A Benchmark Dataset for Retrieval of Point Source CH4 and CO2 Plumes from Imaging Spectrometer Data

Multiple studies have demonstrated successful retrieval of methane (CH4) and carbon dioxide (CO2) plumes from airborne imaging spectrometer data. Comparison of retrieval algorithms and concentration enhancement results has been made more difficult by the lack of common benchmark data. Ten Airborne Visible Infrared Imaging Spectrometer Next Generation (AVIRIS-NG) scenes capturing a variety of CH4 and CO2 point source emissions in the United States and India have been curated into a benchmark dataset. Images were selected to span a wide range of sources, enhancements, geometries, and atmospheric conditions. CH4 point sources captured in the dataset include a range of oil and natural gas facilities, landfills, wastewater treatment, dairy facilities, and a mine. Four fossil fuel power plants comprise the CO2 point sources in the dataset. Each scene within the dataset contains a variety of metadata useful for parameterizing greenhouse gas retrievals. In this presentation, the composition of the benchmark dataset will be described, and concentration enhancements retrieved using a matched filtering technique will be demonstrated. The dataset is publicly available for algorithm development and comparison, and should provide a useful tool for testing of greenhouse gas retrieval methods for future satellite imaging spectrometers.