Many times we find ourselves needing a weather forecast beyond the next couple of days. Maybe you are planning an outdoor party, trip or hoping for a snow day! What is the best way to determine the likely weather for a date further out than the reliable weather forecast?
The first question we ask ourselves is whether I should rely upon long-term forecasts or if understanding the typical weather averages for that area has more value. The answer is not as clear cut in all cases and it almost always relies upon availability and accuracy of the forecast data as well as your needs for the data and timeframe. In this article we will discuss when is the right time to utilize long-term forecast tools.
Weather Forecast Accuracy
Let’s start with the forecast accuracy. The reason to start here is simple: if the long-term forecasts had high accuracy we would simply always choose them. NOAA explore this question and have come up with the following rule of thumb. A 5-day forecast is correct about 90% of the time, 7-day is 80% and beyond that the forecast becomes less reliable. If you read this as a binary right/wrong then statistically speaking this doesn’t say a lot for our forecasting capability outside of one week and you simply should not trust it. But this raises a simple question that is almost impossible to answer which is “what determines if a forecast is accurate?” If a temperature prediction is off by 3 degrees is that consider a failed prediction by the forecasters in a 10-day forecast? Over what period of time are we judging the forecaster to say they are only correct 50% of the time? This article discusses when exactly forecasters are considered wrong and why they are considered to be slightly wrong most of the time. It’s simply impractical for us know the current state of the atmosphere and be able to run a computer model for the whole atmosphere at the molecular level!
For the sake of practical comparison here is a weather forecast screenshot taken from the Visual Crossing Weather Data Query page.
Our example is to compare 2 different forecast styles for 15 days away. In this example we will choose Valentines Day 2020 which is 2 weeks after the date of this article’s creation. The forecast is telling us that temperature will range between 35 and 50F. It is also telling us that it may get 0.9 inches of rain.
Now let’s take a look at a historically-based average scenario. Here is a query using the Visual Crossing Average Weather tool.
These ‘typical’ weather conditions are built from 30 years’ worth of data and you can see it shows a wide variation of possible temperatures ranging from 21-77F. Even in the predominant statistical area that represents 5 our of 7 days per week, the daily maximum temperature range about 25F. As for Precipitation it simply tells us that you can expect it to rain once in a week. From averages though, we can see that the Temperature is statistically likely to be around 42F. This is not far off from our 15-day forecast and if you are tracking this over years to two different calculations will eventually agree with each other.
“The accuracy of a five-day forecast nowadays is comparable to that of a three-day forecast about 20 years ago, and that each decade, we gain about a day’s worth of skill. Essentially, today’s three-day forecasts are as precise as the two-day forecast of ten years ago.” – Jon Shonk Research scientist, University of Reading
Things are improving!
Different Types of Weather
One note to keep in mind is that we are focusing on temperature which is a consistent variable that you can track and average over time with high rates of sampling for every single location on earth. Rain and Snow are much harder. The Historical Averages can only give you general guidance in most places unless you are forecasting an area that has rainy seasons consistently. Without consistent rainfall to compile a history of, only a 15-day forecast can guide you for a specific days’ chance of precipitation.
TLDR… which weather forecast approach is best?
To determine what is best here, we have to find which tool is best for the purpose intended.
15-day Forecast – While everyone agrees that the 15-day forecast can be inaccurate, it still is the only way to find guidance on what general meteorological conditions are leaning toward for a specific day in the near future. Remember, ‘wrong’ may only be a few degrees whereas the Historical Analysis will give you a very wide range of temperatures to consider.
Historical Averages – If no forecast is available, then a statistical weather forecast is your only choice. But there is a role for them to play even within the 15-day window which focuses entirely on your needs. If you are planning a trip, it is good to know what is possible vs what is likely given current conditions. February 14th may get between 57 and 70F 1 out of 7 days in Herndon VA. This is greatly helpful in planning purposes. The 15-day forecast won’t tell you this. Also, if you are planning a longer visit than any single day, your chances of hitting the average weather start to even out statistically.
If knowing the exact weather is critical to your usage such as a business or government use scenario, we recommend using both. Understand what is likely, utilize the variances or inaccuracy but also understand from historical averages what is possible. Knowing the complete picture of odds is invaluable for planning especially longer term and for longer time intervals.