多模式集成的概率天气预报和气候预测研究进展

基于大气的混沌特性,单一的确定性预报逐步向多值的不确定性概率预报转化已成为一种趋势。本文系统地评述了概率天气预报产生的背景,介绍了概率预报的相关概念及国内外的研究状况,着重讨论了多模式集成的概率预报的两种集成方法,即贝叶斯模式平均(Bayesian model averaging,BMA)和多元高斯集合核拟合法(Gaussian ensemble kernel dressing,GEKD),并给出了两个例子的概率预报试验结果。利用BMA方法制作的概率预报的方差较小,减小了预报的不确定性,因此预报结果更接近大气的真实值。作为另一种多模式集成方法,多元高斯集合核拟合法回报的地面气温距平均值及趋势的概率预测结果与实测结果基本一致。利用此方法建立了地面气温年代际变化的概率多模式集合预测模型,并从中提取年代际气候变化特征,对东亚季风区年代际预测具有重要应用价值。     

地面气象要素多模式集成预报研究进展

目前,集合预报已成为天气预报业务的主要支撑。然而,由于数值模式本身的限制与不完善以及集合系统存在初值扰动、集合大小等方面的局限,常存在预报偏差。不同预报模式通常具有不同的物理过程参数化方案、初始条件等,导致其预报能力各有不同。为此,如何纠正预报偏差以及如何充分有效地利用不同模式的预报信息以获得更加准确的天气预报广受关注。近年来,利用统计理论与预报诊断,基于多个集合预报系统的多模式集成预报技术得到快速发展,已成为有效消除预报偏差从而提高天气预报技巧的一种统计后处理方法。针对气温、降水和风3个最基本的地面气象要素,首先依据预报形式将应用范围较广的简单集合平均、消除偏差集合平均、超级集合、贝叶斯模式平均、集合模式输出统计等加权或等权平均多模式集成技术,分成确定性预报和概率预报两大类,并做系统介绍。最后,讨论使用和发展多模式集成技术需要关注的问题,包括考虑参与集成的模式个数、发展降水及风速分级预报模型和发展基于机器学习的多模式集成新技术。     

基于TIGGE多模式降水量预报的统计降尺度研究

利用TIGGE资料中欧洲中期天气预报中心、美国国家环境预报中心、英国气象局以及日本气象厅4个中心,1~7 d预报时效的降水量预报资料,以TRMM/3B42RT降水量作为"观测值",对东亚地区降水量进行统计降尺度处理。首先利用逻辑回归方法将天气分为有雨和无雨,再对有雨的情况,利用线性回归方法对插值后的预报结果进行降尺度订正,最后将4个中心的预报值进行消除偏差集合平均,得到多模式集成的降水量预报场。结果表明:逻辑回归能够有效地改善预报中小雨的空报情况,统计降尺度订正后的预报结果比直接插值更加准确,多模式集成的预报效果优于单模式结果,其改进效果随预报时效的延长逐渐减小。     

A multimodel comparison of centennial Atlantic meridional overturning circulation variability

A mechanism contributing to centennial variability of the Atlantic Meridional Overturning Circulation (AMOC) is tested with multi-millennial control simulations of several coupled general circulation models (CGCMs). These are a substantially extended integration of the 3rd Hadley Centre Coupled Climate Model (HadCM3), the Kiel Climate Model (KCM), and the Max Plank Institute Earth System Model (MPI-ESM). Significant AMOC variability on time scales of around 100?years is simulated in these models. The centennial mechanism links changes in the strength of the AMOC with oceanic salinities and surface temperatures, and atmospheric phenomena such as the Intertropical Convergence Zone (ITCZ). 2 of the 3 models reproduce all aspects of the mechanism, with the third (MPI-ESM) reproducing most of them. A comparison with a high resolution paleo-proxy for Sea Surface Temperatures (SSTs) north of Iceland over the last 4,000?years, also linked to the ITCZ, suggests that elements of this mechanism may also be detectable in the real world.     

The zonal structure of the Hadley circulation

A discussion of the mass transport of the Hadley circulation is presented, with regard to its longitudinal structure. Data from the NCEP/NCAR reanalysis data set for the period 1948–2005 is examined, focusing on the solsticial seasons of June–August and December–February. Quantitative estimates have been extracted from the data to observe connections between the zonal mean of the upper tropospheric north/south mass transports and their relationship to the driving factor of tropical precipitation (implying latent heat release) and subsidence in the subtropical high pressure belts. The longitudinal structure of this flow is then examined with regard to these three main variables. The poleward upper tropospheric transport has four (JJA) or three (DJF) main branches, which link regions of major precipitation with corresponding regions of large subsidence, and one (June, July, August) or two (December, January, February) reverse branches. This structure has remained stable over the past sixty years. Although the total upper tropospheric transport in each season is less than the total sinking transport in the target subtropical high pressure belt, this does not apply to the individual branches, the balance being made up by the upper tropospheric reverse transports. An analysis of correlations between all of these various components shows, however, that the complete picture is more complex, with some precipitation regions being linked to subsidence regions outside their own branch.     

An analysis of the difference between the multiple linear regression approach and the multimodel ensemble mean

An investigation of the difference in seasonal precipitation forecast skills between the multiple linear regression (MLR) ensemble and the simple multimodel ensemble mean (EM) was based on the forecast quality of individual models. The possible causes of difference in previous studies were analyzed. In order to make the simulation capability of studied regions relatively uniform, three regions with different temporal correlation coefficients were chosen for this study. Results show the causes resulting in the incapability of the MLR approach vary among different regions. In the Nino3.4 region, strong co-linearity within individual models is generally the main reason. However, in the high latitude region, no significant co-linearity can be found in individual models, but the abilities of single models are so poor that it makes the MLR approach inappropriate for superensemble forecasts in this region. In addition, it is important to note that the use of various score measurements could result in some discrepancies when we compare the results derived from different multimodel ensemble approaches.     

赤道太平洋高低频纬向环流差异及其物理机制

利用1982—2017年欧洲中期天气预报中心再分析资料和美国国家大气海洋管理局的观测资料,分析了赤道太平洋高、低频海平面气压距平场的特征及差异,并通过诊断赤道太平洋海-气耦合的时间尺度,探讨了导致赤道太平洋高、低频纬向环流差异的原因。结果表明,南方涛动海平面气压东西跷跷板耦合现象只是在低频场中才存在,在高频场中并不存在。低频场上,主要受到热带最明显的年际信号(ENSO)的调控,海平面气压场和海表温度场呈现出东西振荡型。相比而言,在高频场上由于时间短,海、气异常还没有发生较好的耦合,气压场和风场呈现出全海域一致型。高频纬向环流与热带季节内振荡(MJO)紧密联系,具有明显的东传特征,传播速度大约为5 m/s,其变率方差的34%可以由MJO线性解释。相干谱分析表明海-气耦合具有时间依赖性,南方涛动通常只在20候以上的时间尺度才能存在,这与赤道海洋开尔文波横穿太平洋时间相当。     

GRAPES-GEPS环流集合预报的分类释用方法研究与检验

集合预报在数值天气预报体系中具有重要地位,因此如何有效提取集合样本信息以提高集合预报技巧一直是一个重要课题.基于中国全球集合预报业务系统(GRAPES-GEPS)的500?hPa高度场集合资料开展对环流集合预报的分类释用方法研究,并对集合聚类预报结果进行了检验分析.通过在传统Ward聚类法中引入动态聚类的"手肘法"方案...     

Indian Ocean variability in the CMIP5 multi-model ensemble: the zonal dipole mode

The performance of 21 Coupled Model Intercomparison Project Phase 5 (CMIP5) models in the simulation of the Indian Ocean Dipole (IOD) mode is evaluated. Compared to CMIP3, CMIP5 models exhibit a similar spread in IOD intensity. A detailed diagnosis was carried out to understand whether CMIP5 models have shown improvement in their representation of the important dynamical and thermodynamical feedbacks in the tropical Indian Ocean. These include the Bjerknes dynamic air-sea feedback, which includes the equatorial zonal wind response to sea surface temperature (SST) anomaly, the thermocline response to equatorial zonal wind forcing, the ocean subsurface temperature response to the thermocline variations, and the thermodynamic air-sea coupling that includes the wind-evaporation-SST and cloud-radiation-SST feedback. Compared to CMIP3, the CMIP5 ensemble produces a more realistic positive wind-evaporation-SST feedback during the IOD developing phase, while the simulation of Bjerknes dynamic feedback is more unrealistic especially with regard to the wind response to SST forcing and the thermocline response to surface wind forcing. The overall CMIP5 performance in the IOD simulation does not show remarkable improvements compared to CMIP3. It is further noted that the El Niño-Southern Oscillation (ENSO) and IOD amplitudes are closely related, if a model generates a strong ENSO, it is likely that this model also simulates a strong IOD.     

Inverse relationship between the equatorial eastern Pacific annual-cycle and ENSO amplitudes in a coupled general circulation model

We propose a dynamical interpretation of the inverse relationship between the tropical eastern Pacific annual-cycle (AC) amplitude and the El Niño-Southern Oscillation (ENSO) amplitude, based on a pre-industrial simulation of Geophysical Fluid Dynamics Laboratory Couple climate model 2.0 with a fixed concentration of greenhouse gases spanning approximately 500 years. The slowly varying background conditions over more than a decade alternately provided favorable conditions for two opposite regimes, namely the ‘strong AC—weak ENSO regime’ and the ‘weak AC—strong ENSO regime’. For the weak AC—strong ENSO regime, the tropical eastern Pacific shows meridional-asymmetric surface warming with an emphasis on the southern part, leading to weakening of both the zonal trade wind and the cross equatorial southerly wind, as well as deepening of both the thermocline and mixed layer. The deeper mixed layer, weaker southerly wind, and reduced zonal gradient of the mean sea surface temperature due to tropical eastern Pacific warming all acts to reduce the AC. Conversely, the ENSO was intensified by the deeper mixed layer and deeper thermocline depth (thermocline feedback), but suppressed by the deeper thermocline depth (Ekman feedback) and the reduced zonal temperature gradient. We also computed the coupling strengths of the ENSO and AC, defined as the linear regression coefficients of the zonal and meridional wind stresses against the eastern Pacific SST, respectively. The coupling strengths of both the AC and ENSO are larger when they are intensified, and vice versa. All processes for the weak AC—strong ENSO regime operate in the opposite manner for the strong AC—weak ENSO regime.     

Simulating tropical instability waves in the equatorial eastern Pacific with a coupled general circulation model

Satellite observations of SSTs have revealed the existence of unstable waves in the equatorial eastern Pacific and Atlantic oceans. These waves have a 20-40-day periodicity with westward phase speeds of 0.4-0.6 m s-1 and wavelengths of 1000-2000 km during boreal summer and fall. They are generally called tropical instability waves (TIWs). This study investigates TIWs simulated by a high-resolution coupled atmosphere-ocean general circulation model (AOGCM). The horizontal resolution of the model is 120 km in...     

On the origin of equatorial Atlantic biases in coupled general circulation models

Many coupled ocean–atmosphere general circulation models (GCMs) suffer serious biases in the tropical Atlantic including a southward shift of the intertropical convergence zone (ITCZ) in the annual mean, a westerly bias in equatorial surface winds, and a failure to reproduce the eastern equatorial cold tongue in boreal summer. The present study examines an ensemble of coupled GCMs and their uncoupled atmospheric component to identify common sources of error. It is found that the westerly wind bias also exists in the atmospheric GCMs forced with observed sea surface temperature, but only in boreal spring. During this time sea-level pressure is anomalously high (low) in the western (eastern) equatorial Atlantic, which appears to be related to deficient (excessive) precipitation over tropical South America (Africa). In coupled simulations, this westerly bias leads to a deepening of the thermocline in the east, which prevents the equatorial cold tongue from developing in boreal summer. Thus reducing atmospheric model errors during boreal spring may lead to improved coupled simulations of tropical Atlantic climate.     

Application of multimodel ensemble techniques for real time district level rainfall forecasts in short range time scale over Indian region

During the summer monsoon (1 June to 30 September) 2007, real-time district level rainfall forecasts in short-range time scale were generated for Indian region applying multimodel ensemble technique. The pre-assigned grid point weights on the basis of correlation coefficients (CC) between the observed values and forecast values are determined for each constituent model at the resolution of 0.5° × 0.5° utilizing two seasons datasets (1 June to 30 September, 2005 and 2006), and the multimodel ensemble forecasts (day 1 and day 2 forecasts) are generated at the same resolution on a real-time basis. The ensemble forecast fields are then used to prepare forecasts for each district taking the average value of all grid points falling in a particular district. In this paper we examined the performance skill of the multimodel ensemble-based real-time district level short-range forecast of rainfall. It has clearly emerged from the results that the multimodel ensemble technique reported in this study is superior to each ensemble member. District wise performance of the ensemble rainfall forecast reveals that the technique, in general, is capable of providing reasonably good forecast skill over most districts of the country, particularly over the districts where the monsoon systems are dominant. Though the procedure shows appreciable skill to predict occurrence or non-occurrence of rainfall at the district level, it always underestimates rainfall amount, particularly in heavy rainfall events. Possible reasons of this failure may be due to model bias and poor data assimilation procedure.     

Relative roles of the equatorial upper ocean zonal current and thermocline in determining the timescale of the tropical climate system

Summary In this study, it is demonstrated that the amplitude of the equatorial upper-ocean zonal current anomaly induced by the fast-varying wind forcing (shorter than a year) is much greater than that induced by the slowly varying wind forcing (longer than 2 year), and the center of maximum zonal current anomaly shifts from the central Pacific to the western Pacific with an increase in the timescale of wind forcing. As a result, the zonal advective feedback (the zonal advection of mean sea surface temperature by anomalous current) in a slowly varying climate system becomes weaker and barely induces a low-frequency mode such as El Niño-Southern Oscillation. On the other hand, both amplitude and zonal location of the maximum thermocline anomaly are little changed by the change in the timescale of wind forcing – confined at the strong equatorial upwelling region of the eastern Pacific. Accordingly, the thermocline feedback (the vertical advection of anomalous subsurface temperature by the mean upwelling) is more favorable to generate a low-frequency mode.The relative roles of these two feedbacks are further explored under the coupled-system context. The eigen analysis of the stripped-down version of an intermediate ocean-atmosphere coupled model shows that by altering the regime space from the weakly coupled to the strongly coupled, the dominant process that leads the leading eigen mode changes from the zonal advective feedback to the thermocline feedback, and at the same time the frequency of the leading mode also changes from the high-frequency to the low-frequency. It implies that each feedback tends to favor the different timescale coupled mode.     

Probabilistic multimodel ensemble prediction of decadal variability of East Asian surface air temperature based on IPCC-AR5 near-term climate simulations

Based on near-term climate simulations for IPCC-AR5 (The Fifth Assessment Report), probabilistic multimodel ensemble prediction (PMME) of decadal variability of surface air temperature in East Asia (20°-50 °N, 100°-145°E) was conducted using the multivariate Gaussian ensemble kernel dressing (GED) methodology. The ensemble system exhibited high performance in hindcasting the decadal (1981-2010) mean and trend of temperature anomalies with respect to 1961-90, with a RPS of 0.94 and 0.88 respectively. The interpretation of PMME for future decades (2006-35) over East Asia was made on the basis of the bivariate probability density of the mean and trend. The results showed that, under the RCP4.5 (Representative Concentration Pathway 4.5 W m-2 ) scenario, the annual mean temperature increases on average by about 1.1-1.2 K and the temperature trend reaches 0.6-0.7 K (30 yr)-1 . The pattern for both quantities was found to be that the temperature increase will be less intense in the south. While the temperature increase in terms of the 30-yr mean was found to be virtually certain, the results for the 30-yr trend showed an almost 25% chance of a negative value. This indicated that, using a multimodel ensemble system, even if a longer-term warming exists for 2006-35 over East Asia, the trend for temperature may produce a negative value. Temperature was found to be more affected by seasonal variability, with the increase in temperature over East Asia more intense in autumn (mainly), faster in summer to the west of 115°E, and faster still in autumn to the east of 115°E.