Why do forecasts differ?

How come different weather forecasts sometimes differ so drastically in their predictions? This is a complex question, with multiple variables at play, but we’ll tackle the main reason in this article.

When you’re using a forecast to make crucial decisions for day-to-day activities, it is important to use a source of weather information that can be trusted. To do this, the first question that should be asked is “How is the forecast produced?”. There are three primary categories of forecasts: automated, hybrid, and meteorologist-only forecasts.

Forecast Types

An automated forecast is derived entirely from a computer model where there is no human input. These models are known to have biases (e.g. too sunny, not handling thunderstorm activity well, too cold at night, etc). If the forecast is completely automated, these biases go uncorrected as there is no meteorologist correcting them. These forecasts are often used by agencies with limited meteorological expertise – they simply load in raw output from one or more models to produce the desired weather information. To the surprise of many, the forecast you’re reading may not have been written by a meteorologist, it could just be computer output!

A hybrid forecast is where a weather model is used as the base forecast and then a meteorologist adjusts parts of the forecast to improve its accuracy. Depending on the source, meteorologist intervention may be extensive, or limited. Meteorologists are often tasked with focusing on high-impact weather, so their time may be mostly directed at parts of the forecast with the most impact, leaving less important aspects unadjusted. Hybrid forecasts are the most common type of forecast produced by government and private weather agencies.

The final type of forecast is a meteorologist-only forecast. These forecasts are produced entirely by a meteorologist. The meteorologist may use weather models to help produce the forecast, but there is no ‘baseline’ forecast that is adjusted. The meteorologist composes the entire forecast each time. These types of forecasts are rare because they are very time consuming to produce.

The graphical forecast editor is a tool that allows meteorologists to produce hybrid forecasts by adjusting a base forecast from a weather model.

The Role of Forecast Types

The forecast type plays a large role in why forecasts differ. Automated forecasts are produced strictly by models and these models often update four or more times per day. As a result, the forecast can change numerous times per day, often drastically, because no meteorologist is adjusting it to ensure consistency. In addition, there are many different models available. If one forecast uses the American model (called the GFS or Global Forecast System) it may be different than the Canadian model (called the GDPS or Global Deterministic Prediction System). As you get closer to the forecast date, the model’s forecasts often become similar, but not always. During complicated events, model forecasts may differ even as an event begins.

While automated forecasts are most prone to differ, hybrid forecasts can as well. Some large agencies have rolling shifts of different meteorologists, so if a new shift comes on and doesn’t like the previous forecast, it can be changed. However, when meteorologists are overseeing the forecast process, these changes are often less drastic as they can make smaller adjustments. Meteorologist-only forecasts are rarely distributed publicly, so you may not have used one. However, they are often more consistent than other forecast types since there are no underlying models that frequently change.

Weatherlogics forecasts verified against public forecasts.

The difference in forecast types, and the underlying models used, can create discrepancies in forecasts issued. Here at Weatherlogics, we have our own in-house model which combines all available individual models to produce an optimal baseline forecast. Our meteorologists then extensively adjust this model output using their knowledge of local terrain, known model limitations and scientific forecasting techniques. This produces an accurate, consistent, and reliable forecast, which is why our clients find a bit of money spent on better forecasting pays off greatly in better decision making.

Leave a Reply

Your email address will not be published. Required fields are marked *