Can ensembles of regional climate model simulations improve results from sensitivity studies?

Intrinsic variability (IV) in regional climate models (RCMs) is often assumed to be small because at climatological timescales, the model solutions tend to be dominated by the model??s lateral boundary conditions. Recent studies have indicated that this IV may actually be large in certain instances for some variables. Direct interpretation of anomalies from RCM sensitivity studies relies on the assumption that differences between model simulations are entirely due to a physical forcing. However, if IV is as large or larger than the physical signal, then this assumption is violated. Using a 20 member ensemble of RCM simulations, we verify that IV of precipitation within a RCM can be large enough to violate the sensitivity study assumption, and we show that generating ensembles of simulations can help reduce the level of IV. We also present two indicators that can rule out the influence of IV when it is ambiguous whether anomalies within a sensitivity study are due to the sensitivity perturbation or whether they are due to IV.     

Diagnostic budget study of the internal variability in ensemble simulations of the Canadian RCM

Due to the chaotic and nonlinear nature of the atmospheric dynamics, it is known that small differences in the initial conditions (IC) of models can grow and affect the simulation evolution. In this study, we perform a quantitative diagnostic budget calculation of the various diabatic and dynamical contributions to the time evolution and spatial distribution of internal variability (IV) in simulations with the nested Canadian Regional Climate Model. We establish prognostic budget equations of the IV for the potential temperature and the relative vorticity fields. For both of these variables, the IV equations present similar terms, notably terms relating to the transport of IV by ensemble-mean flow and to the covariance of fluctuations acting on the gradient of the ensemble-mean state. We show the skill of these equations to diagnose the IV that took place in an ensemble of 20 3-month (summer season) simulations that differed only in their IC. Our study suggests that the dominant terms responsible for the large increase of IV are either the covariance term involving the potential temperature fluctuations and diabatic heating fluctuations, or the covariance of inter-member fluctuations acting upon ensemble-mean gradients. Our results also show that, on average, the third-order terms are negligible, but they can become important when the IV is large.     

Impacts of Spectral Nudging on the Sensitivity of a Regional Climate Model to Convective Parameterizations in East Asia

The sensitivity of a regional climate model (RCM) to cumulus parameterization (CUPA) schemes in modeling summer precipitation over East Asia has been investigated by using the fifth-generation Pennsylvania State University-National Center for Atmospheric Research Mesoscale Model (PSU-NCAR MM5). The feasibility of physical ensemble and the effect of interior (spectral) nudging are also assessed. The RCM simulations are evaluated against the NCEP/NCAR reanalysis data and NCEP/CPC precipitation data for three summers (JJA) in 1991, 1998, and 2003. The results show that the RCM is highly sensitive to CUPA schemes. Different CUPA schemes cause distinctive characteristics in the modeling of JJA precipitation and the intraseasonal (daily) variability of regional precipitation. The sensitivity of the RCM simulations to the CUPA schemes is reduced by adopting the spectral nudging technique, which enables the RCM to reproduce more realistic large-scale circulations at the upper levels of the atmosphere as well as near the surface, and better precipitation simulation in the selected experiments. The ensemble simulations using different CUPA schemes show higher skills than individual members for both control runs and spectral nudging runs. The physical ensemble adopting the spectral nudging technique shows the highest downscaling skill in capturing the general circulation patterns for all experiments and improved temporal distributions of precipitation in some regions.     

Singular vector decomposition of the internal variability of the Canadian Regional Climate Model

Previous studies have shown that Regional Climate Models (RCM) internal variability (IV) fluctuates in time depending on synoptic events. This study focuses on the physical understanding of episodes with rapid growth of IV. An ensemble of 21 simulations, differing only in their initial conditions, was run over North America using version 5 of the Canadian RCM (CRCM). The IV is quantified in terms of energy of CRCM perturbations with respect to a reference simulation. The working hypothesis is that IV is arising through rapidly growing perturbations developed in dynamically unstable regions. If indeed IV is triggered by the growth of unstable perturbations, a large proportion of the CRCM perturbations must project onto the most unstable singular vectors (SVs). A set of ten SVs was computed to identify the orthogonal set of perturbations that provide the maximum growth with respect to the dry total-energy norm during the course of the CRCM ensemble of simulations. CRCM perturbations were then projected onto the subspace of SVs. The analysis of one episode of rapid growth of IV is presented in detail. It is shown that a large part of the IV growth is explained by initially small-amplitude unstable perturbations represented by the ten leading SVs, the SV subspace accounting for over 70% of the CRCM IV growth in 36?h. The projection on the leading SV at final time is greater than the projection on the remaining SVs and there is a high similarity between the CRCM perturbations and the leading SV after 24–36?h tangent-linear model integration. The vertical structure of perturbations revealed that the baroclinic conversion is the dominant process in IV growth for this particular episode.     

Gradient in the climate change signal of European discharge predicted by a multi-model ensemble

In order to perform hydrological studies on the PRUDENCE regional climate model (RCM) simulations, a special focus was put on the discharge from large river catchments located in northern and central Europe. The discharge was simulated with a simplified land surface (SL) scheme and the Hydrological Discharge (HD) model. The daily fields of precipitation, 2 m temperature and evapotranspiration from the RCM simulations were used as forcing. Therefore the total catchment water balances are constrained by the hydrological cycle of the different RCMs. The validation of the simulated hydrological cycle from the control simulations shows that the multi-model ensemble mean is closer to the observations than each of the models, especially if different catchments and hydrological variables are considered. Therefore, the multi-model ensemble mean can be used to largely reduce the uncertainty that is introduced by a single RCM. This also provides more confidence in the future projections for the multi-model ensemble means. The scenario simulations predict a gradient in the climate change signal over Northern and Central Europe. Common features are the overall warming and the general increase of evapotranspiration. But while in the northern parts the warming will enhance the hydrological cycle leading to an increased discharge, the large warming, especially in the summer, will slow down the hydrological cycle caused by a drying in the central parts of Europe which is accompanied by a reduction of discharge. The comparison of the changes predicted by the multi-model ensemble mean to the changes predicted by the driving GCM indicates that the RCMs can compensate problems that a driving GCM may have with local scale processes or parameterizations.     

Atmospheric and Terrestrial Water Balances of Labrador's Churchill River Basin,as Simulated by the North American Regional Climate Change Assessment Program

In an effort to understand the sources of uncertainty and the physical consistency of climate models from the North American Regional Climate Change Assessment Program (NARCCAP), an ensemble of general circulation models (GCMs) and regional climate models (RCMs) was used to explore climatological water balances for the Churchill River basin in Labrador, Canada. This study quantifies mean atmospheric and terrestrial water balance residuals, as well as their annual cycles. Mean annual atmospheric water balances had consistently higher residuals than the terrestrial water balances due, in part, to the influences of sampling of instantaneous variables and the interpolation of atmospheric data to published pressure levels. Atmospheric and terrestrial water balance residuals for each ensemble member were found to be consistent between base and future periods, implying that they are systemic and not climate dependent. With regard to the annual cycle, no pattern was found across time periods or ensemble members to indicate whether the monthly terrestrial or atmospheric root mean square residual was highest. Because of the interdependence of hydrological cycle components, the complexity of climate models and the variety of methods and processes used by different ensemble members, it was impossible to isolate all causes of the water balance residuals. That being said, the residuals created by interpolating a model's native vertical resolution onto NARCCAP's published pressure levels and the subsequent vertical interpolation were quantified and several other sources were explored. In general, residuals were found to be predominantly functions of the RCM choice (as opposed to the GCM choice) and their respective modelling processes, parameterization schemes, and post-processing.     

Ability of an ensemble of regional climate models to reproduce weather regimes over Europe-Atlantic during the period 1961–2000

One of the main concerns in regional climate modeling is to which extent limited-area regional climate models (RCM) reproduce the large-scale atmospheric conditions of their driving general circulation model (GCM). In this work we investigate the ability of a multi-model ensemble of regional climate simulations to reproduce the large-scale weather regimes of the driving conditions. The ensemble consists of a set of 13 RCMs on a European domain, driven at their lateral boundaries by the ERA40 reanalysis for the time period 1961–2000. Two sets of experiments have been completed with horizontal resolutions of 50 and 25 km, respectively. The spectral nudging technique has been applied to one of the models within the ensemble. The RCMs reproduce the weather regimes behavior in terms of composite pattern, mean frequency of occurrence and persistence reasonably well. The models also simulate well the long-term trends and the inter-annual variability of the frequency of occurrence. However, there is a non-negligible spread among the models which is stronger in summer than in winter. This spread is due to two reasons: (1) we are dealing with different models and (2) each RCM produces an internal variability. As far as the day-to-day weather regime history is concerned, the ensemble shows large discrepancies. At daily time scale, the model spread has also a seasonal dependence, being stronger in summer than in winter. Results also show that the spectral nudging technique improves the model performance in reproducing the large-scale of the driving field. In addition, the impact of increasing the number of grid points has been addressed by comparing the 25 and 50 km experiments. We show that the horizontal resolution does not affect significantly the model performance for large-scale circulation.     

Ensemble Forecasts of Tropical Cyclone Track with Orthogonal Conditional Nonlinear Optimal Perturbations

This paper preliminarily investigates the application of the orthogonal conditional nonlinear optimal perturbations(CNOPs)–based ensemble forecast technique in MM5(Fifth-generation Pennsylvania State University–National Center for Atmospheric Research Mesoscale Model). The results show that the ensemble forecast members generated by the orthogonal CNOPs present large spreads but tend to be located on the two sides of real tropical cyclone(TC) tracks and have good agreements between ensemble spreads and ensemble-mean forecast errors for TC tracks. Subsequently, these members reflect more reasonable forecast uncertainties and enhance the orthogonal CNOPs–based ensemble-mean forecasts to obtain higher skill for TC tracks than the orthogonal SVs(singular vectors)–, BVs(bred vectors)– and RPs(random perturbations)–based ones. The results indicate that orthogonal CNOPs of smaller magnitudes should be adopted to construct the initial ensemble perturbations for short lead–time forecasts, but those of larger magnitudes should be used for longer lead–time forecasts due to the effects of nonlinearities. The performance of the orthogonal CNOPs–based ensemble-mean forecasts is case-dependent,which encourages evaluating statistically the forecast skill with more TC cases. Finally, the results show that the ensemble forecasts with only initial perturbations in this work do not increase the forecast skill of TC intensity, which may be related with both the coarse model horizontal resolution and the model error.     

Investigation of regional climate models’ internal variability with a ten-member ensemble of 10-year simulations over a large domain

Previous investigations on regional climate models’ (RCM) internal variability (IV) were limited owing to small ensembles, short simulations and small domains. The present work extends previous studies with a ten-member ensemble of 10-year simulations performed with the Canadian Regional Climate Model over a large domain covering North America. The results show that the IV has no long-term tendency but rather fluctuates in time following the synoptic situation within the domain. The IV of mean-sea-level pressure (MSLP) and screen temperature (ST) show a small annual cycle with larger values in spring, which differs from previous studies. For precipitation (PCP), the IV shows a clear annual cycle with larger values in summer, as previously reported. The 10-year climatology of the IV for MSLP and ST shows a well-defined spatial distribution with larger values in the northeast of the domain, near the outflow boundary. A comparison of the IV of MSLP and ST in summer with the transient-eddy variance reveals that the IV is close to its maximum in a small region near the outflow boundary. Same analysis for PCP in summer shows that the IV reaches its maximum in most parts of the domain, except for a small region on the western side near the inflow boundary. Finally, a comparison of the 10-year climate of each simulation of the ensemble showed that the IV may have a significant impact on the climatology of some variables.     

Evaluation of the internal variability and estimation of the downscaling ability of the Canadian Regional Climate Model for different domain sizes over the north Atlantic region using the Big-Brother experimental approach

The ability of a nested model to accurately simulate the subarctic climate is studied here. Two issues have been investigated: Model??s internal variability (IV) and the impact of domain size (DS). For this purpose we combine the ??perfect model?? approach, Big-Brother Experiment (BBE) (Denis et al. in Clim Dyn 18:627?C646, 2002) with the ensemble of simulations. The advantage of this framework is the possibility to study small-scale climate features that constitute the main added value of RCM. The effects of the DS on result were studied by employing two Little-Brother (LB) domain sizes. IV has been evaluated by introducing small differences in initial conditions in an ensemble of 20 simulations over each LB. Results confirm previous findings that the IV is more important over the larger domain of integration. The temporal evolution over two domain sizes is rather different and depends strongly on the synoptic situation. Small-scales solution over the larger domain diverges freely from the boundary forcing in some periods. Over the smaller domain, the amplitude of small-scale transient eddies is systematically underestimated, especially at higher altitude characterized by the strongest winds along the storm tracks. Over the larger domain, the amplitude of small-scale transient eddies is better represented. However, the weaker control by the lateral boundaries over the larger domain results in solutions with large internal variability. As a result, the ensemble average strongly underestimates the transient-eddy variance due to partial destructive interference of individual ensemble member solutions.     

Ensemble mean forecast skill and applications with the T213 ensemble prediction system

Ensemble forecasting has become the prevailing method in current operational weather forecasting. Although ensemble mean forecast skill has been studied for many ensemble prediction systems(EPSs) and different cases, theoretical analysis regarding ensemble mean forecast skill has rarely been investigated, especially quantitative analysis without any assumptions of ensemble members. This paper investigates fundamental questions about the ensemble mean, such as the advantage of the ensemble mean over individual members, the potential skill of the ensemble mean, and the skill gain of the ensemble mean with increasing ensemble size. The average error coefficient between each pair of ensemble members is the most important factor in ensemble mean forecast skill, which determines the mean-square error of ensemble mean forecasts and the skill gain with increasing ensemble size. More members are useful if the errors of the members have lower correlations with each other, and vice versa. The theoretical investigation in this study is verified by application with the T213 EPS. A typical EPS has an average error coefficient of between 0.5 and 0.8; the 15-member T213 EPS used here reaches a saturation degree of 95%(i.e., maximum 5% skill gain by adding new members with similar skill to the existing members) for 1–10-day lead time predictions, as far as the mean-square error is concerned.     

分析约束的集合预报初始扰动构造方案的研究

集合预报初始扰动准确描述大气运动的不确定性是集合预报研究的核心问题,合理的扰动结构及振幅应能更好地反映大气运动状态的预报误差特征。随着集合扰动和资料同化的深入研究和理解,集合初始扰动方案与集合同化紧密结合协同发展。本文基于中国气象局数值预报中心自主研发的GRAPES-REPS集合预报系统,针对其初始扰动的结构和振幅与预报误差一致性较差的不合理问题,结合不同空间尺度天气系统预报误差特征,将表征预报不确定性的集合扰动与表达观测和预报不确定性的资料同化分析增量有效结合,研究提出了一种改善集合初始扰动质量的分析约束方案,以实现对集合初始扰动质量进一步改善。分析约束方案充分考虑资料同化分析增量的空间结构和量值特征,分别构造了单一定常和具有一定适应能力的两种分析约束函数,实现对初始扰动中不合理信息的识别和约束调整。试验结果表明,具有适应能力的分析约束方案对集合初始扰动具有良好的调整能力,约束后集合扰动的结构和振幅与中小尺度天气系统的预报误差更为吻合,其集合离散度和扰动能量的空间结构与演变特征更加趋于合理。分析约束方案可有效改善集合初始扰动质量及其预报性能。     

Evaluating uncertainties in regional climate simulations over South America at the seasonal scale

This work focuses on the evaluation of different sources of uncertainty affecting regional climate simulations over South America at the seasonal scale, using the MM5 model. The simulations cover a 3-month period for the austral spring season. Several four-member ensembles were performed in order to quantify the uncertainty due to: the internal variability; the definition of the regional model domain; the choice of physical parameterizations and the selection of physical parameters within a particular cumulus scheme. The uncertainty was measured by means of the spread among individual members of each ensemble during the integration period. Results show that the internal variability, triggered by differences in the initial conditions, represents the lowest level of uncertainty for every variable analyzed. The geographic distribution of the spread among ensemble members depends on the variable: for precipitation and temperature the largest spread is found over tropical South America while for the mean sea level pressure the largest spread is located over the southeastern Atlantic Ocean, where large synoptic-scale activity occurs. Using nudging techniques to ingest the boundary conditions reduces dramatically the internal variability. The uncertainty due to the domain choice displays a similar spatial pattern compared with the internal variability, except for the mean sea level pressure field, though its magnitude is larger all over the model domain for every variable. The largest spread among ensemble members is found for the ensemble in which different combinations of physical parameterizations are selected. The perturbed physics ensemble produces a level of uncertainty slightly larger than the internal variability. This study suggests that no matter what the source of uncertainty is, the geographical distribution of the spread among members of the ensembles is invariant, particularly for precipitation and temperature.     

Combining climatological and participatory approaches for assessing changes in extreme climatic indices at regional scale

This paper combines the climatological and societal perspectives for assessing future climatic extremes over Kangasabati River basin in India using an ensemble of four high resolution (25 km) regional climate model (RCM) simulations from 1970 to 2050. The relevant extreme indices and their thresholds are defined in consultation with stakeholders and are then compared using RCM simulations. To evaluate the performance of RCM in realistically representing atmospheric processes in the basin, model simulations driven with ERAInterim global re-analysis data from 1989 to 2008 are compared with observations. The models perform well in simulating seasonality, interannual variability and climatic extremes. Future climatic extremes are evaluated based on RCM simulations driven by GCMs, for present (1970–1999) and for the SRES A1B scenario for future (2021–2050) period. The analysis shows an intensification of majority of extremes as projected by future ensemble mean. The study suggests that there is a marked consistency in stakeholder observed changes in climate extremes and future predicted trends.     

Regional climate modelling of the 2006 West African monsoon: sensitivity to convection and planetary boundary layer parameterisation using WRF

Regional climate model (RCM) is a valuable scientific tool to address the mechanisms of regional atmospheric systems such as the West African monsoon (WAM). This study aims to improve our understanding of the impact of some physical schemes of RCM on the WAM representation. The weather research and forecasting model has been used by performing six simulations of the 2006 summer WAM season. These simulations use all combinations of three convective parameterization schemes (CPSs) and two planetary boundary layer schemes (PBLSs). By comparing the simulations to a large set of observations and analysis products, we have evaluated the ability of these RCM parameterizations to reproduce different aspects of the regional atmospheric circulation of the WAM. This study focuses in particular on the WAM onset and the rainfall variability simulated over this domain. According to the different parameterizations tested, the PBLSs seem to have the strongest effect on temperature, humidity vertical distribution and rainfall amount. On the other hand, dynamics and precipitation variability are strongly influenced by CPSs. In particular, the Mellor?CYamada?CJanjic PBLS attributes more realistic values of humidity and temperature. Combined with the Kain?CFritsch CPS, the WAM onset is well represented. The different schemes combination tested also reveal the role of different regional climate features on WAM dynamics, namely the low level circulation, the land?Catmosphere interactions and the meridional temperature gradient between the Guinean coast and the Sahel.     

High resolution regional climate model simulations for Germany: Part II—projected climate changes

The projected climate change signals of a five-member high resolution ensemble, based on two global climate models (GCMs: ECHAM5 and CCCma3) and two regional climate models (RCMs: CLM and WRF) are analysed in this paper (Part II of a two part paper). In Part I the performance of the models for the control period are presented. The RCMs use a two nest procedure over Europe and Germany with a final spatial resolution of 7 km to downscale the GCM simulations for the present (1971–2000) and future A1B scenario (2021–2050) time periods. The ensemble was extended by earlier simulations with the RCM REMO (driven by ECHAM5, two realisations) at a slightly coarser resolution. The climate change signals are evaluated and tested for significance for mean values and the seasonal cycles of temperature and precipitation, as well as for the intensity distribution of precipitation and the numbers of dry days and dry periods. All GCMs project a significant warming over Europe on seasonal and annual scales and the projected warming of the GCMs is retained in both nests of the RCMs, however, with added small variations. The mean warming over Germany of all ensemble members for the fine nest is in the range of 0.8 and 1.3 K with an average of 1.1 K. For mean annual precipitation the climate change signal varies in the range of ?2 to 9 % over Germany within the ensemble. Changes in the number of wet days are projected in the range of ±4 % on the annual scale for the future time period. For the probability distribution of precipitation intensity, a decrease of lower intensities and an increase of moderate and higher intensities is projected by most ensemble members. For the mean values, the results indicate that the projected temperature change signal is caused mainly by the GCM and its initial condition (realisation), with little impact from the RCM. For precipitation, in addition, the RCM affects the climate change signal significantly.     

2011年长江中下游梅雨期强降水延伸期集合预报性能初探

针对2011年长江中下游旱涝转换时期的环流形势、强降雨期间的四次强降雨过程对欧洲中心集合预报进行了预报性能初步分析。结果表明：集合平均预报对延伸期预报时效内的大尺度环流调整具有较好的预报性能,预报提前时效可达10~15 d。对强降水过程期间主要影响系统的预报在不同预报时效具有较好的稳定性。随着预报时效的临近,集合预报各个成员对天气系统预报的发散度逐渐减小。长江中下游强降水的发生与低层850 hPa较大的风速有密切关联,集合预报给出的延伸期预报时效内大风速出现的小概率预报信息是有意义的,可以为延伸期强降雨过程预报提供参考。     

Quantifying uncertainty in changes in extreme event frequency in response to doubled CO2 using a large ensemble of GCM simulations

We discuss equilibrium changes in daily extreme surface air temperature and precipitation events in response to doubled atmospheric CO₂, simulated in an ensemble of 53 versions of HadSM3, consisting of the HadAM3 atmospheric general circulation model (GCM) coupled to a mixed layer ocean. By virtue of its size and design, the ensemble, which samples uncertainty arising from the parameterisation of atmospheric physical processes and the effects of natural variability, provides a first opportunity to quantify the robustness of predictions of changes in extremes obtained from GCM simulations. Changes in extremes are quantified by calculating the frequency of exceedance of a fixed threshold in the 2 × CO₂ simulation relative to the 1 × CO₂ simulation. The ensemble-mean value of this relative frequency provides a best estimate of the expected change while the range of values across the ensemble provides a measure of the associated uncertainty. For example, when the extreme threshold is defined as the 99th percentile of the 1 × CO₂ distribution, the global-mean ensemble-mean relative frequency of extremely warm days is found to be 20 in January, and 28 in July, implying that events occurring on one day per hundred under present day conditions would typically occur on 20–30 days per hundred under 2 × CO₂ conditons. However the ensemble range in the relative frequency is of similar magnitude to the ensemble-mean value, indicating considerable uncertainty in the magnitude of the increase. The relative frequencies in response to doubled CO₂ become smaller as the threshold used to define the extreme event is reduced. For one variable (July maximum daily temperature) we investigate this simulated variation with threshold, showing that it can be quite well reproduced by assuming the response to doubling CO₂ to be characterised simply as a uniform shift of a Gaussian distribution. Nevertheless, doubling CO₂ does lead to changes in the shape of the daily distributions for both temperature and precipitation, but the effect of these changes on the relative frequency of extreme events is generally larger for precipitation. For example, around one-fifth of the globe exhibits ensemble-mean decreases in time-averaged precipitation accompanied by increases in the frequency of extremely wet days. The ensemble range of changes in precipitation extremes (relative to the ensemble mean of the changes) is typically larger than for temperature extremes, indicating greater uncertainty in the precipitation changes. In the global average, extremely wet days are predicted to become twice as common under 2 × CO₂ conditions. We also consider changes in extreme seasons, finding that simulated increases in the frequency of extremely warm or wet seasons under 2 × CO₂ are almost everywhere greater than the corresponding increase in daily extremes. The smaller increases in the frequency of daily extremes is explained by the influence of day-to-day weather variability which inflates the variance of daily distributions compared to their seasonal counterparts.     

Internal variability of RCM simulations over an annual cycle

Three one-year simulations generated with the Canadian RCM (CRCM) are compared to each other in order to study internal variability in nested regional climate models and to evaluate the influence exerted by the lateral boundaries information supplied by the nesting procedure. All simulations are generated over a large domain and cover an annual cycle. The simulations use different combinations of surface and atmospheric initial conditions but all of them share the same set of time-dependent lateral boundary conditions taken from a simulation by the Canadian GCM. A first simulation is used as control, the second simulation is launched with different atmospheric and surface initial conditions (IC) and the third simulation is launched taking its surface IC from the control simulation. Comparison of the root-mean-square differences (RMSD) between each pair of simulations shows two distinct seasonal behaviours in the time series of the RMSD. In winter all simulations are almost identical to each other resulting in very low RMSD values while in summer large discrepancies develop between pairs of simulations. For water vapour related fields such as precipitation or specific humidity, these discrepancies are sometimes as large as the monthly averaged variability. However, analysis of the climate statistics shows that, although the evolution of the various summer weather systems is different, the climates of each simulation are similar.     

基于非静力模式物理扰动的中尺度集合预报试验

以GRAPES中尺度有限区模式作为试验模式, 从模式的不确定性方面来构造中尺度的集合预报, 重点考虑物理因子与初始条件的扰动作用。针对2004年7月10日北京城区的突发性暴雨过程进行了36 h的集合预报试验。结果表明:GRAPES模式可有效地捕捉到中尺度过程的信息; 中尺度集合预报是可行的, 可改进中尺度暴雨过程落区、强度的预报; 不同集合方案的预报结果各不相同, 同一方案各个成员的预报结果也有差异, 即存在适宜的离散度; 在离散度分析中发现在北京附近存在一个明显大值区, 且在大气中低层的垂直结构表现出一致性, 表明这一区域的预报不确定性很大。从集合检验结果中得到:单纯考虑模式物理扰动来构造中尺度集合预报系统有一定难度, 当加入初始场不确定信息后, 同时考虑模式的不确定性和初始场的不确定性, 有助于捕捉更多的中尺度系统的不确定信息, 有助于构造更为有效的中尺度集合预报系统。