November 2026

What happens when the planet warms?

Does a warmer Earth mean more storms — or stronger storms?

Our evidence: 30 years of NASA and NOAA records, 1995–2024.

Today’s idea: Frequency and intensity are different questions.

First, Earth has warmed

Hurricanes and typhoons: same kind of storm, different oceans

  • A tropical cyclone is a rotating storm that forms over warm ocean water.
  • Hurricane is used in the Atlantic; typhoon is used in the western North Pacific.
  • We count storms with recorded peak winds of at least 34 knots (tropical-storm strength).
  • We separately count major-intensity storms with peak winds of at least 96 knots.

The next charts use NOAA IBTrACS storm tracks. Counts are not the same as landfalls.

Are tropical storms becoming more frequent?

What about the strongest tropical storms?

What about average peak wind speed?

Tornadoes are different

  • Tornadoes are smaller, short-lived rotating columns of air associated with thunderstorms.
  • They are not the same phenomenon as tropical cyclones.
  • We examine U.S. reports and separately count strong F/EF2–5 tornadoes.
  • Better detection and reporting can change observed totals even without changes in weather.

U.S. tornado reports over time

How do we test the relationship?

Each dot represents one year.

A positive correlation means warmer years tend to have higher values. A negative correlation means warmer years tend to have lower values. A correlation near zero means little linear relationship.

Important: Correlation does not prove that warming caused a particular storm.

Temperature vs. major Atlantic storms

Temperature vs. major western Pacific storms

Temperature vs. strong U.S. tornadoes

What do the correlations actually say?

Storm measure Pearson r p-value Direction
North Atlantic storm count 0.35 0.057 Positive
North Atlantic major storms 0.01 0.947 Positive
North Atlantic average peak wind -0.13 0.479 Negative
Western Pacific storm count -0.26 0.173 Negative
Western Pacific major storms 0.00 0.983 Negative
Western Pacific average peak wind 0.17 0.363 Positive
U.S. tornado count 0.11 0.566 Positive
U.S. strong tornado count 0.03 0.892 Positive

Pearson r ranges from −1 to +1. A small p-value is not proof of causation. These are simple annual comparisons, not causal climate models.

Why can warmer oceans matter?

Warm water is fuel — but not the only ingredient.
  • Warm ocean surfaces can provide heat and moisture to tropical cyclones.
  • Other ingredients — such as wind shear and atmospheric circulation — also matter.
  • The scientific evidence for stronger tropical cyclone rainfall and a higher share of very intense storms is more robust than evidence for increasing total storm counts everywhere.
  • Tornado climate trends remain more uncertain.

Why should we care?

A storm does not need to become more common to become more dangerous.

A very intense storm can cause severe flooding, destructive winds, and lasting damage to homes and infrastructure. Exposure and preparedness also influence the harm a storm causes.

Visual to add: A credited NOAA or FEMA hurricane-damage photograph would help students connect the data to real-world impacts. This slide currently has no photograph.

What did we learn?

  • NASA’s global temperature anomaly changed from 0.44°C in 1995 to 1.29°C in 2024.
  • North Atlantic major storms: correlation with global temperature r = 0.01 (positive).
  • Western Pacific major storms: correlation with global temperature r = 0.00 (negative).
  • U.S. strong tornado count: correlation with global temperature r = 0.03 (positive).

Big takeaway: Global warming is real, but the relationship between warming and storm behavior varies by storm type and measure. We should not confuse more storms with stronger storms — or correlation with causation.

Sources and important limitations

Limitations: Global average temperature is not local ocean temperature; storm detection and intensity estimates vary; IBTrACS USA_WIND coverage is incomplete in some basins; storm seasons and calendar years may differ; tornado ratings/reporting have changed; the 30 annual observations are serially related. These descriptive correlations are not causal estimates.

Thank you!