Weather Services for the 2010 Vancouver Olympic and Paralympic Winter Games

15.0 Lessons Learned, Recommendations and Legacy

15.1 Accommodations and Logistics

15.2 Planning and Management Support

15.3 IT Contracting and Atos Origin

15.4 Weather Observations

15.5 Legacy of Science

15.6 Forecaster Experience and the Introduction of New Technologies

15.7 Probabilistic Forecasts and Client Utility

15.8 Services and Communications Legacy

15.0 Lessons Learned, Recommendations and Legacy

Nearly six years of WSP development invariably allowed for a number of blind alleys and errors that we generally rectified before the Opening Ceremonies, although not, at times, having first suffered probably avoidable delays and setbacks. Having gone through the process, the 2010 WSP management team offers the following suggestions for future Games’ weather service planners:

15.1 Accommodations and Logistics

Since forecaster training for the Games began in the winter of 2005-06 and continued every winter prior to 2010, venue forecaster accommodation and, at times, technician’s accommodation (such as for test events), was required at or near the venues.  Accommodation at Games time for federal employees used a centralized approach, with all accommodations arranged and reserved by the Royal Canadian Mounted Police. However, since MSC forecasters needed to be practicing in venues well before the arrival of the rest of the GoC, EC needed to obtain much of its accommodations through private seasonal leases with Whistler‑area homeowners, which we maintained for several years prior to the Games and during the Games (although at significantly higher prices during the Games than the practicum periods). We recommend this approach for the weather services team, as it offers the benefit of security of tenure and helps the team’s management avoid dealing with the considerable uncertainty of place and location that results from having to depend on another agency to satisfy needs. Forecasters work very long hours at Games time and they need quiet and accessible accommodation, close to their working venues. It is also very helpful to have high-speed Internet access in their accommodations, so they can self-brief before going into work and can respond to telephone requests off-hours.

Technicians can be more remotely located than the forecasters, as they are likely to be more mobile (with their own vehicles). However, team management must ensure that technical staff are granted easy access to venues through the early procurement of a few vehicle access and parking passes, and that proper accreditation is prepared for technicians. The number of accredited technical staff can be kept to a relatively small number as, at Games time, they would be required only in the event of an observing system failure. We obtained technician access through a number of “second-part passes,” but these were awkward and, at each venue, technical staff developed informal agreements with venue managers to permit easier accessibility.

15.2 Planning and Management Support

Weather services planning for 2010 began at the conclusion of the Olympic bid. The structure of the team was very flattened, with the Chief Meteorologist in the role of developing an integrated plan for weather services, and negotiating with other managers in other areas of EC for the services and staff necessary to deliver on the plan. The executive in charge of the project generally had an information-gathering role in the initial stages, which transformed into an oversight and senior management reporting role in the few years prior to the Games. Since the WSP was designed to meet the needs both of VANOC and of the federal agencies that provided services at Games time, there were a considerable number of cross-cutting issues that required coordination. For example, the design of the Olympic weather observing network needed to take into account not only meeting the requirement for specific weather data at the venues, as required by the various international sport federations, but also had to be designed to permit better analyses of current weather patterns over the entire Olympic area as an aid to the quality of the overall forecasting program. It became clear early on that the central coordination role of the Chief Meteorologist was the most significant and important dimension of the job.

There were a large number of key relationships that were made, facilitated and/or maintained by the Chief Meteorologist. These included, on the government and partners side, the following:

• SNOW V-10
• EC Chief Information Officer Branch (CIOB)
• EC MSC Atmospheric Monitoring Division (in PYR and PNR)
• ECCommunications
• EC senior managers
• Team forecasters
• CMC Operations
• CMC Environmental Emergency Response
• The Essential Federal Services Committee (government-wide)
• Local researchers and developers
• The 2010 internal and external weather website developers
• EC MSC Services
• Procurement Review Board
• Public Works and Government Services Canada
• The U.S. National Weather Service
• The Weather Network
• The Pacific Storm Prediction Centre
• EC Legal (contracts and intellectual property).

At VANOC, the Chief Meteorologist developed and maintained relationships with the following entities:

• Sport Department
• Medical Department
• Transport Department
• IT Department
• Atos Origin
• International sport federations
• Sport Production
• Venue Management and Operations
• Official Observers
• Legal Services

Communications were coordinated through the Chief Meteorologist in order to simplify the flow of information and to make connections only where necessary. In some cases, initial connections consisted simply of introductions, and letting the two or more parties follow up to complete their task. Other connections required significant participation, direction and, at times, intervention. The main reason for this structure, however, was that VANOC Sport Services insisted that the Chief Meteorologist be the sole point of contact for all meteorological issues. This provided “one-stop shopping” for the Olympic Games Organizing Committee, and simplified their planning and sourcing of resources and meteorological support.

As workload increased, the Chief Meteorologist developed dozens of connections and productive relationships (ultimately there were more than 240 individuals in EC working on the project). It would have been useful to have a full‑time dedicated administrative assistant as early as three full years before the Opening Ceremonies. By early 2009, as the urgent need for administrative support became obvious, much of that role was taken on by the administrative support staff of the Regional Director. 

Top of page

15.3 IT Contracting and Atos Origin

Atos Origin’s weather data requirements were met by preparing relatively simple in-house software to translate forecast and observation templates into a version of XML that met the format requirements of INFO 2010. Originally it was planned that the conversion of EC data formats into appropriate XML could be completed automatically using the national forecast program’s Scribe database and some built‑in XML output functionality that was part of Scribe’s capabilities. Despite considerable investment in time and effort, it was determined by 2008 that Scribe was not capable of producing weather data that could be successfully ingested by INFO. This was a significant setback, as Scribe was intended, consistent with its design, to be the basis of all forecast production at Games time. To fill the gap, the Q-Device was built, tested and deployed by December 2009. Although the Q-Device worked as intended, it added a few layers of complexity to forecast production. Firstly, forecasters need to become familiar with the operations of the Q-Device in the year prior to the Games, and although this was adequate, given more time, some additional development would likely have been achieved on the basis of forecaster input and evaluation. The Q-Device was somewhat awkward to edit, and this issue could not be rectified in time for the Opening Ceremonies. We recommend that early in their planning period, the Sochi weather team examine the specifications for weather data and forecasts available from Atos Origin and prepare their software solution well in advance. Ideally, all forecasts--for INFO, sport and the public--can be based on a single forecast and observation database, as was our original intention.

15.4 Weather Observations

A number of weather observation innovations were introduced at Games time. These include a higher‑than‑usual observation reporting period (every 15 minutes for the main network and 1 minute for the SNOW‑V10 systems). New data sets were introduced: some to meet the needs of the international sport federations (snow surface temperature) and some to improve forecast quality (15‑minute speed and directional variability of the wind, and 15‑minute instantaneous peak wind). In addition, a comprehensive network of webcams, both at the venues and along the Sea-to-Sky corridor, was installed prior to the Games. These observing‑system innovations proved to be exceptionally useful to the venue meteorologists and ultimately the clients.

Because our approach to weather observation was, with few exceptions, full automation, we could not derive a satisfactory solution to develop a sky condition observation for INFO, as it requires. Sky conditions are measured and classified by human observers. The available automatic equipment generally provides observed data along a line and not with the 360‑degree three-dimensional perspective provided by a trained observer. To compensate for this, venue forecasters produced observed sky condition data every hour while at the venues, and manually entered the data into the Q-Device. During shutdown hours, forecasters at the POD ­provided venue sky condition observations to INFO by evaluating webcam, satellite and other data sets at and over each venue, and then creating a synthetic observation. For the purposes of INFO this worked satisfactorily.

15.5 Legacy of Science

Weather services for the Games represent a considerable investment in money and human resources. As part of the original planning philosophy, it was intended that EC would leverage investments made in advancing the state‑of‑the‑art in weather forecasting for the eventual benefit of the entire national forecasting program. This has come to immediate fruition in a number of areas. Firstly, 1-km resolution NWP guidance for public forecast areas is now routinely available in the Pacific region, although not yet across the country. This is a significant advance from the 2.5-km guidance available prior to the Games and represents a rate of progress some 3-5 years ahead of the schedule of model innovations planned prior to the Olympics. The nationally available 2.5‑km variant of the GEM will be extensively modified in the near future based on lessons learned from the Games‑time prototypes, and eventually national use of the 1.0‑km model will be expanded, depending on computational resources. 

For the time being, SNOW-V10 research continues, although its website (with a large number of forecaster tools), is now offline. Accessibility to high‑resolution models, SNOW-V10 nowcasting tools and the additional wind profiler and Whistler radar observations has created a wealth of data on local conditions and led to the discovery of explanations for weather phenomena never previously observed or perceived in models, such as the “Cypress Chinook.” This occurs when a series of large‑scale events create an easterly flow over the North Shore Mountains, and can lead to rapid snowpack depletion through melt and wind-driven sublimation. It can also create damaging down-slope winds in the southern portion of the Sea-to-Sky corridor, a phenomenon for which, until our new tools were available, we did not have a satisfactory physical explanation. “Blocked flows,” where wind is diverted by topography in situations with precipitation, were for the first time clearly visible on experimental Doppler radar wind and reflectivity imagery. These situations in the past have led to significant snowstorms that were under‑forecast or not forecast at all; the SNOW-V10 technology has greatly improved our likelihood of detecting these phenomena and improving the forecast result.

Top of page

15.6 Forecaster Experience and the Introduction of New Technologies

Perhaps the most significant feedback from forecasters was not about their access to new sets of data and forecaster tools, but about the personal gratification and satisfaction they derived by working closely with venue managers as well as sport and international sport federation officials on day‑to‑day weather forecasts that could and did affect the competition schedule. The new technologies significantly improved their personal confidence in the forecasts they prepared, and they were able to communicate those forecasts and the uncertainty therein in a manner that enabled venue officials to make decisive and timely decisions. Having SNOW-V10 and model experts on hand at the POD also contributed significantly to the success of forecasts, especially during periods where particular radar or model expertise was required to help clarify some forecaster uncertainty.

15.7 Probabilistic Forecasts and Client Utility

In general, weather forecasts contain an element of uncertainty, and these are communicated using different methods. In Canada, the most overt measure of uncertainty is in the probability of precipitation (POP) part of a public forecast, usually expressed as a percentage. The POP is meant to provide information on the likelihood of precipitation in the venue or public forecast area during the period of the forecast. This simple sort of uncertainty statement is, however, not reflective of the much more sophisticated probability data prepared for the Games in 2010. We developed regional ensemble models at high resolution that produced forecast uncertainty information explicitly. However, these Games did not generally benefit from the ensemble approach. There is a controversy in the meteorological and social communications literature on how to communicate uncertainty data in weather forecasts, and we did not resolve this issue for the 2010 Games.

One area where uncertainty information was significantly communicated was in the snowmaking forecasts prepared for the mountain venues prior to the Games. As discussed in section 5.3, both an ensemble forecast graphic and a narrative to discuss its contents were provided to venues and other users on a daily basis between November 2009 and January 2010. Anecdotally, this was used by venue managers to decide on snowmaking locations, elevations and workforce deployments, sometimes days in advance depending on the degree of confidence expressed in the narrative and discernible from the graphic. Although complex compared to POP, clients seemed to grasp the essentials of these forecasts and put them to very good use. For example, not a single opportunity to make snow at Cypress was missed in the months that the forecasts were produced, although there were a few occasions when snowmaking was unsuccessful when it was attempted during periods of borderline forecast conditions.

Several very significant multiple-day adjustments were made to the competition schedule during the Olympics and especially during the Paralympics. These were based on long term (3-7 day) forecasts and derived principally with the help of the North American Ensemble Forecast System.

Sochi­ 2014 and Pyeongchang 2018 should investigate opportunities to communicate a larger fraction of weather forecast information in probabilistic forms. A good place to start evaluating the concept may be here. This is a probabilistic forecasting educational module, prepared by our training partner for the Games, COMET.

15.8 Services and Communications Legacy

EC’s initial steps into social media proved successful and demonstrated the utility of the Internet for disseminating useful information to the public and media. Other approaches involving a broader social‑media marketing approach should be assessed for 2014. The Weatheroffice Olympic web pages and SLF prototype were well received, but would have benefited from a much earlier introduction--at least one year before the Opening Ceremonies rather than the two months as occurred.