Chapter 10: The Forecast Funnel

The forecast funnel model is an effective and systematic approach to weather forecasting. You will start your analysis at the global scale, then, like a funnel, narrow your focus down to the meso-scale. The amount of time you should spend analysing each scale is inversely proportional to the extent of the scale – spending only a little time analysing the global scale, and far more time analysing the meso-scale.  The majority of your time should also be spent analysing the present weather to make a ‘now-cast’, with less time spent the further out you forecast. For short-term forecasts (0-6 hours), you may be able to rely more on extrapolation of weather observations, while long-range forecasts (e.g., 3 days) will involve NWP models to a much greater extent.

Global Scale

At the global scale, you will analyse the jet stream and determine how it will affect your area. For example, a strong South-Westerly flow will likely bring warm and moist air. It is very important to understand what is going on at the global scale before moving to smaller scales.

Questions to Ask:

  • What are the overarching weather patterns, and what effects will they have on my forecast area?
    • Where are the jet streams, and how are they evolving?
    • Where are the Rossby Waves, and how are they evolving?
      • Rossby Ridge (global anti-cyclonic flow); or
      • Rossby Trough (global cyclonic flow).
    • Where are the centres of locally high-speed jets?
    • What air masses are influencing my forecast area?
    • Find troughs and ridges (lows and highs). How are they evolving? Look for areas with tight gradients.

To answer these questions, you can use satellite imagery and upper-air analysis charts. The water vapour image is particularly good at highlighting the flow pattern. The 250hPa and 500hPa charts show troughs and ridges being guided along the jet streams.

Troughs and ridges will either be stationary, progressive, or retrogressive. A stationary pattern is when the troughs and ridges remain in the same pace over several days or longer. A progressive pattern is when the troughs and ridges evolve from West to East. If a system is retrogressive, the jet stream (longwave) undergoes a discontinuous East to West movement whereby an existing trough weakens and moves eastwards. As the trough retrogresses, an upstream shortwave will strongly deepen, developing into a stationary longwave trough West of the original longwave trough position. Discontinuous retrogression is usually associated with strong development of cyclonic circulation (cyclogenesis) as the upstream shortwave intensifies.

Tools to Use:

  • Satellite imagery (water vapour, infrared)
  • Upper air analysis charts (250hPa, 500hPa)

Synoptic Scale

At the synoptic scale, you will analyse the locations of low- and high-pressure centres, and frontal systems.

Questions to Ask:

  • What air masses are affecting my forecast area?
  • Is the flow upslope or downslope?
  • What is the general surface flow direction?
  • Are there areas of low or high thickness (warm or cold cores)?
  • Are there any warm noses (anti-cyclonic curvature in thickness lines)?
  • What is the problem of the day?
    • What and where are the weather-producing features?
    • How are these features likely to evolve?
    • What kind of uncertainty is there?

Determining the Problem of the Day: It is important to relate the various weather features of the synoptic scale to the global scale, and determine if they will affect your forecast area.

After this is done, it is important to check the quality of the NWPs you intend to use. This can be quickly accomplished by comparing satellite and other observation data with the initial panels of the model runs.

Next, the evolution of the features of interest need to be considered. NWPs are the main tools to do this.

Finally, armed with this information, the forecaster will likely have a good perspective of the general problems of the day, including the confidence level of the forecast evolution.

Tools to Use:

  • Satellite imagery (infrared with 500hPa contours)
  • Upper air analysis charts (700hPa, 850hPa)
  • Surface analysis modelling output (GEM, 1000-500hPa thickness panel)
  • Weather radar
  • Soundings

Mesoscale

At the mesoscale, you will determine how the local topography will affect your problem of the day.

Questions to Ask:

  • What are the proximities to mountains, hills, valleys, deserts, and various bodies of water in the forecasted area?
  • What is the direction and magnitude of vertical motion on the local topography?
  • What small-scale influences will there be?
    • Katabatic winds?
    • Gap winds?
    • Land-sea breezes?
  • Is the local air mass going to be wet or dry?
  • How extreme or benign will the weather in my forecast area be?

Writing a Weather Forecast

Most weather forecasts begin with a statement about the general weather pattern. Then it should include key information about weather factors that influence snow instability. This includes information about temperature and freezing level, precipitation (type and amount), and ridgetop winds. In short-term forecasts, you may be able to provide specific values and ranges, or in long-term forecasts only very general statements in the trends of overall weather patterns, temperatures, and precipitations. A good weather forecast should also include a statement about confidence. This is an opportunity for the forecaster to express their degree of certainty in their weather forecast. Forecast confidence is affected by a number of different factors, including:

  • Forecaster knowledge and experience.
  • Familiarity with the forecasted area.
  • Forecast range.
  • Quantity or quality of observation data.
  • Weather model disagreement.

Forecasting Time Table

The tools used to forecast are dependent on the range of the forecast. A short-range forecast (6-12 hours) can rely primarily on a method of extrapolating the existing conditions. The forecast parameters in this case can be very specific and detailed. On the other hand, a long-range forecast (5-14 days) relies on computer models and statistical climatological models.

0 – 6hrs. Essentially nowcasting with a heavy reliance on extrapolation

  • Extrapolation of existing regional conditions and systems.
    • Fog dissipation/formation
    • Time of clearing/clouding
    • Precipitation beginning/ending
    • Time of a certain threshold temperature
  • Reliance upon remote weather stations, observation, and mesoscale analyses.
    • Satellite imagery
    • GFA
    • Remote weather stations
    • METARs
    • Mesoscale models (LAM, WRF)

6 – 12hrs. Equal reliance on extrapolation and NWP models

  • Extrapolation of existing regional conditions and systems + NWP model movement of systems.
    • Precipitation development along a front
    • Influence of shortwave entering region
    • Cloudiness trend
    • Temperature trend
  • Reliance upon remote weather stations, observations, and fine-resolution NWP models
    • Satellite imagery
    • GFAs
    • Remote weather stations
    • METARs
    • Modelled tephigram output (progr-tephis)
    • Mesoscale models (LAM, WRF)

12 – 48hrs. Increasing reliance on NWP models and increasing generality of forecast.

  • Development, intensification, and dissipation of synoptic scale systems become important. When approaching 48 hours, several thousand kilometres must be considered.
    • General trend of long-waves, long-wave movement, long-wave intensification
    • Placement of surface systems, and interference about weather
    • Specific timing of precipitation, and type of precipitation
    • Precipitation type: rain, snow, sleet, and so on
    • Is a shortwave approaching? Will precipitation occur?
    • Expected conditions after fronts
    • Cold or hot air surges, or wind surges
  • Primarily rely on fine-resolution NWP models
    • Mesoscale models (LAM, WRF)
    • Regional models (GEMr)

3 – 4 days. Nearly complete reliance on NWP models, with very general temperature and precipitation statements.

  • Placement of surface systems, inferred from upper level computer models
    • General statement about temperature
    • General statement about precipitation
    • General statement about cloud cover
  • Primarily rely on regional and global NWP models
    • Regional models (GEMr)
    • Global models (GEMg, GFS)

5 – 14 days. Decreasing reliance on computer models, increasing reliance on statistical and ensemble models.

  • Very general statement about probabilities and trends.
    • Above or below normal temperatures
    • Probability of precipitation
  • Primarily rely on ensemble forecast products.
    • EPSgrams
    • Spaghetti plots
    • Temperature anomaly charts
    • Precipitation probability