基于海洋气候数据集的区域海平面变化非线性预测

本文基于中国首套长时间序列、高精度、高时空一致性的全球海洋气候数据集产品,利用1993年1月至2015年12月的山东半岛近海海平面异常数据,构建了基于集合经验模式分解(EEMD)和长短期记忆神经网络(LSTM)的海平面非线性变化组合预测模型。EEMD可以得到海平面异常的各周期项、线性趋势及残差部分,LSTM模型可对其进行逐个预测并重构得到最终的海平面异常预测结果。EEMD-LSTM组合模型海平面异常预测的均方根误差仅为25.87 mm,取得了令人满意的效果。基于该组合模型预测2016—2025年山东半岛近海海平面上升速率将达到3.54 mm·a^(-1)。

海洋气候数据集生成与分析简介

科技部的"十三五"重点研发计划项目"全球气候数据集生成及气候变化关键过程和要素监测"中资助了"海洋气候数据集生成与分析"课题,针对全球气候变化研究对全球海洋环境数据的需求,该课题拟基于国内外多源卫星遥感数据和现场观测资料,发展与全球气候变化相关的海面能量平衡、海洋动力环境参数与过程、海洋水色和海冰等数据产品生成技术,构建全球海洋气候数据集产品生成系统,生成我国首套长时间序列、高精度、高时空一致性的全球海洋气候数据集,为全球气候变化研究与应对提供海洋数据支撑和服务,提升我国在全球气候变化关键过程和要素方面的监测能力。

全球气候数据集生成及气候变化应用研究

科技部在"十三五"期间部署的国家重点研发计划"全球变化及应对"专项资助了"全球气候数据集生成及气候变化关键过程和要素监测"研究项目。项目围绕由全球气候观测系统提出的基本气候变量,完善地空天基观测体系,生成中国首套以遥感数据为主体的涵盖大气、海洋和陆表长时间序列、高精度、高时空一致性的产品,即气候数据集,动态监测全球变化关键过程和要素。

卫星气候数据集的应用研究与发展分析

卫星气候数据集是卫星气候研究的基础。在规范卫星气候数据集基本概念的基础上,针对现有基本气候数据集(FCDR)和专题气候数据集(TCDR)的分类方式,无法反映卫星气候数据特点的问题,认为应将专题气候数据集进一步划分为单一遥感仪器专题气候数据集、多种遥感仪器融合专题气候数据集及卫星与多源资料融合专题气候数据集等几类。这种分类方法便于用户更好地了解和使用卫星气候数据。然后,重点围绕基本气候变量和基本卫星气候变量含义、卫星气候数据集生产规范、国内外主要卫星气候数据生产计划等方面,综述了卫星气候数据集建设及规范化生产已取得的最新研究进展。在此基础上,分析了卫星气候数据集建设和应用中存在的主要问题,展望了卫星气候数据集发展,同时对我国卫星气候数据集建设提出具体建议。

中国1km分辨率生物气候要素空间分布数据集

生物气候要素是一种反映生物与气候间综合关系的指标。在生态学研究中,生物气候要素常被用于阐释物种和植被与气候的相互关系,以及进一步模拟物种和植被的地理分布格局及其功能特征,但区域尺度的生物气候空间数据集却较少见报道。本研究基于ANUSPLIN插值获得的中国1 km分辨率的1951-1980年和1981-2010年两个时间段各30年平均气候要素空间分布数据集(气温、降水和日照百分率),计算了最冷月均温、最热月均温、极端最高温和极端最低温、大于0℃和5℃的年生长度日、生长季降水量、年均干旱指数和年均湿润指数9种生物气候要素,绘制其空间分布图,并统计分析其空间格局和变化趋势。比较分析表明,对应指标间变化幅度极小,且统计指标几近相同,故本数据集的误差主要来源于基础气候要素空间分布数据集,处理过程中的二次误差极小。本数据集可为物种和生态系统与气候关系的研究提供合理的环境机理解释,并为研究者使用生物气候要素模拟物种分布和植被结构与功能提供一条便捷多元的途径。

中国区域地面气象要素数据集在长江上游流域的适用性评估

从中国区域地面要素数据集中选取1980年~2010年中国区域的降水和气温两个要素,与实测降水、气温资料进行对比分析,采用确定性系数、纳什系数、平均误差和均方根误差4个指标分别从时间变化特征和空间分布特征两个方面对地面要素集的降水资料和气温资料在长江上游流域内数据质量进行系统性评估。结果表明,气象网格数据与实测数据总体误差较小,纳什系数接近于1,拟合度高,具有很好的一致性;气温数据具有很好的一致性,而降水数据在不同子流域数据质量不同。

CMIP5全球气候模式统计降尺度数据对辽宁省极端气温模拟能力评估

利用1971−2010年辽宁地区32个气象台站观测资料,采用SS指标和M2指标,评估CMIP5气候预估降尺度数据集中31个模式对辽宁省全年和4个季节极端气温指数时空变化特征的模拟能力,并经MR综合评级,筛选模拟效果最好的模式。极端温度指数选用平均最高气温(TXm)、平均最低气温(TNm)、霜冻日数(FD0)、夏日日数(SU25)、最低气温极小值(TNn)、最高气温极大值(TXx)和极端气温范围(ETR)。结果表明:对单一极端气温指数而言,所有模式对表征气温平均特征的指数模拟效果较好,对表征连续极端气温事件的指数模拟效果一般,而对气温的极端特征模拟能力不足。由于模式对全年、各季节各极端气温指数的排名无一致性,引入MR评级指标进行排序,优选的3个模式为MPI-ESM-LR、GFDL-ESM2M和MIROC-ESM-CHEM。比较发现,优选模式集合平均结果相对于观测值的误差百分率整体上明显优于所有模式的集合平均结果,具有可用性。研究结果可为预估未来区域极端气温提供理论参考,建立的逐日模式资料可为开展辽宁农业气象灾害风险相关研究提供数据支撑。

1995—2010年静止卫星云量数据检验和评价

对利用FY2和GMS静止气象卫星建立的东亚地区气候数据集(EAGSCDR-FY2and GMS GeostationarySatellite Climate Data Record over East Asia)进行了检验和评估,使用的检验源数据包括中国地面气候资料与国际卫星云气候计划ISCCP D2月平均云量数据集。对由上述3种不同观测手段得到的多年平均总云量的空间分布特征分析结果表明:3种资料的总云量分布形势有较好的一致性,但是在40°N以北地区,ISCCP和EAGSCDR得到的总云量在量值上高于地面观测值。用地面观测资料检验华南及长江流域EAGSCDR的云检测产品的结果表明,总的准确率为82.10%,总漏判率6.85%,总误判率为11.05%,秋冬季节准确率偏低。EAGSCDR与ISCCP云量都是由卫星资料处理得到的,二者差异主要来自算法的不同,检验结果表明,EAGSCDR中的云量产品精度优于ISCCP云量,并且其时间分辨率可达到1h,空间分辨率达到5km,由此可见,EAGSCDR的云产品比ISCCP云产品更有优势。

1901~1950年5~9月北半球CRU数据与树轮资料的对比

气候变化的研究依赖于可靠的气候数据。英国东英吉利大学气候研究中心推出的一套依据观测资料插值所得的月平均气候数据集(CRU),是目前使用较为广泛、质量较高的气候数据集。气候观测数据大多始于1950年,已有的研究也表明CRU数据在1950年后更具参考价值,但该数据集在1950年之前的可靠性不清楚。由于1950年之前缺乏器测数据,本研究使用树轮资料,对比每十年树轮重建温度(降水)与树轮附近CRU格点上的温度(降水)的距平和趋势的同号率,来评价CRU温度(降水)数据在不同地区的数据质量(可靠性),由此发现从1930年代开始,CRU温度数据在欧洲和中西伯利亚可信度较高;从1940年代开始,CRU温度数据在美国东部可信度较高;从1940年代开始,CRU降水数据在美国东部与西西伯利亚可信度较高。研究发现,气象站密度对CRU数据质量有很大影响;在气象站密度小的地区,CRU温度或降水数据与树轮序列的差异更为显著,气候数据质量偏低;而数据可信度较高的年代和地区,气象站密度往往较大。

Zonal Mean Mode of Global Warming over the Past 50 Years

Zonal mean annual temperature trends were estimated using four reanalysis and three analysis grid datasets. The trends over land and for the entire globe were estimated from 1958-2001 and 1979-2007, respectively. Estimates of temperature trends over land from Climate Research Unit （CRU） analysis data indicate more intense wanning moving northward, at a rate of about 3.5℃ per century at 65°N, then declining further to the north. CRU estimates indicated dramatic warming over the latitudes of the Antarctic Peninsula, with a localized cooling trend at 45°S. A global estimate was conducted by comparing estimates of the reanalysis datasets. Temperature distribution trends of the reanalysis data were similar to those generated by land observations but with large bias in the Polar Regions. The bias could be reduced by comparing these estimates with those from the analysis data at high latitudes. Extreme warming trends were esti- mated at rates of 2.9℃-3.5℃ per century in the Arctic and 3.2℃-4.7℃ per century in the Antarctic for 1958-2001. Surface warming was even more intense in the Northern Hemisphere for 1979-2007, with extreme arctic warming rates ranging from 8.5℃-8.9℃ per century, as estimated by the analysis and reanalysis datasets. Trends over Antarctica for this period were contradictory, as Japan Meteorological Agency （JMA） reanalysis （JRA-25） indicated a cooling trend at about -7℃ per century, while other reanalysis datasets showed sharp warming over the continent.

用于洞察多时间尺度水文过程的深度学习模型

在流量预测工作中,搭建了由单元、输入门、输出门和遗忘门组成的循环神经网络长短期记忆(RNN LSTM)模型,能学习长期时间依赖和捕获非线性关系,用来预测每日、每周和每月时间尺度的河流流量。利用2006—2014年唐河流域倒马关水文站的多个气候和水文要素数据集,可以对深度学习模型进行训练和验证。通过设置多个时序步长,探讨记忆单元存储信息的长短对不同时间尺度的影响。研究发现,RNN LSTM模型在日尺度预测中表现出较好的预测性能,数据集的粗粒度特性是影响周尺度和月尺度预测性能的关键。

Discussion of Influences on Snow Water Equivalent at Utah Snow Courses

Snow data collection systems in the western United States were originally designed to forecast water supply and may be subject to several sources of bias. In addition to climate change and weather modification effects, site-specific effects may be introduced from vegetation changes, site physical changes, measurement technique, and sensor changes. This paper examines changes in Utah＇s snowpack conditions over the past decade compared with all previous measurement years, focusing on the 15 snow courses with the longest observational record within the state of Utah. Although patterns in snowpack data consistent with those that would be expected due to temperature h as greater declines at lower elevations and latitudes--were not identified, snow water equivalent decreased at sites with significant increases in vegetation coverage. Additionally, we provide a list of 22 snow courses in Utah that are best-suited for long-term climate analysis.

陆表定量遥感反演方法的发展新动态

随着获取的遥感数据越来越多,定量遥感正处于一个飞速发展的时期。本文从反演方法和遥感数据产品生成两个主要方面对近期陆表定量遥感的发展进行评述。由于大气—陆表系统的环境变量数远远超过遥感观测数,定量遥感反演的本质是个病态反演问题。在评述机器学习方法(包括人工神经网络、支持向量回归、多元自适应回归样条函数等)的应用基础上,重点关注克服病态反演的7种正则化方法:多源数据、先验知识、最优化反演的求解约束、时空约束、多反演算法集成、数据同化和尺度转换。定量遥感发展的另外一个显著特征是由数据提供者(比如数据中心)将观测的遥感数据转换成不同的地球生物物理化学参数产品,即遥感高级产品,并服务于数据使用者。概括介绍了北京师范大学牵头研发的GLASS(Global LAnd Surface Satellite)产品的新进展与全球气候数据集的研发情况。

A climatic dataset of ocean vertical turbulent mixing coefficient based on real energy sources

Using data on wind stress, significant height of combined wind waves and swell, potential temperature, salinity and seawater velocity, as well as objectively-analyzed in situ temperature and salinity, we established a global ocean dataset of calculated wind- and tide-induced vertical turbulent mixing coefficients. We then examined energy conservation of ocean vertical mixing from the point of view of ocean wind energy inputs, gravitational potential energy change due to mixing(with and without artificially limiting themixing coefficient), and K-theory vertical turbulent parameterization schemes regardless of energy inputs. Our research showed that calculating the mixing coefficient with average data and artificial limiting the mixing coefficient can cause a remarkable lack of energy conservation, with energy losses of up to 90% and changes in the energy oscillation period. The data also show that wind can introduce a huge amount of energy into the upper layers of the Southern Ocean, and that tidesdo so in regions around underwater mountains. We argue that it is necessary to take wind and tidal energy inputs into account forlong-term ocean climate numerical simulations. We believe that using this ocean vertical turbulent mixing coefficient climatic dataset is a fast and efficient method to maintain the ocean energy balance in ocean modeling research.

An assessment of summer sensible heat flux on the Tibetan Plateau from eight data sets

The eight datasets of the summer (June-August) surface sensible heat (SH) flux over the Tibetan Plateau (TP) are compared on the time scales of the climatology,interannual variability and linear trend during 1980-2006.These data sets include five reanalyses (National Center for Environmental Prediction reanalysis,NCEPR1 and NCEPR2,NCEP climate forecast system reanalysis,CFSR,Japanese 25-year reanalysis,JRA,and European Centre for Medium Range Weather Forecasts reanalysis,ERA40),two land surface model outputs (Noah model data of Global Land Data Assimilation System version 2,G2_Noah,and Simple Biosphere version 2 output by Yang et al.,YSiB2),and estimated SH based on China Meteorological Administration (CMA) station observations,ObCh.The results suggest that the summer SH on the TP differs from one dataset to another due to different inputs and calculations.Climatologically,the ERA40 and JRA distribute rather uniformly while the other six products show similar regional disparities,that is,larger in the west than in the east and stronger in the north and the south than in the middle of the plateau.The mean magnitude of the SH averaged over the 76 stations above the TP varies considerably among each dataset with the difference of more than 20 W m?2 between the maximum (G2_Noah) and minimum (ObCh).Nevertheless,they are consistent in the interannual variability and mostly show a significant decreasing trend corresponding to the weakening surface wind speed,in spite of the distinct trend for the ground-air temperature difference among the different data sets.These two consistencies indicate the particular availability of the SH products,which is helpful to the relevant climate dynamics research.

Climate effects of the GlobeLand30 land cover dataset on the Beijing Climate Center climate model simulations

Land cover is one of the most basic input elements of land surface and climate models. Currently, the direct and indirect effects of land cover data on climate and climate change are receiving increasing attentions. In this study, a high resolution(30 m) global land cover dataset(Globe Land30) produced by Chinese scientists was, for the first time, used in the Beijing Climate Center Climate System Model(BCC_CSM) to assess the influences of land cover dataset on land surface and climate simulations. A two-step strategy was designed to use the Globe Land30 data in the model. First, the Globe Land30 data were merged with other satellite remote sensing and climate datasets to regenerate plant functional type(PFT) data fitted for the BCC_CSM. Second, the up-scaling based on an area-weighted approach was used to aggregate the fine-resolution Globe Land30 land cover type and area percentage with the coarser model grid resolutions globally. The Globe Land30-based and the BCC_CSM-based land cover data had generally consistent spatial distribution features, but there were some differences between them. The simulation results of the different land cover type dataset change experiments showed that effects of the new PFT data were larger than those of the new glaciers and water bodies(lakes and wetlands). The maximum value was attained when dataset of all land cover types were changed. The positive bias of precipitation in the mid-high latitude of the northern hemisphere and the negative bias in the Amazon, as well as the negative bias of air temperature in part of the southern hemisphere, were reduced when the Globe Land30-based data were used in the BCC_CSM atmosphere model. The results suggest that the Globe Land30 data are suitable for use in the BCC_CSM component models and can improve the performance of the land and atmosphere simulations.

Climatologic comparison of HadISST1 and TMI sea surface temperature datasets

Two monthly datasets of sea surface temperature (SST),TMI SST retrieved from satellite observations by Remote Sensing System and HadISST1 (Hadley Centre Sea-ice and Sea-surface Temperature Data Set Version 1) derived from in situ measurements by Hadley Centre,were compared on climatologic multiple time scales over tropical and subtropical areas from 1998 to 2006.Results indicate that there is a good consistency in the horizontal global distribution,with 1.0° resolution on multi-year and multi-season mean scales between the two datasets,and also in the time series of global mean SST anomalies.However,there are still some significant differences between the datasets.Generally,TMI SST is relatively higher than HadISST1.In addition,the differences between the two datasets show not only remarkable regionality,but also distinct seasonal variations.Moreover,the maximum departure occurs in summer,while theminimum takes place in autumn.For all seasons,over 30% of the regions in the Tropical and Subtropical areas have a difference of more than 0.3°C.EOF analysis of the SST anomaly field also shows that there are differences between the two datasets,where HadISST1 has more significant statistical characteristics than TMI SST.On the other hand,results show that the difference between the two datasets is related to the vertical structure of ocean temperatures,as well as other simultaneously retrieved parameters in TMI products,such as wind speed,water vapor,liquid cloud water and rain rates.In addition,large biases between HadISST1 and TMI SST are found in coastal regions,where TMI SST cannot be accurately retrieved because of polluted microwave signals.