Forecasting the uncertainties: Probability

TheWxMeister Wonders

I could simply tell you that it’s going to snow tonight. I’m confident in that. I don’t have uncertainty about it. It’s a deterministic forecast.

Deterministic Forecast: A forecast that does not include expression of the associated uncertainty.

American meteorological society

How useful would that forecast be to you, though? I know that if I don’t tell you anything else, you’re immediately going to ask me, “How much snow are we going to get?” You’ve just entered the world of uncertainty. To give you useful information, I’ve got to provide you with an answer that incorporates the uncertainty that exists in the science.

Probabalistic Forecast: A forecast of the probability of occurrence of one or more categorical events.

american meteorological society

News flash – Meteorology is an inexact science!

We make forecasts for some moment into the future for a very complex system that contains many variables. We’ll never be able to create a perfect forecast; we don’t even have the ability to monitor the current state of the atmosphere perfectly! We simply do not, nor will not during my lifetime, have the ability to measure what every molecule in the atmosphere is doing at this moment. Imperfect data about what is happening right now is going to lead to imperfect forecasts for some point down the road.

That being said, there are elements of uncertainty in each forecast that meteorologists across the world create. Depending on the weather situation, terrain, and rate of change in the atmosphere, the amount of uncertainty may be minute and not really impact what you experience in daily life, or the uncertainties may be significant and we may not be very sure how things are going to play out.

We try to incorporate that uncertainty into our forecast and broadcast presentation a number of different ways, depending on the specific component of the atmosphere we’re uncertain about and the degree to which the uncertainty exists. You’re very familiar with the snow forecast maps that will inevitably show up on every TV broadcast in Colorado when a winter storm is moving in. Here are mine for tonight’s incoming storm.

There’s a general consensus in the broadcast meteorology/news industry after focus groups and research that this is the clearest way to convey snowfall expectations over a large space (tv coverage area), over a certain length of time (duration of the storm), while incorporating uncertainty (notice we’re forecasting ranges). However, is this the extent of the uncertainty that exists in my forecast today? No! It’s more complicated than that.

If I let you into the weather center with me while I’m making my forecast you’d see me looking at a number of different computer models, consecutive runs of some of those models to see how they are changing over time, and searching notes for historical data on how similar storms behaved in the past. My process is lengthy because I’m trying to evaluate the level of uncertainty that exists in the output of our numerical weather prediction (computer models) suite.

Much work is being done across the meteorological enterprise regarding how to effectively use probabilities to provide the most accurate forecasts possible consistently over time. The National Weather Service is in the third year of providing experimental winter weather forecast information. The purpose of the experiment is to provide the meteorological community, emergency managers, and other partners with a range of snowfall possibilities, better communicate forecast uncertainties, and to enhance decisions that are made before and during winter weather events.

Here’s a look at this experimental forecast data from the weather service for our incoming storm. The probabalistic forecast is presented as the probability (that is, the likelihood) that snow will equal or exceed specific amounts during the specified time period of a specific storm. These forecasts are based on many computer model simulations of possible snowfall totals.

This provides a clearer picture as to the amount of uncertainty that’s present in the forecast and hopefully gives you a picture into why sometimes the ranges we present in snow forecasts are smaller or bigger than other times, and an idea of how I landed on my ranges for this storm. For example, in Colorado Springs, you can see there is nearly a 50% probability of getting six or more inches of snow in Colorado Springs. I went with 3″-5″ for the central part of town…because some of it will melt, some of it will compact. Even if we do get six inches, which is most certainly a possibility as you can see in the image above, I have other reasons for you probably not measuring that much or shoveling that much. There is a lot that goes into our forecasts! Here are the probabilities for a few other locations across southern Colorado for Wednesday night and Thursday.

Copyright 2019 Nexstar Broadcasting, Inc. All rights reserved. This material may not be published, broadcast, rewritten, or redistributed.

Trending Stories