Forecast and Research: the Olympic Sochi Testbed (FROST-2014)
Concept paper


The Next Olympic / Paralympic Games «Sochi-2014» (hereinafter referred to as «Olympics») will be held in Sochi, Russia, on February 8-23 / March 7-16, 2014. Roshydromet (the Federal Service for HydroMeteorology and Environmental Monitoring) is responsible for provision of hydrological and meteorological support and services to ensure the safety of the guests and participants and efficient work of involved bodies. All forecast ranges are important for the Olympic meteorological services. However, out of these «seamless» needs the primary focus of the proposed project is on nowcasting and short-range weather forecasting.

FROST-2014 (FROST - Forecast and Research: the Olympic Sochi Testbed) project is proposed for the period from now to the Olympics:
• To develop, enhance and demonstrate capabilities of modern systems of short-range numerical weather prediction (NWP);
• To further develop and demonstrate nowcasting in winter conditions for mountainous terrain and
• To assess the effect of practical use of this information.

Background and heritage

The World Meteorological Organization’s World Weather Research Program (WWRP) has previously organized several Forecast Demonstration Projects (FDP) and Research Development Projects (RDP) to advance and demonstrate state-of-the-art nowcasting and forecasting systems. Mesoscale Alpine Programme (MAP) was the first RDP (1999) and was focused on understanding weather systems in complex terrain and flash flood events. Based on the success of MAP, the MAP-Demonstration Phase was organized in 2005 and focused on demonstrating the utility of prediction systems for hydrological flash flood applications. While in complex terrain, the focus was on precipitation. The Sydney 2000 FDP focused on advancing and demonstrating summer convective nowcasting systems. The Beijing 2008 FDP was conducted with the goal to demonstrate nowcasting advances since 2000 and facilitate technology transfer into operations. In addition, the second project, Beijing 2008 RDP was carried out with focus on mesoscale ensemble prediction. In both projects the focus was on precipitation prediction, convective initiation and summer severe weather. SNOW-V10 (Science of Nowcasting Olympic Weather for Vancouver 2010) project was approved as an RDP due to the novel aspects of the winter nowcasting, particularly in complex terrain, but operated as a blended RDP/FDP. Innovations included nowcasting of rain-snow boundary, snowfall intensity, phase, visibility (fog/low cloud), temperature, and wind.

From the point of view of weather conditions, orography and meteorological needs these Olympics have much in common with the recent «Vancouver-2010» Olympic Games. Similar to Vancouver, high winds, visibility and low cloud, precipitation amount and phase are the critical weather elements. Following the FDP/RDP of the «Vancouver-2010», the present project offers a tremendous opportunity to continue and enhance the progress made there, develop and test new techniques for weather forecasting and nowcasting.

FROST-2014 is expected:
• To extend experience of the B08RDP mesoscale ensemble prediction into mountain environment in winter season;
• To extend the experience of MAP and MAP D-Phase to various weather elements;
• To extend the experience of SNOW-V10, which was focused primarily on nowcasting, to more emphasis on NWP.

The Venues

The city of Sochi is located at approximately 44°N, 40°E at the Black sea coast. Sochi Olympic objects are separated between two clusters: a coastal cluster for ice sport competitions and a mountain cluster for snow sport events. The latter is located at Krasnaya Polyana township about 45 km away from the coast (Fig.1). Mountains of approximately 2 km height are typical for that region. The mountain cluster events are especially weather-sensitive.

Fig.1. The two clusters of «Sochi-2014» Olympic venues:
- The coastal cluster for ice sports competitions (Figure Skating, Short track Skating, Speed Skating, Ice Hockey, Curling, Ice training); Opening and Closing Ceremonies;
- The mountain cluster for snow sports competitions (Alpine skiing, Cross-country skiing, Biathlon, Ski jumping and Nordic Combined, Freestyle and Snowboard, Bobsleigh, Luge and Skeleton).
Weather conditions and challenges

Winter weather conditions in the region of Sochi are mainly determined by the quasi-permanent Black Sea baric depression and the spur of the Asian anticyclone. Rapidly passing cyclones accompanied by strong winds, snow- and rainfall are alternated with cold anti-cyclonic intrusions from the east. Main Caucasian ridge mitigates these cold intrusions and block warm wet landward air flows from the Black sea. The typical sea surface temperature near Sochi in February-March is 8-10ºC.

Multiyear mean minimum near-surface temperatures in the mountain cluster in the period of the Olympics are negative at all altitudes; mean maximum temperatures are below zero at altitudes above 1600 m; and daily mean temperatures are negative at altitudes above 700 m. In some years intense heatwaves might endanger the natural snow cover existence in the lowermost part of the mountain cluster.

On the other hand, heavy snowfalls are also typical for this area. Maximum registered daily snowfall reported by the weather station Krasnaya Poliana (WMO index: 37107. Elevation: 568 m) was 92 cm. Winter maximum is clearly pronounced in the precipitation annual cycle. Mean monthly precipitation totals in Krasnaya Poliana for February and March are equal to 123 and 118 mm respectively. They tend to increase with elevation.

Sharp weather contrasts and high spatial and temporal variability are typical for the region of the Olympics. Steep mountainous terrain and intricate mixture of maritime sub-tropical and Alpine environments make weather forecasting in this region extremely challenging. For the territory of Russia northern Caucasus is among the leaders with respect to the number of annually registered weather hazards (heavy precipitation, strong winds, icing etc). Although not frequent, thunderstorms might also take place in winter season.

Precipitation intensity and type, gusting winds, visibility and cloud ceiling are the primary critical weather elements for the Sochi Olympics.


Goals of the FROST-2014 project

Meteorological support of winter Olympics in mountainous terrain implies both fundamental research and practical forecast demonstration components. A blended RDP/FDP under the auspices of the Nowcasting and Mesoscale Weather Forecasting Research Working Groups of the WWRP is to be an appropriate organizational form for the project. The outputs of the project will be used to enhance the mesoscale and nowcasting services for the Olympics.

• To improve and exploit:
– high-resolution deterministic mesoscale forecasts of meteorological conditions in winter complex terrain environment, including downscaled modeling;
– regional mesoscale EPS (Ensemble Prediction System) forecast products in winter complex terrain environment;
– nowcast systems of high impact weather phenomena (snow levels, wind, visibility, precipitation type and intensity) in complex terrain.
• To improve the understanding of physics of high impact weather phenomena in the region;
• To deliver deterministic and probabilistic forecasts in real time to Olympic weather forecasters and decision makers and assess benefits of forecast improvement.
• To assess benefits of forecast improvement (verification and societal impacts)

As the project evolves these goals will be detailed.

Project components and objectives

Deterministic NWP

NWP is considered to be a backbone of the FROST-2014 project. The very complex region of Sochi provides a stimulating environment for the application of high-resolution meso-scale modeling methods for the purposes of short-term weather forecasting.

There are many scientific issues that should be addressed within this deterministic component of the project, e.g.:
• Impact of horizontal resolution and various physical processes on NWP of high-impact winter weather over a region with mountain terrain;
• Mesoscale data assimilation and its impact on forecasts of winter weather;
• Impact of better surface-atmosphere coupling on predictability of mesoscale phenomena;
• Nowcasting potential of numerical models;
• Predictability of various meteorological parameters/phenomena of winter weather in mountains (precipitation intensity and type, wind speed and direction, visibility, etc.).

Development and testing of physical parameterizations for visibility, wind gusts, precipitation type and snow characteristics along with improvement of formulation of boundary layer and shallow convection should be a part of the project.

The preliminary list of participants of this project component includes: COSMO, AROME, GEM, WRF, GRAPES (?), and HARMONIE. It is planned that in FDP mode FROST-2014 deterministic forecasts will have horizontal resolution of 2-2.5 km or finer, whereas in RDP mode 1 km or less. The groups that wish to go beyond this baseline in terms of resolution are welcome to do so. Besides, various approaches of downscaling of model forecasts to the locations of individual sport venues will be implemented, verified, and intercompared. It is worth noting that at the moment no operational model with resolution about 2 km is implemented for the region in question. Several models are expected to be implemented for the Sochi region with resolution of 1 km or finer (e.g. GEM/LAM, COSMO, AROME).

COSMO will be the basic operational mesoscale NWP system of Roshydromet for the Sochi Olympics.

Ensemble NWP

Many scientific issues should be addressed in the framework of ensemble NWP component of the project, e.g.:
• Forecast error growth and predictability on small spatial and temporal scales over steep terrain;
• Representation of sources of uncertainty in LAM EPS systems. Methodologies for generating perturbations in initial conditions, lateral boundary conditions, surface fields, and model errors;
• Interpretation and configuring of LAM-EPS in the context of high-impact weather in mountains;
• Added value of convection permitting EPSs. Evaluation of the usefulness of coarser resolution EPSs to provide uncertainty information for prediction of weather elements at particular locations (which typically requires much higher horizontal resolution) compared to 1-km or less deterministic models;
• Forecast uncertainty vs observational uncertainty in forecast verification.

The main challenge for regional EPS systems is to accurately assess the probability of High-Impact Weather (HIW) events. With respect to the number of annually registered weather hazards (heavy precipitation, strong winds, icing etc.), northern Caucasus is one of the most affected regions on the territory of Russia. This region might be a good testing ground for the development of new probabilistic forecast products for HIW events. Due to the strong dependence of many winter sport events on weather conditions, HIW in the context of winter Olympics is not necessarily linked with very intense or extreme meteorological phenomena. E.g. for outdoor sport events HIW forecasting also includes accurate representation of cross-zero temperature transitions (especially critical for cross-country skiing), precipitation type and other sensible weather changes with respect to the prescribed decision-making thresholds (Annex 6). Development and demonstration of various HIW-related specific products is part of this ensemble project component.

EPSs with resolution of about 7 km or coarser are planned to be involved in the project in forecast demonstration mode, while EPSs with resolution about 2 km will contribute to the project in research mode. Tentative list of participants: COSMO, ALADIN LAEF, AROME EPS, GLAMEPS, and HARMON EPS. ECMWF officially informed Roshydromet about its readiness to provide lateral boundary conditions for a WMO project associated with «Sochi-2014» Olympics.

Data assimilation

The vast sea area on one side and mountainous terrain on the other side of Sochi impose serious restrictions on configuration of the ground-based observational network in the region. This geography requires the extensive use of satellite, Doppler radar and profiler facilities for sounding of the atmosphere and the underlying surface. This information will be of utmost importance as a source of meso-scale structures for NWP. Therefore, high-resolution data assimilation is a matter of particular interest and a new aspect of WWRP projects.

Several data assimilation approaches for the atmosphere and the underlying surface are expected to be tested. Some project participants will conduct experiments with identical model configurations with and without data assimilation to assess the data assimilation contribution to forecast performance.

Assimilation of radar data in the region of Sochi may only be beneficial for a short forecast range because of lack of radar coverage to the west of Olympic venues. Rapid update cycle capability is therefore needed to exploit the potential benefits. Assimilation of satellite winds (AMV and ASCAT) and radiances over vast areas of the Black Sea is believed to be important as its effect may last longer.

List of non-satellite data that are considered to be useful for assimilation in Sochi project includes:
– Surface observations;
– Radar winds;
– Wind and temperature profiles from local sounders;

Assimilation of radar reflectivity aiming for a better representation of the thermodynamic structure of the atmosphere near precipitating clouds might not be too efficient and useful in winter, but this is an area that may be addressed in the RDP project.

Data assimilation is to be included in both FDP (the existing technologies) and RDP (new developments) FROST-2014 components.

As suggested by the WWRP Working Group on Mesoscale Weather Forecasting Research, a simple downscaling from relatively low-resolution (7 km or coarser) regional data assimilation systems will be the baseline. For example, for COSMO-RU assimilation is to be run at resolution about 7 km with downscaling to 2 or 1 km. But groups that are able to go beyond that and assimilate observations from the «Sochi-2014» network (e.g., radar radial winds, profilers) are encouraged to attempt the kilometric-scale data assimilation.


SNOW-V10 was the first WWRP winter complex terrain nowcasting project. It remains to be demonstrated whether its results are universally applied and can be demonstrated in a different environment or with different observating network. FROST-2014 provides an excellent opportunity to extend the experience of SNOW-V10 project in the scientifically challenging area of winter nowcasting in a region with complex terrain.

Many issues should be tackled within this project component in the RDP mode, e.g.:
- Nowcasting of high impact winter weather and multiple weather elements (wind speed and direction, wind gust, visibility, precipitation intensity and type) in complex terrain;
- Improvement of blending procedures for NWP with time-extrapolated observations for winter;
- Radar retrieval of precipitation type and intensity;
- Diabatic and orographic effects on precipitation nowcasting in complex terrain;
- Assessment and account for observational uncertainty (WGNR mandate).
- Identification of local circulations and clouds controlled by effects of flow blocking, diabatic cooling due to melting snow, and evaporation of precipitation.

A major challenge is the development of nowcasting systems or mesoscale NWP systems to fill the gap in the 4-6 hour lead time and, probably, up to the 12 hour range. Nowcasting potential of participating NWP models (COSMO, HARMONIE, AROME, GEM, and WRF) should be assessed for direct and post-processed (e.g. Kalman filter, 1-D model, MOS) model forecasts. Besides the meso-scale models, specialized nowcasting systems are expected to be used in the project (see Table 1). Capabilities of these technologies and new nowcasting products for winter HIW events are planned to be demonstrated in the FDP mode.

Table 1

SystemForecast PeriodSpatial ResolutionOutput Products
ABOM0-6 h1 kmPrecipitation rate and type; wind speed direction and gust; temperature; humidity; visibility; cloud base
CARDS0-2 h1 kmLocation, intensity and track of storm cell; QPF; hail size; gust; downburst; mesocyclone
INTW0-6 h1 kmPrecipitation rate and type; wind speed direction and gust; temperature; humidity; visbility; cloud base
MeteoExpert0-4 hr1 kmLocation, intensity and track of cloudiness and precipitation zones and dangerous weather (Thunderstorm, hail, microburst, icing and turbulence), QPF.
STEPS0-6 h1 kmQPF and precipitation probability
INCA0-6 h1 kmPrecipitation intensity and type; wind; temperature; humidity; visibility (experimental mode)
WSDDM0-2 h0.2 kmQPF; precipitation type and rate, visibility, temperature

In general, current experience of nowcasting in Russia is very limited and associated mostly with areas of flat terrain. Outcomes of the proposed activity might be a long-term legacy of the project.


Roshydromet will provide the project partners with access to additional in-situ and remote sensing observations not normally available via the GTS. These observations will be available to the participants via Internet with minimal delays.

In 2009-2010 18 automatic meteorological stations (AMS) were installed in the region by Roshydromet. Some of these stations were enhanced with additional sensors (visibility and cloud base) in the autumn 2011 (Annex 1). On account of investors, 10 AMS were installed and 7 AMS will be added in the area of sport venues during the winter 2011/2012. Besides, 8 AMS were installed and about 20 AMS should be mounted on the towers of mobile communication owned by the Megafon corporation. The frequency of observations will be station-dependent (it might be different for the different groups of stations: Roshydromet’s AMSs, AMSs owned by sport venue investors, and AMSs of the Megafon corporation). In general the sampling interval will not exceed 10 minutes. For a subset of the stations it will be substantially higher.

More details on the AMSs (sensors and coordinates) are presented in Annex 1.

Fig.2 AMSs in the region of Sochi (a similar map with zooming is available at Notation: Roshydromet's AMSs are designated by red markers with label "*"; "E" - Roshydromet's air quality control stations; "R" - mountain skiing venue AMSs (owned by the Rosa-Khutor company); "G" - biathlon venue AMSs (owned by the Gazprom company);"M" – AMSs on the cellphone towers of the Megaphone operator; "A" - planned AMSs on the Megaphone cellphone towers. Moored sea buoys are designated by sail boats. Location of Doppler radar on the Akhun mountain is marked by white nested circles. The mountain cluster is outlined by the red cirle.

Data from the new dual polarization Doppler weather radar WRM200 on Akhun mountain will be available by winter 2012/2013. Additional dual-polarization Dopplers in nearby airports might be installed later.

Wind profiler, temperature/humidity profiler and Micro Rain Radar (MRR) will supplement the network by winter 2012/2013. It is quite likely that another MRR will be installed during the winter 2011/2012. In the Krasnaya Poliana valley these vertical soundings will be extremely important for diagnosis of the atmospheric lower layers screened from the Akhun Doppler radar by the mountains.

The nearest to the mountain cluster upper air sounding station is located in Tuapse (approximately 100 km north-west). Soundings in Tuapse will be launched every 6 hours. Other ways to enhance the observational network are looked into (more frequent sounding at other nearest aerological stations including stations in Ukraine, Armenia, Turkey, Bulgaria; receiving of AMDAR data from Adler airport; etc).

Several FROST-2014 participants expressed their readiness to lend additional observational equipment for the project and interest to organization of an instrument intercomparison site. This opportunity is being worked on.

Information Technologies

Development of a comprehensive information resource wih appropriate IT-infrastructure and efficient data facilities is one of the key elements of the project. It is needed for operational data assimilation, forecasting and nowcasting, verification, for posterior diagnostic studies and analysis.

Unification and integration of data flows from multiple observational platforms and forecast systems, timely information generation and delivery, efficient means of information presentation are crucial for the project success. The proposed version of the forecast data exchange protocol is presented in Annex 2.

Data storage with authorized Internet access and capacity for information rescue and memory extention was organized for observation and forecast data exchange between the project participants via FTP and HTTP protocols. Initial version of the project web-site is available at Some ideas for its web-interface that are currently being implemented, originally were taken from web-resources of the previous WWRP projects (e.g. elements of multi-window MAP D-Phase web-interface are being realized on the basis of Google Earth for visualisation of meteorological information).

After the Olympics support of the FROST-2014 http/ftp-server is planned for further retrospective studies (e.g. in the context of data assimilation).

Training and understanding

FROST-2014 is intended as an ‘end-to-end’ project. Its products must be used by local forecasters for meteorological support of the future Olympics and preceding test sport events. Training is critically needed to benefit from FROST-2014 nowcasts and forecasts issued in FDP mode. This need is enhanced by the currently modest experience of Roshydromet in nowcasting, high-resolution NWP and ensemble forecasting in complex terrain. The project will contribute to the capacity building for the Russian Weather Service in these and other related areas.

Annual training courses for future Sochi Olympic forecasters and their practical participation in meteorological support of the test sport events have been practiced since autumn 2010. This practice will continue till the Olympics. It will be intensified as more technical facilities become available for forecasters. To get familiar and develop practical skills with various new forecast products their early availability is of utmost importance. This is also important for training of involved sport managers and decision makers.

Enhanced observational facilities and new NWP instruments are expected to give a new look at the local weather phenomena in the region of the Olympics. The RDP project component can be of great help for better understanding of local HIW and development of conceptual models of local meteorological processes.

Verification and Impact Assessments

Given the probabilistic nature of small-scale weather, traditional verification of deterministic model output often fails to demonstrate the added value of high-resolution mesoscale forecasts. This has provoked an extensive research into alternative mesoscale verification methods. Along with traditional verification methods the substantial attention in FROST-2014 should be given to new probabilistic approaches. Both user-oriented (in particular, with account for the thresholds for decision making) and research-oriented verification approaches will be employed. Spatial verification using remote sensing data will be applied to the predicted precipitation fields. The preliminary version of the project verification plan is presented in Annex 4.

As a complex terrain imposes additional limitations on representativeness of pointwise contact observations, a substantial attention will be paid to account for statistical structure of observed fields and uncertainty in observations.

Periodic formalized surveys among the Olympic forecasters will complement the results of objective verification. Year by year interviewing of the forecasters should help to understand how forecasters’ needs are changing. For example, how understanding of weather processes (conceptual models) or use of EPS products evolve. Analysis of these surveys is a part of social impact assessments, because the forecasters themselves or, e.g., venue managers might be considered as specific users.

Participants and Governance/Project Management

First meeting of potential participants of FROST-2014 project was held in Sochi on 1-3 March 2011 (more details are available in the meeting report at The Science Steering Committee (SSC) composed of the leaders of participating organizations or their representatives was decided to oversee the project.

The tentative SSC membership list is presented below:

Dmitry Kiktev  -  ChairHydrometcentre of Russia, Roshydromet
Detlev MajewskiCOSMO
Tiziana PaccagnellaTIGGE-LAM
Andrea MontaniARPA-SIMC
Paul JoeWWRP Nowcasting Research Working Group
Stephane BelairWWRP Working Group on Mesoscale Weather Forecasting Research
George IsaacSNOW-V10, EC
Roy RasmussenNCAR
Trond IversenHirlam/Aladin
Pertti NurmiJoint Working Group on Forecast Verification Research (JWGFVR)
Yong WangCentral Institute for Meteorology and Geodynamics (ZAMG)
Jian SunCMA
Donghai WangCMA/Center for Analysis and Prediction of Storms, University of Oklahoma
Dmitri MoisseevHelsinki University/Vaisala
Peter RomanovNOAA
Valery LukyanovHydrometcentre of Russia, Roshydromet - Chief Meteorologist of «Sochi-2014» Olympics
Michael TsyrulnikovHydrometcentre of Russia, Roshydromet – NWP

Four Working Groups were established to deal with various components of the project more specifically:

WG1: Observations and nowcasting (including Verification)
Chair: V.Lukyanov/D.Kiktev, International Co-Chair: G.Isaac
WG2: NWP, ensembles and assimilation (including Verification)
Chair: M.Tsyrulnikov/G.Rivin, International Co-Chair: A.Montani
WG3: IT (including graphical tools, formats, archiving and telecommunication)
Chair: D.Kiktev
WG4: Products, training, end user assessment and social impacts
Chair: I.Rozinkina/D.Kiktev

Implementation Plan

Schedule for Implementation includes:
• Concept document;
• Installation of equipment;
• Preparation and testing of forecasting systems;
• Trainings and Workshops;
• Formal Plans.

Winter 2011-12 is the pre-trial period and winter of 2012-13 the trial period, so that all the technologies should be implemented and tested during these two periods. “Pre-olympic” would be the period just before the Olympics where final tuning is done. Data collected during the following winters will be used for high resolution weather forecasting systems testing. The system will be revised and further tested. During the winter of 2013/14, the Olympic Winter, the formal evaluation of nowcasts and forecasts will be made. A summary of lessons learned and results achieved will be presented at the concluding WWRP RDP/FDP meeting after the Olympics.

Preliminary implementation plan is presented in Annex 3. The experience of the pre-trial period (winter 2011-2012) will help to specify further details and needs (sufficiency of the existing data channels, forecasters’ requests for more or better products, modifications of data exchange protocol, etc).

Road safety nowcasts and forecasts are of great interest for Olympic logistics and potentially can be a part of the project. However, it might be difficult to provide efficient support of this activity from the hosting side other than data storage and access facilities, as Roshydromet doesn’t have its own road AMSs and by now this activity has been beyond the traditional scope of Roshydromet.