【作者】 吴佳；

【导师】 高学杰；

【作者基本信息】 中国气象科学研究院 ， 气象学， 2012， 博士

【摘要】 为高分辨率气候模式检验等的需要，论文首先基于2400余个中国地面气象台站（包括基准站、基本站和国家一般气象站）的观测资料，通过插值建立了一套0.25°×0.25°和0.5°×0.5°经纬度分辨率的格点化数据集（CN05.1）。CN05.1包括日平均和最高、最低气温，以及降水4个变量。插值通过常用的“距平逼近”方法实现，首先将计算得到的气候平均场使用薄板样条方法进行插值，随后使用“角距权重法”对距平场进行插值，然后将两者叠加，得到最终的数据集。将CN05.1与CN05、EA05和APHRO三种日气温和降水资料进行对比，分析了它们对气候平均态和极端事件描述上的不同，结果表明几者总体来说在中国东部观测台站密集的地方差别较小，而在台站稀疏的西部差别较大，相差最大的是青藏高原北部至昆仑山西段等地形起伏较大而很少或没有观测台站的地方，反映了格点化数据在这些地区的不确定性，在使用中应予以注意。在国际上正在开展的CORDEX（Coordinated Regional Climate DownscalingExperiment,联合区域气候降尺度试验）框架下，本研究使用两个不同的全球模式输出结果（德国马普气象研究所的ECHAM5_MPI-OM和英国气象局哈德来中心的HadCM3）作为初始和侧边界场，分别驱动区域气候模式RegCM4.0，进行CORDEX-EastAsia区域（以下简称CORDEX-EA区域，包括东亚和东南亚地区）在IPCC SRES A1B温室气体排放情景下1978~2099年，水平分辨率为50km的两组气候变化预估模拟试验（下文分别简称为EdR和HdR）。为简明起见，下文中将以上四组不同的模拟结果（ECHAM5_MPI-OM、HadCM3、EdR和HdR）简称为四个模拟。首先对比并检验了模式对中国区域当代（1980~1999）气候的模拟能力，结果表明：模式对中国地区冬（12~2月）、夏季（6~8月）和年平均地面气温和降水的总体分布型均具有一定的模拟能力。与全球模式相比，区域模式为气温和降水的空间分布提供了更详细的信息。就气温来看，EdR模拟均要好于全球模式，HdR则改进不大。区域模式对降水的改进不明显，EdR仅略好于ECHAM5，HdR则表现比HadCM3要差。模式对中国地区气温和降水年际变率及其趋势变化的模拟较气候平均态均明显要差，模拟结果之间也存在较大差异。区域模式模拟的副热带高压及西南气流均偏弱，造成中国夏季降水模拟偏少。对A1B情景下中国未来气候变化的预估表明，模式所预估的地面气温变化，在21世纪中期和末期冬、夏季和年平均都表现为一致增加，全球模式及其驱动下的区域模式结果在空间分布上较为相似，HadCM3的升温幅度最显著。末期地面气温均在中期基础上进一步升高，ECHAM5和EdR模拟的中国区域年平均升温值分别为4.5°C和4.0°C，HadCM3和HdR则分别为4.7°C和3.7°C。两全球模式模拟的中国21世纪中期冬、夏季和年平均降水变化在大部分区域以增加为主，末期与中期的分布基本一致，但ECHAM5的夏季降水变化有所不同。两区域模式模拟的中期降水变化分布与末期也比较类似，与各自驱动全球模式相比，冬季降水变化分布接近，夏季在中国东部则有较大不同，同时两个区域模式模拟结果间也存在较大差异。21世纪末期，ECHAM5和EdR的区域年平均降水分别增加3.6%和2.5%，HadCM3及HdR增幅相对较大，分别为12.1%和10.5%。模式对中国夏季降水变化的模拟表现出较大差异，主要可能是对夏季平均水汽通量及其散度变化的模拟存在较大的差别和对地形及地形强迫的描述不同引起的。随后还分析了东南亚区域当代气候的模拟情况，结果表明：模式对东南亚地区冬、夏季和年平均气温、降水的总体分布型有一定模拟能力，区域模式由于分辨率较高，提供了气温和降水更详细的空间分布信息。模拟在冬半年均较夏半年要好，EdR和HdR对地面气温的模拟较两全球模式均有一定改进，但降水则改进不大，HdR甚至不如全球模式。模式对东南亚年平均气温和降水年际变率的模拟能力均较气候平均态要差，对气温和降水趋势变化的模拟与观测之间也存在明显差异。区域模式对低纬和海上环流的模拟与NCEP资料差别较大，对赤道以北地区西风气流的模拟偏弱，导致东南亚夏季降水过少。对东南亚未来气候变化的预估表明，21世纪中期，四个模拟的东南亚区域冬、夏季和年平均气温变化均表现为一致增加，HadCM3及HdR的升温幅度较大。末期地面气温均在中期变化的基础上进一步增加，夏季EdR升温值最大。未来东南亚地区气温变化的季节性差异不明显。末期，ECHAM5和EdR模拟的东南亚区域年平均升温值分别为3.3°C和3.5°C，HadCM3和HdR则分别为3.2°C和2.9°C，其总体升温幅度较中国区域平均要小。模式对东南亚地区未来冬、夏季和年平均降水变化的预估存在较大差异。两全球模式模拟的降水变化趋势在中期和末期基本保持一致，但两区域模式模拟结果在中期和末期差别较大。21世纪末期，两全球模式模拟的东南亚区域平均降水变化为较小的增加，分别增加3.8%和3.5%，两区域模式模拟则表现为减少趋势，减少值分别为-6.0%和-4.2%。更多还原

【Abstract】 A new gridded daily dataset with the resolution of0.25°/0.5°latitude by0.25°/0.5°longitude, CN05.1, is constructed for the purpose of high resolution climate model validationover China region. The dataset is based on the interpolation from over2400observing stationsin China, includes4variables: daily mean, minimum and maximum temperature, dailyprecipitation. The “anomaly approach” is applied in this interpolation. The climatology is firstinterpolated by thin-plate smoothing splines and then a gridded daily anomaly derived fromangular distance weighting method is added to climatology to obtain the final dataset.Intercomparison of the dataset with other three daily datasets, CN05for temperature, and EA05and APHRO for precipitation is conducted.For multi-annual mean temperature variables, results show small differences over easternChina with dense observation stations, but larger differences (warmer) over western China withless stations between CN05.1and CN05. The temperature extremes are measured by TX3D(mean of the3greatest maximum temperature in a year) and TN3D (mean of the3lowestminimum temperature). CN05.1in general shows a warmer TX3D over China, while a lowerTN3D in the east and greeter TN3D in the west are found compared to CN05. A greater valueof annual mean precipitation compared to EA05and APHRO, especially for the later, is foundin CN05.1. For precipitation extreme of R3D (mean of the3largest precipitation in a year),CN05.1presents lower value of it in western China compared to EA05.Under the frame of Coordinated Regional climate Downscaling Experiment (CORDEX),high resolution regional climate model simulations are performed, driving by different globalmodel outputs to provide climate change information over East and Southeast region for inputto impact/adaptation work and contribute to the IPCC Fifth Assessment Report (AR5). Twosets of simulation are conducted by RegCM4.0, driving by two different global model output(ECHAM5_MPI-OM from Max-Planck Institute for Meteorology and HadCM3from theHadley Centre) respectively, under IPCC SRES A1B scenarios, form1978to2099(hereaftercalled EdR and HdR). The horizontal resolutions of these simulations are50km. For brevity,the four sets of results are called four simulations.Simulations of present climate from1980to1999over China by two global models andRegCM4.0are evaluated against observations. Results show that all of simulations reproducethe general observed spatial patterns of surface air temperature and precipitation. Compared totwo global model simulations, regional model simulations provide more spatial details ofsurface variables. EdR shows improvement in December-January-February (DJF),June-July-August (JJA) and annual mean air temperature simulations, but it’s not obvious inHdR. The precipitation simulations don’t show any evident improvements in both two RegCM4.0simulations. The EdR simulation is slightly better than ECHAM5, however HdRperforms worse than HadCM3. Different patterns can be found in the interannual variabilityand multi-year-trends of annual mean temperature and precipitation simulations. The abilitiesare not as good as the mean climate. Regional model simulations underestimate the subtropicalhigh and southerwesterly flow, which lead to an underestimation of JJA rainfall over China.The climate change (future-present) projections by the four simulations are analyzed.Significant warming in DJF, JJA and annual mean are simulated by four simulations in themiddle and end of the21st century. RegCM4.0and its driving GCM simulate a consistentchange pattern over China characterized by a greater increasing by HadCM3. The warming areremarkably enhanced in the end of21st century, regional mean increasing of annual meantemperature under A1B scenario simulated by ECHAM5and EdR are4.5°C and4.0°C,HadCM3and HdR increase by4.7°C and3.7°C, respectively.Precipitation change mainly increase in DJF, JJA and annual mean over China by twodriving globle models. The pattern in the end of21st century is similar with the middle of21st’s,but for the JJA precipitation by ECHAM5simulation. The two regional model simulationsshow similar distribution in middle and end of21st century. The regional model and its drivingmodel simulation show a consistent DJF precipitation pattern over China, but a quite differentJJA precipitation, while the change patterns differ across the two regional model simulations. Inthe end of21st century, annual mean precipitation increasing under A1B scenario simulated byECHAM5and EdR by3.6%and2.5%, HadCM3and HdR are12.1%and10.5%. A largerdeviation can be found in the JJA precipitation simulation over China by the four simulations.The weak performance can be found in moisture flux and its convergence change simulation,different topographic representations and topography forcing can be dominant factors.The model performances in present climate simulation over Southeast Asia are analyzed.Results show that the distributions of DJF, JJA and annual mean air temperature andprecipitation are reproduced by all simulations. Overall, the two regional model simulationswith higher horizontal resolution provide more spatial details compared to the driving GCMs.Better performance can be found in DJF of each simulation. Both EdR and HdR improve thesimulation of air temperature pattern compared to the GCMs. But the significant improvementcouldn’t be found in the precipitation simulations, especially in HdR, which even exhibitsweaker. The abilities of the interannual variability and multi-year-trends of annual meantemperature and precipitation simulations over Southeast Asia are not as good as the meanclimate, while a big discrepancy can be found. Similar results are found in present JJAcirculation simulation comapared to NCEP. The two regional model simulations tend to exhibita poor JJA precipitation. Causes of these differences are explained in terms of the different lowlatitude and ocean circulation and weaker westerly flow over norther equator area in the two simulations.The climate change projections over Southeast Asia by the four simulations are estimated.A predominat increasings in DJF, JJA and annual mean are simulated in the middle of the21stcentury with a greater warming by HadCM3and HdR. The warming is remarkably enhanced inthe end of21st century, especially in JJA simulation of EdR. The seasonal difference oftemperature change is not evident. Regional mean increasing of annual mean temperaturesimulated by all simulations, ECHAM5and EdR are3.3°C and3.5°C, the HadCM3and HdRincrease by3.2°C and2.9°C, respectively. Overall, the increasing values are smaller than theChina region’s.Precipitation change simulations in DJF, JJA and annual mean over Southeast Asia arequite different by the four simulations. The pattern of the end of21st century is consistent withthe middle of21st’s. But the discrepancy is big between two regional model simulations. Slightincreasing annual mean precipitation in the end of21st century are simulated by two GCMs,with the values of3.8%and3.5%, respectively, while decreasing pattern been found in bothtwo regional model simulations, with the values of-6.0%and-4.2%.更多还原