Divisional Publications
APD
Data-driven Derivation of Stellar Properties from Photometric Time Series Data Using Convolutional Neural Networks
10.3847/1538-4357/ac7563
https://iopscience.iop.org/article/10.3847/1538-4357/ac7563
Stellar variability is driven by a multitude of internal physical processes that depend on fundamental stellar properties. These properties are our bridge to reconciling stellar observations with stellar physics and to understand the distribution of stellar populations within the context of galaxy formation. Numerous ongoing and upcoming missions are charting brightness fluctuations of stars over time, which encode information about physical processes such as the rotation period, evolutionary state (such as effective temperature and surface gravity), and mass (via asteroseismic parameters). Here, we explore how well we can predict these stellar properties, across different evolutionary states, using only photometric time-series data. To do this, we implement a convolutional neural network, and with data-driven modeling we predict stellar properties from light curves of various baselines and cadences. Based on a single quarter of Kepler data, we recover the stellar properties, including the surface gravity for red giant stars (with an uncertainty of ≲0.06 dex) and rotation period for main-sequence stars (with an uncertainty of ≲5.2 days, and unbiased from ≈5 to 40 days). Shortening the Kepler data to a 27 days Transiting Exoplanet Survey Satellite–like baseline, we recover the stellar properties with a small decrease in precision, ∼0.07 for log g and ∼5.5 days for Prot, unbiased from ≈5 to 35 days. Our flexible data-driven approach leverages the full information content of the data, requires minimal or no feature engineering, and can be generalized to other surveys and data sets. This has the potential to provide stellar property estimates for many millions of stars in current and future surveys.
PHANGS-HST: new methods for star cluster identification in nearby galaxies
https://doi.org/10.1093/mnras/stab3183
https://academic.oup.com/mnras/article/509/3/4094/6425765?login=false
We present an innovative and widely applicable approach for the detection and classification of stellar clusters, developed for the PHANGS-HST Treasury Program, an NUV-to-I band imaging campaign of 38 spiral galaxies. Our pipeline first generates a unified master source list for stars and candidate clusters, to enable a self-consistent inventory of all star formation products. To distinguish cluster candidates from stars, we introduce the Multiple Concentration Index (MCI) parameter, and measure inner and outer MCIs to probe morphology in more detail than with a single, standard concentration index (CI). We improve upon cluster candidate selection, jointly basing our criteria on expectations for MCI derived from synthetic cluster populations and existing cluster catalogues, yielding model and semi-empirical selection regions (respectively). Selection purity (confirmed clusters versus candidates, assessed via human-based classification) is high (up to 70 per cent) for moderately luminous sources in the semi-empirical selection region, and somewhat lower overall (outside the region or fainter). The number of candidates rises steeply with decreasing luminosity, but pipeline-integrated Machine Learning (ML) classification prevents this from being problematic. We quantify the performance of our PHANGS-HST methods in comparison to LEGUS for a sample of four galaxies in common to both surveys, finding overall agreement with 50–75 per cent of human verified star clusters appearing in both catalogues, but also subtle differences attributable to specific choices adopted by each project. The PHANGS-HST ML-classified Class 1 or 2 catalogues reach ∼1 mag fainter, ∼2 × lower stellar mass, and are 2−5 × larger in number, than attained in the human classified samples.
Beyond the Local Volume. I. Surface Densities of Ultracool Dwarfs in Deep HST/WFC3 Parallel Fields
10.3847/1538-4357/ac35ea
https://iopscience.iop.org/article/10.3847/1538-4357/ac35ea
Ultracool dwarf stars and brown dwarfs provide a unique probe of large-scale Galactic structure and evolution; however, until recently spectroscopic samples of sufficient size, depth, and fidelity have been unavailable. Here, we present the identification of 164 M7-T9 ultracool dwarfs in 0.6 deg2 of deep, low-resolution, near-infrared spectroscopic data obtained with the Hubble Space Telescope (HST) Wide Field Camera 3 (WFC3) instrument as part of the WFC3 Infrared Spectroscopic Parallel Survey and the 3D-HST survey. We describe the methodology by which we isolate ultracool dwarf candidates from over 200,000 spectra, and show that selection by machine-learning classification is superior to spectral index-based methods in terms of completeness and contamination. We use the spectra to accurately determine classifications and spectrophotometric distances, the latter reaching to ∼2 kpc for L dwarfs and ∼400 pc for T dwarfs.
Star cluster classification in the PHANGS–HST survey: Comparison between human and machine learning approaches
https://doi.org/10.1093/mnras/stab2087
https://academic.oup.com/mnras/article/506/4/5294/6325189?login=false
When completed, the PHANGS–HST project will provide a census of roughly 50 000 compact star clusters and associations, as well as human morphological classifications for roughly 20 000 of those objects. These large numbers motivated the development of a more objective and repeatable method to help perform source classifications. In this paper, we consider the results for five PHANGS–HST galaxies (NGC 628, NGC 1433, NGC 1566, NGC 3351, NGC 3627) using classifications from two convolutional neural network architectures (RESNET and VGG) trained using deep transfer learning techniques. The results are compared to classifications performed by humans. The primary result is that the neural network classifications are comparable in quality to the human classifications with typical agreement around 70 to 80 per cent for Class 1 clusters (symmetric, centrally concentrated) and 40 to 70 per cent for Class 2 clusters (asymmetric, centrally concentrated). If Class 1 and 2 are considered together the agreement is 82 ± 3 per cent. Dependencies on magnitudes, crowding, and background surface brightness are examined. A detailed description of the criteria and methodology used for the human classifications is included along with an examination of systematic differences between PHANGS–HST and LEGUS. The distribution of data points in a colour–colour diagram is used as a ‘figure of merit’ to further test the relative performances of the different methods. The effects on science results (e.g. determinations of mass and age functions) of using different cluster classification methods are examined and found to be minimal.
Flare Statistics for Young Stars from a Convolutional Neural Network Analysis of TESS Data
10.3847/1538-3881/abac0a
https://iopscience.iop.org/article/10.3847/1538-3881/abac0a
All-sky photometric time-series missions have allowed for the monitoring of thousands of young (tage < 800 Myr) stars in order to understand the evolution of stellar activity. Here, we developed a convolutional neural network (CNN), stella, specifically trained to find flares in Transiting Exoplanet Survey Satellite (TESS) short-cadence data. We applied the network to 3200 young stars in order to evaluate flare rates as a function of age and spectral type. The CNN takes a few seconds to identify flares on a single light curve. We also measured rotation periods for 1500 of our targets and find that flares of all amplitudes are present across all spot phases, suggesting high spot coverage across the entire surface. Additionally, flare rates and amplitudes decrease for stars tage > 50 Myr across all temperatures Teff ≥ 4000 K, while stars from 2300 ≤ Teff < 4000 K show no evolution across 800 Myr. Stars of Teff ≤ 4000 K also show higher flare rates and amplitudes across all ages. We investigate the effects of high flare rates on photoevaporative atmospheric mass loss for young planets. In the presence of flares, planets lose 4%–7% more atmosphere over the first 1 Gyr. stella is an open-source Python toolkit hosted on GitHub and PyPI.
BPS
Space Biology
2021 Workshop on AI and Modeling for Space Biology Workshop
http://dx.doi.org/10.5281/zenodo.7508535
Biological research and self-driving labs in deep space supported by artificial intelligence
doi:10.1038/s42256-023-00618-4
Space biology research aims to understand fundamental spaceflight effects on organisms, develop foundational knowledge to support deep space exploration and, ultimately, bioengineer spacecraft and habitats to stabilize the ecosystem of plants, crops, microbes, animals and humans for sustained multi-planetary life. To advance these aims, the field leverages experiments, platforms, data and model organisms from both spaceborne and ground-analogue studies. As research is extended beyond low Earth orbit, experiments and platforms must be maximally automated, light, agile and intelligent to accelerate knowledge discovery. Here we present a summary of decadal recommendations from a workshop organized by the National Aeronautics and Space Administration on artificial intelligence, machine learning and modelling applications that offer solutions to these space biology challenges. The integration of artificial intelligence into the field of space biology will deepen the biological understanding of spaceflight effects, facilitate predictive modelling and analytics, support maximally automated and reproducible experiments, and efficiently manage spaceborne data and metadata, ultimately to enable life to thrive in deep space.
Biomonitoring and precision health in deep space supported by artificial intelligence.
doi:10.1038/s42256-023-00617-5
Human exploration of deep space will involve missions of substantial distance and duration. To effectively mitigate health hazards, paradigm shifts in astronaut health systems are necessary to enable Earth-independent healthcare, rather than Earth-reliant. Here we present a summary of decadal recommendations from a workshop organized by NASA on artificial intelligence, machine learning and modelling applications that offer key solutions toward these space health challenges. The workshop recommended various biomonitoring approaches, biomarker science, spacecraft/habitat hardware, intelligent software and streamlined data management tools in need of development and integration to enable humanity to thrive in deep space. Participants recommended that these components culminate in a maximally automated, autonomous and intelligent Precision Space Health system, to monitor, aggregate and assess biomedical statuses.
Open Science for the Next Decade of Life and Physical Sciences Research for Deep Space Exploration
Submitted to NASEM for 2023-2032 Decadal Survey on Biological and Physical Sciences Research in Space
http://surveygizmoresponseuploads.s3.amazonaws.com/fileuploads/623127/6392490/103-2bed5af5dd4cc17654149c1d1ac3f34b_ScottRyanT.pdf
All space-related biological, biomedical, and physical sciences research data are precious national data resources. These data need to be as open-access as possible to grow the fields, of reusable quality to empower community-driven informatic research, and collected through researchers using cloud-based digital research notebooks to streamline collection-submission-curation. To fulfill the potential of Open Science research, experiments should be designed to maximize metadata quality, data volume and experimental reproducibility; data should be curated according to acceptable ontologies and standard assay metadata from their respective fields; higher-level data should be generated with consensus processing pipelines; and novel tools for open-access collaborative data networking and exchange should be developed to enable collaboration in a secured manner while removing old NASA security roadblocks. With access to high-end computing and AI/ML modeling tools, this community-driven research approach will be well-positioned to conduct world-class research. As data-intensive, computer-assisted approaches continue to produce essential basic, applied, and operational outcomes, a decade-long investment should be pursued for this campaign of ‘Open Science’ data stewardship in the life and physical space sciences, to support discovery and platform development for deep space exploration.
Machine Learning, Artificial Intelligence and Data Modeling for the Next Decade of Space Biology Research and Astronaut Health Support
Submitted to NASEM for 2023-2032 Decadal Survey on Biological and Physical Sciences Research in Space
http://surveygizmoresponseuploads.s3.amazonaws.com/fileuploads/623127/6392490/153-d4da81bc2b0f81447a9adb0dcb878e64_SandersLaurenM.pdf
We propose a ten-year research campaign to maximally adopt artificial intelligence (AI) and machine learning (ML) capabilities in space biology research and the design of spaceflight health systems. Fully leveraging AI/ML functionalities will enable generating and analyzing large amounts of data while requiring limited human time, and increasing knowledge gain of biological spaceflight effects. Further, integration of AI/ML into support systems for spacecraft, ecosystem, and astronaut health will increase the power of adverse event prediction and mitigation.
Development of New Algorithms for Space Biology
Submitted to NASEM for 2023-2032 Decadal Survey on Biological and Physical Sciences Research in Space
http://surveygizmoresponseuploads.s3.amazonaws.com/fileuploads/623127/6378869/217-4826dc0ada8fb5a9ec7d5a1167b9523b_SandersLaurenM.pdf
Advances in artificial intelligence (AI) algorithms provide the opportunity to both catalyze and be inspired by new space biological discoveries. Currently there is limited adaptation of emerging AI methods to address critical needs for NASA space biology research, and no spaceflight missions are dedicated to developing algorithms inspired by space biology and its needs. This white paper highlights established and emerging algorithms relevant to space biology and recommends investment in space biology inspired algorithm development in the coming decade.
ESD
Potentially Underestimated Gas Flaring Activities -A New Approach to Detect Combustion Using Machine Learning and NASA's Black Marble Product Suite
https://doi.org/10.1088/1748-9326/acb6a7
https://iopscience.iop.org/article/10.1088/1748-9326/acb6a7/meta
Monitoring changes in greenhouse gas (GHG) emission is critical for assessing climate mitigation efforts towards the Paris Agreement goal. A crucial aspect of science-based GHG monitoring is to provide objective information for quality assurance and uncertainty assessment of the reported emissions. Emission estimates from combustion events (gas flaring and biomass burning) are often calculated based on activity data (AD) from satellite observations, such as those detected from the Visible Infrared Imaging Radiometer Suite (VIIRS) onboard the Suomi-NPP and NOAA-20 satellites. These estimates are often incorporated into carbon models for calculating emissions and removals. Consequently, errors and uncertainties associated with AD propagate into these models and impact emission estimates. Deriving uncertainty of AD is therefore crucial for transparency of emission estimates but remains a challenge due to the lack of evaluation data or alternate estimates. This work proposes a new approach using machine learning (ML) for combustion detection from NASA's Black Marble product suite and explores the assessment of potential uncertainties through comparison with existing datasets. We jointly characterize combustion using thermal and light emission signals, with the latter improving detection of probable weaker combustion with less distinct thermal signatures. Being methodologically independent, the differences in ML-derived estimates with existing approaches can indicate the potential uncertainties in detection. The approach was applied to detect gas flaring activities over the Eagle Ford Shale, Texas. We analyzed the spatio-temporal variations in detections and found that approximately 79.04% and 72.14% of the light emission-based detections are missed by ML-derived detections from VIIRS thermal bands and existing datasets, respectively. The region was impacted by the winter storm Uri which resulted in a significant reduction of flaring activities followed by a post-storm resumption. Our method is extendible to combustion events, such as biomass and waste burning, and can be scaled globally for transparent emission estimate reporting
HPD
A Machine-learning Data Set Prepared from the NASA Solar Dynamics Observatory Mission
10.3847/1538-4365/ab1005
https://iopscience.iop.org/article/10.3847/1538-4365/ab1005
In this paper, we present a curated data set from the NASA Solar Dynamics Observatory (SDO) mission in a format suitable for machine-learning research. Beginning from level 1 scientific products we have processed various instrumental corrections, down-sampled to manageable spatial and temporal resolutions, and synchronized observations spatially and temporally. We illustrate the use of this data set with two example applications: forecasting future extreme ultraviolet (EUV) Variability Experiment (EVE) irradiance from present EVE irradiance and translating Helioseismic and Magnetic Imager observations into Atmospheric Imaging Assembly observations. For each application, we provide metrics and baselines for future model comparison. We anticipate this curated data set will facilitate machine-learning research in heliophysics and the physical sciences generally, increasing the scientific return of the SDO mission. This work is a direct result of the 2018 NASA Frontier Development Laboratory Program. Please see the Appendix for access to the data set, totaling 6.5TBs.
A Machine Learning Approach to Predicting SEP Events Using Properties of Coronal Mass Ejections
https://doi.org/10.1029/2021SW002797
https://agupubs.onlinelibrary.wiley.com/doi/10.1029/2021SW002797
Solar energetic particles (SEPs) can cause severe damage to astronauts and their equipment, and can disrupt communications on Earth. A lack of thorough understanding the eruption processes of solar activities and the subsequent acceleration and transport processes of energetic particles makes it difficult to forecast the occurrence of an SEP event and its intensity using conventional modeling with physics-based parameters. Therefore, in order to provide an advance warning for astronauts to seek shelter in a timely manner, we apply neural networks to forecast the occurrence of SEP events. We use the properties of coronal mass ejections (CMEs) archived in the Coordinated Data Analysis Workshops catalog based on SOHO Large Angle and Spectrometric Coronagraph Experiment observations. We also derive some features based on these properties associated with the CME, and analyze the contribution of each feature to the overall prediction. Our algorithm achieves an average True Skill Statistic of 0.906, an average F1 score of 0.246, an average probability of detection of 0.920, and an average false alarm rate of 0.882. An analysis of the features shows that sunspot number and a feature based on Type II radio bursts contribute the most, but when grouped together, CME speed-related features are the most important features.
Solar Flare Intensity Prediction With Machine Learning Models
https://doi.org/10.1029/2020SW002440
https://agupubs.onlinelibrary.wiley.com/doi/10.1029/2020SW002440
We develop a mixed long short-term memory (LSTM) regression model to predict the maximum solar flare intensity within a 24-hr time window 0–24, 6–30, 12–36, and 24–48 hr ahead of time using 6, 12, 24, and 48 hr of data (predictors) for each Helioseismic and Magnetic Imager (HMI) Active Region Patch (HARP). The model makes use of (1) the Space-Weather HMI Active Region Patch (SHARP) parameters as predictors and (2) the exact flare intensities instead of class labels recorded in the Geostationary Operational Environmental Satellites (GOES) data set, which serves as the source of the response variables. Compared to solar flare classification, the model offers us more detailed information about the exact maximum flux level, that is, intensity, for each occurrence of a flare. We also consider classification models built on top of the regression model and obtain better results in solar flare classifications as compared to Chen et al. (2019, https://doi.org/10.1029/2019SW002214). Our results suggest that the most efficient time period for predicting the solar activity is within 24 hr before the prediction time using the SHARP parameters and the LSTM model.
PSD
Flare Statistics for Young Stars from a Convolutional Neural Network Analysis of TESS Data
10.3847/1538-3881/abac0a
https://iopscience.iop.org/article/10.3847/1538-3881/abac0a
All-sky photometric time-series missions have allowed for the monitoring of thousands of young (tage < 800 Myr) stars in order to understand the evolution of stellar activity. Here, we developed a convolutional neural network (CNN), stella, specifically trained to find flares in Transiting Exoplanet Survey Satellite (TESS) short-cadence data. We applied the network to 3200 young stars in order to evaluate flare rates as a function of age and spectral type. The CNN takes a few seconds to identify flares on a single light curve. We also measured rotation periods for 1500 of our targets and find that flares of all amplitudes are present across all spot phases, suggesting high spot coverage across the entire surface. Additionally, flare rates and amplitudes decrease for stars tage > 50 Myr across all temperatures Teff ≥ 4000 K, while stars from 2300 ≤ Teff < 4000 K show no evolution across 800 Myr. Stars of Teff ≤ 4000 K also show higher flare rates and amplitudes across all ages. We investigate the effects of high flare rates on photoevaporative atmospheric mass loss for young planets. In the presence of flares, planets lose 4%–7% more atmosphere over the first 1 Gyr. stella is an open-source Python toolkit hosted on GitHub and PyPI.
Incorporating Physical Knowledge Into Machine Learning for Planetary Space Physics
https://doi.org/10.3389/fspas.2020.00036
https://www.frontiersin.org/articles/10.3389/fspas.2020.00036/full
Recent improvements in data collection volume from planetary and space physics missions have allowed the application of novel data science techniques. The Cassini mission for example collected over 600 gigabytes of scientific data from 2004 to 2017. This represents a surge of data on the Saturn system. In comparison, the previous mission to Saturn, Voyager over 20 years earlier, had onboard a ~70 kB 8-track storage ability. Machine learning can help scientists work with data on this larger scale. Unlike many applications of machine learning, a primary use in planetary space physics applications is to infer behavior about the system itself. This raises three concerns: first, the performance of the machine learning model, second, the need for interpretable applications to answer scientific questions, and third, how characteristics of spacecraft data change these applications. In comparison to these concerns, uses of “black box” or un-interpretable machine learning methods tend toward evaluations of performance only either ignoring the underlying physical process or, less often, providing misleading explanations for it. The present work uses Cassini data as a case study as these data are similar to space physics and planetary missions at Earth and other solar system objects. We build off a previous effort applying a semi-supervised physics-based classification of plasma instabilities in Saturn's magnetic environment, or magnetosphere. We then use this previous effort in comparison to other machine learning classifiers with varying data size access, and physical information access. We show that incorporating knowledge of these orbiting spacecraft data characteristics improves the performance and interpretability of machine leaning methods, which is essential for deriving scientific meaning. Building on these findings, we present a framework on incorporating physics knowledge into machine learning problems targeting semi-supervised classification for space physics data in planetary environments. These findings present a path forward for incorporating physical knowledge into space physics and planetary mission data analyses for scientific discovery.
Machine Learning for the Geosciences: Challenges and Opportunities
10.1109/TKDE.2018.2861006
https://ieeexplore.ieee.org/document/8423072
Geosciences is a field of great societal relevance that requires solutions to several urgent problems facing our humanity and the planet. As geosciences enters the era of big data, machine learning (ML)-that has been widely successful in commercial domains-offers immense potential to contribute to problems in geosciences. However, geoscience applications introduce novel challenges for ML due to combinations of geoscience properties encountered in every problem, requiring novel research in machine learning. This article introduces researchers in the machine learning (ML) community to these challenges offered by geoscience problems and the opportunities that exist for advancing both machine learning and geosciences. We first highlight typical sources of geoscience data and describe their common properties. We then describe some of the common categories of geoscience problems where machine learning can play a role, discussing the challenges faced by existing ML methods and opportunities for novel ML research. We conclude by discussing some of the cross-cutting research themes in machine learning that are applicable across several geoscience problems, and the importance of a deep collaboration between machine learning and geosciences for synergistic advancements in both disciplines.
Machine Learning and Evolutionary Techniques in Interplanetary Trajectory Design
https://link.springer.com/chapter/10.1007/978-3-030-10501-3_8
After providing a brief historical overview on the synergies between artificial intelligence research, in the areas of evolutionary computations and machine learning, and the optimal design of interplanetary trajectories, we propose and study the use of deep artificial neural networks to represent, on-board, the optimal guidance profile of an interplanetary mission. The results, limited to the chosen test case of an Earth–Mars orbital transfer, extend the findings made previously for landing scenarios and quadcopter dynamics, opening a new research area in interplanetary trajectory planning.
Data-driven surface traversability analysis for Mars 2020 landing site selection
10.1109/AERO.2016.7500597
https://ieeexplore.ieee.org/document/7500597
The objective of this paper is three-fold: 1) to describe the engineering challenges in the surface mobility of the Mars 2020 Rover mission that are considered in the landing site selection processs, 2) to introduce new automated traversability analysis capabilities, and 3) to present the preliminary analysis results for top candidate landing sites. The analysis capabilities presented in this paper include automated terrain classification, automated rock detection, digital elevation model (DEM) generation, and multi-ROI (region of interest) route planning. These analysis capabilities enable to fully utilize the vast volume of high-resolution orbiter imagery, quantitatively evaluate surface mobility requirements for each candidate site, and reject subjectivity in the comparison between sites in terms of engineering considerations. The analysis results supported the discussion in the Second Landing Site Workshop held in August 2015, which resulted in selecting eight candidate sites that will be considered in the third workshop."}">The objective of this paper is three-fold: 1) to describe the engineering challenges in the surface mobility of the Mars 2020 Rover mission that are considered in the landing site selection processs, 2) to introduce new automated traversability analysis capabilities, and 3) to present the preliminary analysis results for top candidate landing sites. The analysis capabilities presented in this paper include automated terrain classification, automated rock detection, digital elevation model (DEM) generation, and multi-ROI (region of interest) route planning. These analysis capabilities enable to fully utilize the vast volume of high-resolution orbiter imagery, quantitatively evaluate surface mobility requirements for each candidate site, and reject subjectivity in the comparison between sites in terms of engineering considerations. The analysis results supported the discussion in the Second Landing Site Workshop held in August 2015, which resulted in selecting eight candidate sites that will be considered in the third workshop.