0.53, while that of ERAinterim SWE product and ESA SWE product are 0.32, 0.38, respectively. Then the newSWE product is used to analyze the trend of cumulative snow water equivalent in theManas River Basin. Results show that glaciers regions had higher annual SWE, wherecumulative SWE was more than 300 mm. The valley region in the middle basin has thesmallest cumulative SWE which is less than 100 mm. The cumulative SWE is morethan 200mm in downstream. There are most snowfall in late winter and early summer.The region with the most significant increase SWE is located in the high-altitude areain the southeast corner of the basin, while the region with the most obvious decreasetrend in SWE is in the lower plains.(3) Based on the measured precipitation data, TRMM 3B43 data, vegetation indexdata, and reanalyzed data, using Bayesian Neural Network model and data fusiontechnology, this study proposed a TRMM spatial downscaling and correction model.This model can not only improve the spatial resolution of TRMM precipitation data,but also improve the precision of precipitation products and the spatial distribution is more reasonable. Then, the distribution of average annual, seasonal, and monthlyprecipitation with elevation was obtained which based on the 1998-2014 new monthlyTRMM precipitation data produced by the TRMM spatial downscaling and correctionmodel. The results show that the annual precipitation changes obviously with elevation.The precipitation centers are mainly distributed in glacial areas and high mountain areas.The variation trend of annual precipitation with elevation is "increasing-decreasingincreasing". In spring and autumn, precipitation mainly occurs in the midmountainareas, while precipitation increase with elevation in summer, in reverse, precipitationdecreases with altitude in winter. There are three possible reasons for the existence ofthe second largest precipitation zone in the river basin: low temperature, high altitude,and water absorption effects of glaciers.(4) Snowmelt runoff is an important runoff component in the Manas River Basin.Accurate simulation of snow cover runoff is of great significance for snowmelt floodforecasting and water resources management. Here we set three experiments: (1)calibration to runoff alone; (2) calibration to both runoff and new snow cover product;(3) calibration to both runoff and new SWE product. It was found that the calibrationto both runoff and new snow cover product can significantly improve the simulationaccuracy of the model, not only in the snowmelt runoff process but also in the spatialdistribution of SWE. Finally, the runoff composition of the Manas River was analyzed.From 2003 to 2012, the Manas River was dominated by ice-melt runoff, accounting for32.61% of the total runoff, followed by snowmelt runoff at 26.35%, and then rainfallrunoff and baseflow, accounted for 19.91%, 21.13%.
Subject Area自然地理学
Language中文
Document Type学位论文
Identifierhttp://ir.xjlas.org/handle/365004/15386
Collection中国科学院新疆生态与地理研究所
研究系统
Affiliation中国科学院新疆生态与地理研究所
First Author Affilication中国科学院新疆生态与地理研究所
Recommended Citation
GB/T 7714
任伟伟. 玛纳斯河流域积雪参数反演及径流模拟研究[D]. 北京. 中国科学院大学,2020.
Files in This Item:
There are no files associated with this item.
Related Services
Recommend this item
Bookmark
Usage statistics
Export to Endnote
Google Scholar
Similar articles in Google Scholar
[任伟伟]'s Articles
Baidu academic
Similar articles in Baidu academic
[任伟伟]'s Articles
Bing Scholar
Similar articles in Bing Scholar
[任伟伟]'s Articles
Terms of Use
No data!
Social Bookmark/Share
All comments (0)
No comment.
 

Items in the repository are protected by copyright, with all rights reserved, unless otherwise indicated.