Simulated HYSPLIT Plume Output Resampled to TROPOMI Pixel Footprints

This dataset contains a synthetic set of methane plume observations, simulated with HYSPLIT and resampled to TROPOMI pixel footprints. The dataset was developed for the evaluation of ML-SPERE (the Machine Learning-Superemitting Plume Emission Rate Estimate), a convolutional neural network framework for estimating methane emission rates from TROPOspheric Monitoring Instrument (TROPOMI) plumes. Please see the corresponding associated publication (listed below) for a complete description of preprocessing steps taken in transforming raw HYSPLIT simulation output into the resampled netCDF4 files presented here. This dataset consists of 438 synthetic methane plume scenes simulated using the HYSPLIT atmospheric transport model, driven by ERA5 meteorological data. Plumes were simulated at 224 global locations associated with persistent methane emissions identified through satellite observations and reported to the International Methane Emissions Observatory (IMEO). For each simulation, emission started at 0:00 UTC the day before the corresponding overpass to ensure at least 24 hours of emission history. The outputgrid was 5 degrees latitude by 6 degrees longitude at 0.01 degree resolution and sampled at 15 minute snapshot intervals, where the interval closest to the corresponding TROPOMI overpass was selected for resampling. NUMPAR was set to 40000, MAXPAR to 1000000, the final data was taken as the average over 8 parallel runs. Multiple plume realizations were generated at different timestamps in 2019, with emission rates sampled from the distribution of observed methane super-emissions reported in Schuit et. al. (2023). To represent realistic TROPOMI observation conditions, each simulated plume is embedded within a 32 × 32 pixel TROPOMI methane observation background. These backgrounds were selected from TROPOMI observations from 2023–2024 using an automated screening approach to minimize the likelihood of containing real methane plumes. The final dataset includes the resampled HYSPLIT plume abundances, the TROPOMI background observation, a scene where the two are combined, and corresponding meteorological and metadata fields. To replicate the environment used to run the demo.ipynb notebook, use pip to create a virtual env with the python version specified in python-version.txt, and install the packages specified in requirements.txt.