Inside the lab

For more than thirty-five years in this business I have wrestled with the conundrum of predicting summer time showers and thunderstorms. Accurately forecasting a shower or thundershower hinges on monitoring the ever changing atmospheric boundaries and vertical profile.

In the warm season, convective storms seem to randomly blossom in the heat of the day and then drift in the relatively light steering winds overhead. A valued colleague in Indianapolis once declared that thunderstorm formation is not random; there is a trigger to ignite the towering cumulus cloud. Boundaries can be subtle. Identifying where a specific boundary might form two days out is next to impossible.

Regrettably, meteorologists are still left with providing clients the proverbial thirty percent chance when it comes to crafting a forecast on sultry summer day.

So, decades after I began my forecasting career, the precipitation probability dilemma remains. If you tend to be an optimist, you could infer that there is a seventy percent chance for no rain.

Forecasters in National Weather Service office collaborate the seven day forecast to mosaic the weather regionally and nationally. Twice a day they present their case for the decisions they have made in formulating your forecast.

<Forecast discussion


  • bsimon

    That reminds me. As the metro area expands, does your job get more complicated or less?

    If, for instance, you predict a 20% chance of thunderstorms tomorrow, does that mean there’s a 20% chance the whole area will receive thunderstorms, or there’s a 100% chance that 20% of the metro will get thunderstorms, but we don’t know which 20%?

  • Craig Edwards

    Very astute. Yes there can be resolution added to the first and second period of the forecast for the expanded metro area. within 12 hours forecasters can determine boundaries that will ignite thunderstorms. Thus for the afternoon and evnening hours, the meteorologist is able to say, for example, there is a 50 percent chance for thunderstorms in Blaine and a 20 percent chance in Burnsville.

    Beyond 18 hours, the best forecast is made with a broader bursh.

    When you hear a forecast that says a 30 percent chance for thunderstorms it means your specific location has a one in three chance of getting measurable precipitation. Over time, relieability verification suggests that you will be dry seven out of ten times.

    Thanks for asking. Craig

  • Chris B. Critter

    In meteorological modeling, it’s not too difficult to pull out variables for a point (grid cell, actually) for various parameters such as temperature, wind, humidity, pressure. Rainfall amount is a bit more difficult, but not too tough, from what I can assume.

    But I’ve been recently interested in how the models compute *percent* of precipitation (POP). That’s essentially a probability. How is that done?