兰州大学机构库
中国典型区积雪遥感监测及其时空变化特征研究
Alternative TitleSnow Remote Sensing Monitoring and Spatiotemporal Variation Characteristics in Typical Regions in 中文na
王云龙
Subtype博士
Thesis Advisor黄晓东
2020-11-25
Degree Grantor兰州大学
Place of Conferral兰州
Degree Discipline草学
Keyword积雪 MODIS 去云算法 林区 降尺度 气候 响应
Abstract政府间气候变化专门委员会(Intergovernmental Panel on Climate Change, IPCC)第五次评估报告表明,近年来全球出现加速升温趋势毋庸置疑。积雪作为地球表面最为活跃的自然要素之一,对气候变化具有最为敏感的反馈。中国疆域辽阔,积雪覆盖范围广,在全球逐年升温的大背景下,深入研究中国区域积雪时空变化以及积雪对气候变化的动态响应,对于我国雪水资源可持续利用、生态环境保护、区域气候变化以及自然灾害预测等具有重要的研究意义。 结合光学遥感具有较高的时间和空间分辨率,且被动微波遥感可穿透云层、不受不良天气的影响等特点,本研究基于多源遥感数据,利用MODIS逐日积雪标准产品、AMSR-E雪水当量产品以及IMS雪冰产品,发展了一套逐日无云积雪制图算法,生成了2000-2018年中国区域逐日无云积雪范围产品。同时,基于被动微波AMSR2亮度温度数据和光学Landsat数据,分别提出了多因素雪深降尺度算法和林区积雪制图算法。在此基础上,基于中国区域气象台站的气候数据,生成了2000-2018年间中国区域气温和降水栅格数据集,并系统分析了2000-2018年不同时空尺度下中国区域的积雪、气温和降水的时空分布特征,以及年际和各季节的动态变化趋势,最后采用Person相关分析法研究了积雪对气温和降水变化的动态响应关系。本研究结果表明: (1)MODIS逐日积雪标准产品受到云的影响较为严重,无法直接使用该产品进行积雪监测,利用本研究去云算法可以达到了完全去除云像素的目的。但不同的地表覆盖类型条件对本研究合成的逐日无云积雪产品的精度影响不同。整体上,本研究得到的中国区逐日无云积雪产品整体精度较高,Kappa均值达到0.570,接近高度一致性,在完全去除云干扰的条件下有效提高了大尺度准确监测积雪覆盖范围的能力。 (2)我国青藏高原地区积雪深度受到多种因素的影响,包括地理位置,积雪覆盖日数,地形和地表亮度温度因素。通过输入多个变量,雪深降尺度模型的精度得到了极大的改善,其中表现最好的雪深降尺度模型是乘幂模型。整体上,本研究提出的降尺度雪深数据集的均方根误差和平均绝对误差分别为2.00 cm和0.25 cm,均优于其他已存在的青藏高原雪深数据集。在雪深小于3 cm的浅雪区,本研究降尺度雪深数据集精度较高,均方根误差仅为0.58 cm,这是非常有意义的,总体上达到了理想的雪深降尺度效果。 (3)利用NDVI和NDFSI指数对森林地区的积雪有很好的探测潜力,设置NDFSI和NDVI的阈值分别为0.35和0.25。与landsat8 OLI二值积雪图像相比,改进后的林区积雪制图算法的平均偏差为1.24,虚警率为14.34%,分别降低了2.09和33.72%。整体平均精度达到80.67%,提高了22.89%。基于MODIS数据的积雪分类方案综合了NDFSI,NDVI和NDSI多种指数,算法简单有效,在改善我国东北典型森林地区的积雪自动化监测方面作用显著。 (4)2000-2018年中国区域平均年积雪覆盖日数和年均积雪深度的分布具有一定的纬度和海拔地带性特征,即年积雪覆盖日数和积雪深度较大的区域则纬度和海拔相对较高。在2000-2018年间,中国在春季和夏季平均积雪覆盖日数呈减少趋势,而秋季和冬季平均积雪覆盖日数呈增加趋势。积雪深度年际变化略有不同,中国在春季、秋季和冬季平均积雪深度呈增加趋势。整体而言,在多数季节性积雪区的年积雪日数和积雪深度逐年增加,而分布在高纬度地区和高海拔山区的多年积雪区的年积雪日数和积雪深度逐年减少。2000-2018年中国区域平均气温在空间分布上具有明显的海拔地带性,即海拔越高的区域偏向于越低的气温。而中国区域的年降水量在我国南、北方分布差异较大,整体呈现出由东南沿海向西北内陆递减的分布格局。2000-2018年间,中国区气温和降水均呈波动上升趋势,趋势倾向率分别为 -0.024℃/year和4.065 mm/year。 (5)2000-2018年间,中国范围内积雪变化对气温和降水的响应存在明显的区域差异。整体而言,在中国东北地区,降水量是该地区积雪增多的主导性因子。在中国南方部分地区,气温是该地区积雪增多的主导性因子,结合气温的季节变化特征可知,是因为中国南方地区在冬季和春季气温的逐年降低趋势造成了积雪呈现增加趋势的主要原因。在中国新疆北部地区,逐年降低的降水量是积雪表现减少趋势的主导性因子。在青藏高原东部地区,逐年增加的降水量是积雪表现增加趋势的主导性因子,青藏高原东部气候朝暖湿化趋势发展。而在青藏高原南部地区,积雪减少是气温升高和降水量减少的综合作用结果,降水量的减少造成水汽来源不足、积雪累积量减少,逐年的升温导致积雪消融的进一步加快,进而造成积雪的逐年减少趋势,使得该地区气候朝着暖干化趋势发展。
Other AbstractThe fifth assessment report of the Intergovernmental Panel on Climate Change (IPCC) pointed out that there is no doubt that the global warming trend has accelerated in recent years. As one of the most active natural elements in the cryosphere, snow has the most sensitive feedback to climate change. 中文na has a vast territory and a wide range of snow cover. In the context of global climate change, in-depth research on the temporal and spatial variations of snow and the dynamic response of snow to climate change have important research significance in the aspect the sustainable use of snow water resources, ecological environment protection, environmental protection, regional climate change and natural disaster prediction. Based on the characteristics of optical remote sensing with high temporal and spatial resolution and passive microwave remote sensing not be affected by bad weather, we combined the MODIS standard daily snow cover products MOD10A1 and MYD10A1 with the passive microwave remote sensing data of the AMSR-E SWE product and the data of the multi-source remote sensing product IMS to develop a new snow cover mapping algorithm, and generated daily cloud-free snow cover product for 中文na from 2000 to 2018. At the same time, based on AMSR2 brightness temperature products and optical Landsat data, a multi-factor snow depth downscaling algorithm based on a multi-factor approach and snow cover mapping in forested areas are respectively proposed. On this basis, based on the climate data of national meteorological stations, the raster datasets of 中文na's regional temperature and precipitation from 2000 to 2018 are generated, and the temporal and spatial distribution characteristics of snow, temperature and precipitation in 中文na under different scales during 2000-2018 are systematically analyzed, as well as the dynamic trend of inter-annual and seasonal changes. Finally, the Person correlation analysis method is used to study the dynamic response relationship of snow to temperature and precipitation changes. The results of this study indicate: (1)The MOD10A1 and MYD10A1 images are greatly affected by clouds and cannot be directly used for snow monitoring. The cloud removal algorithm in this study can a中文eve the goal of completely removing cloud pixels, but different land cover types can affect the accuracy of the product. Overall, the MODIS daily cloud-free snow products obtained in this study have a high overall accuracy, with the average Kappa being 0.581, which is close to a high consistency. By completely removing cloud interference, we effectively improved the capability to correctly monitor snow cover range at a large scale. (2)In the Tibetan Plateau, snow depth is greatly affected by geographic location, snow-covered days, terrain and brightness temperature difference. By inputting multiple variables, the accuracy of snow depth downscaling model is greatly improved, and the best model is the power model. Overall, the downscaled snow depth datasets proposed in this study has a high accuracy, with root mean square error (RMSE) and mean absolute error of 2.00 cm and 0.25 cm respectively, both of which are better than other existing snow depth products. Furthermore, the downscaled snow depth datasets exhibit good accuracy (RMSE = 0.58 cm) in shallow snow areas where snow depth is less than 3 cm, which is very meaningful. (3)The NDFSI index has good potential to detect snow cover in forested areas with the aid of NDVI index. The threshold value of NDFSI and NDVI is set to be 0.35 and 0.25, respectively. Compared with the snow cover measured by Landsat 8 OLI images, the average BIAS and FAR values of this results are 1.24 and 14.34%, which are reduced by 2.09 and 33.72%, respectively. The overall accuracy of 80.67% is reached, which is improved by 22.89%. The snow classification scheme combining the NDFSI, NDVI and NDSI indexes based on MODIS data used in this work is simple and very effective in improving automatic snow cover mapping in the typical forested areas of Northeast 中文na. (4)The distribution of average annual snow cover days and annual snow depth in 中文na from 2000 to 2018 has certain characteristics of latitude and altitude zonality, that is, areas with more snow cover days and larger snow depth have relatively high latitude. From 2000 to 2018, the average snow cover days in spring and summer showed a decreasing trend in 中文na, while the average snow cover days in autumn and winter showed an increasing trend. The inter-annual variation of snow depth is slightly different. The average snow depth in 中文na in spring, autumn and winter shows an increasing trend. On the whole, the annual snow cover days and snow depth of seasonal snow cover areas in 中文na show an increasing trend, while the annual snow cover days and snow depth in high latitude areas and high-altitude mountainous areas show a decreasing trend. The regional average temperature in 中文na from 2000 to 2018 has obvious altitude zonality in spatial distribution, that is, regions with higher altitudes tend to have lower temperatures. The distribution of annual precipitation in 中文na has obvious differences from north to south, and the overall spatial distribution shows a decreasing distribution pattern from the southeast coast to the northwest inland. From 2000 to 2018, both temperature and precipitation in 中文na showed a fluctuating upward trend, and the trend rates were -0.024°C/year and 4.065 mm/year, respectively. (5)From 2000 to 2018, there are obvious regional differences in the response of snow variations to temperature and precipitation in 中文na. On the whole, in Northeast 中文na, precipitation is the leading factor for the increase in snow in this area. In the parts of southern 中文na, temperature is the leading factor in the increase of snow in this area. Combined with the seasonal variations of temperature, it can be seen that the annual decrease in temperature in winter and spring in southern 中文na has caused the main reason for the increase in snow. In the northern region of Xinjiang, 中文na, the annual decrease in precipitation is the leading factor in the decreasing trend of snow. In the eastern part of the Tibetan Plateau, the annual increase in precipitation is the leading factor in the increasing trend of snow cover, and the climate in the eastern part of the Tibetan Plateau is developing towards a warm and humid trend. In the southern part of the plateau, the decrease in snow is the result of the combined effect of rising temperature and decrease in precipitation. The decrease in precipitation has resulted in insufficient water vapor sources and reduced snow accumulation, and snow melting is further accelerated due to the warming temperature. As a result, the snow has been decreasing year by year, and the climate in this area is developing towards warming and drying.
Pages103
URL查看原文
Language中文
Document Type学位论文
Identifierhttp://ir.lzu.edu.cn/handle/262010/448606
Collection兰州大学
Affiliation草地农业科技学院
Recommended Citation
GB/T 7714
王云龙. 中国典型区积雪遥感监测及其时空变化特征研究[D]. 兰州. 兰州大学,2020.
Files in This Item:
There are no files associated with this item.
Related Services
Recommend this item
Bookmark
Usage statistics
Export to Endnote
Altmetrics Score
Google Scholar
Similar articles in Google Scholar
[王云龙]'s Articles
Baidu academic
Similar articles in Baidu academic
[王云龙]'s Articles
Bing Scholar
Similar articles in Bing Scholar
[王云龙]'s Articles
Terms of Use
No data!
Social Bookmark/Share
No comment.
Items in the repository are protected by copyright, with all rights reserved, unless otherwise indicated.