Data Sources & Methodology
Detailed information about all datasets, citations, and processing methodology.
About This Analysis
This climate-storms analysis integrates three authoritative datasets from the United States government to examine long-term correlations between rising global temperatures and storm activity. Data spans from 1950 to 2025, with overlap periods carefully analyzed to identify climate patterns. All data is publicly available and downloadable from the sources cited below.
Global Temperature Anomalies
NASA Goddard Institute for Space Studies (GISS)
Temperature anomalies represent deviations from the 1951-1980 baseline, allowing comparison of warming trends across time. Data is aggregated to annual averages for correlation analysis with storm metrics.
Tornado Data
NOAA National Centers for Environmental Information (NCEI)
Data comes from 79 gzip-compressed CSV files (one per month from 1950 onwards). This study consolidates all tornado records and aggregates by year for correlation analysis. Death counts and frequency metrics are tracked separately.
Hurricane Data
NOAA International Best Track Archive for Climate Stewardship (IBTrACS)
IBTrACS is the authoritative source for historical and current tropical cyclone data, maintained by NOAA. Data comes from multiple agencies (UNISYS, JMA, CMA, others) harmonized into a unified format. This study uses maximum sustained wind speeds to track intensity over time and correlate with temperature anomalies.
Analysis Methodology
1. Data Consolidation
79 NOAA StormEvents files (monthly gzip CSVs) were decompressed and merged into a single tornado dataset. Duplicate records removed based on event date and location.
2. Hurricane Shapefile Extraction
IBTrACS shapefile data (.shp, .dbf format) was extracted using GeoPandas with pyogrio backend. All geographic features and storm attributes converted to tabular CSV format for easier analysis.
3. Temporal Aggregation
Daily temperature data aggregated to annual averages. Tornado events counted and summed per year. Hurricane data aggregated to annual maximums and averages of wind speed, pressure, and point density.
4. Overlap Period Selection
All three datasets merged on year field. Temperature (1880+) and tornado (1950+) data fully overlap. Hurricane data begins 1980, creating three analysis windows: 1950-1980 (tornado-temp), 1980-2025 (all three).
5. Correlation Calculation
Pearson correlation coefficients computed between temperature anomalies and: (a) tornado frequency, (b) tornado deaths, (c) max hurricane wind speed, (d) average hurricane wind speed. Range: -1.0 to +1.0.
6. Visualization
Canvas-based spiral plot maps hurricane data (1980-2025) in polar coordinates: radius = year, angle = month (60° per month), point size = wind speed, color = Saffir-Simpson category.
Data Quality & Limitations
- ⚠️Completeness: Tornado reporting improved significantly post-1990. Early frequency numbers likely undercount.
- ⚠️Geographic Bias: Data covers US territory only. Global storms (Atlantic/Pacific) included; Indian Ocean excluded.
- ⚠️Instrumentation: Early storm measurements less precise than satellite-era data (post-1970s).
- ⚠️Correlation ≠ Causation: Statistical correlations show relationships, not necessarily direct cause-effect.
- ⚠️Confounding Factors: Population growth, urban development, improved detection also drive storm reporting increases.
Processing Tools
All data processing performed in Python using open-source libraries:
Project Files
Download the full project bundle with all files needed to recreate the Story5 site locally, including data, processing scripts, and a step-by-step README.
Download Story5 Project (ZIP)Attribution & Acknowledgments
This analysis is a student project for data science and climate education. All data comes from official US government sources:
- • NASA GISS: Part of NOAA's climate monitoring and research mission
- • NOAA NCEI: National Centers for Environmental Information, part of the National Weather Service
- • NOAA IBTrACS: International partnership archiving tropical cyclone data since 1851
Project created as part of CUNY Data Science curriculum. Data sources and methods follow scientific standards.