Storm Watch: Trends in Extreme US Weather Events

Yanjun (Penny) Liao and Margaret A. Walls

Climate change is increasing the frequency and intensity of extreme weather events in the United States. How has this trajectory looked in recent decades, and what are the costs of the resulting damages?

Special Series: Weather Volatility in the United States

Climate change is increasing the frequency and intensity of extreme weather events in the United States. In this series of blog posts, experts examine trends in extreme weather, impacts on communities, and policies that potentially could mitigate damage from extreme weather events.

The 2024 hurricane season officially kicked off June 1, and by most accounts, the season is expected to be a big one. Forecasters at the National Weather Service (NWS), a program housed within the National Oceanic and Atmospheric Administration (NOAA), released their predictions for the 2024 season in May. They expect 17 to 25 named storms in 2024—above the 30-year average of 14.4 named storms. Independent hurricane forecasts from 24 other university, government, and private-sector groups all predict above-average numbers for the year, as well.

Climate change is increasing the frequency of very intense hurricanes (classified as category 4 or category 5), according to climate scientists, and sea level rise is exacerbating floods from hurricane storm surges. Population growth in high-risk coastal areas also significantly contributes to the increasing costs of storms: with more assets in harm’s way, property damage and the overall economic cost of storms are on the rise.

Intense hurricanes are not the only type of extreme weather events that are becoming more common. Floods, winter storms, wildfires, droughts, tornadoes, and heat waves all can damage communities and ecosystems. Many of these weather extremes are presenting challenges for communities across the United States as these communities struggle to find ways to adapt and respond to low-probability, but high-cost, events.

In a series of three blog posts, we’ll summarize the trends and geographic patterns of extreme weather events in the United States, along with the impact of these events in the forms of property damage, crop damage, and deaths.

Storm (Data) Chasing

We used the Storm Events Database operated by NOAA to study these trends. The Storm Events Database is a comprehensive database of storms and other significant weather events, such as wildfires, tornadoes, and extreme heat, that are recorded by NWS. NWS operates 123 forecast offices across the country that receive information about storm events from a variety of sources, including emergency management officials at the federal, state, and local levels; local law enforcement; trained volunteers in the Skywarn spotters program; NWS damage surveys; the insurance industry; and the general public. The data include estimates of both insured and uninsured property damage, crop damage, deaths, and injuries.

The raw data are not well-suited for analyses such as identification of trends or mapping for several reasons. For one thing, the geographical information that’s reported varies for individual events; sometimes a weather event is reported at the county level, but oftentimes, the report is in an NWS “forecast zone,” which can be either at the subcounty level or cover multiple counties. A single event also can hit multiple locations at different times.

We processed and aggregated the data to account for these complicating factors. In addition, we combined similar types of weather events into broader categories for clarity, because the 48 types of events that are listed in the raw data are highly specific. For example, NWS lists flash floods, lakeshore floods, coastal floods, and floods; our flood category encompasses all of these types of floods. We dropped a few minor types of weather events, as well (e.g., landslides and volcanoes). For the purposes of this blog series, we decided on eight main categories: drought; flood; heat; hurricane; thunderstorm, wind, or hail; tornado; wildfire; and winter weather. We also organized the raw data by type of weather event, county, and year. (Detailed documentation of our procedures for data cleaning and aggregation is available upon request.)

The Storm Events Database differs in three main dimensions from the widely cited data on billion-dollar climate and weather disasters, which also is tracked by NOAA. First, the Storm Events Database includes all recorded significant events, including those with aggregate damage below $1 billion. Second, the database includes geographical information, which is unavailable in the data on billion-dollar disasters. Third, the Storm Events Database focuses on direct damage and does not account for indirect costs, such as interruptions to business activity, while the data on billion-dollar disasters includes direct and indirect costs and thus reports much larger total losses from disasters. (A recent government report about the costs of flooding in the United States finds that indirect costs from flooding are 11 times greater than direct property damage.) Another source of data on weather events is available from the Spatial Hazard Events and Losses Database for the United States (SHELDUS), a county-level data set that also uses the Storm Events Database as its starting point. (Data from SHELDUS are available with a paid subscription.)

In general, collecting data on deaths and damage from disasters is inherently difficult, as a study from 2009 highlights. The authors of the study compared four major sources of information about losses from disasters and find that all four exhibit inconsistencies in reporting. Damage can be underreported in the Storm Events Database, since the NWS is not required to report monetary damage if an estimate of the damage is not available (except for floods). The challenge of tallying death totals from storm events is highlighted in the controversy that arose over death tolls from Hurricane Maria, which struck Puerto Rico in 2017. Official estimates showed 64 deaths; subsequent studies estimated the number of deaths to be in the thousands.

Nevertheless, the Storm Events Database presents relatively consistent records over a long period of time. The database reflects official efforts by NOAA to collect data, and the database has been used in a number of peer-reviewed studies, including research that assesses trends related to severe thunderstorms and tornadoes, analyzes damage from windstorms, measures public concern about extreme weather, and assesses the benefits of coastal wetlands for reducing floods.

Checking the Damage

The Storm Events Database reports four main categories of direct losses from disaster events: death, injury, property damage (including damage to infrastructure), and crop damage. Here, we’ll focus on death and damage to property and crops from 1995 to 2022 (Table 1).

Table 1. Direct Losses by Type of Weather Event, 1995–2022

Between 1995 and 2022, extreme weather events collectively caused 13,857 deaths, $339 billion (2020$) in property damage, and $48.8 billion (2020$) in crop damage.

These numbers reveal that different types of disasters tend to cause different amounts and kinds of damage. Floods and hurricanes accounted for almost 70 percent of total property damage over this 28-year period. And although floods and hurricanes threaten human lives and damage crops substantially, other types of weather events caused even more significant losses in those categories. Drought accounted for 46 percent of crop damage. Heat caused the most deaths, while winter weather was the second-largest cause of death.

Tracking Extreme Weather Events by Year: What Are the Trends?

The most common type of extreme weather event is our category that includes thunderstorms, wind, and hail, with a range of 20,000 to more than 40,000 events each year between 1995 and 2022. Winter weather events are second-most common, occurring about half as frequently as events in our category for thunderstorms, wind, and hail. Floods and droughts, which occur thousands of times each year, are the next-most-common types of events. The remaining five types occur with lower frequency.

Figure 1. Occurrences of Extreme Weather Events by Type, 1995–2022

Notes: In the Storm Events Database, one entire storm system can be associated with weather events of different types and in different locations. For example, if a tropical cyclone causes flooding and tornadoes in two counties, the record will show a total of four events—two for each county. The figure reflects this method of recording data.

Impacts of Extreme Weather Events

Extreme weather events are not equal in the damage they cause (Figure 2). Some events that occur more often, like thunderstorms, are not nearly as damaging as hurricanes, for example.

Figure 2. Property Damage, Crop Damage, and Deaths by Type of Weather Event, 1995–2022

We’ve noted six significant findings from the data presented in Figure 2.

A key characteristic of the data on property and crop damage is the spikiness. In a few especially intense years, catastrophic events created a large amount of damage, whereas other years saw much less damage. This irregular pattern of damage across years contrasts with the frequency of events from 1995 to 2022 (Figure 1)—the amount of damage from weather events tends to be much more volatile from year to year than the occurrence of extreme weather events.

Hurricanes and flooding drove most of the spikes in property damage. The largest losses were caused by Hurricanes Harvey, Irma, and Maria in 2017; Hurricanes Katrina and Rita in 2005; and flooding associated with Superstorm Sandy in 2012.

Crop damage overwhelmingly was caused by drought, except for a couple of years in which hurricanes caused significant crop damage. For example, Hurricane Michael caused over $4 billion in crop damage in 2018, primarily in Georgia and Florida.

Wildfires generally cause less property damage than other extreme events, but the spike in 2018, which mainly is associated with the Camp Fire in Paradise, California, shows the large potential impact of fires. When wildfires hit developed areas, they tend to be extremely destructive. The damage from these wildfires raises concerns for population growth in areas where human development borders wildlands, given that these areas tend to be more vulnerable to wildfires.

The number of deaths from storm events is more consistent across years (i.e., less spiky) than property and crop damage. In almost all the years we evaluated, heat and winter weather were responsible for the most deaths. Heat-related deaths have shown an alarming upward trend in the past decade, corresponding to a similarly steady increase in the number of heat-related events (Figure 1). The US Centers for Disease Control and Prevention, a government agency that compiles data on deaths from multiple causes, shows even higher numbers of deaths from heat and heat-related illnesses than the NWS. The highest number of heat-related deaths in the United States on record occurred in 2023, according to the agency.

Tornadoes do not account for a significant share of damage and deaths in most years, but 2011 was the second-most severe year for tornadoes in US history. In that year, 2,595 tornadoes were reported, and 567 deaths and $7.9 billion in property damage were attributed to tornadoes.

To Be Continued: Examining Regional Trends

That storm events do not strike all regions of the country equally is important context for our findings. The Gulf Coast is prone to hurricanes, and California is highly susceptible to wildfires. Drought mainly occurs in the Southwest, winter weather events in the Midwest.

In our next blog post, we will dig into these regional differences in more detail and present findings about where the damage from weather events is occurring. These findings will explore the challenges that some regions face in building adaptation and resilience to a changing climate—and highlight the volatile weather that a changing climate brings. Stay tuned.

Our data and detailed documentation of our related methodology are available in the Harvard Dataverse, a free and public data repository where researchers from various disciplines share, archive, cite, access, and explore research data.

Related People