Analyzing Cloud Seeding Using Data Science

Abstract

Cloud seeding is a weather modification method that attempts to increase precipitation by adding particles like silver iodide to suitable clouds. Although cloud seeding has been used in the United States for decades the activities and effectiveness are difficult to evaluate across historical records because outcomes depend heavily on atmospheric conditions. This project uses data science and artificial intelligence to organize historical cloud-seeding records and identify patterns in their location, timing, and seeding agents, providing a more structured way to examine existing evidence. A dataset of 832 cloud-seeding projects was extracted from NOAA reports (2000–2025) using a PDF-processing and LLM-based pipeline and achieves a 98.38% extraction accuracy on a manual review of 200 sampled records. Basic statistical analysis of the dataset shows that cloud-seeding activity is concentrated in the western United States with silver iodide as the most common seeding agent and snowpack enhancement as the most common stated purpose. Activity declined through the 2010s before increasing again after 2021. Overall, these findings demonstrate how data science and AI can organize historical cloud-seeding records for large-scale analysis. However, the dataset reflects reported activities rather than proven effectiveness so future work should apply rigorous methods to evaluate causal effects.