Determining uncertainty: a review of hurricane intensity predictability

In recent years, meteorologists have made strides in predicting the trajectory of tropical cyclones as models and forecasts have become more accurate. However, forecasting intensity remains a much larger challenge; small changes in atmospheric conditions can have a large impact on forecasts. To better understand why some forecasts are more accurate, scientists at NOAA’s Atlantic Oceanographic and Meteorological Laboratory (AOML) are studying the environmental factors that influence tropical cyclone intensity forecasts.

When meteorologists make predictions about a storm, their confidence is measured as “forecast uncertainty.” When a storm is challenging to predict, its forecast is characterized as having high uncertainty. Conversely, when  the models better agree, the forecasts are characterized as “low uncertainty.” Atmospheric conditions can change quickly, often influencing the formation, strength, and trajectory of a storm. But what factors contribute to whether the forecast for a system has high or low uncertainty? AOML meteorologist Jason Sippel and University of Miami researcher Michael Fischer decided to answer this question.

In their paper “Tropical cyclone intensity predictability,” Sippel and Fischer reviewed the literature to analyze what factors have the greatest influence on the level of uncertainty in intensity forecasting. Their findings suggest there are two primary drivers: vertical wind shear and atmospheric moisture.

Vertical wind shear describes the changes in wind speed and direction throughout the  atmosphere. This shear affects how well thunderstorms organize around the tropical cyclone center that drive the hurricane “engine.” Strong wind shear tends to prevent intensification or causes a storm to weaken as its structure is pulled apart. Atmospheric moisture, or humidity, is the gas that fuels a storm, enabling tropical cyclones to spin up. Without an abundance of moisture, a tropical cyclone loses its energy source and either weakens or dissipates. However, the most unpredictable tropical cyclones tend to exist in environments with moderate amounts of vertical wind shear and humidity. These “high uncertainty” storms are more challenging to forecast, and small differences in the tropical cyclone or its environment can determine whether it intensifies.

 Figure shows the relationship between vertical wind shear and humidity, and their relative role in contributing to intensity forecast uncertainty. In the top left corner, red indicates a high likelihood of intensification in an environment with low shear and high humidity. In the bottom right corner, blue represents a low likelihood of intensification in an environment with high shear and low humidity. The figure demonstrates that the most unpredictable storms tend to be moderate in both shear and humidity.
 Figure shows the relationship between vertical wind shear and humidity and their relative role in contributing to intensity forecast uncertainty. The figure highlights that “middle of the road” scenarios with moderate moisture and moderate shear is the most difficult to accurately forecast.

Nonetheless, intensity forecasts have drastically improved over the last several decades. Improvement in hurricane models, developments in meteorological research, and discoveries about the mechanics behind hurricane formation have given scientists a clearer picture of storm behavior. Recent research has helped lower model forecast errors for both intensity and track guidance across all lead times, resulting in more accurate and precise forecasts from the National Hurricane Center (NHC). 

Trendlines for NHC forecast error in terms of TC track (a) and intensity (b) from 1990-2023. The graphic displays results for three lead times: 72 hours (dashed line), 48 hours (dotted line) and 24 hours (solid line). The trendlines show a decrease in forecast error across both track and intensity across all lead times.
Trendlines for NHC forecast error in terms of TC (a) track and (b) maximum winds  in the North Atlantic basin from 1990–2023. Trendlines show a decrease in forecast error across both intensity and track, and across all lead times.

While significant progress has been made in the accuracy of hurricane intensity forecasts, uncertainty remains a key challenge. As NOAA researchers continue to explore these storm dynamics, our understanding of tropical cyclone behavior improves. Advancements in forecasting models have significantly reduced errors, and continued hurricane research will further enhance predictions, enabling communities to better prepare for the impacts of powerful storms.