Knowledge Resource Center for Ecological Environment in Arid Area
| 陆面过程模型模拟径流评价及改进 | |
| 其他题名 | Evaluation and improvement of land surface models against the observed runoff |
| 周新尧 | |
| 出版年 | 2013 |
| 学位类型 | 博士 |
| 导师 | 杨永辉 |
| 学位授予单位 | 中国科学院大学 |
| 中文摘要 | 陆面过程模型是地球科学领域最先进、应用最广泛的一类模型，它通过解析多种地气界面通量，为陆地生态系统的尖端研究、工程技术及政策管理等提供指导，提高陆面过程模型描述地气界面通量的精度，对于我们更加准确地理解地表的物质和能量循环过程、应对气候变化具有重大意义。 径流是地表过程中监测时间最长、范围最广的地气界面通量，与人类社会生产生活息息相关。本文选择径流来评价陆面过程模型，分析模型模拟径流的不确定性，并通过尝试改进陆面过程模型，提高陆面过程模型模拟径流的精度。 首先，在全球尺度上，收集了150个流域1986-1995年的实际观测年径流以及利用6个Budyko模型和18个陆面过程模型模拟的年径流结果。其中13个陆面过程模型的模拟径流从GSWP-2（the second Global Soil Wetness Project）的网站下载，4个陆面过程模型的模拟径流从GLDAS（Global Land Data Assimilation System）的网站下载， 并借助GSWP-2的气象数据和地表边界数据利用CABLE（Community Atmosphere Biosphere Land Exchange）陆面过程模型模拟了全球各流域的径流，借助GSWP-2和GLDAS的降水数据以及CRU (Climate Research Unit)的气象数据利用6个Budyko模型模拟了全球各流域的径流。 其次，在全球多年平均尺度上，从参数不确定性、结构不确定性和降水不确定性三个方面分析了陆面过程模型模拟径流结果不确定性的来源。 最后，在区域尺度上，通过改进CABLE模型的产流模块，利用澳大利亚东南部6个小流域的实测数据验证了模型改进前后径流受结构不确定性影响的变化。 结论如下： 首先，在多年平均尺度上，陆面过程模型的模拟径流与观测径流达到了显著相关，且相关系数很高，GLDAS实验中的陆面过程模型模拟径流的精度要高于GSWP-2实验中陆面过程模型的模拟精度。但各模型模拟径流之间的差异性很大，如GSWP-2数据驱动的SWAP、NSIPP、LaD模型和GLDAS数据驱动的CLM模型模拟径流的精度较高，而GSWP-2数据驱动的CLMTOP、COLASSiB模型和GLDAS数据驱动的Mosaic模型模拟径流的精度较低。Budyko模型模拟径流与观测径流的相关系数同样很高，也达到了显著相关，而且在稳定性和精度上都要优于陆面过程模型，表明该类模型在年尺度的径流模拟方面仍有一定优势。陆面过程模型在干旱区的表现要比在湿润区的表现差，其原因可能是在干旱区，径流对降水的强度和历时比降水量的多少更为敏感。 其次，径流的不确定性研究表明，降水不确定性是径流不确定性的主要原因，GSWP-2与GPCC降水的差异与其驱动的Budyko模型径流的差异达到了显著相关，相关系数为0.95，GLDAS与GPCC降水的差异与其驱动的Budyko模型径流的差异也达到了显著相关，相关系数为0.82。且降水的季节变化也会影响模拟径流的季节变化。在多年平均尺度上，径流对与土壤有关的参数最为敏感。模型结构对径流有影响，但影响不大：在全球尺度上，表现较好的模型径流系数在0.4-0.45之间，且干旱区的径流系数低、湿润区径流系数高。非地表径流占总径流比值的大小对模型的表现影响不大，但干旱区非地表径流占总径流比值低、湿润区非地表径流占总径流比值高时，模型表现更好，表明模型的蒸发和非地表径流结构对径流有一定影响。 最后，从CABLE模型改进前后径流过程的对比来看，在湿润区，改进前后径流过程的变化不大，而在干旱半干旱地区，具有超渗产流结构的CABLE模型比蓄满产流结构的CABLE模型能更好地模拟出洪峰，这反映了干旱区径流产流以超渗产流为主的特征。就两种蓄满产流结构设计来讲，土壤蓄水容量曲线结构比桶状结构更好。统计分析表明，在湿润区，改进前后的模拟径流无显著差异，在干旱半干旱区，改进后（超渗产流结构）的模拟径流精度明显提高。分析其原因是在湿润流域，土壤中的水呈饱和状态，小部分降水转化成土壤水储存起来，大部分降水转化成径流，因此无论径流的峰值能否被模拟出来，径流的总量变化较小。而在干旱半干旱流域，土壤处于不饱和状态，大部分降水转化成土壤水储存起来，小部分降水转化成径流，因此径流的总量变化很大。所以，土壤水模拟是干旱半干旱地区径流正确模拟的关键。 |
| 英文摘要 | The state-of-the-art Land Surface Models (LSMs) are widely used in the simulation of various land surface processes to help us understanding mass and energy circulation in land surface, making water and climate policies. It is meaningful to help us to understand earth process better and face climate change by improving LSMs. Surface runoff relates to human life closely. Meanwhile, surface runoff is widely observed and has the longest time series. In this research, to evaluate LSMs and analyse uncertainties, simulated LSMs runoff were compared with observed runoff. Then a LSM was improved for better results. First of all, to evaluate LSMs, mean annual simulated runoffs in the period 1986 – 1995 from18 LSMs and were compared with those from 6 Budyko-type models and mean annual observed runoff in the same period across 150 global basins. Of the 18 LSMs runoffs, 13 are from GSWP-2 (the second Global Soil Wetness Project) website and 4 are from GLDAS (Global Land Data Assimilation System) website. One LSM runoff was calculated by CABLE (Community Atmosphere Biosphere Land Exchange) using GSWP-2 dataset. The 6 Budyko-type models runoffs were calculated using precipitation from GSWP-2, GLDAS and evapotranspiration from CRU (Climate Research Unit) meteorological data. Secondly, analyse the impact of precipitation uncertainty, parameters uncertainty and structure uncertainty on runoff uncertainties. Finally, the runoff generation structure of a LSM, CABLE, was improved. Runoffs (before and after improvements) were compared in 6 Australian basins. Results are as follows: First of all, all LSMs runoffs have significant correlation with observed runoffs. GLDAS LSMs runoffs are better than GSWP-2 LSMs runoffs. The difference amongst LSMs is huge. Some LSMs outperform the other LSMs, such as SWAP, NSIPP, LaD, and CLM, while some LSMs underperform the other LSMs, such as CLMTOP, COLASSiB, Mosaic. Same as LSMs, Budyko-type models runoffs have significant correlation with observed runoffs. However, Budyko-type models runoffs outperform most LSMs and are more stable than LSMs. Considering different climatic regions, LSMs in wet basins outperform that in dry basins. Secondly, uncertainties of simulated mean annual runoff are analysed. Three factors contribute to the uncertainty. Based on this research, parameters and structures are less important factors, precipitation is the most important factor. Parameters related to soil have the most influences on runoff. However, variation of runoffs is less than 5% while that of parameters is 20%. Runoff ratios and baseflow ratios also have influences on runoff. LSMs are better when the range of runoff ratios is 0.4-0.45, and runoff ratios are low in dry region and high in wet region. There is no relationship between LSMs and baseflow ratio. However, LSMs are better when baseflow ratios are low in dry region and high in wet region. The correlation between precipitation difference and runoff difference from Budyko-type models is significant (0.95 for GSWP-2 dataset and 0.82 for GLDAS dataset). The monthly variation of precipitation is also influence on that of runoff. Finally, in wet basins, the runoffs after CABLE improvement don’t show large difference with those before. In semi-arid and arid basins, the runoffs after CABLE improvement do show some difference with those before, especially peak. Statistic data shows runoffs are best in wet basins and worst in semi-arid basins. The key is on soil moisture. In wet basins, soil moisture is almost saturate, so most precipitation becomes runoff. While in semi-arid and arid basins, most precipitation becomes soil water, so the runoff-generation method is important for runoff. Therefore, it is important to get the accurate soil water content in semi-arid and arid region. |
| 中文关键词 | 陆面过程模型 ; Budyko模型 ; 径流 ; 不确定性 ; 模型评价 |
| 英文关键词 | land surface models Budyko models runoff uncertainties model evaluation |
| 语种 | 中文 |
| 国家 | 中国 |
| 来源学科分类 | 生态学 |
| 来源机构 | 中国科学院遗传与发育生物学研究所 |
| 资源类型 | 学位论文 |
| 条目标识符 | http://119.78.100.177/qdio/handle/2XILL650/287320 |
| 推荐引用方式 GB/T 7714 | 周新尧. 陆面过程模型模拟径流评价及改进[D]. 中国科学院大学,2013. |
| 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 |
| [周新尧]的文章 |
| 百度学术 |
| 百度学术中相似的文章 |
| [周新尧]的文章 |
| 必应学术 |
| 必应学术中相似的文章 |
| [周新尧]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。
