The Toronto 2015 Pan and Parapan American Games Experience

From an Information Technology (IT) perspective, the Games represented an opportunity to showcase the enhanced software and systems developed over the past few years within ECCC to create and manage meteorological data and products. The challenges to support the Games were enormous; namely, greatly increasing the number of observations being collected and processed in the region of the Games (i.e., close to 60 new atmospheric monitoring stations reporting every minute), creating weather forecasts and alerts for new locations, and supporting forecaster workstations located internally and externally at the MOC and UCC. 

The approach was simple: reuse and enhance existing systems. Current systems were reconfigured in innovative ways. This was combined with the addition of a few carefully selected new technologies to meet the unique operating requirements of the Games while still managing risk.

Existing Integrated Forecaster Workstations (IFWs) were designed, modified, built and supported for use by the operational forecasters, weather briefing teams and research scientists in the OSPC. The Integrated Showcase Workstation (ISW) was used by the forecasters to produce venue-specific forecasts and alerts while the briefers at the MOC and UCC used an Integrated Pan Am Workstation (IPAW). Integrated Research Workstations (IRWs) were used at the two Research Support Desks in the OSPC for real-time scientific support to operations. The IFW would have been used as a contingency should the IPAW have failed.

6.1 Creation of New Integrated Workstations for the Games

The requirement to have fully functional forecaster workstations operating at the OSPC, TO2015 MOC, UCC and in the Storm Prediction Centre contingency office in Winnipeg, Manitoba, represented a significant challenge. It involved the following considerations:

1. Moving large volumes of specialized meteorological data within and outside our internal networks;
2. Reconfiguring the forecaster workstation to operate as a single unified entity rather than the client/server architecture used internally; and
3. Meeting ECCC’s security requirements and those of the external organizations.

To meet this challenge, the IT team modified the existing operational forecaster workstation, IFW, to create three different workstations: ISW, IPAW and IRW. All of the modified workstations featured enhancements to hardware including additional memory and replacement of standard hard drives with state-of-the-art high-speed solid-state drives. In addition, the client and server portions of the forecasting application were combined onto a single workstation and a virtual machine was introduced for further flexibility. The net result was a series of workstations that were reliable and performed significantly better than a traditional workstation by enabling local storage of data.

In addressing the challenge of moving data from within ECCC’s systems to external offices, a new approach was taken. Traditionally, data is pushed from servers to client systems. This was not possible due to the security restrictions of the hosting organizations. To resolve this problem an enhanced client-driven pull data movement technique was implemented that used standard Web protocol technology as well as a messaging protocol borrowed from stock trading systems known as “AMQP.” The net result of this implementation was that all internal forecast information was made available in real time and with enhanced performance characteristics to the weather briefers operating the IPAWs at the MOC and UCC. In addition, AMQP allowed the client to configure and manage the data being viewed. Data access was strictly limited to the viewing needs as documented by clients. For any bandwidth issues, the amount of data transmitted could be reconfigured, thus allowing for successful transmission of the most important data without either overloading the system or compromising the timeliness of data delivery.

One key requirement was the capability to provide forecasts and alerts tailored to each sporting venue. In addressing this requirement, the recently implemented warning systems were reconfigured. This demonstrated the value and benefit of the recent investments in renewing our warning production systems. Also, as the format of the data in the new system was XML, it allowed for the data to be shared with our suite of partners, easily allowing them to work in parallel to meet tight and inflexible deadlines.

6.2 Production and Capture of Minutely Data

The capture and processing of minutely observation data represented a significant test of established data management capabilities. The many challenges included how to collect, process and represent this data to each client in a manner to which they were accustomed.

In implementing the solution, a new data collection mechanism was deployed. Instead of dialing via traditional telephone modems to remote observation sites, wireless cellular modems were used, and the data was sent via the Internet to our systems.

From a data processing perspective, some key decisions were made. In order to manage workload and reduce risk, a decision was made that the output for each minutely observation would be in the same format as the data the forecaster traditionally saw at the top of the hour. While this represented a data assimilation challenge, it allowed IT experts to reuse existing software for formulas.

Requirements were gathered and specifications documented for the development of software to decode weather reports that were transmitted every minute, sent in new formats and included additional data elements from new sensors. Data elements needed to be derived that were not previously produced by ECCC, such as black globe temperature used in the evaluation of heat stress. All the decoded data, derivations and disseminated products were thoroughly documented and validated to ensure they met specifications and could be signed off well before the commencement of the Games.

To address these increased demands, computational capacity was increased by 25% and special arrangements were made to manage backlog processing if necessary. The net result of these changes was a successful solution that allowed IT experts to process and create all derived products within ten seconds of receiving an observation.

The IPAW design proved robust and flexible. When an additional and unplanned request to set up another remote office at the UCC became a requirement, it was solved simply by recreating another instance of an IPAW and deploying it.