Knowledge Resource Center for Ecological Environment in Arid Area
| 卫星降水尺度问题及产品检验策略研究 | |
| 其他题名 | Scale-relevant Validation of Satellite Precipitation Products |
| 郭瑞芳 | |
| 出版年 | 2018 |
| 学位类型 | 博士 |
| 导师 | 杨井 |
| 学位授予单位 | 中国科学院大学 |
| 中文摘要 | 自然界降水具有强烈的时空变异特征，准确的降水测量是水文气象领域颇具挑战性的科学问题之一。卫星遥感反演是提供全球降水数据的重要途径。尺度问题是遥感降水反演、降水产品检验以至多源降水数据联合使用所面临的首要问题。论文围绕尺度这一核心问题，从量化遥感降水数据的尺度效应出发，重点发展高时空分辨率降水数据的降尺度方法，设计并提出遥感降水数据多尺度陆面检验策略，为融合多源卫星遥感数据和地面观测数据提供研究基础，最终建立多源降水数据检验的完备理论和方法体系。定量研究了尺度差异对卫星降水产品精度的影响。在华南地区（25°N~30°N，115°E~120°E），以区内100个地面站点观测数据为参照，采用卫星数据升尺度和站点测量数据聚合的多尺度比较方法，定量评估了热带降雨观测卫星（TRMM）3B43数据和3B42实时数据的精度与尺度之间的关系。研究结果显示，随着粒度尺度的增大，遥感降水数据的空间变异性减小；遥感数据和站点数据的各自的均值相当；二者的一致性增大，差异性减小，且系统误差与随机误差均呈现减小趋势。在某一粒度尺度上，降水数据的系统误差趋于稳定，随机误差接近于0。总之，随着粒度尺度的增大，遥感数据与站点数据的一致性增强，差异性减弱，表明遥感数据精度随粒度尺度的增大而提高。此外，利用加法误差模型和乘法误差模型得到的结果具有良好的一致性。首次提出了基于累积直方图匹配的高时空分辨率遥感降水数据的降尺度方法（Downscaling based on Cumulative Distribution of Frequency，DCDF）。该方法假设降水速率和红外波段云顶辐射温度（Tb）之间具有相似的累积频率分布特征。论文以中国地区的六个5°×5°不同气候区为研究对象，基于2014年气候预测中心算法生成的降水数据（CMORPH）和FY-2E热红外波段的Tb数据，在10天1°×1°范围内建立降水速率和Tb之间的定量关系，最终将3小时、0.25°分辨率的CMORPH数据提高到1小时、0.05°分辨率。研究表明，降水速率和Tb之间存在显著的幂函数关系，该关系可以有效地反应降水的特点。降尺度后的降水产品能更好地反映降水的细节和运动过程。当降水速率为1~10 mm·h?1时，降尺度数据和CMORPH数据与雨滴谱数据有良好的一致性。在日尺度上，降尺度数据和CMORPH数据在区域尺度上的精度相近；在点位尺度上，降尺度数据的精度更高。定量检验结果显示，在对流雨为主的区域，降尺度数据的精度最高（Bias = 7.35% ~ 10.35%；R = 0.48 ~ 0.60），相比之下，CMORPH数据的精度为Bias = 20.82% ~ 94.19%，R = 0.31 ~ 0.59。在锋面雨为主的区域，降尺度数据和CMORPH数据的精度相当。在地形雨为主的区域、干旱区、地形复杂区域或者旱季，二者的精度均较低。考虑降水的气象气候学特征与遥感降水数据的多尺度特性，提出陆面遥感降水数据的多尺度检验策略。降水的气象和气候学特征决定了开展检验的空间范围（幅度尺度），同时限定了待检验数据的分辨率（粒度尺度）。针对中小尺度天气系统下的降水产品检验，主要是范围小于200公里、生命期短的对流雨，通常开展小时或日数据的检验。针对中间天气系统尺度下的降水产品检验，主要是范围为200~2000公里、生命期较长的温带锋面雨，通常开展日或者几日数据的检验。针对气候尺度下的降水产品检验，主要是某一区域、气候区或者气候带的月、年遥感数据检验。全球尺度检验是指兼顾降水类型和降水分布规律，包括纬度、海陆和地形等因素对遥感降水精度的影响。最后，以中国区域为研究对象，开展遥感降水数据多尺度检验，检验遥感降水数据多尺度检验策略的有效性。论文首先以华南部分区域为例，从分析遥感降水数据的尺度效应着手，揭示了卫星遥感数据精度随粒度尺度的变化规律；进一步以中国区六大气候区域为研究对象，提出并验证了基于累积直方图匹配的高时空分辨率遥感降水数据降尺度方法，在充分认识数据精度与粒度尺度的基础上，提高了卫星遥感降水数据的时空分辨率；最后以整个中国区域为落脚点，涉及并提出了遥感降水数据陆面多尺度检验策略，为多源卫星遥感降水数据之间、卫星遥感降水数据与地面观测数据的联合使用奠定了理论和方法基础。研究结果同样适用于全球降水数据。 |
| 英文摘要 | Precipitation is governed by a variety of factors, showing strong spatio-temporal variations. Therefore, an accurate estimate of precipitation poses a scientific challenge to the scientific community in the research field of hydrology and meteorology. Satellite remote sensing provides an important means to deriving global precipitation data. However, the issue of scale differences should be firstly tackled for retrieval, validation and combination of multiple precipitation data, because the spatial resolutions of precipitation data vary from ground observations at site scale to satellite observations at a scale of tens of kilometers. Therefore, this study was based on the analyses of scale effects in satellite precipitation data, and developed a precipitation downscaling method, and designed a validation strategy targeted to multi-scale satellite precipitation data over land surface. The results were expected to provide a research basis for the synergy of multi-satellite multi-sensor precipitation data and ground based precipitation data, and offers a self-contained theoretical and methodological system for a comprehensive validation of satellite precipitation data.The effects of scale difference on the accuracy of satellite precipitation data were first investigated. The dependencies between the Tropical Rainfall Measuring Mission (TRMM) 3B43/3B42 real time data and the grain scale were determined using the upscaling (for satellite data) and aggregation (for ground data) methods over the southern China (25°N~30°N，115°E~120°E), which was supported by precipitation data from 100 ground stations. Results showed that the variability of satellite data decreased with the grain scale. Moreover, the mean values were closer to each other, and the consistency (difference) increased (decreased), and the systematic and random error decreased with the grain scale. The syetematic error would finally remain stable and the random error approach to zero at a given grain scale. This grain scale was 2.5° as in this case for the TRMM 3B43. Overall, the similarities (differences) between the satellite and ground data increased (decreased) with the grain scale. In addition, the results were similar for the additive error model and the multiplicative error model.The downscaling based on cumulative distribution of frequency (DCDF) method was proposed to obtain precipitation data with high spatial and temporal resolutions. The basic assumption was that the precipitation rate and the radiant temperature (Tb) shared similar the similar cumulative probability distribution. The Climate Prediction Center morphing technique (CMORPH) data (3 h and 0.25°) in 2014 were downscaled by reference to the Tb data in the FY-2E infrared band, over the six 5°×5° climatic zones in China. A quantitative relationhip was first developed between the precipitation rate and Tb for each 1°×1° subregion within ten days, and then the relationships were applied for the updated precipitation data (1 h and 0.05°). The precipitation rate and Tb showed a significant power function, which was in accordance with the behavior of precipitation. The updated precipitation data can well describe the detailes and motion trajectory of precipitation. Both the updated data and the CMORPH data were in well agreement with the data from the disdrometer, when the precipitation rate was 1 ~ 10 mm·h?1. At daily scale, the accuracies of the updated data and the CMORPH data were similar at regional scale, while the updated data were observed with better accuracies (Bias = 7.35% ~ 10.35%; R = 0.48 ~ 0.60 vs. Bias = 20.82% ~ 94.19%; R = 0.31 ~ 0.59). The updated data performed better over the convective precipitating regions, and behaved similarly for the regions with frontal rain systems. Both data performed poorly over mountainous or hilly areas where orographic rain systems dominated or arid regions or seasons.In view of the meteorological and climatological features of precipitation and the multi-scale satellite precipitation estimates, a multi-scale evaluation strategy for satellite precipitation data was proposed in terms of both grain and extent scales. For the evalution of meso-and micro-scale weather systems characterized by<200 km and short-lived convective precipitation, houly or daily data should be evaluated. For the evalution of intermediate scale systems characterized by 200~2000 km and long-lived temperate frontal precipitation, daily to weakly data should be evaluated. The monthly to yearly data should be evaluated at climate scale over a specific region, climatic region or climatic zone. However, the type and distribution of precipitation should be considered for global scale evaluation. The detailed factors to be considered included latitude, the location of land and sea and topography. Moreover, the evaluation stategy was validated over China.This study revealed the dependency between satellite precipitation data accuracy and the grain scale over part of southern China, based on the understanding of scale effects in satellite precipitation data. Then, the DCDF method was poposed to obtain precipitation data with high spatial and temporal resolutions, and successfully applied to six climatic zones in China. Lastly, a comprehensive evaluation strategy targeted to multi-scale satellite precipitation data was proposed, and tested over the whole China. The results provided a theoretical and methodological basis for the combined use of multi-satellite multi-sensor precipitation data and ground precipitation data. It was expected that the results also applied to the global precipitation data. |
| 中文关键词 | 遥感降水 ; 尺度 ; 精度 ; 降尺度 ; 检验策略 |
| 英文关键词 | Satellite precipitation Scale Downscaling Accuracy Evaluation strategy |
| 语种 | 中文 |
| 国家 | 中国 |
| 来源学科分类 | 地图学与地理信息系统 |
| 来源机构 | 中国科学院南京地理与湖泊研究所 |
| 资源类型 | 学位论文 |
| 条目标识符 | http://119.78.100.177/qdio/handle/2XILL650/288101 |
| 推荐引用方式 GB/T 7714 | 郭瑞芳. 卫星降水尺度问题及产品检验策略研究[D]. 中国科学院大学,2018. |
| 条目包含的文件 | 条目无相关文件。 | |||||
| 个性服务 |
| 推荐该条目 |
| 保存到收藏夹 |
| 导出为Endnote文件 |
| 谷歌学术 |
| 谷歌学术中相似的文章 |
| [郭瑞芳]的文章 |
| 百度学术 |
| 百度学术中相似的文章 |
| [郭瑞芳]的文章 |
| 必应学术 |
| 必应学术中相似的文章 |
| [郭瑞芳]的文章 |
| 相关权益政策 |
| 暂无数据 |
| 收藏/分享 |
除非特别说明，本系统中所有内容都受版权保护，并保留所有权利。
