When Predictions Become Unpredictable: An Uncertain Harvest Weather Forecast

The three layers of forecasting

First up, the forecast funnel. This tool describes the way forecasters focus their attention when looking to predict future weather patterns. Imagine a giant funnel, split into 3 horizontal sections. The top—and biggest section—is the global scale. This is the easiest to predict because there are no geographical boundaries within the numerical models we run. Any variable that could affect the weather—carbon emissions, temperature, dew point, humidity, radiation—does so first on a global scale.

If you take a step down in the funnel to the middle section, complexity increases as regional boundaries and the potential movement of weather events come into play, typically covering a three-month timeframe with higher uncertainty.

Now let’s step down even further into the smallest section of the funnel. This section encompasses area-specific forecasts, and is what you see being used for the 7-day forecasts on the news and your weather apps. Here, forecasters have to include information about the region they’re looking at when they run the numbers.

What’s the ocean got to do with it?

You may have heard of jet streams, the narrow bands of strong wind in the upper atmosphere that encircle the globe. Ocean temperatures can affect how and where the jet stream moves, causing intense weather events even in areas thousands of miles away from any ocean. This is because jet streams move from east to west. Since El Niño is located in the Pacific, it pushes the jet stream towards the U.S., which in turn shifts everything else on the mainland.

You probably remember the extreme temperatures we experienced across the country this summer. Believe it or not, this was due to the jet stream being located farther north of where it might typically be for a long period of time, which in turn created a ridge, or a high pressure system, in the central U.S. Now, the location of this ridge has shifted, but it’s difficult to predict where exactly it will move. Most forecasts predict that it will be directly over the U.S., while others show it located somewhere completely different. This contributes even more to the uncertainty surrounding this season’s forecast.

The top—and biggest section—is the global scale. Forecasters first look at the weather globally, which is actually the easiest to predict. This is because there are no geographical boundaries within the numerical models we run; it’s simply a matter of how much energy goes into the model and how much comes out. 

Think of it like balancing a checkbook. Any variable that could affect the weather—carbon emissions, temperature, dew point, humidity, radiation—does so first on a global scale. For example, I know that this year will likely be the hottest year on record, and next year will likely surpass it, because of these one-dimensional global models. It’s simply a matter of balance. 

If you take a step down in the funnel to the middle section, this is where things start to get tricky. Here we have to introduce regional boundaries to the equation, and our timeframe gets smaller. The middle section encompasses about three months into the future, and we now get into the subseasonal predictability gap. This gap describes the period of time when we’re not perfectly certain what will happen, and things get harder to predict. Based on trends and how consistent conditions have been, we can tell that this harvest will be wetter than last year, and that things will continue to be warm, but that’s as far as our assurance goes. 

Now let’s step down even further into the smallest section of the funnel. This section encompasses area-specific forecasts, and is what you see being used for the 7-day forecasts on the news. Here, forecasters have to include information about the region they’re looking at when they run the numbers.