One of the most common, but least understood, elements of a weather forecast is the probability of precipitation. You're probably used to seeing forecasts that claim a "30% chance of rain" or a "70% chance of snow."

What does the "probability of precipitation" exactly mean?

The basic definition is the likelihood of measurable precipitation (0.01 inches or 0.254 mm) occurring at a location during a specific timeframe.

However, there are some factors we need to break down further, as there are a couple of ways these probabilities can be determined. Also, there is a difference between human-generated probabilities and model-generated probabilities.

Defining the Timeframe

The phrase "30% chance of rain" means nothing if a timeframe is not also provided. When you see a National Weather Service forecast or a news station forecast, typically you'll see a percentage chance of precipitation for a daytime period and a nighttime period over the first few days.

Daytime and nighttime periods as defined in weather terms typically will cover 12-hour intervals (6 am to 6 pm is common). So if you see a 30% chance of rain forecasted for Thursday night, that means that there is a 30% likelihood that rain will fall in the forecast location sometime during the overnight period.

Here is an example of a 12-hour U.S.-wide probability of precipitation forecast for a nighttime period.

Further out (beyond 3-5 days), forecast services will usually provide a probability of precipitation forecast for an entire 24-hour period rather than separate day/night periods. This is because confidence in the timing of precipitation decreases the further out in time you go.

Forecaster Confidence x Areal Coverage

To actually determine a precipitation probability, we must first know the size of the area we are forecasting. This could include an entire county, a city, a ski resort, a single point on a map, or a mountain summit. 

National Weather Service forecasts cover 2.5 km grids, which is an area of roughly 1.5 square miles. However, news stations and media outlets typically forecast precipitation probabilities for an entire city or town when they present graphics to their users, since they have to cover a much larger geographic area.

Next, we must factor in a forecaster's confidence of rain or snow occurring in a given forecast area during a defined interval.

Once the confidence level is determined, a forecaster develops a probability of precipitation percentage by using the following formula: Probability (%) = Confidence (%) x Area (%).

Let's say there is a 40% chance of snow on Saturday night. Here are three different examples of how a forecaster could determine this percentage.

1) The forecaster is 100% confident that snow will fall over 40% of the forecast area. (100% x 40% = 40%)

2) The forecaster is 40% confident that snow will fall over 100% of the forecast area. (100% x 40% = 40%)

3) The forecaster is 80% confident that snow will fall over 50% of the forecast area. (80% x 50% = 40%)

Using forecaster confidence and areal coverage is a good method for human forecasters. But what about weather model-generated probability of precipitation forecasts?

Weather Model Probability of Precipitation Forecasts

Weather models are a little bit different because they involve forecasts that are generated by computer models every 1 to 6 hours depending on the model. Each model also covers different-sized grid points, ranging from 2.5 km to 27 km.

So the probability of precipitation for a model-defined gridpoint is simply the percentage chance of 0.01 inches of rain or melted snow occurring within that grid point in a given timeframe.

At OpenSnow, we have developed our own proprietary model blend where we take numerous simulations of both high-resolution and low-resolution weather models (both of which have their advantages and disadvantages) to determine the probability of precipitation "downscaled" to highly localized points.

This is important in mountain environments, where precipitation and other types of weather can vary significantly by elevation over short distances.

Interpreting Precipitation Probability in OpenSnow

We provide the probability of precipitation forecast by the hour for the next 10 days to help subscribers get more dialed in on planning outdoor adventures. 

These probabilities are also for very precise points, so unlike the previous human-forecasted method, we can say this probability of precipitation forecast is simply the confidence level of rain or snow falling at a given hour.

In the example above, the probability of precipitation forecast is in the 25-35% range during the morning hours before increasing to roughly 40-50% after noon.

If you wanted to go hiking, you could use this information to determine that the morning would be a better time to get outside than the afternoon, although you'd still want to pack a rain jacket in the morning.

---

Questions? Send an email to [email protected] and we'll respond within 24 hours. You can also visit our Support Center to view frequently asked questions and feature guides.

Alan Smith