H. Ritchie and G. Brunet

• Co-investigators:
  
  1. G. Brunet (RPN)
     
  2. J. Fyfe (CCCMA)
  
  
  
• Collaborator: B. Dugas (RPN)

A new improved Empirical Normal Mode (ENM) analysis software package has been developed by G. Brunet and B. Dugas. It is currently being used to study tropospheric-stratospheric NCEP and UKMO analyses and primitive equation model output (Princeton/SKYHI and RPN/MC2 models - see conference list). A set of programs has been developed to perform ENM analyses and should be made available with a documentation booklet during the coming year.

2. Sub-project: Diagnostics and predictions of the low-frequency variability

• Co-investigators:
  
  1. R. Wang (RPN)
     
  2. G. Brunet (RPN)
     
  3. B. Dugas (RPN)
     
     
• Collaborators:
  
  1. J. Derome (McGill)
     
  2. S. Edouard (LMD, Paris)
     
  3. J. Fyfe (CCCMA)
     
  4. G. Plaut (INL, Nice)
     
  5. H. Ritchie (RPN)
     
  6. H. Sheng (RPN)
     
  7. R. Vautard (LMD, Paris)
  
  
  

This ENM diagnostic project has been in part refocused on seasonal predictions to meet Dr. Jacques Derome's historical forecast project (HFP) objectives. The dedicated PDF on this sub-project Dr. Wang, has then redirected his efforts to the HFP in place of the ENM diagnostic project which is now G. Brunet's and B. Dugas' responsibility.

A. Empirical Normal Mode Diagnostics

A predictability study using a shallow-water model has shown the advantage of using ENMs as a statistical basis to predict the dynamics of the flow with an auto-regressive model for each principal component (Brunet and Vautard, 1996). We are now extending this study to more complex time series, i.e. NCEP re-analyses and primitive equation model integrations. Progress on these ENM analyses has been slowed due to the reallocation of R. Wang to the HFP but interesting results have still been obtained with the NCEP re-analyses and an article is currently in preparation (Brunet et al., 1997). In preparation for our ENM diagnostics of the low-frequency variability with potential vorticity maps, we have completed a three-dimensional climatology of potential vorticity in isentropic coordinates and the results have been accepted for publication in QJMRS (Edouard et al., 1997).

B. Historical Forecast Project

Seasonal prediction has become one of the major topics under the CLIVAR GOALS project. Efforts devoted to developing numerical forecasting schemes based on ensemble integrations of GCMs have led to significant progress in seasonal predictions, though there is still a long way to go before useful forecasting can be achieved with this method. On the other hand, empirical models with dynamical insights derived from observational studies and GCM experiments can often produce useful forecasts, not only for operational purposes but also for theoretical investigations in the sense that they provide a benchmark for potential users and GCM approaches.

It is widely accepted that the atmospheric dynamics is typically a complex of regularity and randomness that unfolds on a broad spectrum of space-time scales. The predictive skill of an empirical model is derived either from the internal dynamics (such as the observed low-frequency variability of the space-time coherent structures) or from some other slowly varying components (such as SST) which impact on the atmospheric processes. There are, therefore, many different schemes in the literature, each designed to utilize particular aspects of the predictive skills such as the persistence of the weather regimes, space-time teleconnections, linear regression with SSTs with a possible time lag, etc. Careful comparison of the different schemes may yield useful insight into the underlying system and provide further guidelines for improvement. Moreover, the available techniques are no longer restricted to classical tools due to the development of the dynamical system theory and the associated techniques for nonlinear time series analysis.

Along these lines, G. Brunet, G. Plaut, R. Vautard and R. Wang have done a preliminary analysis of the dynamical (RPN/SEF) seasonal predictions performed by H. Sheng and H. Ritchie. The approach was to compare the performance of the SEF integrations with two empirical models for monthly and seasonal forecasts over the Canadian region. A technical report (Vautard et al., 1996) is available.

We intend to redo this study for the North American area with the numerical integrations of the HFP (Brunet et al., 1997). The contribution of R. Wang (NSERC fellow) to this project is important. Much of the effort of the visiting fellow during the past year has been devoted to studying the literature, testing algorithms and preparing papers (Wang, 1997 and Wang et al., 1997). He has learned to use the software package provided by R. Vautard, including ST-PC analysis, auto and multivariate regressive models and the program to search for analogues. Further research activities will include customizing the routine and establishing a data bank for practical seasonal forecasting based mainly on the package provided by Vautard and testing the possibility of hybridization with the GCMs. He has collaborated also on the paper "Monthly and seasonal prediction of Canadian surface air temperature" submitted to Journal of Climate (Vautard et al., 1997). This work shows that the methodology promoted by our group is significantly better than all available empirical methods, including the CCA in its actual state.

The planned contributions of R. Wang to this project for the next fiscal 1997-1998 year are :

1. To compare systematically the design and performance of different prediction schemes with respect to predictive skills and their sources. The outcome will be a review paper that summarizes the history of seasonal prediction, its state of the art, still open questions and possible improvements by combination of different schemes and perhaps the incorporation of nonlinear methods.
   
   
2. The development of a nonlinear approach to seasonal predictions in collaboration with H. Van den Dool of the US National Centre for Environmental Prediction (NCEP).
   
   
3. Helping to establish a routine prediction scheme at RPN/CMC in collaboration with Vautard's group from LMD/CNRS, Paris, with the focus on forecasting the seasonal temperature and precipitation over Canada and the United States.
   
   
4. Participation in developing a prediction scheme on the basis of hybridization of empirical models and GCMs, in collaboration with G. Brunet.
   
   

• Co-investigators
  
  1. H. Sheng (McGill and RPN)
     
  2. G. Brunet (RPN)
     
  3. B. Dugas (RPN)
     
• Collaborator:
  
  1. J. Derome (McGill)

1. Sensitivity to horizontal resolution, semi-Lagrangian and Eulerian schemes:

   We previously reported that in order to assess the sensitivity of the low-frequency variability in seasonal forecasts to the horizontal resolution in semi-Lagrangian and Eulerian schemes, northern winter (December, January and February) simulations for ten years (the AMIP period, 1979 - 1988) have been performed using various combinations of model resolution and formulations in the operational Canadian global spectral forecast model. Comparisons have been performed between semi-Lagrangian (SL) versions with linear unaliasing and Eulerian (EU) versions with quadratic unaliasing running with triangular truncation on the same Gaussian grids, namely: SLT31(EUT21), SLT63(EUT42) and SLT95(EUT63). In the current reporting period an article entitled "Sensitivity of seasonal forecasts to horizontal resolution in semi-Lagrangian and Eulerian schemes" has been finalized and submitted to the Journal of Climate. An important improvement made to the article was the inclusion of a series of Student t-tests comparing the simulations with each other as well as with the NMC analyses. The benefit of increasing the horizontal resolution is reflected in the fact that the area with significant differences compared to the analyses reduces as the resolution increases and especially for fields that are strongly affected by the parameterization of physical processes. As one would expect from this result, a significant sensitivity to resolution is found in comparing the various simulations amongst themselves. A refinement of these conclusions is that, while the middle latitude dynamics may have largely converged in the T63 and T95 simulations, the tropical convergence zone convection parameterization is still undergoing changes with increasing resolution
   
   

2. Improved numerical algorithms for stratospheric modelling:
   
   

   One of the most evident deficiencies in the original RPN-AMIP simulation was the inadequate representation of the stratosphere. As previously reported, in preparation for a more detailed examination of the sensitivity of low-frequency variability to the vertical resolution and numerical treatment of the stratosphere, modifications have been made to introduce a hybrid vertical coordinate and to raise the model top to approximately the stratopause with an associated increase in the number of vertical levels. During the current reporting period we have continued testing hybrid configurations of the model with complete physical parameterizations and work is in progress to redo the RPN-AMIP simulation with this version of the model. An associated presentation has been submitted for the CMOS Congress in Saskatoon.

   This model will also be used to investigate the dynamical influence of the stratosphere on the low-frequency variability of the troposphere. A preparatory observational study has been completed based on the UK Meteorological Office UARS (Upper Atmospheric Research Satellite) data which extends from the earth's surface to the stratopause. A sudden stratospheric warming event during the winter of 1994-1995 has been analysed in detail, including analyses of zonal mean temperature, isentropic potential vorticity and Eliassen-Palm fluxes. Two stratospheric sudden warmings occurred in this northern hemisphere winter. Inspecting the time and space development of the 10 mb height fields reveals that a first anomalous amplification of wave number one took place in late December and early January during a minor warming. A second anomalous amplification of wave number 1 occurred in the middle of January concurrently with a minimum of wave number 2 as a precursor of a major warming in late January. This study was presented at the Workshop on Climate Variability, February 27 - March 1, 1997, in Victoria, B.C. In the near future we plan to simulate this event using the improved version of the model
   
   

3. Impact of alternate dynamical formulations:
   
   

   The previously mentioned improved version of the model also includes a dynamical modification which increases the accuracy of the time integration scheme and is expected to have an impact on the low-frequency variability via a more precise treatment of the planetary and synoptic scale waves. Even with its semi-Lagrangian semi-implicit time integration scheme, the time step of the three-time-level Canadian global spectral forecast model is limited by a stability constraint imposed by the explicit treatment of the Coriolis terms.

   This limitation can be removed by a three-dimensional extension of an implicit treatment that has been used in a previous spectral shallow water model. In addition to removing the stability constraint, an analysis and experiment with a shallow water model show that the implicit treatment has much smaller time truncation errors for all wave lengths in semi-Lagrangian spectral models. This treatment has also subsequently permitted an easy conversion of the three-time-level Canadian global spectral forecast model into a two-time-level one, almost doubling its efficiency (Ritchie and Beaudoin, 1997). Tests are in progress to assess the impact of these dynamical improvements in terms of simulating the atmosphere's low-frequency variability. Preliminary results indicate that the implicit treatment of the Coriolis term produces a striking improvement in seasonal simulations, particularly in the equatorial stratosphere (Ritchie and Sheng, 1997).
   
   

PUBLICATIONS

PUBLICATIONS IN PREPARATION:

REPORTS AND TECHNICAL NOTES:

ORAL PRESENTATIONS:

INVITED PRESENTATIONS:

FUTURE ORAL PRESENTATIONS:

RESEARCH PLAN FOR 1997-98

1. Compare systematically the design and performance of different prediction schemes with respect to predictive skills and their sources. The outcome will be a review paper that summarizes the history of seasonal prediction, its state of the art, still open questions and possible improvements by a combination of different schemes and perhaps the incorporation of nonlinear methods.
   
   
2. Develop a nonlinear approach to seasonal prediction in collaboration with H. Van den Dool of the US National Centre for Environmental Prediction (NCEP).
   
   
3. Helping to establish a routine prediction scheme at RPN/CMC in collaboration with Vautard's group from LMD/CNRS, Paris, with the focus on forecasting the seasonal temperature and precipitation over Canada and the United States.
   
   
4. Participate in developing a prediction scheme on the basis of hybridization of empirical models and GCMs, in collaboration with G. Brunet.
   
   

   Dr. Hua Sheng will continue to work with Dr. H. Ritchie and J. Derome of McGill University on the influence of the stratosphere on low-frequency variability. Dr. Sheng's funding is requested by J. Derome at McGill University.
   
   return to Network Page
   
   return to Home Page