"RAMS-Simulated Meteorological and Atmospheric Tracer Dataset for Ensemble Learning-Based Methane Detection and Intensity Prediction "

"This dataset contains RAMS (Regional Atmospheric Modeling System)-simulated meteorological and atmospheric tracer concentration data generated for ensemble learning-based methane (CH\u2084) detection and intensity prediction. The simulation was conducted over a 125-meter grid on a 7.6 \u00d7 7.6 km domain centered on the Savannah River Site (SRS), Aiken, South Carolina, USA, running for six hours starting February 10, 2022, at 1200 UTC. The dataset includes spatiotemporal grid indices, geographic coordinates, and the following meteorological variables: temperature, relative humidity, vertical velocity, pressure, water vapor, turbulent kinetic energy, precipitation rate, sensible heat flux, latent heat flux, and horizontal wind components. Derived predictors include wind speed, directional sine and cosine embeddings, and diurnal periodic features. The target variable is atmospheric tracer concentration representing fugitive methane. The dataset was preprocessed to ensure consistency in variable names and data types, and was checked for duplicate records and missing values. It is intended to support the development and evaluation of machine learning models for environmental methane detection (classification) and intensity prediction (regression)."