Spectral Nudging Research Articles

The Concept of Large-Scale Conditioning of Climate Model Simulations of Atmospheric Coastal Dynamics: Current State and Perspectives

We review the state of dynamical downscaling with scale-constrained regional and global models. The methodology, in particular spectral nudging, has become a routine and well-researched tool for hindcasting climatologies of sub-synoptic atmospheric disturbances in coastal regions. At present, the spectrum of applications is expanding to other phenomena, but also to ocean dynamics and to extended forecasting. Additionally, new diagnostic challenges are appearing such as spatial characteristics of small-scale phenomena such as Low Level Jets.

Developing configuration of WRF model for long-term high-resolution wind wave hindcast over the North Atlantic with WAVEWATCH III

The spatial resolution of wind forcing fields is critical for modeling ocean surface waves. We analyze here the performance of the non-hydrostatic numerical weather prediction system WRF-ARW (Weather Research and Forecasting) run with a 14-km resolution for hindcasting wind waves in the North Atlantic. The regional atmospheric model was run in the domain from 20° N to 70° N in the North Atlantic and was forced with ERA-Interim reanalysis as initial and boundary conditions in a spectral nudging mode. Here, we present the analysis of the impact of spectral nudging formulation (cutoff wavelengths and depth through which full weighting from reanalysis data is applied) onto the performance of the modeled 10-m wind speed and wind wave fields for 1 year (2010). For modeling waves, we use the third-generation spectral wave model WAVEWATCH III. The sensitivity of the atmospheric and wave models to the spectral nudging formulation is investigated via the comparison with reanalysis and observational data. The results reveal strong and persistent agreement with reanalysis data during all seasons within the year with well-simulated annual cycle and regional patterns independently of the nudging parameters that were tested. Thus, the proposed formulation of the nudging provides a reliable framework for future long-term experiments aiming at hindcasting climate variability in the North Atlantic wave field. At the same time, dynamical downscaling allows for simulation of higher waves in coastal regions, specifically near the Greenland east coast likely due to a better representation of the mesoscale atmospheric dynamics in this area.

Incorporating a Large‐Scale Constraint Into Radar Data Assimilation to Mitigate the Effects of Large‐Scale Bias on the Analysis and Forecast of a Squall Line Over the Yangtze‐Huaihe River Basin

AbstractWhen there is an obvious large‐scale bias between a regional simulation and its driving global analysis, the regional model will provide inaccurate background information for radar data assimilation, which may eventually yield location errors associated with predicted precipitation. A case study of a squall line over the Yangtze‐Huaihe river basin presents such a situation. In this regard, we propose an approach to incorporate a large‐scale constraint into radar data assimilation to mitigate the effects of large‐scale bias on analysis and forecast results, in which global analysis data are introduced into the regional model using the spectral nudging technique to improve the quality of the first guess and background error statistics in radar data assimilation. A series of experiments are conducted with the Weather Research and Forecasting model and its three‐dimensional variational system to investigate the effectiveness of the proposed approach to introduce a large‐scale constraint into radar data assimilation. The experimental results demonstrate that the introduction of global analysis data can effectively correct the large‐scale bias and significantly improve the forecast skill of large‐scale patterns and convection initiation. The background error covariance (BE) obtained with the large‐scale constraint plays an important role in improving the assimilation effect. The length scales of BE are reduced after the large‐scale bias is removed, which represents a partial solution to the overestimation of BE reported in previous studies. In addition, applying a larger nudging wave number to radar data assimilation domain is not appropriate because the use of a larger wave number can negatively impact the three‐dimensional variational analysis.

A Maieutic Exploration of Nudging Strategies for Regional Climate Applications Using the WRF Model.

The use of nudging in the Weather Research and Forecasting (WRF) Model to constrain regional climate downscaling simulations is gaining in popularity because it can reduce error and improve consistency with the driving data. While some attention has been paid to whether nudging is beneficial for downscaling, very little research has been performed to determine best practices. In fact, many published papers use the default nudging configuration (which was designed for numerical weather prediction), follow practices used by colleagues, or adapt methods developed for other regional climate models. Here, a suite of 45 three-year simulations is conducted with WRF over the continental United States to systematically and comprehensively examine a variety of nudging strategies. The simulations here use a longer test period than did previously published works to better evaluate the robustness of each strategy through all four seasons, through multiple years, and across nine regions of the United States. The analysis focuses on the evaluation of 2-m temperature and precipitation, which are two of the most commonly required downscaled output fields for air quality, health, and ecosystems applications. Several specific recommendations are provided to effectively use nudging in WRF for regional climate applications. In particular, spectral nudging is preferred over analysis nudging. Spectral nudging performs best in WRF when it is used toward wind above the planetary boundary layer (through the stratosphere) and temperature and moisture only within the free troposphere. Furthermore, the nudging toward moisture is very sensitive to the nudging coefficient, and the default nudging coefficient in WRF is too high to be used effectively for moisture.

Effects of spectral nudging on precipitation extremes and diurnal cycle over CORDEX‐East Asia domain

In this study, by using the Weather Research and Forecasting (WRF) model, the impact of spectral nudging (SN) on simulating the precipitation extremes and diurnal cycle of precipitation over Coordinated Regional Climate Downscaling Experiment‐East Asia (CORDEX‐EA) domain is evaluated. The simulations are conducted during 1989–2007 driven by the ERA‐Interim reanalysis data with different SN experiments. It is clear that WRF has the ability to reproduce the long‐term climatology as well as the extremes over Asia, while biases are generated in the experiment without SN (NOSN). The simulations with SN, especially SN6 (SN with wavenumber 6) and SN3‐UVT (SN with wavenumber 3 applied to wind and temperature), can efficiently improve the model's skill by reducing the biases of precipitation frequency and intensity as well as the extreme indices except for the consecutive dry days (CDD). The diurnal harmonics of precipitation frequency and amount are well produced by WRF, and the SN experiments also perform better than the NOSN experiment, with SN3‐UVT showing the best skill. However, the SN only with winds shows little improvement in simulating the diurnal cycles, and the additional nudging of temperature can improve the simulation only over south China and Indo‐China. It can be concluded that the strength of nudging parameters can significantly affect the simulation of precipitation, with better performance under the stronger nudging. Moreover, additional nudging of temperature can also improve the simulation.

WRF hindcasts of cold front passages over the ARM Eastern North Atlantic Site: a sensitivity study.

The present study explores the ability of the Weather Research and Forecasting (WRF) Model to accurately reproduce the passage of extratropical cold fronts at the DOE-ARM Eastern North Atlantic (ENA) observation site on the Azores Islands. An analysis of three case studies is performed in which the impact of the WRF domain size, position of the model boundary relative to the ENA site, grid spacing, and spectral nudging conditions are explored. The results from these case studies indicate that model biases in the timing and duration of cold front passages change with the distance between the model domain boundary and the ENA site. For these three cases, if the western model boundary is further than 1500 km from the site, the front becomes too meridional and fails to reach the site, making 1000 km or 1500 km the optimal distances. In contrast, integrations with small distances (e.g., 500 km) between the site and domain boundaries have inadequate spatial spin-up, i.e. the domain is too small for the model to properly stabilize. For all 3 cases, regardless of domain size, the model has biases in its upper-level circulation that impact the position and timing of the front. However, this issue is most serious for 4000 km2 domains and larger. For these domains, prolonged spectral nudging can correct cold front biases. As such, this analysis provides a framework to optimize the WRF model configuration necessary for a realistic hindcast of a cold front passage at a fixed location centered in a domain as large as computationally possible.

Does convection-permitting simulate better rainfall distribution and extreme over Guinean coast and surroundings?

This research work focuses on the problem of climate simulation of rainfall over West Africa and particularly over coastal countries of the Gulf of Guinea by Regional Climate Models. The sensitivities of Weather Research and Forecasting (WRF) Model were tested for changes in horizontal resolution (convection permitting versus parameterized) on the replication of West African monsoon for the year 2014. Sensitivity test was also performed for response of rainfall to changes in microphysics (MP) and planetary boundary layer (PBL) schemes. Generally, the result shows that WRF is able to replicates rainfall distribution with an adequate representation of the dynamical features of West African monsoon system. The high-resolution (wrf-4km) shows dry bias along the coast of the Gulf of Guinea but generally outperforms wrf-24km run especially in replication of the extreme rainfall distribution. The dry bias along the coastal area is suggested to not to be only related to convection but mostly to microphysics and PBL parameterisation schemes. Differences were noticed between the dynamics of WRF and ERA-interim outputs despite the use of spectral nudging in the experiment which then suggest strong interactions between scales. These differences were observed to be restricted mainly to the low-layer monsoon flow in JJA. Both runs at 24 and 4 km hardly simulate the typical diurnal distribution of rainfall. The sensitivity of WRF to MP (only sophisticated MP were tested) and PBL reveals a stronger impact of PBL than MP on rainfall distribution and the most significant added value over the Guinean coast and surroundings area was provided by the configurations using non-local PBL scheme (as ACM2). The changes in MP and PBL schemes in general seem to have less effect on the explicit runs (wrf-4km) in the replication of the rainfall over the Gulf of Guinea and the surroundings seaboard.

A Simple Ensemble Simulation Technique for Assessment of Future Variations in Specific High‐Impact Weather Events

AbstractTo investigate future variations in high‐impact weather events, numerous samples are required. For the detailed assessment in a specific region, a high spatial resolution is also required. A simple ensemble simulation technique is proposed in this paper. In the proposed technique, new ensemble members were generated from one basic state vector and two perturbation vectors, which were obtained by lagged average forecasting simulations. Sensitivity experiments with different numbers of ensemble members, different simulation lengths, and different perturbation magnitudes were performed. Experimental application to a global warming study was also implemented for a typhoon event. Ensemble‐mean results and ensemble spreads of total precipitation, atmospheric conditions showed similar characteristics across the sensitivity experiments. The frequencies of the maximum total and hourly precipitation also showed similar distributions. These results indicate the robustness of the proposed technique. On the other hand, considerable ensemble spread was found in each ensemble experiment. In addition, the results of the application to a global warming study showed possible variations in the future. These results indicate that the proposed technique is useful for investigating various meteorological phenomena and the impacts of global warming. The results of the ensemble simulations also enable the stochastic evaluation of differences in high‐impact weather events. In addition, the impacts of a spectral nudging technique were also examined. The tracks of a typhoon were quite different between cases with and without spectral nudging; however, the ranges of the tracks among ensemble members were comparable. It indicates that spectral nudging does not necessarily suppress ensemble spread.

Spatiotemporal characteristics of heat waves over China in regional climate simulations within the CORDEX-EA project

Using the Weather Research and Forecasting (WRF) model, this paper analyzes the spatiotemporal features of heat waves in 20-year regional climate simulations over East Asia, and investigates the capability of WRF to reproduce observational heat waves in China. Within the framework of the Coordinated Regional Climate Downscaling Experiment (CORDEX), the WRF model is driven by the ERA-Interim (ERAIN) reanalysis, and five continuous simulations are conducted from 1989 to 2008. Of these, four runs apply the interior spectral nudging (SN) technique with different wavenumbers, nudging variables and nudging coefficients. Model validations show that WRF can reasonably reproduce the spatiotemporal features of heat waves in China. Compared with the experiment without SN, the application of SN is effectie on improving the skill of the model in simulating both the spatial distributions and temporal variations of heat waves of different intensities. The WRF model shows advantages in reproducing the synoptic circulations with SN and therefore yields better representations for heat wave events. Besides, the SN method is able to preserve the variability of large-scale circulations quite well, which in turn adjusts the extreme temperature variability towards the observation. Among the four SN experiments, those with stronger nudging coefficients perform better in modulating both the spatial and temporal features of heat waves. In contrast, smaller nudging coefficients weaken the effects of SN on improving WRF’s performances.

Assessment of the Performance of Three Dynamical Climate Downscaling Methods Using Different Land Surface Information over China

This study aims to assess the performance of different dynamical downscaling methods using updated land surface information. Particular attention is given to obtaining high-resolution climate information over China by the combination of an appropriate dynamical downscaling method and updated land surface information. Two group experiments using two land surface datasets are performed, including default Weather Research and Forecasting (WRF) land surface data (OLD) and accurate dynamically accordant MODIS data (NEW). Each group consists of three types of experiments for the summer of 2014, including traditional continuous integration (CT), spectral nudging (SN), and re-initialization (Re) experiments. The Weather Research and Forecasting (WRF) model is used to dynamically downscale ERA-Interim (reanalysis of the European Centre for Medium-Range Weather Forecast, ECMWF) data with a grid spacing of 30 km over China. The simulations are evaluated via comparison with observed conventional meteorological variables, showing that the CT method, which notably overestimates 2 m temperature and underestimates 2 m relative humidity across China, performs the worst; the SN and Re runs outperform the CT method, and the Re shows the smallest RMSE (root means square error). A comparison of observed and simulated precipitation shows that the SN simulation is closest to the observed data, while the CT and Re simulations overestimate precipitation south of the Yangtze River. Compared with the OLD group, the RMSE values of temperature and relative humidity are significantly improved in CT and SN, and there is smaller improved in Re. However, obvious improvements in precipitation are not evident.

Cyclone Activity in the Arctic From an Ensemble of Regional Climate Models (Arctic CORDEX)

AbstractThe ability of state‐of‐the‐art regional climate models to simulate cyclone activity in the Arctic is assessed based on an ensemble of 13 simulations from 11 models from the Arctic‐CORDEX initiative. Some models employ large‐scale spectral nudging techniques. Cyclone characteristics simulated by the ensemble are compared with the results forced by four reanalyses (ERA‐Interim, National Centers for Environmental Prediction‐Climate Forecast System Reanalysis, National Aeronautics and Space Administration‐Modern‐Era Retrospective analysis for Research and Applications Version 2, and Japan Meteorological Agency‐Japanese 55‐year reanalysis) in winter and summer for 1981–2010 period. In addition, we compare cyclone statistics between ERA‐Interim and the Arctic System Reanalysis reanalyses for 2000–2010. Biases in cyclone frequency, intensity, and size over the Arctic are also quantified. Variations in cyclone frequency across the models are partly attributed to the differences in cyclone frequency over land. The variations across the models are largest for small and shallow cyclones for both seasons. A connection between biases in the zonal wind at 200 hPa and cyclone characteristics is found for both seasons. Most models underestimate zonal wind speed in both seasons, which likely leads to underestimation of cyclone mean depth and deep cyclone frequency in the Arctic. In general, the regional climate models are able to represent the spatial distribution of cyclone characteristics in the Arctic but models that employ large‐scale spectral nudging show a better agreement with ERA‐Interim reanalysis than the rest of the models. Trends also exhibit the benefits of nudging. Models with spectral nudging are able to reproduce the cyclone trends, whereas most of the nonnudged models fail to do so. However, the cyclone characteristics and trends are sensitive to the choice of nudged variables.

Precipitation Dynamical Downscaling Over the Great Plains

AbstractDetailed, regional climate projections, particularly for precipitation, are critical for many applications. Accurate precipitation downscaling in the United States Great Plains remains a great challenge for most Regional Climate Models, particularly for warm months. Most previous dynamic downscaling simulations significantly underestimate warm‐season precipitation in the region. This study aims to achieve a better precipitation downscaling in the Great Plains with the Weather Research and Forecast (WRF) model. To this end, WRF simulations with different physics schemes and nudging strategies are first conducted for a representative warm season. Results show that different cumulus schemes lead to more pronounced difference in simulated precipitation than other tested physics schemes. Simply choosing different physics schemes is not enough to alleviate the dry bias over the southern Great Plains, which is related to an anticyclonic circulation anomaly over the central and western parts of continental U.S. in the simulations. Spectral nudging emerges as an effective solution for alleviating the precipitation bias. Spectral nudging ensures that large and synoptic‐scale circulations are faithfully reproduced while still allowing WRF to develop small‐scale dynamics, thus effectively suppressing the large‐scale circulation anomaly in the downscaling. As a result, a better precipitation downscaling is achieved. With the carefully validated configurations, WRF downscaling is conducted for 1980–2015. The downscaling captures well the spatial distribution of monthly climatology precipitation and the monthly/yearly variability, showing improvement over at least two previously published precipitation downscaling studies. With the improved precipitation downscaling, a better hydrological simulation over the trans‐state Oologah watershed is also achieved.

Large-scale secondary circulations in a limited area model – the impact of lateral boundaries and resolution

Within their domain, regional climate and weather forecasting models deviate from the driving data. Small-scale deviations are a desired effect of adding regional details. There are, however, also deviations of the large-scale circulation, which can be caused by orographic effects and depend on the large-scale flow condition. These ‘secondary circulations’ (SCs) are confined to the model domain due to the prescribed boundary conditions. Here, the impact of different regional model configurations on the SC is analysed in a case study for the European region using an ensemble approach. It is shown that at 500 hPa, vortices of the SC have diameters on the order of several thousand kilometres and are related to wind speed anomalies of more than 5 m/s and geopotential height anomalies of more than 5 dam. The spatial structure and the amplitude of the SC strongly depend on the location of the lateral boundaries. The impact of the boundary location on the anomalies is on the same order of magnitude as the anomalies themselves. The resolution of the regional model, as well as the application of spectral nudging and a smoothed topography, affects mainly the amplitude of the SC, but not the spatial structure.

The Use of a Spectral Nudging Technique to Determine the Impact of Environmental Factors on the Track of Typhoon Megi (2010)

Sensitivity tests based on a spectral nudging (SN) technique are conducted to analyze the effect of large-scale environmental factors on the movement of typhoon Megi (2010). The error of simulated typhoon track is effectively reduced using SN and the impact of dynamical factors is more significant than that of thermal factors. During the initial integration and deflection period of Megi (2010), the local steering flow of the whole and lower troposphere is corrected by a direct large-scale wind adjustment, which improves track simulation. However, environmental field nudging may weaken the impacts of terrain and typhoon system development in the landfall period, resulting in large simulated track errors. Comparison of the steering flow and inner structure of the typhoon reveals that the large-scale circulation influences the speed and direction of typhoon motion by: (1) adjusting the local steering flow and (2) modifying the environmental vertical wind shear to change the location and symmetry of the inner severe convection.

A Combined Approach to Improving the Regional Model Forecasts for the Rainy Season in China

A combined forecasting methodology, into which the spectral nudging, lateral boundary filtering, and update initial conditions methods are incorporated, was employed in the regional Weather Research and Forecasting (WRF) Model. The intent was to investigate the potential for improving the prediction capability for the rainy season in China via using as many merits of the global model having better predictability as it does for the large-scale circulation and of the regional model as it does for the small-scale features. The combined methodology was found to be successful in improving the prediction of the regional atmospheric circulation and precipitation. It performed best for the larger magnitude precipitation, the relative humidity above 800 hPa, and wind fields below 300 hPa. Furthermore, the larger the magnitude and the longer the lead time, the more obvious is the improvement in terms of the accumulated rainfall of persistent severe rainfall events.

High‐Resolution Regional Reanalysis in China: Evaluation of 1 Year Period Experiments

AbstractGlobally, reanalysis data sets are widely used in assessing climate change, validating numerical models, and understanding the interactions between the components of a climate system. However, due to the relatively coarse resolution, most global reanalysis data sets are not suitable to apply at the local and regional scales directly with the inadequate descriptions of mesoscale systems and climatic extreme incidents such as mesoscale convective systems, squall lines, tropical cyclones, regional droughts, and heat waves. In this study, by using a data assimilation system of Gridpoint Statistical Interpolation, and a mesoscale atmospheric model of Weather Research and Forecast model, we build a regional reanalysis system. This is preliminary and the first experimental attempt to construct a high‐resolution reanalysis for China main land. Four regional test bed data sets are generated for year 2013 via three widely used methods (classical dynamical downscaling, spectral nudging, and data assimilation) and a hybrid method with data assimilation coupled with spectral nudging. Temperature at 2 m, precipitation, and upper level atmospheric variables are evaluated by comparing against observations for one‐year‐long tests. It can be concluded that the regional reanalysis with assimilation and nudging methods can better produce the atmospheric variables from surface to upper levels, and regional extreme events such as heat waves, than the classical dynamical downscaling. Compared to the ERA‐Interim global reanalysis, the hybrid nudging method performs slightly better in reproducing upper level temperature and low‐level moisture over China, which improves regional reanalysis data quality.

Dynamical downscaling with the fifth-generation Canadian regional climate model (CRCM5) over the CORDEX Arctic domain: effect of large-scale spectral nudging and of empirical correction of sea-surface temperature

As part of the CORDEX project, the fifth-generation Canadian Regional Climate Model (CRCM5) is used over the Arctic for climate simulations driven by reanalyses and by the MPI-ESM-MR coupled global climate model (CGCM) under the RCP8.5 scenario. The CRCM5 shows adequate skills capturing general features of mean sea level pressure (MSLP) for all seasons. Evaluating 2-m temperature (T2m) and precipitation is more problematic, because of inconsistencies between observational reference datasets over the Arctic that suffer of a sparse distribution of weather stations. In our study, we additionally investigated the effect of large-scale spectral nudging (SN) on the hindcast simulation driven by reanalyses. The analysis shows that SN is effective in reducing the spring MSLP bias, but otherwise it has little impact. We have also conducted another experiment in which the CGCM-simulated sea-surface temperature (SST) is empirically corrected and used as lower boundary conditions over the ocean for an atmosphere-only global simulation (AGCM), which in turn provides the atmospheric lateral boundary conditions to drive the CRCM5 simulation. This approach, so-called 3-step approach of dynamical downscaling (CGCM-AGCM-RCM), which had considerably improved the CRCM5 historical simulations over Africa, exhibits reduced impact over the Arctic domain. The most notable positive effect over the Arctic is a reduction of the T2m bias over the North Pacific Ocean and the North Atlantic Ocean in all seasons. Future projections using this method are compared with the results obtained with the traditional 2-step dynamical downscaling (CGCM-RCM) to assess the impact of correcting systematic biases of SST upon future-climate projections. The future projections are mostly similar for the two methods, except for precipitation.

Does Spectral Nudging Have an Effect on Dynamical Downscaling Applied in Small Regional Model Domains?

Spectral nudging is a method that was developed to constrain regional climate models so that they reproduce the development of the large-scale atmospheric state, while permitting the formation of regional-scale details as conditioned by the large scales. Besides keeping the large-scale development in the interior close to a given state, the method also suppresses the emergence of ensemble variability. The method is mostly applied to reconstructions of past weather developments in regions with an extension of typically 1000–8000 km. In this article, the authors examine if spectral nudging is having an effect on simulations with model regions of the size of about 700 km × 500 km at midlatitudes located mainly over flat terrain. First two pairs of simulations are compared, two runs each with and without spectral nudging, and it is found that the four simulations are very similar, without systematic or intermittent phases of divergence. Smooth fields, which are mainly determined by spatial patterns, such as air pressure, show hardly any differences, while small-scale and heterogeneous fields such as precipitation vary strongly, mostly on the gridpoint scale, irrespective if spectral nudging is employed or not. It is concluded that the application of spectral nudging has little effect on the simulation when the model region is relatively small.

Development of a Model Performance Metric and Its Application to Assess Summer Precipitation over the U.S. Great Plains in Downscaled Climate Simulations

Abstract Several dynamically downscaled climate simulations with various spatial resolutions (24, 12, and 4 km) and spectral nudging strengths (0, 600, and 2000 km) have been run over the contiguous United States from 2000 to 2009 using the high-resolution NASA Unified Weather and Research Forecasting (NU-WRF) regional model initialized and constrained by the Modern-Era Retrospective Analysis for Research and Applications, version 2 (MERRA-2). This paper summarizes the authors’ efforts on the development of a model performance metric and its application to assess summer precipitation over the U.S. Great Plains (USGP) in these downscaled climate simulations. A new model performance metric T was first developed that uses both the linear correlation coefficient and mean square error and is consistent with other commonly used metrics, but gives a bigger separation between good and bad simulations. This metric T was then applied to the summer mean precipitation spatial pattern, diurnal Hovmöller diagram, and diurnal spatial pattern over the USGP from the simulations focusing on the summer precipitation diurnal cycle related to mesoscale convective systems (MCSs). The metric T skill scores increase significantly from the control simulation to the nudged simulations and from the nudged simulations with shorter wavelengths to the nudged simulations with longer wavelengths, but do not change much from MERRA-2 to the downscaled simulations or between the various downscaled simulations with different spatial resolutions. Thus, there is some credibility, but no significant value added compared to MERRA-2, of the downscaled climate simulations of the summer precipitation over the USGP.

Internal variability of a dynamically downscaled climate over North America

This study investigates the internal variability (IV) of a regional climate model, and considers the impacts of horizontal resolution and spectral nudging on the IV. A 16-member simulation ensemble was conducted using the Weather Research Forecasting model for three model configurations. Ensemble members included simulations at spatial resolutions of 50 and 12 km without spectral nudging and simulations at a spatial resolution of 12 km with spectral nudging. All the simulations were generated over the same domain, which covered much of North America. The degree of IV was measured as the spread between the individual members of the ensemble during the integration period. The IV of the 12 km simulation with spectral nudging was also compared with a future climate change simulation projected by the same model configuration. The variables investigated focus on precipitation and near-surface air temperature. While the IVs show a clear annual cycle with larger values in summer and smaller values in winter, the seasonal IV is smaller for a 50-km spatial resolution than for a 12-km resolution when nudging is not applied. Applying a nudging technique to the 12-km simulation reduces the IV by a factor of two, and produces smaller IV than the simulation at 50 km without nudging. Applying a nudging technique also changes the geographic distributions of IV in all examined variables. The IV is much smaller than the inter-annual variability at seasonal scales for regionally averaged temperature and precipitation. The IV is also smaller than the projected changes in air-temperature for the mid- and late twenty-first century. However, the IV is larger than the projected changes in precipitation for the mid- and late twenty-first century.