The Surface Biology and Geology global imaging spectrometer is primarily designed to observe the chemical fingerprint of the Earth’s surface. However imaging spectroscopy across the visible to shortwave infrared (VSWIR) can also provide important atmospheric observations of methane point sources, highly concentrated emissions from energy, waste management and livestock operations. Relating these point-source observations to greenhouse gas inventories and coarser, regional methane observations from sensors like the European Space Agency (ESA) TROPOMI will contribute to reducing uncertainties in local, regional and global carbon budgets. We present the Multi-scale Methane Analytic Framework (M2AF) that facilitates disentangling confounding processes by streamlining analysis of cross-scale, multi-sensor methane observations across three key, overlapping spatial scales: 1) global to regional scale, 2) regional to local scale, and 3) facility (point source scale). M2AF is an information system that bridges methane research and applied science by integrating tiered observations of methane from surface measurements, airborne sensors and satellite. Reducing uncertainty in methane fluxes with multi-scale analyses can improve carbon accounting and attribution which is valuable to both formulation and verification of mitigation actions. M2AF lays the foundation for extending existing methane analysis systems beyond their current experimental states, reducing latency and cost of methane data analysis and improving accessibility by researchers and decision makers. M2AF leverages the NASA Methane Source Finder (MSF), the NASA Science Data Analytics Platform (SDAP), Amazon Web Services (AWS) and two supercomputers for fast, on-demand analytics of cross-scale, integrated, quality-controlled methane flux estimates.

To address discrepancies between bottom-up and top-down inventories of methane emissions, the detection and quantification of methane point source emissions is of critical concern. Multiple airborne imaging spectrometer campaigns have identified the heavy-tailed distribution of point source methane emissions. The quantification of point source plumes is a two-part problem requiring the detection and delineation of plumes, followed by an accurate accounting of their total methane enhancement. Plume detection and delineation has often relied on manual or statistical methods, but automated methods taking into account plume morphology are essential as the volume of imaging spectrometer data rapidly increases. Fully convolutional neural networks (FCNNs) represent a robust solution to this problem allowing for the detection and delineation of methane point source emissions with minimal analyst input. This work demonstrates the applicability of FCNNs for accurate quantification of methane point source emissions by training a model on data from a 2019 Permian Basin survey by the Next Generation Airborne Visible InfraRed Imaging Spectrometer (AVIRIS-NG). FCNNs were trained using plumes that were manually interpreted from matched filter retrievals of methane enhancements. Our methodology was able to accurately detect and delineate methane plumes, and did so with fewer false positives than statistical methods. Given the anticipated satellite imaging spectrometer missions capable of global mapping of point sources, automated deep learning methods will be necessary to deal with methane plume detection in very large volumes of data.

Methane is an important greenhouse gas, and anthropogenic methane emissions from point sources are a frequent target for emissions reductions. Airborne imaging spectrometers measuring shortwave infrared radiance are becoming regular sources of data for methane point source plume detection and flux estimation. Accurate and efficient detection and delineation of methane plumes will play an essential role in quantifying point source fluxes. Methane plumes are highly variable in space and time, whereas surfaces that are typically “false positive” detections in methane enhancement retrievals are more regularly shaped and change on longer time scales. This work aims to take advantage of plume variability by applying a fully convolutional network (FCN) to detection and delineation of methane plumes within imaging spectrometer time series data. Using a time series of matched filter methane retrieval products derived from Airborne Visible and InfraRed Imaging Spectrometer Next Generation (AVIRIS-NG) data, we demonstrate the ability of a FCN to classify methane plumes at each time step. Comparison with plume detection and delineation using conventional statistical methods demonstrates the value of this approach. Automated approaches incorporating deep learning will become increasingly important as future global satellite missions greatly increase the frequency at which methane point sources are imaged.

Despite methane’s important role as a greenhouse gas, the contribution of individual sources to rising methane concentrations in Earth’s atmosphere is poorly understood. This is in part due to the lack of frequent measurements on a global scale, required to accurately quantify fugitive methane sources. Future missions such as Earth Surface Mineral Dust Source Investigation (EMIT), Surface Biology and Geology (SBG), and Carbon Mapper promise to provide global, spatially resolved spectroscopy observations that will allow us to map methane sources. However, the detection and attribution of individual methane sources is challenged by retrieval artifacts and noise in retrieved methane concentrations. Additionally, manual methane plume detection is not scalable to global space-borne observations due to the sheer volume of data generated. A robust automated system to detect methane plumes is needed. We evaluated the performance and sensitivity of several methane plume detection methods on 30m to 60m hyperspectral imagery, downsampled from airborne campaigns with AVIRIS-NG. To aid the training of the plume detection models, we explored supplementing downsampled airborne imagery with Large Eddy Simulations (LES) of methane plumes. We compared baseline methods such as thresholding and random forest classifiers, as well as state-of-the-art deep learning methods such as convolutional neural networks (CNNs) for classification and conditional adversarial networks (pix2pix) for plume segmentation.

Methane (CH4) emissions from climate-sensitive ecosystems within the northern permafrost region represent a large but highly uncertain source, with current estimates spanning a factor of seven (11 – 75 Tg CH4 yr-1). Accelerating permafrost thaw threatens significant increases in pan-Arctic CH4 emissions, amplifying the permafrost carbon feedback. We used airborne imaging spectroscopy with meter-scale spatial resolution and broad coverage to identify a previously undiscovered CH4 hotspot adjacent to an intensively studied thermokarst lake in interior Alaska. Hotspot emissions were confined to < 1% of the 10 ha study area. Ground-based chamber measurements confirmed average daily fluxes of 1,170 mg CH4 m-2 d-1, with extreme daily maxima up to 24,200 mg CH4 m-2 d-1. Ground-based geophysics measurements revealed thawed permafrost at and directly beneath the CH4 hotspot, extending to a depth of ~15 m, indicating that the intense CH4 emissions likely originated from recently thawed permafrost. Emissions from the hotspot accounted for ~40% of total diffusive CH4 emissions from the entire study area. Combining these results with hotspot statistics from our 70,000 km2 airborne survey across Alaska and northwestern Canada, we estimate that terrestrial thermokarst hotspots currently emit 1.1 (0.1 – 5.2) Tg CH4 yr-1, or roughly 4% of the annual pan-Arctic wetland budget from just 0.01% of the northern permafrost land area. Our results suggest that significant proportions of pan-Arctic CH4 emissions originate from disproportionately small areas of previously undetermined thermokarst emissions hotspots, and that pan-Arctic CH4 emissions may increase non-linearly as thermokarst processes increase under a warming climate.

Methane’s high heat trapping potential has made it a priority for quantification and mitigation efforts worldwide. Ground-based surveys and in-situ measurement techniques to quantify natural and fugitive methane emission sources are time-consuming, expensive, and often lead to sparse measurements. Failure to accurately quantify emissions at the point-source scale have thus led to poorly constrained emission estimates. Airborne imaging spectrometers such as the Airborne Visible/Infrared Imaging Spectrometer (AVIRIS-NG) and the Global Airborne Observatory (GAO) have been employed to map the often stochastic and intermittent point-source emissions from a diverse set of source types including oil and gas, dairy, etc. A matched filter is applied to the methane-absorption relevant spectral features of the instrument’s radiance cube. Machine learning models are then trained to recognize methane plumes from these column-matched filter methane maps. However, current Convolutional Neural Network (CNN) models suffer from a high false-positive rate and poorly generalize to new scenes. False-positive detections are primarily due to methane absorption-mimicking surface spectroscopic features, as well as a lack of training data. To supplement the available training data, we utilize Large Eddy Simulations (LES) of methane point-source emissions to train a Convolutional Neural Network (CNN) on a plume-classification task. We observe a significant distribution shift between LES and AVIRIS-NG plumes, primarily caused by high LES plume enhancements. Through a series of image transforms verified through an adversarial approach using a discriminator network, we minimize the distribution shift between synthetic LES plumes and plumes observed by AVIRIS-NG and GAO. CNNs trained on a mixture of LES and real-world plumes, and tested on flightlines from multiple campaigns exhibit an error reduction compared to previous models. The reduction in false-positive plume detections demonstrates that supplementing the limited training data of real methane plumes with LES provides an avenue to make automatic detection more robust for future airborne and spaceborne missions such as SBG, EMIT, and Carbon Mapper.