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.     

Evaluation of the regional climate model ALADIN to simulate the climate over North America in the CORDEX framework

In this study, an ensemble of four multi-year climate simulations is performed with the regional climate model ALADIN to evaluate its ability to simulate the climate over North America in the CORDEX framework. The simulations differ in their driving fields (ERA-40 or ERA-Interim) and the nudging technique (with or without large-scale nudging). The validation of the simulated 2-m temperature and precipitation with observationally-based gridded data sets shows that ALADIN performs similarly to other regional climate models that are commonly used over North America. Large-scale nudging improves the temporal correlation of the atmospheric circulation between ALADIN and its driving field, and also reduces the warm and dry summer biases in central North America. The differences between the simulations driven with different reanalyses are small and are likely related to the regional climate model’s induced internal variability. In general, the impact of different driving fields on ALADIN is smaller than that of large-scale nudging. The analysis of the multi-year simulations over the prairie and the east taiga indicates that the ALADIN 2-m temperature and precipitation interannual variability is similar or larger than that observed. Finally, a comparison of the simulations with observations for the summer 1993 shows that ALADIN underestimates the flood in central North America mainly due to its systematic dry bias in this region. Overall, the results indicate that ALADIN can produce a valuable contribution to CORDEX over North America.     

Simulating the impact of the large-scale circulation on the 2-m temperature and precipitation climatology

The impact of the simulated large-scale atmospheric circulation on the regional climate is examined using the Weather Research and Forecasting (WRF) model as a regional climate model. The purpose is to understand the potential need for interior grid nudging for dynamical downscaling of global climate model (GCM) output for air quality applications under a changing climate. In this study we downscale the NCEP-Department of Energy Atmospheric Model Intercomparison Project (AMIP-II) Reanalysis using three continuous 20-year WRF simulations: one simulation without interior grid nudging and two using different interior grid nudging methods. The biases in 2-m temperature and precipitation for the simulation without interior grid nudging are unreasonably large with respect to the North American Regional Reanalysis (NARR) over the eastern half of the contiguous United States (CONUS) during the summer when air quality concerns are most relevant. This study examines how these differences arise from errors in predicting the large-scale atmospheric circulation. It is demonstrated that the Bermuda high, which strongly influences the regional climate for much of the eastern half of the CONUS during the summer, is poorly simulated without interior grid nudging. In particular, two summers when the Bermuda high was west (1993) and east (2003) of its climatological position are chosen to illustrate problems in the large-scale atmospheric circulation anomalies. For both summers, WRF without interior grid nudging fails to simulate the placement of the upper-level anticyclonic (1993) and cyclonic (2003) circulation anomalies. The displacement of the large-scale circulation impacts the lower atmosphere moisture transport and precipitable water, affecting the convective environment and precipitation. Using interior grid nudging improves the large-scale circulation aloft and moisture transport/precipitable water anomalies, thereby improving the simulated 2-m temperature and precipitation. The results demonstrate that constraining the RCM to the large-scale features in the driving fields improves the overall accuracy of the simulated regional climate, and suggest that in the absence of such a constraint, the RCM will likely misrepresent important large-scale shifts in the atmospheric circulation under a future climate.     

Dynamical downscaling of ERA-40 in complex terrain using the WRF regional climate model

Results from a first-time employment of the WRF regional climate model to climatological simulations in Europe are presented. The ERA-40 reanalysis (resolution 1°) has been downscaled to a horizontal resolution of 30 and 10?km for the period of 1961?C1990. This model setup includes the whole North Atlantic in the 30?km domain and spectral nudging is used to keep the large scales consistent with the driving ERA-40 reanalysis. The model results are compared against an extensive observational network of surface variables in complex terrain in Norway. The comparison shows that the WRF model is able to add significant detail to the representation of precipitation and 2-m temperature of the ERA-40 reanalysis. Especially the geographical distribution, wet day frequency and extreme values of precipitation are highly improved due to the better representation of the orography. Refining the resolution from 30 to 10?km further increases the skill of the model, especially in case of precipitation. Our results indicate that the use of 10-km resolution is advantageous for producing regional future climate projections. Use of a large domain and spectral nudging seems to be useful in reproducing the extreme precipitation events due to the better resolved synoptic scale features over the North Atlantic, and also helps to reduce the large regional temperature biases over Norway. This study presents a high-resolution, high-quality climatological data set useful for reference climate impact studies.     

Evaluation of Precipitation Indices over North America from Various Configurations of Regional Climate Models

This study presents the evaluation of simulations from two new Canadian regional climate models (RCMs), CanRCM4 and CRCM5, with a focus on the models’ skill in simulating daily precipitation indices and the Standardized Precipitation Index (SPI). The evaluation was carried out over the past two decades using several sets of gridded observations that partially cover North America. The new Canadian RCMs were also compared with four reanalysis products and six other RCMs. The different configurations of the Canadian RCM simulations also permit evaluation of the impact of different spatial resolutions, atmospheric drivers, and nudging conditions. The results from the new Canadian models show some improvement in precipitation characteristics over the previous Canadian RCM (CRCM4), but these differ with the seasons. For winter, CanRCM4 and CRCM5 have better skill than most other models over all of North America. For the summer, CRCM5 0.44° performs best over the United States, while CRCM4 has the best skill over Canada. Good skill is exhibited by CanRCM4 and CRCM4 in simulating the 6-month SPI over the Prairies and the western US Corn Belt. In general, differences are small between runs with or without large-scale spectral nudging; differences are small when different boundary conditions are used.     

东亚区域气候变化的长期数值模拟试验

文中利用NCAR的中尺度模式MM 5V3对东亚地区进行了 10a的长期积分模拟试验 ,并着重对冬、夏两季东亚区域气候变化特征进行了分析。分析结果表明 :(1)模式能够合理地模拟出 10a冬、夏平均的区域气候特征。模拟的 10a冬季平均降水的分布和强度与实际比较一致 ,对夏季降水分布特征的模拟也比较合理 ,但模拟的夏季华北降水偏多。模式对冬季平均场的模拟要优于对夏季的模拟 ;(2 )模式对降水、地面气温年际变率的模拟较为合理 ,模拟的中高层环流、温度场等要素的距平相关系数都比较高 ;(3)模式对不同ElNi no年对东亚区域气候变化影响的模拟能力有所不同 ,模拟的 1992 ,1995年的结果比较合理 ,但对 1998年模拟得不理想 ;(4)MM5V3模式具备一定的区域气候模拟能力。     

Atmospheric and terrestrial water budgets: sensitivity and performance of configurations and global driving data for long term continental scale WRF simulations

Driving data and physical parametrizations can significantly impact the performance of regional dynamical atmospheric models in reproducing hydrometeorologically relevant variables. Our study addresses the water budget sensitivity of the Weather Research and Forecasting Model System WRF (WRF-ARW) with respect to two cumulus parametrizations (Kain–Fritsch, Betts–Miller–Janji?), two global driving reanalyses (ECMWF ERA-INTERIM and NCAR/NCEP NNRP), time variant and invariant sea surface temperature and optional gridded nudging. The skill of global and downscaled models is evaluated against different gridded observations for precipitation, 2 m-temperature, evapotranspiration, and against measured discharge time-series on a monthly basis. Multi-year spatial deviation patterns and basin aggregated time series are examined for four globally distributed regions with different climatic characteristics: Siberia, Northern and Western Africa, the Central Australian Plane, and the Amazonian tropics. The simulations cover the period from 2003 to 2006 with a horizontal mesh of 30 km. The results suggest a high sensitivity of the physical parametrizations and the driving data on the water budgets of the regional atmospheric simulations. While the global reanalyses tend to underestimate 2 m-temperature by 0.2–2 K, the regional simulations are typically 0.5–3 K warmer than observed. Many configurations show difficulties in reproducing the water budget terms, e.g. with long-term mean precipitation biases of 150 mm month^?1 and higher. Nevertheless, with the water budget analysis viable setups can be deduced for all four study regions.     

Evaluation of regional climate model simulations versus gridded observed and regional reanalysis products using a combined weighting scheme

This study presents a combined weighting scheme which contains five attributes that reflect accuracy of climate data, i.e. short-term (daily), mid-term (annual), and long-term (decadal) timescales, as well as spatial pattern, and extreme values, as simulated from Regional Climate Models (RCMs) with respect to observed and regional reanalysis products. Southern areas of Quebec and Ontario provinces in Canada are used for the study area. Three series of simulation from two different versions of the Canadian RCM (CRCM4.1.1, and CRCM4.2.3) are employed over 23?years from 1979 to 2001, driven by both NCEP and ERA40 global reanalysis products. One series of regional reanalysis dataset (i.e. NARR) over North America is also used as reference for comparison and validation purpose, as well as gridded historical observed daily data of precipitation and temperatures, both series have been beforehand interpolated on the CRCM 45-km grid resolution. Monthly weighting factors are calculated and then combined into four seasons to reflect seasonal variability of climate data accuracy. In addition, this study generates weight averaged references (WARs) with different weighting factors and ensemble size as new reference climate data set. The simulation results indicate that the NARR is in general superior to the CRCM simulated precipitation values, but the CRCM4.1.1 provides the highest weighting factors during the winter season. For minimum and maximum temperature, both the CRCM4.1.1 and the NARR products provide the highest weighting factors, respectively. The NARR provides more accurate short- and mid-term climate data, but the two versions of the CRCM provide more precise long-term data, spatial pattern and extreme events. Or study confirms also that the global reanalysis data (i.e. NCEP vs. ERA40) used as boundary conditions in the CRCM runs has non-negligible effects on the accuracy of CRCM simulated precipitation and temperature values. In addition, this study demonstrates that the proposed weighting factors reflect well all five attributes and the performances of weighted averaged references are better than that of the best single model. This study also found that the improvement of WARs’ performance is due to the reliability (accuracy) of RCMs rather than the ensemble size.     

Nudging方法对中国西北强降水过程的模拟试验研究

通过在WRF(Weather Research and Forecasting)模式中运用松弛逼近方法(Nudging)同化NCEP-FNL资料,对中国西北地区夏季3次对流降水过程进行了模拟试验,检验了模式运用谱逼近(Spectral Nudging,SN)和格点逼近(Grid Nudging,GN)方案后对研究区域内降水过程的模拟性能。结果表明:SN和GN试验相对于控制试验,模拟的降水结果更接近于站点观测。在降水落区和量级的模拟效果上都有显著提高,且SN试验所得结果优于GN试验。通过分析两个同化试验在降水时段的基本要素场(湿度、温度和风速)变化,在近地面层,GN试验模拟的风速和温度较SN试验更接近观测;但700 hPa上,SN试验中风速、风向、温度、湿度的增量场变化及分布同降水的模拟结果存在较好的对应关系。最后,从物理诊断量—水汽通量散度的空间垂直变化来看,SN试验中600 hPa和700 hPa的水汽通量散度的正负分布有效调节降水的空间分布,更逼近观测。因此700 hPa的物理量场变量增量是降水模拟效果改善的主要原因之一。     

Numerical Simulation of Long-Term Climate Change in East Asia

A 10-yr regional climate simulation was performed using the fifth-generation PSU/NCAR Mesoscale Model Version 3 (MM5V3) driven by large-scale NCEP/NCAR reanalyses. Simulations of winter and summer mean regional climate features were examined against observations. The results showed that the model could well simulate the 10-yr winter and summer mean circulation, temperature, and moisture transport at middle and low levels. The simulated winter and summer mean sea level pressure agreed with the NCAR/NCEP reanalysis data. The model could well simulate the distribution and intensity of winter mean precipitation rates as well as the distribution of summer mean precipitation rates, but it overestimated the summer mean precipitation over North China. The model's ability to simulate the regional climate change in winter was superior to that in summer. In addition, the model could simulate the inter-annual variation of seasonal precipitation and surface air temperature. Geopotential heights and temperature at middle and high levels between simulations and observations exhibited high anomaly correlation coefficients. The model also showed large variability to simulate the regional climate change associated with the El Nino events. The MM5V3 well simulated the anomalies of summer mean precipitation in 1992 and 1995, while it demonstrated much less ability to simulate that in 1998. Generally speaking, the MM5V3 is capable of simulating the regional climate change, and could be used for long-term regional climate simulation.     

Testing the influence of western Pacific subtropical high...

In this paper, two seasonal scale simulations were conducted for the abnormal climate event in China in the summer of 1998 using a regional climate model (RegCM3). One is the control run, the other is nudging run, which was performed for zonal and meridional wind components, temperature, and humidity data for the region east of 120° E in the model domain to ensure that the simulated activity of western Pacific subtropical high (WPSH) in summer followed those of reanalysis data, while the interaction between the WPSH and the surrounding circulation systems was still maintained partially. Comparisons between the simulated regional circulation systems and the extension/withdrawal of the rain belt over eastern China as well as the activity of the WPSH were carried out. The results show that the relationship between the precipitation over eastern China and WPSH can be reproduced well in the nudging run. However, though the extension/withdrawal of the rain belt over eastern China is mainly dominated by that of WPSH, as pointed out by so many research works, the detailed precipitation scenario is not solely determined by the intensity and position of WPSH, and the precipitation discrepancy between simulation and observation is significant to some extent, which suggests that it is important to improve the precipitation physical process of the model in simulating the detailed precipitation scenario over eastern China.     

Regional model simulation of summer rainfall over the Philippines: Effect of choice of driving fields and ocean flux schemes

Summary The latest version of the Abdus Salam International Centre for Theoretical Physics (ICTP) regional model RegCM is used to investigate summer monsoon precipitation over the Philippine archipelago and surrounding ocean waters, a region where regional climate models have not been applied before. The sensitivity of simulated precipitation to driving lateral boundary conditions (NCEP and ERA40 reanalyses) and ocean surface flux scheme (BATS and Zeng) is assessed for 5 monsoon seasons. The ability of the RegCM to simulate the spatial patterns and magnitude of monsoon precipitation is demonstrated, both in response to the prominent large scale circulations over the region and to the local forcing by the physiographical features of the Philippine islands. This provides encouraging indications concerning the development of a regional climate modeling system for the Philippine region. On the other hand, the model shows a substantial sensitivity to the analysis fields used for lateral boundary conditions as well as the ocean surface flux schemes. The use of ERA40 lateral boundary fields consistently yields greater precipitation amounts compared to the use of NCEP fields. Similarly, the BATS scheme consistently produces more precipitation compared to the Zeng scheme. As a result, different combinations of lateral boundary fields and surface ocean flux schemes provide a good simulation of precipitation amounts and spatial structure over the region. The response of simulated precipitation to using different forcing analysis fields is of the same order of magnitude as the response to using different surface flux parameterizations in the model. As a result it is difficult to unambiguously establish which of the model configurations is best performing.     

Modeling and analysis of the potential impacts on regional climate due to vegetation degradation over arid and semi-arid regions of China

The regional climate effects of vegetation change in arid and semi-arid regions of China, which has experienced serious grassland degradation, are investigated in this study using the Weather Research and Forecasting (WRF) regional climate model. Two long-term simulation experiments (from January 1, 1980 to March 1, 2010), one with the land cover derived from the original United States Geological Survey’s (USGS) data (denoted as CTL) and the other (denoted as SEN) with a modification of the former one by vegetation degradation in arid and semi-arid regions of China, are undertaken to investigate the influence of land cover change on regional climate over arid and semi-arid regions of China. The possible mechanisms of how land cover change affects the regional climate in arid and semi-arid regions of China are also examined. The simulation results indicate that when compared with the observation datasets, the WRF model simulates the spatial pattern of observed temperature and precipitation quite well. After vegetation degradation over the arid and semi-arid regions of China, the net radiation and evaporation are reduced mainly within the degraded areas in summer, consistent with the reduction in precipitation and the increase in 2-m air temperature (T2 m).     

区域气候模式REMO对东亚季风季节变化的模拟研究

将欧洲区域气候模式REMO首次应用于东亚区域,利用该模式对1980年和1990年东亚季风季节变化进行了模拟研究,并将模拟结果与NCEP再分析资料进行比较,以检验该模式对东亚季风的模拟能力.研究表明,区域气候模式REMO能够较好地模拟出东亚地区高、低空的大气环流特征,模拟的高度场、流场和温度场与NCEP再分析资料场都比较一致.模拟结果显示了东亚季风的月变化和季节转换特征.模拟的降水场与GPCC降水资料的对比结果表明,REMO能较为成功地模拟出东亚地区降水的空间分布,并能较好地反映降水的季节变化及主要降水趋势,夏季降水模拟偏大,整个区域平均的降水量偏差约为18%左右.     

四川盆地夏季降水日变化的数值模拟

利用区域气候模式RegCM3对1991-2004年四川盆地夏季降水进行了数值模拟,通过模拟结果和NCEP/NCAR再分析资料的对比,评估了模式对四川盆地夏季＂夜雨＂现象的模拟能力。结果表明,RegCM3模式能较好地模拟出四川盆地夏季降水的空间分布和日变化规律,四川盆地夏季＂夜雨＂现象的形成与该地区的地形分布有密切关系。...     

Optimal nudging strategies in regional climate modelling: investigation in a Big-Brother experiment over the European and Mediterranean regions

The objective of this work is to gain a general insight into the key mechanisms involved in the impact of nudging on the large scales and the small scales of a regional climate simulation. A “Big Brother experiment” (BBE) approach is used where a “reference atmosphere” is known, unlike when regional climate models are used in practice. The main focus is on the sensitivity to nudging time, but the BBE approach allows to go beyond a pure sensitivity study by providing a reference which model outputs try to approach, defining an optimal nudging time. Elaborating upon previous idealized studies, this work introduces key novel points. The BBE approach to optimal nudging is used with a realistic model, here the weather research and forecasting model over the European and Mediterranean regions. A winter simulation (1 December 1989–28 February 1990) and a summer simulation (1 June 1999–31 August 1999) with a 50 km horizontal mesh grid have been performed with initial and boundary conditions provided by the ERA-interim reanalysis of the European Center for Medium-range Weather Forecast to produce the “reference atmosphere”. The impacts of spectral and indiscriminate nudging are compared all others things being equal and as a function of nudging time. The impact of other numerical parameters, specifically the domain size and update frequency of the large-scale driving fields, on the sensitivity of the optimal nudging time is investigated. The nudged simulations are also compared to non-nudged simulations. Similarity between the reference and the simulations is evaluated for the surface temperature, surface wind and for rainfall, key variables for climate variability analysis and impact studies. These variables are located in the planetary boundary layer, which is not subject to nudging. Regarding the determination of a possible optimal nudging time, the conclusion is not the same for indiscriminate nudging (IN) and spectral nudging and depends on the update frequency of the driving large-scale fields τ _a . For IN, the optimal nudging time is around τ = 3 h for almost all cases. For spectral nudging, the best results are for the smallest value of τ used for the simulations (τ = 1 h) for frequent update of the driving large-scale fields (3 and 6 h). The optimal nudging time is 3 for 12 h interval between two consecutive driving large-scale fields due to time sampling errors. In terms of resemblance to the reference fields, the differences between the simulations performed with IN and spectral nudging are small. A possible reason for this very similar performance is that nudging is active only above the planetary boundary layer where small-scale features are less energetic. As expected from previous studies, the impact of nudging is weaker for a smaller domain size. However the optimal nudging time itself is not sensitive to domain size. The proposed strategy ensures a dynamical consistency between the driving field and the simulated small-scale field but it does not ensure the best “observed” fine scale field because of the possible impact of incorrect driving large-scale field. This type of downscaling provides an upper bound on the skill possible for recent historical past and twenty-first century projections. The optimal nudging strategy with respect to dynamic downscaling could add skill whenever the parent global model has some level of skill.     

Sensitivity of a regional climate model on the simulation of high intensity rainfall events over the Arabian Peninsula and around Jeddah (Saudi Arabia)

As a step towards the adoption and use of the regional climate model (RegCM3) for the simulation of intense rainfall events over the Arabian Peninsula, this study examines its sensitivity to domain size, boundary location, forcing fields, and resolution. In the climatological results, RegCM3 performs well in reproducing the annual and the seasonal mean precipitation as well as the contrast between wet and dry years in terms of the amounts and locations of the rainbands. In addition, simulations are performed for two cases of intense rainfall events in the Jeddah area and surroundings using a combination of three domains and two boundary forcings at 50?km. The results show that different combinations of these parameters provide different skills for the regional model. However, RegCM3 performs relatively better when ERA40 (NNRP2) is used at the boundaries in the smaller domain (larger domain), indicating the importance of the stronger (relatively weaker) influence of boundary forcing needed to capture these intense rainfall events around Jeddah. This may be explained by the fact that around that region, RegCM3 produces, in the smaller domain, higher relative humidity and stronger wind vectors closer to the reanalyses when nested within the ERA40, while it shows its best performance with the larger domain when driven by NNRP2. It is also shown that the use of high resolution does not systematically improve the simulation of such events, although some encouraging results were produced.     

Improvement of an Extreme Heavy Rainfall Simulation Using Nudging Assimilation

From 21 to 22 July 2012, Beijing and its surrounding areas suffered from an extreme precipitation event that was unprecedented relative to the past 61 years, and the event caused 79 deaths and reported direct economic losses of11.64 billion Yuan. However, current models have difficulty to simulate the spatial and temporal distribution characteristics of such events. Therefore, improved simulations of these extreme precipitation processes are needed. In this study, nudging methods, including grid nudging(GN) and spectral nudging(SN), and more accurate surface type data retrieved from remote sensing were used in the Weather Research and Forecasting(WRF) model to simulate this extreme precipitation case. When the default city underlay surface of the WRF model was replaced by a more accurate urban surface(NU), the precipitation intensity could be better simulated, but the peak moment of precipitation seriously lagged. Although the peak precipitation intensity simulated by the GN experiment was weak, the simulated precipitation time was basically consistent with the observations. Using GN in only the outside domain could better simulate precipitation peaks, while using GN in both the inside and outside domains could better simulate the spatial distribution characteristics of precipitation. Additionally, the precipitation from GN could be better simulated than that from SN. Overall, the two nudging methods could contribute to better simulations of this case because the nudging methods could improve the simulations of 500-hPa geopotential height, 850-hPa water vapor transport, and low-level weather systems, which are the key factors in adjusting the spatial and temporal distributions of precipitation. This study is the basis for the investigation of the mechanism and attribution of extreme precipitation processes,and the results are of great significance for promoting understanding of and mitigating disasters caused by extreme precipitation.     

On the capability of RegCM to capture extremes in long term regional climate simulation – comparison with the observations for Czech Republic

Summary We analyze daily precipitation and temperature extremes over the Czech Republic in a regional climate simulation for the 40-year period of 1961–2000 carried out with the RegCM3 regional climate model. The model is run at 45 km grid interval and is driven by NCEP/NCAR reanalysis lateral boundary conditions. Comparison with station data shows that the model performs reasonably well in simulating the frequency of daily precipitation events of medium to high intensity as well as the precipitation intensities (return levels) of long return periods, with the exception of mountain stations. While this is attributed mainly to the relatively coarse representation of topography across the area of the Czech Republic, the parameterization of convection can be another reason. The model underestimates daily maximum temperature (especially in the warm seasons) and as a result the occurrence of heat waves (high temperature episodes). The performance of the model improves in the simulation of daily minimum temperature and cold wave events. In order to apply this regional model to the simulation of extreme events over the complex terrain as for Czech Republic we recommend that a higher resolution is used in order to better describe the topography of the Czech Republic and that the daily maximum temperature bias is reduced.     

Domain choice in an experimental nested modeling prediction system for South America

Summary The purposes of this paper are to evaluate the new version of the regional model, RegCM3, over South America for two test seasons, and to select a domain for use in an experimental nested prediction system, which incorporates RegCM3 and the European Community-Hamburg (ECHAM) general circulation model (GCM). To evaluate RegCM3, control experiments were completed with RegCM3 driven by both the NCEP/NCAR Reanalysis (NNRP) and ECHAM, using a small control domain (D-CTRL) and integration periods of January–March 1983 (El Ni?o) and January–March 1985 (La Ni?a). The new version of the regional model captures the primary circulation and rainfall differences between the two years over tropical and subtropical South America. Both the NNRP-driven and ECHAM-driven RegCM3 improve the simulation of the Atlantic intertropical convergence zone (ITCZ) compared to the GCM. However, there are some simulation errors. Irrespective of the driving fields, weak northeasterlies associated with reduced precipitation are observed over the Amazon. The simulation of the South Atlantic convergence zone is poor due to errors in the boundary condition forcing which appear to be amplified by the regional model. To select a domain for use in an experimental prediction system, sensitivity tests were performed for three domains, each of which includes important regional features and processes of the climate system. The domain sensitivity experiments were designed to determine how domain size and the location of the GCM boundary forcing affect the regional circulation, moisture transport, and rainfall in two years with different large scale conditions. First, the control domain was extended southward to include the exit region of the Andes low level jet (D-LLJ), then eastward to include the South Atlantic subtropical high (D-ATL), and finally westward to include the subsidence region of the South Pacific subtropical high and to permit the regional model more freedom to respond to the increased resolution of the Andes Mountains (D-PAC). In order to quantify differences between the domain experiments, measures of bias, root mean square error, and the spatial correlation pattern were calculated between the model results and the observed data for the seasonal average fields. The results show the GCM driving fields have remarkable control over the RegCM3 simulations. Although no single domain clearly outperforms the others in both seasons, the control domain, D-CTRL, compares most favorably with observations. Over the ITCZ region, the simulations were improved by including a large portion of the South Atlantic subtropical high (D-ATL). The methodology presented here provides a quantitative basis for evaluating domain choice in future studies.