Canadian Meteorological Centre Research Articles

A simple strategy for linearizing complex moist convective schemes

AbstractVariational data assimilation methods and related applications depend on the validity of the tangent‐linear approximation, which is truly challenging when applied to deep moist convection. A simple strategy for linearizing complex moist convective schemes in numerical weather prediction models is proposed. This strategy represents a trade‐off between code development and maintenance on the one hand and expected benefit on the other hand. The generic linearized scheme described hereafter can be used in conjunction with any nonlinear moist convective scheme, thus eliminating the need to linearize complex codes. The universality of the scheme stems from the fact that conditional triggers and cloud vertical extent are supplied by the trajectory of the nonlinear scheme. In active columns, the convective tendencies cancel the large‐scale dynamical tendencies. Potential uses of the scheme include variational data assimilation and diagnostic studies such as tropical singular vectors and key analysis errors with precipitation. The relevance of the methodology is examined with the Global Environmental Multiscale model using the Kain–Fritsch mass‐flux scheme that is operational at the Canadian Meteorological Centre. It is shown that the tangent‐linear approximation is improved when compared to a linearized convection scheme having its own trigger functions. The improvement is particularly noticeable for the partition between stratiform and convective components of surface precipitation. Finally the examination of adjoint sensitivities of surface precipitation with the simplified linearized scheme triggered by the Kain–Fritsch scheme reveals a more pronounced sensitivity to midlatitude baroclinic instability and less predictability for tropical systems when compared to a Kuo‐type scheme. Copyright © 2009 Royal Meteorological Society and Crown in the Right of Canada

Measurement and modelling of radioxenon plumes in the Ottawa Valley

Since 2001 a real-time radiation monitoring network of Canadian nuclear facilities and major population centres has been implemented for response to nuclear incidents including a possible terrorist attack. Unshielded NaI(Tl) spectroscopic detectors are employed to measure gamma radiation from airborne radioactivity and radioactivity deposited on the ground. These detectors are composed of a standard 3″ × 3″ cylindrical NaI(Tl) spectrometers with data storage and integrated telemetry. Some of the detectors have been deployed in the Ottawa Valley near Chalk River Laboratories and Ottawa, which has a complex radioxenon environment due to the proximity of nuclear power reactors, and medical isotope facilities. Although not a health threat, these releases have provided an opportunity for the Canadian Meteorological Centre and the Commissariat à l'Energie Atomique to validate their meteorological models. The meteorological models of the two organizations are in good agreement on the origin and the source terms of these releases.

Impact of observational error on the validation of ensemble prediction systems

AbstractEnsemble prediction systems (EPSs) are usually validated under the assumption that the verifying observations are exact. In this paper, two methods are considered for taking observation errors into account. In the ‘perturbed‐ensemble’ method, which has already been studied by other authors, the predicted ensemble elements are randomly perturbed in a way that is consistent with the assumed observation error. In the ‘observational‐probability’ method, which is new, a verifying observation is considered as defining, together with the assumed associated error, a probability distribution. All standard scores for evaluation of EPSs (reliability diagram, Brier score, ranked probability score (RPS), continuous RPS (CRPS), relative‐operating‐characteristics (ROC) curve area), with the exception of the rank histogram, remain defined in this second method. In particular, the classical reliability–resolution decomposition of the Brier score, and of its extension to the RPS and CRPS, remain defined.Numerical simulations, partially supported by theoretical considerations, show that, with respect to the case when observation errors are ignored, the perturbed‐ensemble method improves reliability, as well as the ROC score, while it has no significant impact on resolution, as measured by the Brier score. The observational‐probability method, on the other hand, degrades reliability and the ROC score, but improves resolution.With respect to the ‘real’ performance of the system (i.e. the one that would be diagnosed if no error were present), reliability is unchanged in the perturbed‐ensemble method, while resolution and the ROC score are degraded. The observational‐probability method degrades reliability and the ROC score. As for resolution, an optimum value of the observational error is found, below which resolution is improved.Diagnostics performed on the operational EPS of the Canadian Meteorological Centre confirm the results of the simulations as to the consequences of ignoring observation errors, or on the contrary of taking them into account through either of the two methods. The significance of those various results is discussed. This article replaces a previously published version (Q. J. R. Meteorol. Soc. 134(631): 509–521, DOI: 10.1002/qj.221). Copyright © 2008 Royal Meteorological Society

The MAGS Water and Energy Budget Study

Abstract This study represents the first attempt at developing a comprehensive climatology of atmospheric and surface water and energy budgets for the Mackenzie River basin (MRB). Different observed, remotely sensed, (re)analyzed, and modeled datasets were used to obtain independent estimates of the budgets. In particular, assimilated datasets, including the National Centers for Environmental Prediction Global Reanalysis 2 (NCEP-R2), the global 40-yr European Centre for Medium-Range Weather Forecasts Re-Analysis (ERA-40), the NCEP North American Regional Reanalysis (NARR), and the Canadian Meteorological Centre (CMC) operational regional analysis as well as results from the Canadian Regional Climate Model (CRCM) simulations, are used in the study. Apart from the development of state-of-the-art budget estimates for the MRB, the relative merits of current models, data assimilation systems, and global blended datasets in representing aspects of the water and energy cycle of this northern and data-sparse region were also assessed. In addition, the levels of uncertainty in assessing the budgets as well as their sources are discussed. The regional water budget for the MRB is closed within 10% of the observed runoff by using the moisture flux convergence from ERA-40, NARR, CMC, or CRCM. While these are noted improvements over previous water closure assessments for the region, magnitudes of the residuals in balancing the budgets are often comparable to the budget terms themselves in all the model and analysis datasets, and the spreads of budget estimates from the different datasets are also typically large, suggesting that substantial improvements to the models and observations are needed before the assessments of water and energy budgets for this northern region can be vastly improved.

Excitation of Rossby-wave trains: optimal growth of forecast errors

Singular vectors (SVs) with appropriate norms and time scales are used in this study to capture the excitation of Rossby-wave trains (RWTs). Using global analysis data provided by the Canadian Meteorological Centre, a long-lasting RWT is selected beginning on 19 Nov 2002. Ten SVs are calculated using an optimization time interval of 48 h and a tangent linear model including a complete set of simplified physical parameterizations. At initial time, the global total-energy norm is used. For the final time, the total energy (TE) and the rotational kinetic energy (RKE) norms over three different restricted horizontal and vertical domains are used. These different configurations are examined to enable an appropriate configuration to be selected for our future work on using SVs to study the excitation of RWTs. The SVs are computed and non-linear forecasts are performed using the Canadian Global Environment Multiscale model. Results obtained show that when the analysis is perturbed with the pseudo-inverse of the 48 h forecast error in the SV subspace, the perturbation propagates with the group velocity of the RWT and the error in the non-linear forecast over the life of the wave train (6 days) is significantly reduced. Comparison between using a final-time norm based on either TE or RKE showed little impact.

An observation‐based formulation of snow cover fraction and its evaluation over large North American river basins

Snow cover strongly interacts with climate through snow albedo feedbacks. However, global climate models still are not adequate in representing snow cover fraction (SCF), i.e., the fraction of a model grid cell covered by snow. Through an analysis of the advanced very high resolution radiometer (AVHRR) derived SCF and the Canadian Meteorological Centre (CMC) gridded snow depth and snow water equivalent (SWE), we found that the SCF–snow depth relationship varies with seasons, which may be approximated by variations in snow density. We then added snow density to an existing SCF formulation to reflect the variations in the SCF–snow depth relationship with seasons. The reconstructed SCF with the gridded snow depth and SWE by employing this snow density–dependent SCF formulation agrees better with the AVHRR‐derived SCF than other formulations. The default SCF formulation in the National Center for Atmospheric Research community land model (CLM), driven by observed near‐surface meteorological forcings, simulates a smaller SCF and a shallower snow depth than observations. Implementation of the new SCF formulation into the NCAR CLM greatly improves the simulations of SCF, snow depth, and SWE in most North American (NA) river basins. The new SCF formulation increases SCF by 20–40%, decreases net solar radiation by up to 20 W m−2, and decreases surface temperature by up to 4 K in most midlatitude regions in winter and at high latitudes in spring. The new scheme reproduces the observed SCF, snow depth, and SWE in terms of interannual variability and interbasin variability in most NA river basins except for the mountainous Columbia and Colorado River basins. It produces SCF trends similar to that of AVHRR. However, it produces greater decreasing trends in ablation seasons and smaller increasing trends in accumulation seasons than those of the CMC snow depth and SWE.

A New Spatial-Scale Decomposition of the Brier Score: Application to the Verification of Lightning Probability Forecasts

Abstract A new scale decomposition of the Brier score for the verification of probabilistic forecasts defined on a spatial domain is introduced. The technique is illustrated on the Canadian Meteorological Centre (CMC) lightning probability forecasts. Probability forecasts of lightning occurrence in 3-h time windows and 24-km spatial resolution are verified against lightning observations from the North American Lightning Detection Network (NALDN) on a domain encompassing Canada and the northern United States. Verification is performed for lightning occurrences exceeding two different thresholds, to assess the forecast performance both for modest and intense lightning activity. Observation and forecast fields are decomposed into the sum of components on different spatial scales by performing a discrete 2D Haar wavelet decomposition. Wavelets, rather than Fourier transforms, were chosen because they are locally defined, and therefore more suitable for representing discontinuous spatial fields characterized by the presence of a few sparse nonzero values, such as lightning. Verification at different spatial scales is performed by evaluating Brier score and Brier skill score for each spatial-scale component. Reliability and resolution are also evaluated on different scales. Moreover, the bias on different scales is assessed, along with the ability of the forecasts to reproduce the observed-scale structure.

Verification of an Ensemble Prediction System against Observations

Abstract A verification system has been developed for the ensemble prediction system (EPS) at the Canadian Meteorological Centre (CMC). This provides objective criteria for comparing two EPSs, necessary when deciding whether or not to implement a new or revised EPS. The proposed verification methodology is based on the continuous ranked probability score (CRPS), which provides an evaluation of the global skill of an EPS. Its reliability/resolution partition, proposed by Hersbach, is used to measure the two main attributes of a probabilistic system. Also, the characteristics of the reliability are obtained from the two first moments of the reduced centered random variable (RCRV), which define the bias and the dispersion of an EPS. Resampling bootstrap techniques have been applied to these scores. Confidence intervals are thus defined, expressing the uncertainty due to the finiteness of the number of realizations used to compute the scores. All verifications are performed against observations to provide more independent validations and to avoid any local systematic bias of an analysis. A revised EPS, which has been tested at the CMC in a parallel run during the autumn of 2005, is described in this paper. This EPS has been compared with the previously operational one with the verification system presented above. To illustrate the verification methodology, results are shown for the temperature at 850 hPa. The confidence intervals are computed by taking into account the spatial correlation of the data and the temporal autocorrelation of the forecast error. The revised EPS performs significantly better for all the forecast ranges, except for the resolution component of the CRPS where the improvement is no longer significant from day 7. The significant improvement of the reliability is mainly due to a better dispersion of the ensemble. Finally, the verification system correctly indicates that variations are not significant when two theoretically similar EPSs are compared.

A Canadian Precipitation Analysis (CaPA) Project: Description and Preliminary Results

Abstract A Canadian Precipitation Analysis (CaPA) project for producing 6 h rainfall accumulations at a resolution of 15 km over North America in real‐time is described. The spatial interpolation technique is based on statistical interpolation using short‐range precipitation forecasts from the Canadian Meteorological Centre's (CMC) regional model as the background field with rain‐gauge measurements from the surface network and radar derived rain rates as observations. A pilot study was undertaken over the province of Québec at a resolution of 10 km using additional rain‐gauge observations from a cooperative network for August 2003. This test period allowed the development of methodologies for an objective estimation of background and observation error statistics and for improving the overall quality of rain‐gauge and radar data. The improved skill of the analysis with respect to the model short‐range forecasts is assessed against radar precipitation. The use of additional rain gauges from the surface cooperative network in Québec significantly increased the quality of the resulting analysis.

A Canadian precipitation analysis (CaPA) project: Description and preliminary results

A Canadian Precipitation Analysis (CaPA) project for producing 6 h rainfall accumulations at a resolution of 15 km over North America in real‐time is described. The spatial interpolation technique is based on statistical interpolation using short‐range precipitation forecasts from the Canadian Meteorological Centre's (CMC) regional model as the background field with rain‐gauge measurements from the surface network and radar derived rain rates as observations. A pilot study was undertaken over the province of Québec at a resolution of 10 km using additional rain‐gauge observations from a cooperative network for August 2003. This test period allowed the development of methodologies for an objective estimation of background and observation error statistics and for improving the overall quality of rain‐gauge and radar data. The improved skill of the analysis with respect to the model short‐range forecasts is assessed against radar precipitation. The use of additional rain gauges from the surface cooperative network in Québec significantly increased the quality of the resulting analysis.

The Characteristics of Key Analysis Errors. Part I: Dynamical Balance and Comparison with Observations

Abstract The characteristics of the initial corrections obtained from the Canadian Meteorological Centre (CMC) energy-norm-based key analysis error algorithm that minimizes short-range (24 h) forecast errors were investigated for four specific CMC operational analyses. The results show that both the rotational and the divergent components of the initial corrections are strongly out of balance. Some dispersive modes are also present in the mass component of the initial corrections. The results from one experiment where the initial state errors were known suggest that the current algorithm always selects a set of unbalanced initial corrections with more mass correction than wind correction, regardless of the characteristics of the real initial condition errors. Comparison with observational data showed that the corrected analysis is systematically farther away from the observations than the control analysis even in large forecast error events where most of the forecast errors are believed to have originated from errors in the initial state.

The Characteristics of Key Analysis Errors. Part II: The Importance of the PV Corrections and the Impact of Balance

Abstract This study examines a few approaches to isolate the balanced component of the initial corrections from the Canadian Meteorological Centre energy-norm-based key analysis error algorithm, in an attempt to capture the part of the key analysis errors responsible for short-range forecast errors. The best results were obtained with the nonlinear balance potential vorticity (PV) inversion technique. It was shown that the PV component of the initial corrections contains the essential information for reducing short-range forecast errors. The remaining imbalance part of the initial corrections does not grow in time and does not contribute to the improvement of the forecast. The removal of the imbalance part of the initial corrections makes the corrected analysis slightly closer to the observations, but remains systematically farther away as compared with the original analysis. Thus the balanced part of the key analysis errors cannot justifiably be associated to analysis errors. A methodology to balance the divergent part of the initial corrections, which reduces significantly the spinup in the vertical motion corrections, is also presented. Finally, in light of the results presented in this paper, some recommendations to improve the key analysis error algorithm are proposed.

Daily Forecast Skill of Multi-Center Grand Ensemble

In this study, we investigate the daily forecast skill of Multi-Center Grand Ensemble (MCGE), consisting of three operational medium-range ensemble forecast data by the Japan Meteorological Agency (JMA), the National Centers for Environmental Prediction (NCEP), and the Canadian Meteorological Center (CMC). The skill is evaluated by comparison among the daily RMSE of ensemble mean forecasts for 500 hPa geopotential height over the Northern Hemisphere (20°N-90°N) from August 2005 to February 2006.It is found that MCGE with the same ensemble size as that of the JMA ensemble is more skillful than JMA ensemble for about 75% in frequency both in autumn and winter. Reduction of error with MCGE has little dependence on the atmospheric flow. The RMSE of MCGE can be reduced up to about 20% whether the atmospheric field is easily-predictable or not. Even for the case that MCGE is not more skillful than JMA, the RMSE is increased at most 10%. We argue that the major benefit of MCGE is to avoid the poorest forecast.

One-Dimensional Variational Data Assimilation of SSM/I Observations in Rainy Atmospheres at MSC

Abstract Currently, satellite radiances in the Canadian Meteorological Centre operational data assimilation system are only assimilated in clear skies. A two-step method, developed at the European Centre for Medium-Range Weather Forecasts, is considered to assimilate Special Sensor Microwave Imager (SSM/I) observations in rainy atmospheres. The first step consists of a one-dimensional variational data assimilation (1DVAR) method. Model temperature and humidity profiles are adjusted by assimilating either SSM/I brightness temperatures or retrieved surface rain rates (derived from SSM/I brightness temperatures). In the second step, 1DVAR column-integrated water vapor analyses are assimilated in four-dimensional variational data assimilation (4DVAR). At the Meteorological Service of Canada, such a 1DVAR assimilation system has been developed. Model profiles are obtained from a research version of the Global Environmental Multi-Scale model. Several issues raised while developing the 1DVAR system are addressed. The impact of the size of the observation error is studied when brightness temperatures are assimilated. For two case studies, analyses are derived when either surface rain rate or brightness temperatures are assimilated. Differences in the analyzed fields between these configurations are discussed and shortcomings of each approach are identified. Results of sensitivity studies are also provided. First the impact of observation error correlation between channels is investigated. Second, the size of the background temperature error is varied to assess its impact on the analyzed column-integrated water vapor. Third, the importance of each moist physical scheme is investigated. Finally, the portability of moist physical schemes specifically developed for data assimilation is discussed.

Forecasting storm surges along the east coast of Canada and the North‐Eastern United States: The storm of 21 January 2000

A powerful storm passed over the coastal waters of eastern Canada on the 21 and 22 January 2000 causing significant damage to coastal infrastructure. The storm generated a large (>1.4 m) storm surge in the southern Gulf of St. Lawrence that unfortunately coincided with a high spring tide. This resulted in record high water levels in the southern Gulf of St. Lawrence (e.g., the highest level at Charlottetown since records began in 1911) and severe flooding around Prince Edward Island and along the eastern shore of New Brunswick. During January 2000, a recently developed storm surge forecast system was running in pre‐operational mode at Dalhousie University. The core of the forecast system is a depth‐averaged, non‐linear, barotropic ocean model driven by forecast winds and air pressures produced by the Canadian Meteorological Centre's regional atmospheric forecast model. In this study we assess the forecast skill of the surge model for the 21 January storm by comparing its 24‐hour forecasts with two independent hourly dataseis: (i) sea levels recorded by 12 tide gauges located in eastern Canada and the north‐eastern United States, and (ii) depth‐mean currents recorded by an acoustic Doppler current profiler deployed on the outer Scotian Shelf. Overall, the forecasts of coastal sea level and depth‐mean currents are reasonable and have forecast errors below about 0.1 m and 0.1 m s−1 respectively.

The 15‐km version of the Canadian regional forecast system

A new mesoscale version of the regional forecast system became operational at the Canadian Meteorological Centre on 18 May 2004. The main changes to the regional modelling system include an increase in both the horizontal and vertical resolutions (15‐km horizontal resolution and 58 vertical levels instead of 24‐km resolution and 28 levels) as well as major upgrades to the physics package. The latter consist of a new condensation package, with an improved formulation of the cloudy boundary layer, a new shallow convection scheme based on a Kuo‐type closure, and the Kain and Fritsch deep convection scheme, together with a subgrid‐scale orography parametrization scheme to represent gravity wave drag and low‐level blocking effects. The new forecast system also includes a few changes to the regional data assimilation such as additional radiance data from satellites. Objective verifications using a series of cases and parallel runs, along with subjective evaluations by CMC meteorologists, indicate significantly improved performance using the new 15‐km resolution forecast system. We can conclude from these verifications that the model exhibits a marked reduction in errors, improved predictability by about 12 hours, better forecasts of precipitation, a significant reduction in the spin‐up time, and a different implicit‐explicit partitioning of precipitation. A number of other features include: sharper precipitation patterns, better representation of trace precipitation, and general improvements of deepening lows and hurricanes. In mountainous regions, several aspects are better represented due to combined higher‐resolution orography and the low‐level blocking term.

Spatial Forecasts of Maximum Hail Size Using Prognostic Model Soundings and HAILCAST

Abstract Forecasting the occurrence of hail and the maximum hail size is a challenging problem. This paper investigates the feasibility of producing maps of the forecast maximum hail size over the Canadian prairies using 12-h prognostic soundings from an operational NWP model as input for a numerical hail growth model. Specifically, the Global Environmental Multiscale model run by the Canadian Meteorological Center is used to provide the initial data for the HAILCAST model on a 0.5° × 0.5° grid. Maps depicting maximum hail size for the Canadian prairies are generated for 0000 UTC for each day from 1 June to 31 August 2000. The forecast hail-size maps are compared with radar-derived vertically integrated liquid data over southern Alberta and surface hail reports. Verification statistics suggest that the forecast technique is skillful at identifying the occurrence of a hail day versus no-hail day up to 12 h in advance. The technique is also skillful at predicting the main threat areas. The maximum diameter of the hailstones is generally forecast quite accurately when compared with surface observations. However, the technique displays limited skill when forecasting the distribution of hail on a small spatial scale.

Multi-Center Grand Ensemble using Three Operational Ensemble Forecasts

In this study, we investigate the impact of Multi-Center Grand Ensemble (MCGE) forecasts, consisting of three operational ensemble forecasts by the Japan Meteorological Agency (JMA), the National Centers for Environmental Prediction, and the Canadian Meteorological Center. We verified the skill of MCGE forecasts in comparison with that of JMA ensemble forecast using root mean square error, anomaly correlation, and Brier skill score for 500 hPa geopotential height and 850 hPa temperature in the Northern Hemisphere (20°N-90°N) in September 2005.Our results show that MCGE forecasts are more skillful than single-center ensemble forecast without considering weight among ensemble members and bias corrections. This implies that considering weight or bias corrections may result in further improvement of MCGE forecasts, specifically in probabilistic forecasts.

Warm Season Lightning Probability Prediction for Canada and the Northern United States

Abstract Statistical models valid May–September were developed to predict the probability of lightning in 3-h intervals using observations from the North American Lightning Detection Network and predictors derived from Global Environmental Multiscale (GEM) model output at the Canadian Meteorological Centre. Models were built with pooled data from the years 2000–01 using tree-structured regression. Error reduction by most models was about 0.4–0.7 of initial predictand variance. Many predictors were required to model lightning occurrence for this large area. Highest ranked overall were the Showalter index, mean sea level pressure, and troposphere precipitable water. Three-hour changes of 500-hPa geopotential height, 500–1000-hPa thickness, and MSL pressure were highly ranked in most areas. The 3-h average of most predictors was more important than the mean or maximum (minimum where appropriate). Several predictors outranked CAPE, indicating it must appear with other predictors for successful statistical lightning prediction models. Results presented herein demonstrate that tree-structured regression is a viable method for building statistical models to forecast lightning probability. Real-time forecasts in 3-h intervals to 45–48 h were made in 2003 and 2004. The 2003 verification suggests a hybrid forecast based on a mixture of maximum and mean forecast probabilities in a radius around a grid point and on monthly climatology will improve accuracy. The 2004 verification shows that the hybrid forecasts had positive skill with respect to a reference forecast and performed better than forecasts defined by either the mean or maximum probability at most times. This was achieved even though an increase of resolution and change of convective parameterization scheme were made to the GEM model in May 2004.

Evaluation of new estimates of background‐ and observation‐error covariances for variational assimilation

AbstractIn this study new estimates for the observation‐ and background‐error covariances in a three‐dimensional variational analysis system at the Canadian Meteorological Centre are evaluated. In the current system, the observation errors are assumed uncorrelated with the variances estimated following a more or less ad hoc approach for each instrument type. The background‐error covariances are computed using the so‐called NMC method. The new estimate for the observation error variances is obtained using a practical approach developed at Météo‐France that computes the maximum likelihood estimate of the variances. Two new estimates for the stationary background‐error covariances are evaluated. The first simply involves tuning the variances in the operational background‐error covariances. For the second, the NMC method is replaced by a Monte Carlo approach applied to the existing analysis system.Both the variances and spatial covariances of the innovations are computed and compared with the corresponding quantities derived from each set of error covariances used by the analysis system. This comparison shows a significantly improved consistency for radiosonde data when using the new error covariances, especially with the Monte Carlo approach. In contrast, the interchannel innovation covariances for ATOVS AMSU‐A observations are more consistent when using background‐error covariances obtained with the NMC method. In addition, modest forecast improvements are obtained by using the new observation‐ and background‐error covariance estimates, most notably for the tropics. © Crown copyright, 2005