WORLD METEOROLOGICAL ORGANIZATION

Pretoria, South Africa, 8-12 November 1999

1. OPENING OF THE MEETING (agenda item 1)

1.1 The meeting of the Implementation Co-ordination Team on Data Processing and Forecasting Systems chaired by Angèle Simard (Canada) was at the kind invitation of the South African Weather Bureau (SAWB) held at the Forum Building in Pretoria, SA, from 8 to 12 November 1999. Dr Winifred Jordaan welcomed the participants, indicated facilities available for the meeting and gave an overview of the SAWB facilities housed within the Forum Building. She gave a brief overview of the research and operational forecasting activities at the Bureau and indicated the various SAWB researches, operations and long range forecasting groups who will participate in the meeting. Dr Jordaan informed the meeting planned social functions and wished the meeting success in its deliberations.

1.2 Mr Morrison Mlaki, Chief Data Processing Division, WWW Department of WMO, on behalf of the WMO Secretariat, thanked and expressed appreciation to Mr G. Schulze, the Permanent Representative of South Africa with WMO and through him the SAWB for hosting the meeting, and welcomed the participants. He recalled that CBS-Ext. (98) in restructuring it working modalities established four Open Programme Area Groups (OPAGs) on the activities of CBS grouped under four main programme areas: (a) Integrated Observing System; (b) Information Systems and Services; (c) Data-processing and Forecasting Systems and (d) Public Weather Services.

1.3 Mr Mlaki pointed out that the three major issues before the meeting in accordance with the terms of reference and deliverables of the team relate to:

• Identification of Members' emerging requirements for applications and use of GDPS outputs (including atmospheric transport products, long-range forecasts, severe weather, ensemble forecasting);

• determine how GDPS centres can best contribute to fulfil emerging requirements;

• identify needs for workshop/ training.

1.4 Mr Mlaki further noted that with respect to long range forecasting infrastructure issues, just before Cg-XIII, an Initial Planning Meeting on Infrastructure for Seasonal to Interannual Prediction was held in Geneva. The initial meeting recommendations were subsequently endorsed by Cg-XIII. Cg-XIII took note of among others, agreement that in the SI arena the components of the activities that clearly fit within the World Weather Watch/Global Data-processing System framework are the Climate Monitoring and Diagnostic Activities and the Objective Numerical Prediction Activities. Several options were considered regarding designation of SI centres within the GDPS framework and a preference expressed for the option to utilise the present RSMC designation and when appropriate create RSMCs with Climate Monitoring and Prediction Specialization. The WCP/WCASP/CLIPS will focus on the development of interface procedures and methods to provide the international mechanism to carry out, based on GDPS framework global and regional models output, the applications driven and downscaled SI Forecast and Outlook function. The NMHSs are the crucial players at the national level and CLIPS' primary task will be to develop their capacity.

1.5 Mr Mlaki noted that as the agenda of the meeting suggest, the focus of this meeting is to review current activities and on the basis of new emerging requirements, propose recommendations on best response strategies to fulfil the requirements. It will also address related workshops and training needs and develop implementation action plans to address the requirements. He to paid tribute to all contributors to documentation for the meeting for their contributions and a wealth of recommendations, which highly facilitated the work of the meeting. He especially thanked contributors who could neither attend nor be represented. These includes Dr Tracton et al (NCEP), Dr H.E. Brooks et al, (NOAA), Ms Hui Sian TAN (Singapore), Dr K Prasad (India), and Ms Lynda Jones (UK).

2.1.1 The meeting adopted the agenda given in Appendix I.

2.2.1 The meeting agreed on its working hours, mechanism and work schedule.

2.2.2 There were 20 participants at the meeting as indicated in the list of participants given in Appendix II.

3.1        Ensemble forecasting methodology and progress

3.1.1 The first two centres to develop ensemble forecasting techniques were NCEP and ECMWF in the early 1990s. Significant progress has been made over recent years in improving these techniques, developing new approach (Canada), transferring technology to other centres (SAWB system based on NCEP), and developing a variety of products than can be used operationally. For example Météo-France, and several European countries, are using ECMWF ensemble forecast as input to their worded medium-range forecast. The meeting was informed of the activities taking place in Canada, South African Weather Bureau, NCEP and Météo-France.

3.1.2 Different methodologies are being used. The method used at the Canadian Meteorological Centre (CMC) consists of running independent assimilation cycles that use perturbed sets of observations. The first guess is obtained from 8 different versions of a T95 spectral model that are mainly different in their physical parameterization. The size of the perturbed analysis thus produced is doubled by taking opposite pairs. The system consists of 16 members run to 10days every day. The output has been used to generate several products such as spaghetti plots of the 500hPA heights, probability of precipitation charts for various thresholds (2, 5,10, 25mm and more), day to day temperature forecasts, 10-day mean temperature. The spaghetti plots show the 570dam contour lines for each members of the ensemble on the same graph. The probability of precipitation for a particular threshold is function of the number of members forecasting a precipitation amount above that threshold. Verifications are made to assess the skill of products. It is found that the RMS error for the ensemble mean is smaller than the RMS error for the high-resolution model. The ROC score as approved by CBS-Ext. (98) is used for the verification of probability of precipitation. The ROC score shows more skill in the winter forecast than in the summer. This could be explained by insufficient resolution to properly capture the convective nature of summer precipitation. It is planned in the future to use probabilities obtained from the ensemble forecast system in the production of worded forecasts up to day 10.

3.1.3 The SAWB has adopted the two-week forecasting system from NCEP where a 16 member ensemble is created daily using the breeding of perturbations method. It runs for 7 days during the week but extend to 14 days over the weekend. Combining the bred ensemble from successive initial times using the lagged average forecasting technique, the ensemble suite can be increased to 43 members once a week. Updated two weeks forecast for this system are produced every Monday. Various forecast products include probability calculation and cluster analysis of forecast temperature, wind and cloud cover. An average of the largest populated cluster is used to produce the final forecast. Products include: probability of precipitation amount exceeding a threshold value based on the proportion of ensemble members that project the event, weighted forecast maximum and minimum temperatures from the largest cluster at each model grid-point are used to calculate regional values, wind speeds which are determined from an average of the cluster members, cloud cover based on the maximum percentage for three levels (low, middle and high) is given in three categories i.e. clear (below 25%), partly cloudy (25% to 75%) and cloudy (beyond 75%).

3.1.4 NCEP is using the breeding method. In this technique, perturbations that grew very fast during the analysis cycles leading time, and thus are likely to dominate errors, are used as initial ensemble perturbations. The breeding method is an attempt to capture the fastest-growing analysis errors that are most likely to be responsible for the error in the control forecast. NCEP currently has an ensemble of 17 forecasts out to 16 days. Of all the graphical products available to the forecaster, the most heavily used is the spaghetti diagram, where a single selected contour of a variable is plotted for each ensemble member. The ensemble mean field is also available to the forecasters. This field offers a forecast that is on average, better than the control (or any ensemble member) forecast in terms of RMS errors. At longer lead times, the small-scale features whose positions are uncertain are naturally filtered out from the ensemble mean, which is used as guidance for day 6-8 and beyond. Clusters that are formed of ensemble members with similar solutions based on an objective algorithm are also used. The current worded forecasts do not explicitly address the possibility of alternate scenarios.

3.1.5 NCEP applies the concept of ensemble forecasting exploited for medium range prediction to short range (0-3 days). The problems for generating useful SREF are potentially much more difficult. Beyond scientific concerns, practical issues relevant to an operational setting must be dealt with, such as trade-offs between model resolution and ensemble size, and development of user-friendly products developed from ensemble output. Results from a pilot project and from a prototype were encouraging. As a result, NCEP expects to run as SREF for approximately 6-12 months in a test and evaluation mode, probably beginning in the first quarter of 2000.

3.1.6 Météo-France products are based on the ECMWF system. ECMWF uses the single vector approach in which the fastest growing perturbations after 72 hours are determined. Météo-France medium range forecasts are made no 2-day periods. Forecasters in France have been associated with the evaluation of the system and participated in the development of methods to make use of it in the medium-range. The Ensemble Prediction System (EPS) currently has 51 members at a resolution of T159. Météo-France produces worded forecasts based on EPS for day-4 to 7 and includes some information about the uncertainty of the forecast.

3.1.7 All centres have plans to improve their technique, and develop new products. Ensemble technique approach is also used by several centres in the production of monthly forecasts and seasonal outlook.

3.2.1 The meeting noted with satisfaction the status of progress made in EER activities including in particular, areas and means of co-operation with CTBTO, exploring other and better means of transmitting the information to the NMSs that have the required technology (e-mail, www, ftp, etc.) although faxing would remain the official means of transmitting the information to the other Countries. It was noted that the list for Delegated Authorities and NMS operational contact continues to be regularly updated on line by Members and accessible from the WMO web site. A regulatory text on notification together with other updates addressing new requirements has been proposed for inclusion in the Manual on the GDPS in Appendix I-3. It was noted that a WMO/IAEA global exercise was planned for 2000 to test the new procedures and as an opportunity for NMSs to get some training and to demonstrate the role they could play in EER and that WMO would participate in other inter-agency exercises (INEX-2001) only if the WMO exercise objectives would be met.

3.2.2 The meeting noted that the ET on EER concluded that the CBS Ext. (98) guidance for response to chemical incidents supplemented by input on KNMI experience in setting up an emergency response organization on serious chemical accidents would suffice for Members’ response activities. This guidance will shortly be included in the Documentation on RSMC support for Environmental Emergency response (TD/No 778).

3.2.3 Recalling the recent nuclear incident in Japan, the meeting observed that no detailed procedures have yet been established to activate additional observation in the event of a nuclear accident although such additional observing networks have been adopted under the GOS programme. The meeting invited the ET on EER and related activities in collaboration with the OPAG on IOS to address this issue.

3.2.4 The ET on EER addressed the issue of products to made available and procedures as they relate to backup NMCs and RSMCs for meteorological support services to humanitarian missions. It was noted that WMO had already established a Web Server which contained links to the NMSs and the list of the above-mentioned contact points which has been implemented under the Public Weather Programme and that the RSMC back up access includes under the GDPS programme links to RSMCs web pages. RSMCs with geographical specialization and relevant activity specialization were invited to consider providing for their areas of responsibility the products specified in Appendix I-5 to the Manual on the GDPS to be accessed through their web site. NMSs with web facilities should also consider providing this service as an integrated part of its public weather service on their web site.

3.2.5 As regards the climate database to support humanitarian missions envisaged by CBS-Ext. (98), the ET on EER felt that it could not discuss constructively data content and possible access mechanisms as long as there was not a clearer requirement provided from UN/OCHA.

3.2.7 The meeting noted views expressed that action is required for designing arrangements and standards for EER products and methods of dissemination of non nuclear response products, on a pattern similar to Appendix II-7 of the Manual on GDPS (relating to nuclear emergency), despite the limitations associated with non-nuclear emergencies like haze from wide spread persistent forest fires, chemical incidents and volcanic ash for non aviation purpose. It noted that the ET on EER and related activities has already addressed this issue but requested that the issue be referred to the Chair of the ET for consideration as appropriate.

3.3.1 The meeting noted that among the conclusions of the Initial Planning Meeting on the Co-ordination of Infrastructure Needs for Seasonal to Inter-annual (SI) Climate Prediction held in Geneva, April 1999 was the fact that in the Sl arena the components of the activity that clearly fit within the WWW/GDPS framework are the Climate Monitoring and Diagnostic activities and the Objective Numerical Prediction activities. The SI Forecast and Outlook component is addressed in the framework of the World Climate Programme and its CLIPS project taking into account existing GDPS infrastructure and procedures. It was noted that Congress XIII generally endorsed the relevant conclusions of the initial planning meeting as recorded in Resolutions 2 (Cg XIII) – WWW Programme for 2000-2003 and 8 (Cg XIII)-CLIPS project.

3.3.2 The meeting noted with appreciation progress achieved in development of verification system for long-range forecasts including the ET on DVSLRF proposed recommendations on updates to the CBS adopted verification system. It noted the specified arrangements and procedures to operationally use and exchange verification scores including a mechanism for preparation of a specification document that will describe in detail the mechanism for verification of long-range forecasts and the calculation of all proposed scores including status of availability of data. It endorsed the ET recommendation to the President of CBS to designate a lead GDPS Centre on implementation and operational aspects of verification system for long-range forecasts.

3.3.3 The meeting took note and encouraged GDPS and other centres engaged in long range forecasting to actively participate in carrying out activities envisaged in the ET’s co-ordinated implementation action plan covering activities for the period from fourth quarter of 1999 with start of experimental exchange of LRF verification scores between all participating Centres and Institutes by the first quarter of 2001. For ease of reference the full report of the meeting of the Expert Team to Develop a Verification System for Long-range Forecasts can be accessed on the WMO web site at: http://www.wmo.ch/web/www/reports/ECMWF-AUG-99.html

3.4.1 The meeting was informed of the progress made by the expert team. The first meeting of the team is planned to be held in Toulouse, 9-11 March 2000, back to back with the COSNA workshop. The team has been asked for two deliverables initially:

• a report on how analyses of NWP verification can be used to assess positive and negative impact of changes to the GOS on the operation of the GDPS, particularly NWP,

• a set of guidelines to be used where action is required to minimise the impact of a loss of observations on the operation of the GDPS.

On the face of it, the use of NWP verification statistics to detect the impact of changes in the GOS would appear to be straightforward. Many Observing System Experiments (OSEs) or Observing System Simulation Experiments (OSSEs) have shown potential impact of new or improved observation types and have been used to justify the development or extension of observation platforms. Predictions based on the NCEP and ECMWF re-analyses show increasing skill as the observation base has improved. However, systematic retrospective use of verification statistics appears to be less common. Such assessments would be useful in confirming the anticipatory studies and demonstrating the real impact of both improvements and deteriorations in the observing system.

In summary, some factors making it difficult to use the verification statistics are:

• OSEs indicate that the change in observations must be significant in scale and geographical coverage for a signal to be detected; conclusive results cannot be obtained when the volume of observations is small,

• the performance of NWP systems is affected by other factors such as changes in the analysis and prediction components,

• there is significant natural variability in predictability of the atmosphere,

• the NWP centers differ in the types and quantity of observations used,

• there are still some differences in methods for computing the scores.

The following has been suggested as a general methodology:

(i) list planned, possible, and actual changes to the GOS, which we need to assess.

(ii) survey existing observation impact experiments, for results relating to the type of changes listed in (i).

(iii) apply our judgement to extrapolate the results from (ii), to predict the probable impact of changes in (i).

(iv) look at operational verification results before and after the changes, to see if they reflect (within normal variability) the changes predicted in (iii).

(b) A Case study has been carried out based on various measures taken by different centers following failure of the TOVS instruments on NOAA-11 on 26 February 1999. Preliminary results, based on the rms error of the 72 hour predictions of geopotential height computed from analysis, indicate that there is a significant increase in rms error from February to April for five of the seven centres represented in the study. The two centres that showed a decrease were using NOAA-15 data by April. These are preliminary results only. A more comprehensive analysis is required including other measures.

Further analysis of the scores and discussion among the Team are required to establish the usefulness of this approach for this case and to assess its potential for broader application.

3.4.2 The ET came up with some guidelines to minimize the impacts of the loss of data in a variety of situations such as the planned shutdown of Omega System on 30Sept 1997, the unplanned failure of satellite soundings such as NOAA-11, the reduction of network due to economic factors or Year 2000 and others. The guidelines are largely based on experience with the shutdown of the Omega system and the action taken on Year 2000 compliance. It addresses several aspects: alerting; assessment of the problem; problem prevention where possible; investigation of mitigation strategies; finding resources to ameliorate the problem; monitoring the problem; post-event review and administrative aspects. These guidelines, addressed to members, will be refined and clearly identify who does what.

• review what comes out of the ET on the redesign of GOS, examine its impact on the GDPS

• enhance communication among GDPS centres in relation to the use of observations in NWP and other systems, to assist in providing a prompt and pro-active response to any problems or developments

• provide an established group to respond to requests from CBS to investigate specific changes to the GOS which have international implications for the operation of the GDPS

• maintain a pool of expertise to provide advice from an operational setting on likely impacts of planned changes to the GOS

• continue to refine and enhance the exchange of NWP verification statistics

• preparing a review each year, identifying any impacts on the operations of the GDPS associated with changes to the GOS

• The change to the year 2000 may produce a loss of some observations that may warrant investigation by the team.

3.5.1 The meeting reviewed the current status of preparedness of WMO Data Processing and Forecasting Systems (DPFS) to the millennium change. It noted that meteorological data processing and weather forecasting systems include a number of sub-systems such as pre-processing of observations (ingestion of raw data, decoding, file transfers, copying, data monitoring, quality control and archiving), data objective analysis and assimilation, numerical weather prediction, post-processing of the NWP model output (digital representation, animation, dissemination and products archiving). Many problems exist in connection with forthcoming millennium change. They include assuring that the following will have been done:

• to check for compliance using hardware (PCs, WSs, servers, mainframes, supercomputers);

• to check for compliance local area networks;

• to check for compliance operating systems;

• to check for compliance home developed real-time data processing and forecasting systems;

• to asses compliance third party developed meteorological applications;

• to test on-line and off-line archives software;

• to provide staff training.

3.5.2 To achieve the goal of smooth transition lead data processing centres have developed comprehensive plans to address this issue. Key steps usually include:

• conduct of an inventory of all computer-based systems;

• definition of priorities;

• development of the schedule for resolving problems;

• identification of resources to effect solutions;

• contact the vendors which provided the meteorological application systems;

• test application even if vendors state compliant;

• modification and test home-developed software including scripts that run operational jobs;

• estimation of dependence real-time meteorological systems from the data that are not of real time, e.g. forecast validation systems, forecasting diagnostics, restoration scripts;

• test interactions between the systems;

• development of the contingency plans.

3.5.3 It is clear from the above mentioned that the Y2K is not only a technical problem but also a management and business problem than just a technical problem.

3.5.4 The meeting noted that Lead centres are making good progress in preparing their DPFSs to a smooth transition to the next millennium. Based on published information one can draw a conclusion that three WMCs - Washington, Melbourne and Moscow, most of RSMCs with geographical specialization and most of RSMCs with specialization in activity have compliant operational data processing and forecasting systems. Very large and successful work has been done at the WMC Washington, RSMCs Bracknell and ECMWF. A few RSMCs have not yet provided any information on the status of their Y2K activities.

3.5.5 As of 15 October 1999, WMO Secretariat has received information from 147 countries of the 185 WMO Member regarding the Y2K compliance in their NMHSs (see the WMO Web page http:/www.wmo.ch/web/www/y2k-info.html). Forty-four NMHSs have indicated that their systems are Y2K compliant. Fifty-four indicated that they have projects underway to ensure the compliance. Thirteen countries have requested the assistance. Eleven have responded, but did not give any indication of their Y2K status. The other non-responded 38 NMHSs are a source of concern.

3.5.6 It was noted that the Ad hoc meeting on the Year 2000 Problem developed the International WMO Y2K Monitoring and Contingency Plan. Plan is available on the WMO Web page http:/www.wmo.ch/web/www/y2k-info.html(Y2K-Monit-Cont-Plan.doc). This provides for real-time data monitoring by WMO lead centres for a few days surrounding 1 January 2000 to determine if significant outages of critical data occur. Each of the three WMO World Meteorological Centres and the two World Area Forecast Centres has agreed to act as a Y2K Situation Centre. The contingency plan includes:

1. actions to be taken before late December 1999 (telecommunication failures, loss of electric power, equipment failure, interruption on flow of data or products, etc.)

2. actions to be taken during the few days immediately before and after the new year (ensuring that members of local crisis management team are present or available on call, evaluation response to the change to the new year, etc.)

3. longer term follow-up actions (monitoring the situation around 29 February 2000, make data available via Internet password protected FTP, etc.)

3.5.7 As a backup for possible communication outages the NMCs Melbourne, Offenbach, Tokyo and Washington will make observational data received at their centres available via FTP in the few weeks surrounding 1 January 2000.

4.1 The meeting discussed user requirements, examined the best strategy to meet these requirements including ways to exchange and disseminate the Ensemble Prediction System (EPS) outputs, identified the needs for training, and proposed an action plan. It should be noted that the meeting is mainly concerned with medium-range forecasting.

4.2 Two types of requirements have been identified. There is a need for some centres to get access to the Ensemble members in GRIB format in order to develop their own products, others only need access to the end products. Some of these products are available on Internet. Large centres may have a more specific interest in exchanging their products in GRIB format to work into the direction of super-ensembles.

4.3 The NMHSs requirements are still not precise enough to enable this meeting to specify the list of products that could and should be made available. Different attempts to define requirements had limited success. It has therefore been recommended that the requirements should be refined on a regional basis, taking into account the different capabilities of NMHSs. As centres develop expertise in ensemble forecasting, it is expected that new products will be developed and requirements will be further refined.

4.4 In order to integrate EPS in their own forecasting activities, the NMHSs need as a first step to have a better knowledge of the products available and how to interpret them. In addition there is a need for producing centers to know more about other centers products. In particular there is a need for information from the producing centers on :

• Available products (areas, parameters, levels, lead times, format), and their skill where available.

• How to use and interpret some specific products (mean fields, products resulting from particular clustering techniques, fields or charts of probabilities, plumes, spaghetti plots).

• Future products that could be generated.

• How to access the products as well as some specifications on hardware and software needed to process or display these products.

4.5 Available products may differ from one center to the other. Some of these products are targeted on the EPS running centers zone of interest, and are therefore of a limited utility for other centers or NMHSs and their skill may have not been evaluated elsewhere. Evaluation will need to be done on the areas of the requesting NMHSs.

4.6 Despite these difficulties, existing products can at least be considered as demonstrative products of what EPS outputs are. In some cases supplementary products could be generated through a small adjustment of the post-processing programs. However centers producing EPS products will need feedback from NMHSs on the skill of these products. In some others, the amount of resources that would be needed to generate the requested product or to evaluate their quality, could be critical.

4.7 It was agreed that a basic list of EPS products to be distributed should include such products as:

• probability of precipitation,

• ensemble mean at 500 hPa,

• some indication of variability (e.g. spaghetti plots, spread).

Ranges to be covered are at least D4 (96 h) to D7 (168 h).

4.8 This should not be considered as an exclusive list. Other products may be made available upon request. It is expected that this will evolve as NMHSs gain experience in ensemble forecasting.

4.9 Considering the fact that the producing centers may have different views about the distribution of their products, the meeting recommends that as a first step the dissemination should be done on the basis of individual request from NMHSs to centres running EPS. These products could be made available on web sites or by FTP. This will allow NMHSs to familiarize themselves with the products while details for further exchange and dissemination of products are being worked out.

4.10 Recommendation:

1) Countries generating EPS products covering some parts of the world are encouraged to produce the same kind of results over other areas as requested.

2) Products shouldn’t be used operationally without an adequate previous evaluation. Evaluations should be delivered together with the products. The recipient centers should not redistribute the information to other end users until it has been evaluated.

3) The public should be informed of how to interpret operational forecast based on EPS products before they are being issued operationally. The public needs to understand that medium-range forecast do not have the same level of accuracy than short-range forecast, as well as to understand how to interpret probabilistic forecast and/or the level of confidence associated with EPS forecasts. It is recommended that this role be carried on by the OPAG on PWS.

4) Some difficulty arise from the fact that the skill of EPS may vary from one area to the other and that the resources or observations that would be needed to make the evaluations will not necessarily be available at the centers running the EPS.

Centers running the EPS should deliver at least basic evaluations of the gridded fields (skill scores like RMSS or ACC for the ensemble mean on standard domains already used for short range forecasts evaluation) while further evaluation needed on specific products or areas should be conducted by those NMHSs who are interested in it.

5) Centers requesting the products should also make their evaluation (ROC curves for instance) and give a feedback to producing centers. However some centers may be willing to make this evaluation for other countries.

It was underlined that feedback from NMHSs to producing centers would be of a great utility to help improve future products.

6) It was also noticed that the post-processing of EPS products, like the calibration of probabilities for specific locations, will also be needed in some cases and should also be conducted by the receiving centers. The centers producing EPS products should at least provide NMHSs guidance on how to calibrate the probabilities. In some cases part of the software to be used could also be made available.

7) Considering the variety of potential products and the total number of members in the EPS run by different centers, the problem of the dissemination of these products and the telecom impact on the GTS of this dissemination will have to be carefully studied when a more precise list of requirements will be defined.

4.11 The training should be made before NMHSs make use of the product to avoid misinterpretations. However, this training should not be done too early, in order to be most beneficial. Needs for training will depend on the use of EPS products. Some countries will need training to help interpret EPS outputs and build their own end-products, some others will need training for the use of EPS end products.

4.12 It was also noted that training will be needed on the skill of the forecast and on the best ways to deliver the medium-range forecasts based upon EPS products to the public. It must be acknowledged that medium-range forecasts cannot be presented in the same way as short range forecast. The public itself will have to be educated in order to clearly understand forecasts based on EPS.

Three types of training are proposed

• Regional WMO workshops to explain the EPS approach, its usefulness, and its limitations. It should concentrate on the products which are available. These workshops would be mainly useful for those who intend to use EPS end-products.

• Technical cooperation type of training for those who intend to make their own products and/or who will need more specific training about products or the methodology of the forecast. Training could be organized on individual request or through WMO voluntary cooperation arrangements.

• Workshops or seminars developed by Centres running EPS. These centers are encouraged to open them to a wide range of participants. Co-sponsorship with WMO should be considered.

• It is proposed that the CBS newsletter be used as a mean to publicise web sites where information on ensemble forecasting and products are available.

• As a first step, and starting immediately, it is proposed that requirements should be refined by regional associations and that initial individual request from NMHSs and interested EPS running centres be acted upon. Refining the requirement could be done under the form of questionnaires but also through meetings.

• It is recommended that WMO plans regional workshops with focus on EPS training.

• It is recommended that CBS creates under the OPAG on DPFS an expert team on EPS to :

1. Review the list of fields and products that should be distributed and could meet regional refined user requirements.

2. Propose an update to the manual on the GDPS(Appendix II-6) concerning the list of output products available for international exchange and dissemination.

3. Establish procedures for exchange of EPS GRIB data, including the needs of large centres to exchange their ensemble.

4. Study the appropriate means for the dissemination of these products, including the possible use of GTS, FTP, Internet.

5. Help evaluate the telecommunication impact of the dissemination.

This team should include members from each regional association, users of EPS and a specialist in telecom .

5.1 Recognizing the need for severe weather guidance, CBS-Ext.(98)approved an extension of NWP output to provide more explicit guidance on the occurrence of severe weather. The team reviewed user requirements, proposed recommendations to best fulfil these requirements, identified training needs and developed an action plan. The team recognises that the problem of nowcasting and forecasting of severe weather is much wider than Numerical Weather Prediction but decided to focus its deliberations to fulfil the above request from CBS as endorsed by CG-XIII.

5.2 An excellent paper submitted by H. E. Brooks et. al (USA) was presented. It reviews severe weather associated with thunderstorms (tornadoes, hail, high winds, and flash floods). It notes that the declining death rates due to severe weather in North America is associated with improved forecasting and detection of events, improved awareness in the general public, and improved communication of meteorological and safety information.

5.3 It was stressed again that the definition of severe weather events differ in various parts of the world. Forecasting of most of the severe weather events also depends heavily on the local knowledge of the climate and geography.

5.4 Although the definition of severe weather varies, the severe weather terms can be put forward and the meeting did group these events into two categories. Category number one concerns severe events that are quite common right around the world and number two are events that are more localized. It is acknowledged that the thresholds of the general category will differ from country to country and no attempt was made to generalise it. These thresholds could be established by the National weather centres themselves.

5.5 General severe weather include events like:

Heavy rain
Strong wind/ wind gusts
Hail
Lightning
tornadoes
flash floods
extreme temperature
.....

5.6 The more localized events are suggested as:

Snow storms
dust/sand storms
sea swell/ tsunamis/ storm surge
extended area of fog for transport (aviation especially)
.....

5.7 Designated weather centres already deliver services like tropical cyclone forecasts and the meeting suggested that guidance should be sought from these centres. Also, the meeting did not discuss the problem of volcanic ash.

5.8 As local knowledge is extremely important for category number 2 (localized events), it was decided to concentrate on category one (general severe weather) as this is the area where the best guidance can be given, based on scientific knowledge.

5.9 Different user requirements were identified, recognising that different levels of expertise and capabilities exist due to economic limitations.

5.10 The meeting concentrated on centres needing assistance in severe weather forecasting. Although most of the comments below are concentrated on Numerical Weather Predictions (NWP) products it is acknowledged that data and Nowcasting are important.

5.11 User requirements will vary depending on the expertise and capabilities of different centres. These requirements are defined according to three categories:

1)    Centres that have the capabilities to set up their own limited NWP suite but need help in achieving this status. These centres should also enhance their own observational network to complement the NWP.

2)    Centres that can receive NWP Grid Point Value (GPV) data and have the means or capabilities to develop their own post- processing.

3) Centres that do not have the means to receive NWP GPV products but that can only receive either fax, satellite based receivers or data from a very slow speed line. These centres do not have the ability to do post processing as they have only static displays. Some of these centres have Internet and ftp facilities. However for severe weather forecasting, where real-time dissemination of information is crucial, Internet and ftp facilities may not be reliable enough but may be used where this is the only means available.

5.12 Other important points that came out of the discussions include other user requirements like:

4) Harmonization of severe weather programmes between neighbouring countries

5) Use of Ensemble forecasts for severe weather forecasting

6) Improvement of severe weather records

7) Dissemination of severe weather warnings to end users

8) Monitoring of the impact of observations like radar and satellite on products and exchange of data

9) Development of a good severe weather service

10) Further research and information on relation ship between severe weather events to long term variations.

5.13   Recommendations to meet these user requirements:

It was noted by the meeting that contacts between CBS and the CAS/WWRP group should be strengthened to utilize the latest science and technology in the field of severe weather forecasting.

5.14    All these requirements were discussed and the following recommendations were made:

5.14.1 Setting up of own NWP system

Not every centre will be able to run their own NWP system. However, a number of centres have been successful in doing so over the past few years.

The following recommendations were put forward:

• The centre should undertake a feasibility study so that the shortcomings can be addressed. The shortcomings might include the initial and ongoing cost of hardware and software, telecommunications cost, maintenance cost, human resources and scientific capability.

• The centre then may contact one of the leading centres for initial and boundary conditions.

It must be realized by the centres that this is a long term commitment and it will take years to establish.

5.14.2 Centres that can do post processing:

These centres do receive the GRIB data but might not be able to exploit the use of this data to their benefit yet. Other centres that do not have these capabilities should be encouraged to strive for this solution. The following guidelines were put forward:

• As needed, these centres may request additional GRIB data from major centres to assist in the development of products for severe weather forecasting.

• In the case where a Weather service wants to start developing products for severe weather forecasting based on GRIB data, the results of the upcoming survey on forecasting workstations could assist in the selection of a workstation for post processing. Some attention must also be given to training of forecasters on the use of workstations.

• Training of forecasters in the meteorological aspects of severe weather, including model interpretation and use of post-processed products is crucial.

• Research into local severe weather is recommended

• Development of own indices for severe weather warnings based on research is recommended

It must be realised that it will take time to set up such a system as well.

5.14.3 The static display system

These centres must be encouraged to upgrade to a post processing system because static display has several limitations.

If the centre stays dependent of these static displays the following is recommended:

• The special products needed to forecast severe weather must be requested by the individual NMC. An agreement by RSMCs to provide NWP guidance already exists and members should be encouraged to utilize the products made available

• The centre providing diagnostic products relevant to severe weather, based on NWP, will not take into account local conditions. The centre that uses these calculated indices must be aware that these products are solely NWP-based with no human intervention or quality control. It is recommended that these products be carefully evaluated.

• Diagnostic products relevant to severe weather may need to be adapted to local conditions, following evaluation and case studies. Examples of diagnostics products were noted during CBS-Ext. (98).

• The specialised centres should then monitor requests and provide the special products agreed upon.

It is recommended that the above requirements should also be addressed by the Regional Associations where appropriate.

5.14.4 Harmonization of severe weather programmes between neighbouring countries

Neighbours are encouraged to exchange severe weather warnings and forecasts.

It is recommended that some arrangements be made between neighbours on how to exchange these severe weather warnings and forecasts. These warnings and forecasts should clearly indicate the originating centre and include clarification concerning their validity period, updates, area of coverage etc.

Members should be encouraged to exchange information about procedures and definitions of their severe weather programme.

Close consultation with OPAG/PWS should be maintained.

5.14.5 Ensemble forecasts of severe weather

The leading centres should be encouraged to do more research into the subject.

5.14.6 Improvement of severe weather records

Members should be encouraged to build more up to date climate records of severe weather events. This should be referred to CCl.

5.14.7 Dissemination of severe weather warnings to end users

The meeting identified two main streams of end users:

1) The official centres of disaster management

2) The ordinary citizen where appropriate.

Where a warning of severe weather is provided to the general public, it should include suggestions of what to do in the stated crises.

5.14.8 Monitoring of the impact of observations like radar and satellite on products and exchange of data.

It was noted by the meeting that this is important and members should be encouraged to monitor any degradation in the forecasting performance due to lack of observations and strive to improve the observational network by exchange of data with neighbouring countries if necessary. It was also mentioned that this is a Regional Association matter.

5.14.9 Development of a good severe weather service

It is recognized that the training of severe weather forecasters is essential for a good severe weather service.

5.14.10 Further research and information on relation ship between severe weather events to long term variations.

It was noted that this is an ongoing research field.

The following training needs were identified:

• Need for training to implement a model.

• How to handle the post processing system acquired.

• Applications and interpretation of model products.

• How to develop conceptual models.

The following action plan is suggested:

6.1 The meeting noted that the issue of an appropriate infrastructure for the provision of long-range forecasts was an active concern within several Commissions of WMO. The issue had been discussed extensively at the Initial Planning Meeting on the Co-ordination of Infrastructure Needs for Seasonal and Interannual Climate Prediction held in Geneva in April 1999. Arising from this meeting specific action had been included in two resolutions at WMO Congress-XIII, namely Resolution 2 (World Weather Watch Programme for 2000-2003) and Resolution 8 (Climate Information and Prediction Services (CLIPS) Project). The meeting noted the particular role assigned in both resolutions to the OPAG on DPFS in designating RSMCs specializing in the area of climate monitoring and prediction.

6.2 The meeting has taken note of these resolutions in its consideration of the need to design an LRF infrastructure. The meeting discussed a consultant’s report on the implementation of these resolutions and has provided comments and recommendations as given below.

6.3 The main concern is the proliferation of long-range forecasting products being made generally available from a broad range of institutions. Many of the products are broad-scale if not global in coverage leading to overlapping in available predictions. As the skill in many regions is low the forecasts often vary considerably in their predictions. The products are often made available with no quality assessment or record of past performance of the technique, making it difficult for NMHSs and end-users to assign a level of confidence to the product. The end-result is confusion and uncertainty about the information being offered.

6.4 The problem is compounded by the difference in nature of long-range forecasts from typical deterministic weather forecasts. Long-range forecasts are often couched in probabilistic terms requiring training in the proper interpretation and application of such forecasts. The appropriate training and guidance material is often not provided with the products.

6.5 The general availability of long-range forecasting products has sometimes placed National Meteorological Services, who have the ultimate national responsibility for issuing forecasts, in an awkward position, and has led to the feeling that their national responsibility is being infringed. They have been caught unaware by comments in the media relating to their area of responsibility. They may not have the skills to comment on these predictions or to co-ordinate comment through the media.

6.6 Given the issues noted above both NMHSs and the general user community recognize that there is a need for co-ordination of the provision of long-range forecasts. However, the particular characteristics of long-range forecasting need to be recognized. In particular, there are many contributors outside the traditional NMHS framework. The valuable contribution of these organisations is recognised and needs to be included in any proposed coordination framework.

6.7 Any co-ordination framework needs at a minimum to:

• Provide a structure that will facilitate the preparation of authoritative and credible forecasts

• Provide an opportunity for all organisations with a demonstrated capacity in the provision of long-range forecasting products to contribute their expertise to the delivery of a consensus product.

• Recognise the particular peculiarities of long-range prediction products and their mode of delivery

• Ensure and strengthen the role of NMHSs in the provision of long-range forecasts

• Provide users with a credible forecast backed by appropriate performance statistics

6.8 While these broad requirements have been recognized and discussed in the meetings mentioned above, the Team noted that the more specific requirements at the sub-regional level still needed clarification.

6.9 The meeting noted, however, that the broad-scale issues have now been extensively canvassed. The issue has existed for several years now and the need for coordination has been regarded as pressing. Given the clear direction from WMO Congress the time does appear right for action to be initiated to implement an appropriate infrastructure at least at the proposed RSMC level.

6.10 The meeting welcomed the draft proposals, contained in a document for the meeting, prepared by Mr Hubert Allard (Canada), WMO consultant. The document presents draft recommendations for an infrastructure for the generation of seasonal and interannual prediction/forecast based on the WWW/GDPS and complemented by the WCP CLIPS project. Initial discussions on the proposals had been held with representatives of several WMO Commissions and Secretariat, and with representatives of organizations involved in long-range forecasting.

6.11 The meeting endorsed the general approach proposed for a Global Long Range Forecasting System (GLRFS) and particularly

(a) the three level approach:

• RSMCs with activity specialisation in long-range forecasting providing global and/or broad-scale regional products (Level 1)

• A second level establishing entities or mechanisms to develop a consensus forecast (Level 2)

• NMHSs having ultimate responsibility for preparation and delivery of long-range forecasts (Level 3)

(b) the key role for CCl/CLIPS

(c) the clear role for RAs

(d) explicit recognition of the role of institutions other than NMHSs in long-range forecasting, and of the possibility of the inclusion of some such institutions as potential RSMCs

6.12    The meeting considered that the proposal should emphasise the elements of flexibility, adaptability to regional circumstances, and the need to evolve to meet changing circumstances. The meeting was concerned to ensure that a rigid structure was not intended, but rather a loose enabling mechanism to meet the coordination requirements noted above and provide a basic infrastructure in support of the programmes of WMO and other international organisations. Many of the details in the Annexes should be regarded as "experimental" initially and subject to further refinement after experience with the arrangements.

6.13   While the arrangements for Level 1 seem clear, there needs to be considerable flexibility in the proposals for Level 2. Indeed, there may not be a need for this level in some regions and in others it may be an interim measure. The need for this intermediate level will depend on factors such as:

• The presence of several countries within broadly homogeneous climate areas

• National capabilities for interpretation and coordination of the broad-scale guidance

• The state of the science and the level of skill in LRF.

6.14   It was noted that a wide range of mechanisms or entities could fulfil this role. The RAs with the support of CLIPS should take a strong role in defining the Level 2 structure. In some regions entities and/or mechanisms already exist. The arrangements may not necessarily coincide with the RA boundaries but should be based on the patterns of climate variability and the requirement for coordination. As an example, the Climate Outlook Fora have addressed this need in several regions since their commencement in the late 1990s. Other institutions such as ACMAD and the DMCs have made similar contributions.

6.15    Reservations were raised about the term "Regional Climate Board". The Team suggested retention of the term in the proposal but with the inclusion of an interpretative note emphasizing the range of possibilities implicit in the term.

6.16    The meeting noted, however, that the requirements for the Level 1 structure had become clear enough and given the clear direction from WMO Congress it recommends that action be initiated to implement an appropriate infrastructure at least at the proposed RSMC level. The meeting recommended that a revised proposal be presented to other Commissions and other relevant agencies and institutions, with a view to presenting a recommendation along the lines of that in the draft proposal to CBS-XII in November 2000.

6.17    Some specific comments on the details of the proposal were made by the meeting for consideration by the consultant. Only the more significant are mentioned below.

6.17.1 The meeting felt that the draft proposal needed more background discussion, particularly on the user requirements and the reasons for the intermediate level relating to the proposed Regional Climate Boards. Some of this discussion may have been included in reports of previous meetings but it would be helpful to include some of that background in the proposal. The purpose of the arrangements in strengthening the role of the NMHSs could be made more strongly.

6.17.2 The meeting felt that some of the references to other bodies needed further clarification, particularly in relation to the climate service and research areas. This was particularly apparent in Sections 2.5, 4 and 6, and the Implementation section of the draft proposal.

6.17.3 The Team supported the provisions in the draft proposal to allow NMHSs advance access to the LRF products made available by the RSMCs although details of the timing may be revised after further consideration. The proposal is in accord with the direction from WMO Congress-XIII. It may also be desirable to adopt, as an additional measure, the practice of alerting NMHSs to the release date of such LRF information. Reservations was expressed about the details of the provisions for advance access (3.5 and Annex 1) and some doubts raised about their feasibility. The reasons for advance access are to prevent problems noted above of NMHSs being caught unaware of forecasts for their area, and to strengthen the role of NMHSs.

However it was not considered feasible to maintain an embargo until a consensus forecast had been prepared at the Regional Climate Board level due to the variety of issue times from participating organisations. It was considered that advance access by NMHSs to LRF products at the Level 1 would be adequate.

It was noted, however, that LRF information can be used commercially and so procedures for public release are a sensitive issue. Any legal issues in applying such an embargo need to be considered. A potential problem was noted of leakage of information which had been provided on a confidential non-disclosure basis to NMSs. Such leakages would undermine the embargoes and procedures which are applied by some organisations. Measures to ensure the integrity of the advance access procedures may be required.

6.17.4 The role of CCl/CLIPS in furthering the development of the GLRFS needs to be emphasised and expressed more generally than in the draft recommendation (para. 3).

6.17.5 The importance of developing and making available surface forcing data sets should be expressed more strongly (para. 6 of draft recommendation). This is an objective of GCOS but the recommendation needs to be addressed to the right body.

6.17.6 Para. 7 of the recommendation referring to satellite data sets is addressed to CBS in general. In particular, the requirements of long-range forecasting need to be included by the OPAG on IOS in their review and specification of satellite data requirements.

6.17.7 The details in the Annexes should be regarded as tentative examples, subject to further refinement, rather than definitive at this stage.

6.17.8 Because the precise form of the Level 2 structure is not clear, it is premature to specify the training needs. However, the (potential) RSMCs could be encouraged to include participants in existing training opportunities.

7.1.1 The meeting noted views expressed that the following action plan on atmospheric modelling need to be considered and implemented by Members:

• Developing NMHSs' capability in handling chemical incidents by promoting through VCP activities use of PC based models; preparation of inventories of potential sites for chemical incidents in each Member country and setting up of observational systems at such sites.

• Requesting through VCP as appropriate relevant software from members offering such software.

• Maintaining inventory of Country focal points in WMO and by neighbouring countries for rapid exchange of information in case of trans-boundary incidents;

• Identification and implementation of communication facilities for rapid exchange of information within each country.

7.1.2 It was recalled that the eleventh session of CBS (1996) identified new requirements of GDPS products and services including drought-monitoring products such as drought indices. There however do not seem to be any standard procedures in existence for preparation and dissemination of such products as a global or regional arrangement. Some countries prepare such products, for example India Meteorological Department, for national use. There was a proposal that it will be useful to designate activity centres in each WMO Region with an assigned area of responsibility on the pattern of RSMC with activity specialization for preparing and disseminating drought-monitoring products. The meeting noted that the manual on the GDPS already provides for the possibility of designation of RSMCs with activity specialization in provision of drought monitoring products. Further more RA I has already recommended that the Drought Monitoring Centres (DMCs), Nairobi and Harare once they are satisfied that they have built sufficient capability may demonstrate their capabilities to CBS with a view to their designation as RSMCs with activity specialization in this field. It was therefore noted that Regional Associations with requirements for designation of such RSMCs might do so in accordance with the CBS designation procedures.

7.1.3 In the interim it was suggested that the DMCs might consult with other centres generating drought-monitoring products with a view to developing proposals for standard procedure for dissemination of drought monitoring products on GTS/Internet.

7.2.1 The meeting noted the report on the progress of the Rapporteur on this issue involving the project to gather and analyse information on the use of Forecaster Workstations in NMHS’s. It reviewed the Rapporteurs’ draft questionnaire, which she plans to issue to Members soon in order to gather the information, and provided suggestions for improvements to be communicated directly to the Rapporteur.

8.1 The meeting was closed o Friday 12 November 1999.

Appendix I

 

AGENDA

 

1. OPENING OF THE MEETING

2. ORGANIZATION OF THE MEETING

2.1 Approval of the agenda

2.2 Working arrangements for the meeting

3. REVIEW OF CURRENT ACTIVITIES

3.1 Ensemble forecasting methodology and progress

3.2 Atmospheric transport products

3.3 Seasonal to interannual forecasting (including verifications)

3.4 Impact of changes to the GOS on NWP

3.5 Y2K and GDPS

4. PROVISION OF ENSEMBLE PREDICTION SYSTEM PRODUCTS

5. NWP GUDANCE ON THE OCCURRENCE OF SEVERE WEATHER

6. LONG-RANGE FORECASTING INFRASTRUCTURE

7. OTHER EMERGING REQUIREMENTS

8. CLOSURE OF THE MEETING