If you spend much time on social media, you have probably seen screaming headlines on Facebook, Twitter, or elsewhere about impending extreme weather. In the last few weeks I have seen wild stories about a late-season tropical storm forming in the Caribbean and an Arctic outbreak and snowstorm heading through the eastern US all the way to the Gulf. Neither of these had any real chance of occurrence but the people who post them are looking for clicks and attention. Garden Professors’ blog readers already know about misleading information in social media (Epsom salts and cardboard mulch, for example) but may not know much about weather forecasting and how it is misused in these click-bait posts to gain attention for weather that probably will never happen. This blog post will describe how weather forecasts are made so that you can understand which forecasts are the most reliable and useful for gardeners and others who work and play outdoors.
Types of weather forecasts
In general, forecasts can be categorized in several ways. They can be categorized by time period (nowcasting, short-range, medium-range, long-range) or by method (based on evolving weather conditions, statistics of past weather, or numerical weather prediction by computers). Forecasts related to time just indicate how close to the actual weather occurrence you are making the prediction. A “nowcast” is a forecast for the immediate future up to about six hours ahead. Short-range forecasts usually cover 1-3 days, medium-range 3-7 days, and long-range forecasts are made more than a week ahead. Long-range forecasts tend to provide more general descriptions of climate patterns and departures from climate conditions rather than specific weather conditions because accuracy decreases as you go farther ahead in time. This is one clue that a specific social media forecast map is likely to be unreliable because we cannot make specific predictions of atmospheric conditions more than about a week ahead due to lack of sufficient data, simplifications in the computer models used to make predictions, and chaos in atmospheric conditions that cannot be easily described by the input data to forecast model programs.
Methods for making weather forecasts can also vary. The simplest one says that today’s weather will be the same as what happened yesterday. It works well in regions and seasons when the weather does not change much from one day to the next but is poor where more dynamic weather occurs. One step up is what we call an advective forecast, which predicts changes in temperature, moisture, and other properties due to the horizontal movement of air by wind. Forecasters determine this by looking at wind direction and the temperature and other conditions of the air that will be arriving from upstream. If air is blowing in from the Gulf, for example, the weather is likely to turn warmer and more humid, while air arriving from the Arctic is likely to be colder and drier than what is already present. Similarly, you can sometimes use observations of clouds to make simple weather forecasts based on how they change over time. Climatological forecasts look at past weather occurrences on the dates for which you are forecasting and describe the statistical likelihood of specific temperatures and rainfall conditions. Some also look at analogs from previous weather situations that look like the current conditions and predict that the future weather conditions will occur again based on past performance.
Using computers to predict weather
Most forecasts used by modern meteorologists use numerical weather prediction to forecast future weather patterns by entering observational atmospheric data into complicated computer models that use the physical equations of motion to determine how the atmosphere will change in response to moving air, development of clouds and precipitation, and interactions with land and ocean surfaces at the ground and put the results into map formats that provide a graphical depiction of what the weather will be like in the future that can be interpreted by meteorologists to provide the expected local conditions (these are called deterministic forecasts). There are multiple weather models that are run by different groups, including US-based models as well as those from Europe, Canada, Japan, and other countries with large computing facilities.
Most models are run for a variety of different starting conditions to give a range of possible outcomes, and the combination of different model runs and different models lead to what is commonly called “spaghetti models” that show each individual model solution on the same map. The closer together all the runs are in the future, the more confidence we have that the models are in agreement, but if they are spread out, forecasters have low confidence in what will occur. Expected impacts at any one location change depending on which computer run is used. Many of the extreme weather events shown in maps show by social “media-rologists” are from a single computer run far in the future showing the worst possible case, even though there may be 99 other computer solutions that show something much less severe. The social media folk are showing whatever extreme case will get them the most attention (and clicks), which allows them to monetize their accounts by promoting the worst-case event, no matter how unlikely, not the most likely one. Long-range forecasts which look at all the solutions may use statistics to determine where the storm is most likely to go and base the probability on the range of individual model runs we see (these are called probabilistic forecasts).
How accurate are weather forecasts?
Despite the common belief that weather forecasts are seldom right, the statistics show that they are very accurate. People generally only remember the few occasions when they are wrong, not the many when they are correct. Today’s forecasts are better than in the past because computers are getting larger and more complex and more data are being collected to feed into the models, including satellite data that fill in holes in surface data over the oceans. A daily forecast now is likely to be good at least 7 days in the future, compared to 5 days twenty years ago. In the future, we may see even better results using artificial intelligence (AI) to fine-tune forecasts, as we did this year with hurricane track forecasts. But we will never be able to provide an accurate daily weather map 90 days in the future, so don’t count on the long-range forecasts to tell you what to expect on the day of your summer garden party when you are planning in January or even in April). If you need that forecast to plan for the likely weather, using climatological information is your best bet.
The best weather forecasts for gardeners
Your best source of accurate weather forecasts in the United States is the National Weather Service. Most other countries provide forecasts from their own government weather services. If you need to plan for specific times for gardening projects such as spraying products that require dry conditions or mowing, hourly forecasts for up to 6 days ahead can be found using the information at https://site.extension.uga.edu/climate/2018/03/where-to-get-hourly-weather-forecast-information/. Keep in mind that the farther ahead in time you get, the less accurate they will be. Weather information from private forecasters, including broadcast meteorologists and commercial companies that provide handy apps for cell phones, mainly base their forecasts on NWS predictions but often add value by providing descriptions of conditions, specific impacts on sectors like agriculture and transportation, or pretty graphics to make the forecasts look more appealing. Gardeners should keep in mind that most weather apps on phones are fine for daily planning but are not suitable for rapidly changing weather conditions like severe weather, since they are not updated often enough to factor in these short-term events.
Gardeners who want to maximize their outdoor activities need access to accurate and believable forecasts so they can plan the use of their time effectively. By understanding the nature of weather forecasts and the dangers of exaggerated predictions of likely future weather conditions, they can make the best use of their time and enjoy their garden work without worrying about overhyped extreme events that likely won’t happen.
