If you were living in the Tampa, Florida area during Hurricane Irma in 2017, you may have seen news forecasts showing the track of the tropical storm before landfall. These early forecasts indicated a track over Florida’s southeastern coast and would have made you confident that your area would not be hit, although now, we know that to be false. So what happened? How did early forecasts get the storm track so wrong—leaving residents residents along the western coast of Florida wholly unprepared?
The answer is that what viewers were given was a deterministic forecast, the type most often utilized by the media. This is because deterministic forecasts are straightforward and easy to understand as they tell us what is most likely to happen. However, they do not show the probabilities of other, potentially likely outcomes.
But there is another type of forecast that provides insight into other possibilities that could occur. This is known as a probabilistic forecast. Probabilistic forecasting is a concept that takes some explaining, which is why it isn’t shown as much in the media. But it does provide a full range of possible outcomes, making it the best option for businesses and individuals who need to make crucial decisions during heavy storms like hurricanes.
To understand the difference between the two types of forecasts, let’s look at how weather forecasts are created. First, everything that affects the atmosphere has to be measured. For this purpose, we have hundreds of thousands of weather stations on the surface of the Earth, as well as weather balloons for upper-air observation. These are placed across hundreds of locations globally and their data is collected a few times per day. Data is also obtained from satellites, radar, ground or soil observations, the oceans and elsewhere.
Every day supercomputers collect and organize billions of measurements from these observation points. This data covers initial conditions for variables such as temperature, air pressure, moisture, wind, water levels and much more.
Using complex mathematical equations, these supercomputers perform three quadrillion calculations per second to numerically compute weather forecasts out to 15 days. Meteorologists refer to the results as deterministic model output.
Since our numerical models are based on high level math and physics, we should expect 100% accuracy, right? But we cannot, simply because the Earth’s atmosphere is chaotic. Deterministic weather models are sensitive to the most miniscule changes to initial conditions, which are difficult to accurately measure. This means errors are inevitable, and we need to account for these slight changes by implementing ensemble modeling.
As ‘ensemble’ basically means a set or a group, an ensemble forecast is a set of forecasts (each with a slight variation in initial conditions) created using the same numerical model. The result is a group of forecasts that represents a range of possible future weather conditions, aka a probabilistic forecast—showing the possibilities that may not be the most likely to occur (deterministic forecast) but could happen nonetheless.
Let’s say your business is sensitive to freezing temperatures. In this case, probabilistic forecasts provide insight into all risks, which a deterministic forecast (showing the one most likely possibility) can’t.
For example, your weather forecast from an app, TV or online provides you with a deterministic forecast that says tomorrow’s low will be 35° F (the most likely outcome). Looking at this same forecast probabilistically, we see there is a possibility of variance, both warmer and colder. What your forecast doesn’t tell you is that there is also a 30% chance temperatures will fall to freezing or below.
For your business, the loss associated with freezing temperatures, even just three out of ten times, may present too much risk. This is where a mitigating strategy may need to be implemented.
Many of StormGeo’s solutions contain a wealth of probabilistic information on everything from wave height to thunderstorms. We also have a new Long Range Planner tool that allows clients to define their own unique weather hazards and probabilistic criteria.
Deterministic forecasts lend themselves to simple communication, but they only provide insight into the most likely outcome.
With a probabilistic forecast, we no longer have one possible outcome. By generating a range of possible outcomes, the ensemble method shows how likely different scenarios are, including those that would be detrimental to your home or business. StormGeo provides a broad range of probabilistic forecast products that decision makers can use to take optimal decisions for their business.
StormGeo's lead forecaster, Chris Hebert provides the forecast for the 2023 Northwest Pacific Typhoon Season.Shipping | Oil & Gas | Renewables & Energy Markets | Other Industries
In this regularly updated forecast, StormGeo's Hurricane Forecasting team looks at current conditions across the Atlantic and Pacific as well as long-range patterns to identify...Shipping | Oil & Gas | Renewables & Energy Markets | Other Industries