Analyzing 32,000 Tornado Records with LogSheet

The Story Hidden in the Data

The NOAA Storm Prediction Center maintains a public dataset of every tornado recorded in the United States since 1950. Thousands of rows, each containing the date, state, Enhanced Fujita rating, fatality count, estimated property damage, and geographic coordinates.

The dataset is available as a CSV file. But opening it in a spreadsheet does not show which tornadoes account for most fatalities and damage.

Preparing the Dataset

The analysis focused on the 21st century: January 2000 through June 2026. After filtering, the dataset contained 32,332 tornado events:

• Multi-segment tornadoes (storms recorded across multiple counties) were deduplicated by identifier to avoid double counting;
• Unrated events, representing 4.6% of the total, were excluded from EF share calculations.

The data was structured as a LogSheet journal with typed fields for date, state, EF rating, fatalities, and damage.

The Inversion

The dashboard showed one clear pattern.

Of the 32,332 tornadoes:

• 948 (2.94%) were rated EF3 or higher;
• 31,384 (97.06%) were rated EF0 through EF2.

Those 948 tornadoes accounted for:

• 1,533 fatalities (82% of all deaths);
• $27.4 billion in damage (71% of the total).

The remaining 31,384 tornadoes accounted for:

• 338 fatalities (18% of all deaths);
• $11.25 billion in damage (29% of the total).

Less than 3% of events produced more than four-fifths of all fatalities and property losses.

Why This Matters for Reporting

EF0 and EF1 tornadoes make up 88% of all recorded events. A monthly summary based on total tornado count tells almost nothing about actual risk exposure.

The dashboard shows three metrics side by side: count, deaths, and damage, broken down by EF rating. The categories that barely register in the count column dominate the fatality and damage columns.

From Raw Data to Operational Insight

LogSheet is used to organize tabular data into structured journals, dashboards, and reports. The tornado dataset is one example of this workflow in practice.

Data source: NOAA Storm Prediction Center, spc.noaa.gov/wcm. File: 1950-2025_all_tornadoes.csv + 2026 events through June 16, 2026.

Note: Damage values are NWS field-office estimates and may understate actual losses.