By far, the most popular question I am asked at any agricultural conference is about the weather and how it pertains to the upcoming growing season. Due to the popularity of that question, the ag industry has been inundated with extended range forecasts from weather enthusiasts to optimistic growers with a passion for weather. You see it at conferences and on Twitter; when a forecast is posted by anyone, a professional meteorologist or not, folks are insisting that this forecast is the most accurate. Creating extended range forecasts with no scientific backing isn’t a recent trend, just look at the Farmers’ Almanac.
How are the forecasts created? How accurate are they? Does the forecaster have a degree in meteorology? Unfortunately, these are questions that rarely get asked about extended range forecasts.
I will be answering many of these issues in the coming weeks, as well as giving an exclusive look into the upcoming growing season at our company’s (Weather Decision Technologies) webinar. For now, I will explain how the forecasts work and how they are created in the hopes that growers start to identify what is scientifically backed and what is not.
Climate science has advanced to the point where long-range forecasts provide a measure of skill out to many months in advance, at least when compared to baseline climatology. Even though meteorologists cannot predict with accuracy whether a particular farm will be above or below normal during an entire season, we can predict this for an area. In general, skill increases as the size of the area of interest increases and the time scale increases, particularly with precipitation which can vary dramatically across short distances and over short time scales. So, a forecast for the southern half of the U.S. for the entire June-August period will be much more accurate than a forecast for Dallas in mid-July made five months in advance.
While many portions of the climate system are chaotic, and thus hard to predict, large scale weather patterns tend to be driven most strongly by more stable variables, such as large-scale ocean temperature patterns.
The biggest driver of weather patterns across North America is the location and intensity of tropical thunderstorms over the central and western Pacific Ocean or at times the Indian Ocean. That region of the world has the largest body of warm water on the planet and is the primary heat engine for tropical thunderstorms. These tropical thunderstorms transport ocean heat into the upper levels of the atmosphere. The location of that tropical convection strongly influences the jet stream pattern over the Pacific, which in turn affects the pattern across North America.
As the climate warms, particularly across the high latitudes of the Northern Hemisphere, the general tendency has been for the jet stream to remain farther north than normal. At the same time, the strength of the flow has been weaker leading to more extreme patterns (bigger more sustained troughs and ridges). The net result is that the overall trend during the last couple of decades has been for a greater frequency of above-normal temperatures in most regions. However, the areas that do end up colder than normal have tended to stay that way for longer periods of time leading to more prolonged cold extremes.
By focusing on portions of the climate pattern that are slightly more predictable, such as ocean temperature and tropical convection patterns, and combining that with climate tendencies, seasonal forecasts can provide reasonable skill, particularly over larger areas. Climate models also continue to improve, and while individual models may have poor skill, by combining multiple models together and focusing on the portions of their forecasts that are more skillful, further forecast accuracy gains are possible.
When viewing a forecast, before even comparing the “accuracy,” make sure the forecasters talk about Pacific Ocean temperatures, tropical convection (thunderstorms), and if they use ensemble climate models. There is no such thing as a “secret formula” for performing sound extended forecasts. Anyone that isn’t doing these things or discussing their sources is neglecting the tools that are peer reviewed and probably shouldn’t be trusted. At that point, one can start comparing the accuracy of various extended forecasts.