WORLD METEOROLOGICAL ORGANIZATION
COMMISSION FOR ATMOSPHERIC SCIENCES
MANAGEMENT group, second session
Oslo, Norway, 24-26 september 2007
(14. VIII. 2007)
REPORT FROM THE CAS/JSC WGNE
(Submitted by Martin Miller, chair of the WGNE)
The following text briefly review the main activities of WGNE in support of CAS objectives, emphasizing items arising at its twenty second session which was kindly hosted by the National Centre for Atmospheric Research, Boulder, Colorado USA, 24 -27 October 2006. On this occasion the 24 October was a joint session with the WMP. Emphasis has been given to activities where international coordination is paramount and facilitated by the working group’s existence, or where new scientific initiatives are involved.
1. Role of WGNE in support of CAS and WCRP
WGNE, as a joint working group of CAS and the JSC/WCRP, has the basic responsibility of fostering the development of atmospheric models for use in weather prediction and climate studies on all space and timescales. In the WCRP, WGNE is at the core of the global modelling effort and co-ordination between WGNE, WGCM and WGSIP is maintained primarily through ex officio meeting attendances. WGNE also works in close conjunction with the WCRP Global Energy and Water Cycle Experiment (GEWEX) particularly in the development of atmospheric model parametrizations, with WGNE sessions held jointly with the GMPP (but not in 2006). The WGNE Chair attends the new JSC of the WWRP, the CAS Management Group and the THORPEX ICSC, he is also a member of the WCRP/WMP, with WGNE represented on WOAP also. WGNE also has specific THORPEX sessions at its meetings. The close relationship that exists between WGNE and operational (NWP) centres underpins many of the activities of WGNE, and it is the work of these centres that provides much of the impetus for the development and refinement of the physics and dynamics of atmospheric models.
2. Studies and comparisons of atmospheric model simulations
Model inter-comparison exercises are a key element in meeting a basic WGNE objective of identifying errors in atmospheric models, appreciating their causes and reducing or eliminating these errors.
PCMDI, CMIP and a Workshop on Model systematic errors
WGNE congratulated PCMDI for continuing to maintain and enhance a valuable infrastructure for processing model outputs at PCMDI and establishing efficient data formats etc for such exchanges of model simulations. The recent outstanding achievements in the context of the IPCC/AR4 are of particular note. PCMDI has offered to receive high resolution NWP AMIP-type runs to complement their ongoing CMIP activities.
PCMDI was the local host for a pan-WCRP/CAS workshop on Model systematic errors in February 2007 attended by about 170 people. This was organized by PCMDI and WGNE with input from WGCM and GMPP, and the programme was structured by timescales to emphasis the ‘seamlessness’ of many model errors. A full report on this very successful meeting is available.
The goal of the WGNE-Transpose AMIP is to obtain the benefits for climate model development and evaluation that have been invaluable for weather prediction model development, by applying climate models to weather forecasting. The method allows direct comparison of parametrized variables such as clouds and precipitation with synoptic observations, satellite and field programmes. In general, development of a complete analysis system is not needed with initial conditions obtained from NWP (re-)analyses. The method allows direct comparison of parametrized variables such as clouds and precipitation with observations including field programmes (such as ARM), early in the forecast while the model state is still near that of the real atmosphere. This WGNE initiative was prototyped jointly by PCMDI and NCAR and known as CAPT. The intention is to encourage climate modelling groups to implement this forecast strategy into their development process. The formal proposal for Transpose AMIP has been sent to climate modelling groups.
Aqua-Planet Experiments (APE)
WGNE recognizes the value of applying atmospheric models to simplified surface conditions for examining the behaviour of physical parametrizations and the interactions of parametrizations with the dynamical cores. In particular, "aqua-planet" experiments with a basic sea surface temperature distribution offer a useful vehicle in this regard. The details of the experiment and schedule are available at http://www.met.reading.ac.uk/~mike/APE. The experiment is designed to provide a benchmark of current model behaviour and to stimulate research to understand differences arising from: (1) different subgrid-scale parametrization suites, (2) different dynamical cores, and (3) different methods of coupling model dynamics and parametrizations. Using the APE database, analysis of the APE experiments is continuing for another year. Following the workshop held in April 2005, a second workshop is planned to discuss the more complete analyses in late 2007 at the University of Tokyo.
The basic experiments are deliberately done at "climate model" resolutions but a few groups are examining convergence with resolution and the results and interpretation of resolution studies will be an important outcome of this work.
Regional Climate Modelling
Following the WMO/WCRP sponsored RCM Workshop in Lund, Sweden in 2004, the Transferability Working Group (TWG) was created. The aims of this group are to assess the global applicability of RCMs in regions remote from their home domain of development. Particular emphasis is being placed on the simulation of regional scale water and energy cycles in a wide variety of climatic regimes and the Inter Continental Scale Experiment Transferability Study (ICTS) focussing on GEWEX Continental Scale Experiment sites is in progress. A second RCM workshop is planned for 2008, probably in Trieste (ICTP). WGNE also discussed results from SGMIP (Stretched Grid Model Inter-comparison Project). It will continue to monitor the developments in this area in its future sessions.
3. Climate Model Metrics
WGNE has been involved in developing standard climate model diagnostics and metrics for some years. The goal of such metrics is to objectively measure model quality or skill and suitable metrics depend on the intended applications. The application for climate models includes the prediction of future climates for which no verification data will be available within the lifetime of the model. WGNE discussed the issue of climate model metrics at some length with many questions and issues resulting. A sub group with a member from each of PCMDI, WGCM, WGNE, GMPP and the JWGV (Joint Working Group on Verification) will define the climate model metrics and standard verification data sets with the intention of asking WCRP to encourage usage of these metrics for climate models. It was decided to ensure some emphasis on climate model metrics at the February 2007 model systematic errors workshop.
The need for good metrics for climate-type models is under discussion. WGNE will discuss this further also in the context of the new ‘unified’ prediction systems.
4. Physical parametrizations in models
WGNE’s close working relationship with GMPP (the GEWEX modelling and prediction panel), provides the focus for the development, refinement and evaluation of atmospheric model parametrizations, notably those of cloud and radiation, land surface processes and soil moisture, and the atmospheric boundary layer. WGNE reiterated the value of the interaction with GMPP for parametrization work, particularly with GCSS. A joint WGNE/GCSS model intercomparison study of a Pacific cross section (GPCI) to evaluate physical parametrizations along the atmospheric cross section following the trade winds is in progress, with excellent support from both NWP and climate modeling groups. The need for an expert group on parametrization to advise both WCRP and WWRP (and their Working Groups) was discussed, and further consideration is being given to this in consultation with the GMPP. (As an agenda item at the forthcoming WGNE/GMPP meeting in October)
5. Numerical weather prediction
Reanalysis projects and data assimilation
The ERA-40 reanalysis at ECMWF is complete and an “interim reanalysis” has begun. This is running from 1989 onwards. It contains improvements that greatly alleviate deficiencies identified in ERA-40. A comprehensive atlas of the atmospheric general circulation as depicted by ERA-40 has been produced in collaboration with the Meteorology Department of the University of Reading. The Japanese 25-year Reanalysis Project (JRA-25, 1979-2004) has been completed. WGNE reiterated its strong support for the reanalysis work, the desirability of maintaining a core of experts without excessive duplication of effort and ensuring efficient phasing of these efforts.
Earth System assimilation
The new developments in the assimilation of parameters pertinent to the Earth System but not routinely analysed by current data assimilation systems are being monitored by WGNE. These include analyses of greenhouse gases, aerosols and reactive gases. Earth system science such as the GEMS (Global and regional Earth-system Monitoring using Satellite and in-situ data) project will increasingly demand cross-project liaison within WCRP and CAS.
WGNE noted the substantial improvements in the resolution of global and deep convection permitting forecast models in progress or planned in the next few years. There exists a dichotomy of opinion regarding the use and interpretation of grid-lengths of several kms for forecasting. These resolutions will become affordable for GCM use in the coming years, and the prospect of climate simulations with grids of order one kilometre is an issue of international activity and debate, and WGNE will continue to monitor such developments. Recent results showing the need for model resolutions of 100 kms or better to properly define the statistics of extra-tropical storm tracks were noted. This contrasts with typical climate model resolutions substantially poorer than this, a matter of serious concern to the group.
WGNE noted that plans for unified (coupled) forecast systems that will provide forecasts from days out to seasons, typically by progressively degrading the resolution with forecast range, will provide new opportunities for ensemble techniques, including initial perturbations, stochastic parametrizations and metrics, and bring even closer collaboration between the NWP and climate communities.
WGNE will contribute to the TFSP meeting in Barcelona, June, 2007 including a report on the Systematic Errors workshop.
Performance of the main global operational forecasting models
WGNE routinely reviews the skill of daily forecasts from a number of the main operational centres in terms of verification scores (such as anomaly correlation and root mean square error) for various fields at different lead times. For most centres, a distinct increase in skill continues.
With global models attaining much higher resolutions, and mesoscale models being routinely run at most operational centres, consideration is being given to additional skill scores to the conventional ones that are more appropriate for such resolutions. Furthermore there is an increasing requirement to provide measures of model performance for predicting weather elements and severe weather events. The joint WGNE/WWRP working group on verification (JWGC) is now considering this important subject.
There are a number of WGNE projects involved with the validation of forecasts. New developments were discussed including the development of methods to verify high resolution spatial forecasts; verification methods for rare events; incorporation of scaling methods into verification processes; approaches to account for observational uncertainty in verification measures and analyses; development of methods that are customer dependent and appropriate for studies of forecast value; and verification of probability distribution functions.
Inter-comparison of typhoon track forecasts
The inter-comparison of forecasts of typhoon tracks has been an ongoing project that has been conducted by the Japan Meteorological Agency on behalf of WGNE for a number of years. This now includes all ocean basins, and data from operational forecasts is now available from eight Centres. The overall gradually improving performance of these models in predicting cyclone tracks over the past few years has been maintained. In future statistics will be gathered to assess the skill in intensity forecasts and forecasts of cyclone genesis. Many results related to typhoon track forecast including a multi-model ensemble are presented on the web site
(http://nwp-verif.kishou.go.jp/wgne_tc/index.html (user id and password are required)).
Verification and inter-comparison of precipitation forecasts
This WGNE initiative is being conducted at the DWD, NCEP, BMRC, CMA, JMA, CMC, the Met Office and Meteo-France. Quantitative global precipitation forecasts from the above are being verified against surface stations in these relatively data rich areas (some Centres also include their limited area model forecasts in the verification). A series of scores such as bias, Heike skill score, equitable threat score are used. It was noted that there is clear evidence from several Centres that the skill of precipitation forecasts in mid-latitudes was increasing.
Model-derived estimates of ocean-atmosphere fluxes (SURFA)
SURFA will evaluate and inter-compare global surface flux products (over ocean and land) from the operational products of a number of the main NWP centres and this will provide a good opportunity for estimating and determining the quality of model surface fluxes, of considerable relevance to atmospheric and coupled modelling communities and oceanographers. Following a joint session at WGNE-22 with the WCRP Working Group on Surface Fluxes (WGSF) it has been agreed to revitalize SURFA, and an agreed set of NWP fields etc will be routinely archived at the National Climate Data Centre from a number of NWP Centres (after a preliminary pilot study currently in progress).
The WGNE ‘Blue book’
A key WGNE publication for many years has been the WGNE "blue cover" numerical experimentation report series which continues to be popular with the modelling community. and is prepared on behalf of WGNE by Recherche en Prevision Numerique (RPN), Montreal since its inception, and the latest annual summary of research activities in atmospheric and oceanic modelling (No. 36) has been released. This publication is facilitated by use of e-mail contact and the website at RPN, (www.cmc.ec.gc.ca/rpn/wgne)
At the 22nd WGNE meeting there was a session which reviewed the status and plans of THORPEX and the wide-ranging opportunities for collaboration and synergy with WCRP and other bodies. The plans for T-PARC were of particular note, and this ‘campaign’ promises to make a major contribution to our understanding of meteorology in the Pacific basin.
The use of ensemble methods now forms a cornerstone of forecasting on all timescales, and WGNE hoped that the rapidly progressing TIGGE project will help accelerate the effective use of ensemble forecasting information.
7. A Year of Tropical Convection
WGNE discussed the proposal for ‘A Year of Tropical Convection’ (YOTC) which as currently envisaged, is aiming to assemble a dataset that will enable focussed research on many aspects of tropical convection, which in turn should lead to significant/important advances in our NWP abilities on all timescales currently labelled under 'seamless' prediction. The discussions strongly supported the idea but felt that it was less clear how the aims of the YOTC would be achieved. Some concern was also expressed that the proposed timescales were somewhat too tight.
As this YOTC dataset will be a judicious combination of many existing datasets in a variety of forms and repositories, questions were asked as to whether this is an opportunity to harness the powers of the new WMO Information System (WIS), and what was the YOTC relationship to other planned 'global' activities such as IPY and a possible Monsoon' focus. It was suggested that WWRP and WCRP should consider these questions and the efficacy of having a working group and/or a workshop in 2007. (Meeting now planned for November)
Recognizing that convection is central to many problems in modelling research on almost all space and time scales, WGNE/GMPP were already jointly considering a high resolution modelling experiment specifically directed towards aiding and accelerating parametrization development. This could be part of a coordinated effort to benefit the entire community.