Meteorology 5540
Mini-Lecture 9: Temperature Forecasting

I. Factors to consider when making a temperature forecast

  • Climatology: Consider time of year and location
  • Persistence: Understand and take advantage of what has happened recently
  • Airmass: 1000-500 hPa thickness or top of boundary layer temperatures characterize the general thermal character of the airmass.
  • Cloud cover: Effects diurnal range. Consider extent, altitude, timing, etc...
  • Winds: Mechanical mixing limits min AND max temp. Impact on min temperature most pronounced. Local mesoscale circulations such as lake or sea breezesalso influence temperature.
  • Precipitation: Tends to send temperature and dew-point to the wet-bulb temperture. Limits the diurnal cycle. Convective downdrafts can be colder than morning min in some cases.
  • Terrain: High (low) relative to surrounding terrain means a smaller (larger) diurnal cycle. Subtle differences in elevation on the order of tens of meters can have drastic effects, particulary on min temperature. Exposure also important.
  • Other orographic features: Proximity to bodies of water, snow cover, irrigation, etc... that can have pronounced local effects.
  • Consensus does well...use a number of methods
  • Forecasting is very intuitive. There is no one "right" way to forecast. Good forecasters think for themselves and develop their own strategies and methods for various weather patterns.

    II. Relating temperature to "free" atmosphere temperature

  • Use a level near or in the boundary layer (850 mb in the east, low plains, and Pacific coast, 700 mb in the Rockies and high plains)
  • Use analysis or model forecast temperatures
  • Bring down temperature dry adiabatically to estimate maximum temperature
  • Adjust based on individual temperature bias and local effects
  • When detailed sounding data is available, adjust level from which dry adiabatic lapse rate is estimated based on the vertical temperature profile.
  • Method assumes enough surface heating for a well-mixed boundary layer to form. Cloud cover, precipitation, etc... all tend to result in lower maximum temperatures then estimated with this method.
  • For first guess for minimum temperature, bring down temperature along the moist adiabat; adjust for local effects and ambient conditions.

    III. Relating temperature to "FOUS" data

  • During more subtle airmass changes, one can use FOUS output from the models to derive maximum or minimum temperature forecasts
  • First guess for tomorrows max is todays observed (or estimated) max
  • Examine how ETA or NGM T1 temperatures change from today to tomorrow
  • Forecast for tomorrow is todays observed (or estimated) max adjusted by the amount that the model warms the T1 layer by.
  • Assumes similar cloud cover, winds, etc... from day to day. Consider how such features would impact temperature.
  • Can also be used for min temperatures.

    IV. Analogous thickness and pressure method

  • Use most recent analysis chart that is valid for the same time of day as your forecast.
  • Find the location on the analysis chart that has the same thickness (or top of boundary layer temperature) and pressure as that on the model forecast chart for the point and time you are forecasting for.
  • Determine temperature at the analog location on the analysis chart.
  • Adjust analog temperature for changes in elevation
  • Adjust for changes in cloud cover, etc... between the sites.
  • Can be used for max or min temperature
  • A dangerous method during rapidly evolving flow patterns

    IV. Strategies for beating MOS

  • When you expect MOS errors to be small, use MOS
  • When you expect MOS errors to be large, adjust in the direction of the expected errors, but don't go too far out on the limb
  • Know mesoscale and local effects that are not handled well by MOS
  • Be alert for extreme or unusual patterns and phenomena
  • Be very careful in adjusting MOS during "chaotic" weather periods when temperatures can be greatly influenced by random phenomena such as convection.
  • Look for times when ETA is performing significantly better than NGM and/or AVN or periods that MOS may not handle well.

    Updated May 8, 1997