Search NASA⌕ Search

SEARCH · Search NASA

Results for “CLASSES”

Search indexed NASA NTRS and DOE OSTI research on propulsion, heat transfer, battery materials and energy systems. Follow report and document links to the original sources.

Quote a phrase for an exact phrase match. Source license links do not imply unrestricted reuse.

At least 451 records · Page 25
All sourcesNASA NTRSDOE OSTI
DOE OSTI2021

Body Mass Variation in an Introduced Wild Pig Population with Changing Ancestry

Total body mass variation was studied in a single wild pig (Sus scrofa) population that originated from three introductions of varying ancestry. Body mass data from 12,257 wild pigs collected over more than five decades were analyzed for differences due to the effects of sex, age class, reproductive status, season, and historic subpopulation groupings. Total body mass data from this population averaged 38 kg and ranged from 0.511 to 189 kg. The overall observed range of body mass data was consistent with previous studies. Very heavy individuals (i.e., >150 kg) made up only a small percentage of the overall population. Males were significantly larger than females, being on average 1.09 times heavier. This significant sexual dimorphic difference was seen in all but the piglet age class. This percent dimorphic difference increased with age. Body mass variation and overlap between the age classes also increased with age. Males were more variable than females within the same age class. The study area’s wild pigs were on average born at slightly less than one percent of their adult body mass. Percent weight gain toward the adult body mass increased with age and was slightly greater in females than in males. In contrast, absolute weight gain between age classes was greater in males. Pregnant sows were significantly heavier than non-pregnant females in all age class comparisons; however, this difference decreased with age. There was a seasonal shift in body mass with age, with younger age classes being heaviest in the fall-winter period and the older age classes being heaviest in spring-summer. Having originated from different ancestral stocks, two major subpopulations of wild pigs on the study area differed significantly in body mass in all sex and age class comparisons. The SRS wild pig population showed a decrease in body mass during the 1990s that was significant. This change may have been the result of both ancestral and population density variation, as well as unspecified environmental factors

59 BASIC BIOLOGICAL SCIENCES↗
DOE OSTI2023

High-resolution (30-m) urban land cover projections for Los Angeles California Urban Area: 2010 to 2100 under SSP5

These data represent simulations of future land use and land cover (LULCC) for Los Angeles urban area (U.S. Census Bureau defined area) as raster tiff images at a 30-m pixel resolution and at decadal time steps from 2010 to 2100. LULCC classes in this product follow the National Land Cover Dataset (NLCD) classification. NLCD 21-24 correspond to open developed, low developed, medium developed, and high developed urban land classes, respectively. Only urban land cover classes (NLCD class 21, 22, 23, and 24) are dynamic over time; however, all NLCD classes are included in the final product. Therefore, NLCD classes that do not convert to an urban class will be similar to year 2000. The products were developed using a hybridized statistical and cellular automata approach. Linear mixed models (LMMs) were used to estimate future urban land budgets based on 1-km urban land fraction projections from Gao and Pesaresi (2021), whereas separate generalized linear mixed models (GLMMs) were used to estimate shifts in urban land intensities based on retrospective shifts in NLCD urban class intensities over a 20- year period. Based on urban land allocations from the statistical models, a cellular-automata and downscaling routine was used to simulate dynamic urban land expansion at a 30-m resolution based on suitability criteria. Scenarios of future urban landcover change projections include variant solutions for the Shared Socioeconomic Pathway 5 (SSP5) based on different population assumptions, different land use intensification assumptions, variable land zoning constraints, and iterative adjustments to correct for over allocation of urban expansion across decadal time periods from 2010 to 2100. This results in 320 raster products.

Land↗
DOE OSTI2023

High-resolution (30-m) urban land cover projections for Los Angeles California Urban Area: 2010 to 2100 under SSP3 and SSP5 [Updated simulations based on population-driven urban intensity transitions]

These data (v3) are updated from previous versions (1 and 2) in that they include consider the effects of population on transitions in urban land intensity. This leads to more reasonable differences in urban land projections under variant SSPs. For the present dataset, both SSP3 and SSP5 are provided. These data represent simulations of future land use and land cover (LULCC) for Los Angeles urban area (U.S. Census Bureau defined area) as raster tiff images at a 30-m pixel resolution and at decadal time steps from 2010 to 2100. LULCC classes in this product follow the National Land Cover Dataset (NLCD) classification. NLCD 21-24 correspond to open developed, low developed, medium developed, and high developed urban land classes, respectively. Only urban land cover classes (NLCD class 21, 22, 23, and 24) are dynamic over time; however, all NLCD classes are included in the final product. Therefore, NLCD classes that do not convert to an urban class will be similar to year 2000. The products were developed using a hybridized statistical and cellular automata approach. Linear mixed models (LMMs) were used to estimate future urban land budgets based on 1-km urban land fraction projections from Gao and Pesaresi (2021), whereas separate generalized linear mixed models (GLMMs) were used to estimate shifts in urban land intensities based on retrospective shifts in NLCD urban class intensities over a 20- year period. Based on urban land allocations from the statistical models, a cellular-automata and downscaling routine was used to simulate dynamic urban land expansion at a 30-m resolution based on suitability criteria. Scenarios of future urban landcover change projections include variant solutions for the Shared Socioeconomic Pathway 5 (SSP5) and SSP 3 based on different population assumptions, different land use intensification assumptions, variable land zoning constraints, and iterative adjustments to correct for over allocation of urban expansion across decadal time periods from 2010 to 2100. This results in 320 raster products.

Land↗
DOE OSTI2022

Semi-Supervised Domain Adaptation for Cross-Survey Galaxy Morphology Classification and Anomaly Detection

In the era of big astronomical surveys, our ability to leverage artificial intelligence algorithms simultaneously for multiple datasets will open new avenues for scientific discovery. Unfortunately, simply training a deep neural network on images from one data domain often leads to very poor performance on any other dataset. Here we develop a Universal Domain Adaptation method DeepAstroUDA, capable of performing semi-supervised domain alignment that can be applied to datasets with different types of class overlap. Extra classes can be present in any of the two datasets, and the method can even be used in the presence of unknown classes. For the first time, we demonstrate the successful use of domain adaptation on two very different observational datasets (from SDSS and DECaLS). We show that our method is capable of bridging the gap between two astronomical surveys, and also performs well for anomaly detection and clustering of unknown data in the unlabeled dataset. We apply our model to two examples of galaxy morphology classification tasks with anomaly detection: 1) classifying spiral and elliptical galaxies with detection of merging galaxies (three classes including one unknown anomaly class); 2) a more granular problem where the classes describe more detailed morphological properties of galaxies, with the detection of gravitational lenses (ten classes including one unknown anomaly class).

79 ASTRONOMY AND ASTROPHYSICS↗
DOE OSTI2024

Developing a Mapping Inquiry Tool and Geodatabase to Support Characterization of Candidate Geologic Carbon Storage Sites

Poster to showcase the Class VI Mapping Inquiry Tool (C6 MapIT) and support geodatabase available on EDX. C6 MapIT and the underpinning Class VI Data Support Tool Geodatabase serve as comprehensive resources to aid with understanding data availability and supporting the initial stages of Class VI siting and permit applications. In addition to enabling users to explore available spatial data, additional resources are provided to empower users to find more information, such as links to documentation to support project goals, a query tool for USGS publications, and links to relevant sections of the EPA Class VI documentation. Safe and appropriate selection of UIC Class VI well sites is key as the U.S. makes the energy transition. The Class VI Mapping Inquiry Tool and Class VI Data Tool Geodatabase are valuable, intuitive resources that empower commercial and industry stakeholders to explore, analyze, and acquire Class VI related data.

Pantaleone, Scott↗
DOE OSTI2025

Total cost of ownership of vehicle electrification and fuel switching options for light-duty and heavy-duty vehicles

Projecting the transition from combustion engines to battery-based powertrains is complex becuase it involves numerous interdependent decisions. This study estimates total cost of ownership (TCO) to assess the economic viability of powertrain electrification, focusing exclusively on advances in vehicle and fuel technologies. Under two bounding technology-progress scenarios, we develop vehicle designs and fuel cost trajectories, which serve as inputs to TCO projections for selected classes from 2021 to 2050. We analyzed a small sport utility vehicle (SUV) to represent the light-duty vehicle (LDV) sector, and four medium- and heavy-duty vehicle (MHDV) classes: Class 6 box delivery, Class 8 drayage, Class 8 long-haul, and Class 8 transit bus. For each class, we compared the TCO of battery electric vehicles (BEVs) and fuel cell hybrid electric vehicles (FCHEVs) against conventional internal combustion engine vehicles (ICEVs). The results show that modern ICEVs generally have lower TCO; however, BEVs and FCHEVs could match or have lower TCOs than ICEVs over time, depending on technological progress. In LDVs, BEV300 is projected to deliver the lowest TCO by 2050, particularly under the high-progress scenario. In MHDVs, both BEVs and FCHEVs could become more cost-competitive than ICEVs by 2050 in the high-progress case. Beyond these results, the findings suggest further investigation is warranted for BEV charging infrastructure, FCHEV hydrogen refueling infrastructure, and MHDV charging strategies. In conclusion, these factors could reduce the fuel-cost share of TCO and enhance the competitiveness of BEVs and FCHEVs relative to ICEVs.

Battery electric vehicle↗
DOE OSTI2022

Cellulose synthesis complexes are homo-oligomeric and hetero-oligomeric in Physcomitrium patens

The common ancestor of seed plants and mosses contained homo-oligomeric cellulose synthesis complexes (CSCs) composed of identical subunits encoded by a single CELLULOSE SYNTHASE (CESA) gene. Seed plants use different CESA isoforms for primary and secondary cell wall deposition. Both primary and secondary CESAs form hetero-oligomeric CSCs that assemble and function in planta only when all the required isoforms are present. The moss Physcomitrium (Physcomitrella) patens has seven CESA genes that can be grouped into two functionally and phylogenetically distinct classes. Previously, we showed that PpCESA3 and/or PpCESA8 (class A) together with PpCESA6 and/or PpCESA7 (class B) form obligate hetero-oligomeric complexes required for normal secondary cell wall deposition. Here, we show that gametophore morphogenesis requires a member of class A, PpCESA5, and is sustained in the absence of other PpCESA isoforms. Further, PpCESA5 also differs from the other class A PpCESAs as it is able to self-interact and does not co-immunoprecipitate with other PpCESA isoforms. These results are consistent with the hypothesis that homo-oligomeric CSCs containing only PpCESA5 subunits synthesize cellulose required for gametophore morphogenesis. Analysis of mutant phenotypes also revealed that, like secondary cell wall deposition, normal protonemal tip growth requires class B isoforms (PpCESA4 or PpCESA10), along with a class A partner (PpCESA3, PpCESA5, or PpCESA8). Thus, P. patens contains both homo-oligomeric and hetero-oligomeric CSCs.

59 BASIC BIOLOGICAL SCIENCES↗
DOE OSTI2022

Fully differential rail-to-rail output amplifier with inverter-based input pair

A fully differential rail-to-rail-output amplifier includes a differential input inverter pair, folded cascode pair, class AB control pair, and class AB output rail-to-rail pair. A drain associated with the folded cascode pair is operatively coupled to the class AB control pair, and the drain associated with the folded cascode pair is unconnected to the current source associated with the class AB control pair. A method of providing fully differential rail-to-rail-output amplification includes coupling a folded cascode pair operatively to a differential input inverter pair, coupling a drain associated with the folded cascode pair operatively to a class AB control pair, and coupling a class AB output rail-to-rail pair operatively to the class AB control pair.

Dabrowski, Mieczyslaw M.↗
DOE OSTI2022

Dynamical evolution in the D1D5 CFT

It is interesting to ask: how does the radial space direction emerge from the CFT in gauge-gravity duality? In this context we resolve a long-standing puzzle with the gravity duals of two classes of states in the D1D5 CFT. For each class the CFT states are in the untwisted sector, suggesting that the energy gap should be 1/R y where R y is the radius of the circle on which the D1D5 CFT is compactified. For one class of states, the gravity dual indeed has exactly this gap, while for the other class, the gravity dual has a very deep throat, leading to an energy gap much smaller than 1/R y . We resolve this puzzle by showing that for the latter class of states, perturbing the CFT off its free point leads to the formation of a band structure in the CFT. We also explain why such a band structure does not arise for the first class of states. Thus for the case where a deep throat emerges in the gravity description, the dynamics of falling down this throat is described in the CFT as a sequential ‘hopping’ between states all of which have the same energy at the free point; this hopping amplitude converts an integer spaced spectrum into a closely spaced band of energy levels.

72 PHYSICS OF ELEMENTARY PARTICLES AND FIELDS↗
DOE OSTI2024

Deficiency in PHD2-mediated hydroxylation of HIF2α underlies Pacak-Zhuang syndrome

Pacak-Zhuang syndrome is caused by mutations in the EPAS1 gene, which encodes for one of the three hypoxia-inducible factor alpha (HIFα) paralogs HIF2α and is associated with defined but varied phenotypic presentations including neuroendocrine tumors and polycythemia. However, the mechanisms underlying the complex genotype-phenotype correlations remain incompletely understood. Here, we devised a quantitative method for determining the dissociation constant (K d ) of the HIF2α peptides containing disease-associated mutations and the catalytic domain of prolyl-hydroxylase (PHD2) using microscale thermophoresis (MST) and showed that neuroendocrine-associated Class 1 HIF2α mutants have distinctly higher K d than the exclusively polycythemia-associated Class 2 HIF2α mutants. Based on the co-crystal structure of PHD2/HIF2α peptide complex at 1.8 Å resolution, we showed that the Class 1 mutated residues are localized to the critical interface between HIF2α and PHD2, adjacent to the PHD2 active catalytic site, while Class 2 mutated residues are localized to the more flexible region of HIF2α that makes less contact with PHD2. Concordantly, Class 1 mutations were found to significantly increase HIF2α-mediated transcriptional activation in cellulo compared to Class 2 counterparts. These results reveal a structural mechanism in which the strength of the interaction between HIF2α and PHD2 is at the root of the general genotype-phenotype correlations observed in Pacak-Zhuang syndrome.

59 BASIC BIOLOGICAL SCIENCES↗
DOE OSTI2024

Strategies for Developing High-Volume Fly Ash Concrete with High Early-Age Strength for Precast Applications

Partial replacement of portland cement with supplementary cementitious materials (SCMs), such as fly ash, is an effective strategy for improving durability and reducing the CO 2 footprint of concrete. However, using high-volume fly ash (HVFA) binders in precast and prestressed concrete is currently limited; largely due to reduced early-age strength development that impedes rapid production and prestressing of precast concrete. To investigate and address this challenge, HVFA mortars with a minimum of 40% fly ash by mass of cementitious materials were developed and tested in this study. Two fresh fly ashes (an ASTM C618 Class F and a Class C) and a landfilled fly ash (Class F) were included. Various strategies for improving the early strength were evaluated, including gypsum optimization, chemical accelerators, steam curing, use of CSA cements, and adding other reactive SCMs like silica fume, calcined clay, and slag cement. Steam curing and the use of CSA cement at high dosages (40% of total binder) were found to be the most successful strategies across all three fly ashes. Additionally, significant improvements were observed with gypsum optimization (for Class C fly ash) and the use of accelerators (for Class F fly ashes), and these strategies are likely to be more feasible considering later-age strength and economic viability. Interestingly, HVFA mixtures made with the landfilled fly ash used in this study were able to achieve high early strengths with water-to-cementitious materials ratio adjustment alone. As a result, these HVFA mixtures were also found to be less responsive to accelerators when compared to the fresh Class F fly ash, highlighting an important distinction between the materials despite the similarity in chemical composition.

42 ENGINEERING↗
DOE OSTI2023

DeepAstroUDA: semi-supervised universal domain adaptation for cross-survey galaxy morphology classification and anomaly detection

Abstract Artificial intelligence methods show great promise in increasing the quality and speed of work with large astronomical datasets, but the high complexity of these methods leads to the extraction of dataset-specific, non-robust features. Therefore, such methods do not generalize well across multiple datasets. We present a universal domain adaptation method, DeepAstroUDA , as an approach to overcome this challenge. This algorithm performs semi-supervised domain adaptation (DA) and can be applied to datasets with different data distributions and class overlaps. Non-overlapping classes can be present in any of the two datasets (the labeled source domain, or the unlabeled target domain), and the method can even be used in the presence of unknown classes. We apply our method to three examples of galaxy morphology classification tasks of different complexities (three-class and ten-class problems), with anomaly detection: (1) datasets created after different numbers of observing years from a single survey (Legacy Survey of Space and Time mock data of one and ten years of observations); (2) data from different surveys (Sloan Digital Sky Survey (SDSS) and DECaLS); and (3) data from observing fields with different depths within one survey (wide field and Stripe 82 deep field of SDSS). For the first time, we demonstrate the successful use of DA between very discrepant observational datasets. DeepAstroUDA is capable of bridging the gap between two astronomical surveys, increasing classification accuracy in both domains (up to 40 % on the unlabeled data), and making model performance consistent across datasets. Furthermore, our method also performs well as an anomaly detection algorithm and successfully clusters unknown class samples even in the unlabeled target dataset.

79 ASTRONOMY AND ASTROPHYSICS↗
DOE OSTI2024

Toward a classification of PT-symmetric quantum systems: From dissipative dynamics to topology and wormholes

Studies of many-body non-Hermitian parity-time (PT)-symmetric quantum systems are attracting a lot of interest due to their relevance in research areas ranging from quantum optics and continuously monitored dynamics to Euclidean wormholes in quantum gravity and dissipative quantum chaos. While a symmetry classification of non-Hermitian systems leads to 38 universality classes, we show that, under certain conditions, PT-symmetric systems are grouped into 24 universality classes. We identify 14 of them in a coupled two-site Sachdev-Ye-Kitaev (SYK) model and confirm the classification by spectral analysis using exact diagonalization techniques. Intriguingly, in 4 of these 14 universality classes, AIII ν , BDI ν † , BDI + + ν , and CI − − ν , we identify a basis in which the SYK Hamiltonian has a block structure in which some blocks are rectangular, with ν ∈ N the difference between the number of rows and columns. We show analytically that this feature leads to the existence of ν robust purely eigenvalues, whose level statistics follow the predictions of Hermitian random matrix theory for classes A, AI, BDI, and CI, respectively. We have recently found that this ν is a topological invariant, so these classes are topological. By contrast, nontopological real eigenvalues display a crossover between Hermitian and non-Hermitian level statistics. Similarly to the case of Lindbladian dynamics, the reduction of universality classes leads to unexpected results, such as the absence of Kramers degeneracy in a given sector of the theory. Another novel feature of the classification scheme is that different sectors of the PT-symmetric Hamiltonian may have different symmetries. Published by the American Physical Society 2024

Astronomy & Astrophysics↗
DOE OSTI2020

Structure–activity relationship-based chemical classification of highly imbalanced Tox21 datasets

Abstract The specificity of toxicant-target biomolecule interactions lends to the very imbalanced nature of many toxicity datasets, causing poor performance in Structure–Activity Relationship (SAR)-based chemical classification. Undersampling and oversampling are representative techniques for handling such an imbalance challenge. However, removing inactive chemical compound instances from the majority class using an undersampling technique can result in information loss, whereas increasing active toxicant instances in the minority class by interpolation tends to introduce artificial minority instances that often cross into the majority class space, giving rise to class overlapping and a higher false prediction rate. In this study, in order to improve the prediction accuracy of imbalanced learning, we employed SMOTEENN, a combination of Synthetic Minority Over-sampling Technique (SMOTE) and Edited Nearest Neighbor (ENN) algorithms, to oversample the minority class by creating synthetic samples, followed by cleaning the mislabeled instances. We chose the highly imbalanced Tox21 dataset, which consisted of 12 in vitro bioassays for > 10,000 chemicals that were distributed unevenly between binary classes. With Random Forest (RF) as the base classifier and bagging as the ensemble strategy, we applied four hybrid learning methods, i.e., RF without imbalance handling (RF), RF with Random Undersampling (RUS), RF with SMOTE (SMO), and RF with SMOTEENN (SMN). The performance of the four learning methods was compared using nine evaluation metrics, among which F 1 score, Matthews correlation coefficient and Brier score provided a more consistent assessment of the overall performance across the 12 datasets. The Friedman’s aligned ranks test and the subsequent Bergmann-Hommel post hoc test showed that SMN significantly outperformed the other three methods. We also found that a strong negative correlation existed between the prediction accuracy and the imbalance ratio (IR), which is defined as the number of inactive compounds divided by the number of active compounds. SMN became less effective when IR exceeded a certain threshold (e.g., > 28). The ability to separate the few active compounds from the vast amounts of inactive ones is of great importance in computational toxicology. This work demonstrates that the performance of SAR-based, imbalanced chemical toxicity classification can be significantly improved through the use of data rebalancing.

Idakwo, Gabriel↗
DOE OSTI2021

CSPlib - A Software Toolkit for the Analysis of Dynamical Systems and Chemical Kinetic Models

CSPlib is an open source software library for analyzing general ordinary differential equation (ODE) systems and detailed chemical kinetic ODE systems. It relies on the computational singular perturbation (CSP) method for the analysis of these systems. The software provides support for: General ODE models (gODE model class) for computing source terms and Jacobians for a generic ODE system; TChem model (ChemElemODETChem model class) for computing source term, Jacobian, other necessary chemical reaction data, as well as the rates of progress for a homogenous batch reactor using an elementary step detailed chemical kinetic reaction mechanism. This class relies on the TChem [2] library; A set of functions to compute essential elements of CSP analysis (Kernel class). This includes computations of the eigensolution of the Jacobian matrix, CSP basis vectors and co-vectors, time scales (reciprocals of the magnitudes of the Jacobian eigenvalues), mode amplitudes, CSP pointers, and the number of exhausted modes. This class relies on the Tines library; A set of functions to compute the eigensolution of the Jacobian matrix using Tines library GPU eigensolver; A set of functions to compute CSP indices (Index Class). This includes participation indices and both slow and fast importance indices.

97 MATHEMATICS AND COMPUTING↗
DOE OSTI2023

A genome-informed higher rank classification of the biotechnologically important fungal subphylum Saccharomycotina

The subphylum Saccharomycotina is a lineage in the fungal phylum Ascomycota that exhibits levels of genomic diversity similar to those of plants and animals. The Saccharomycotina consist of more than 1 200 known species currently divided into 16 families, one order, and one class. Species in this subphylum are ecologically and metabolically diverse and include important opportunistic human pathogens, as well as species important in biotechnological applications. Many traits of biotechnological interest are found in closely related species and often restricted to single phylogenetic clades. However, the biotechnological potential of most yeast species remains unexplored. Although the subphylum Saccharomycotina has much higher rates of genome sequence evolution than its sister subphylum, Pezizomycotina, it contains only one class compared to the 16 classes in Pezizomycotina. The third subphylum of Ascomycota, the Taphrinomycotina, consists of six classes and has approximately 10 times fewer species than the Saccharomycotina. These data indicate that the current classification of all these yeasts into a single class and a single order is an underappreciation of their diversity. Our previous genome-scale phylogenetic analyses showed that the Saccharomycotina contains 12 major and robustly supported phylogenetic clades; seven of these are current families (Lipomycetaceae, Trigonopsidaceae, Alloascoideaceae, Pichiaceae, Phaffomycetaceae, Saccharomycodaceae, and Saccharomycetaceae), one comprises two current families (Dipodascaceae and Trichomonascaceae), one represents the genus Sporopachydermia, and three represent lineages that differ in their translation of the CUG codon (CUG-Ala, CUG-Ser1, and CUG-Ser2). Using these analyses in combination with relative evolutionary divergence and genome content analyses, we propose an updated classification for the Saccharomycotina, including seven classes and 12 orders that can be diagnosed by genome content. This updated classification is consistent with the high levels of genomic diversity within this subphylum and is necessary to make the higher rank classification of the Saccharomycotina more comparable to that of other fungi, as well as to communicate efficiently on lineages that are not yet formally named.

59 BASIC BIOLOGICAL SCIENCES↗
DOE OSTI2022

A lipidomics platform to analyze the fatty acid compositions of non-polar and polar lipid molecular species from plant tissues: Examples from developing seeds and seedlings of pennycress (Thlaspi arvense)

The lipidome comprises the total content of molecular species of each lipid class, and is measured using the analytical techniques of lipidomics. Many liquid chromatography-mass spectrometry (LC-MS) methods have previously been described to characterize the lipidome. However, many lipidomic approaches may not fully uncover the subtleties of lipid molecular species, such as the full fatty acid (FA) composition of certain lipid classes. Here, we describe a stepwise targeted lipidomics approach to characterize the polar and non-polar lipid classes using complementary LC-MS methods. Our “polar” method measures 260 molecular species across 12 polar lipid classes, and is performed using hydrophilic interaction chromatography (HILIC) on a NH2 column to separate lipid classes by their headgroup. Our “non-polar” method measures 254 molecular species across three non-polar lipid classes, separating molecular species on their FA characteristics by reverse phase (RP) chromatography on a C30 column. Five different extraction methods were compared, with an MTBE-based extraction chosen for the final lipidomics workflow. A state-of-the-art strategy to determine and relatively quantify the FA composition of triacylglycerols is also described. This lipidomics workflow was applied to developing, mature, and germinated pennycress seeds/seedlings and found unexpected changes among several lipid molecular species. During development, diacylglycerols predominantly contained long chain length FAs, which contrasted with the very long chain FAs of triacylglycerols in mature seeds. Potential metabolic explanations are discussed. The lack of very long chain fatty acids in diacylglycerols of germinating seeds may indicate very long chain FAs, such as erucic acid, are preferentially channeled into beta-oxidation for energy production.

59 BASIC BIOLOGICAL SCIENCES↗
DOE OSTI2022

AICCA: AI-Driven Cloud Classification Atlas

Clouds play an important role in the Earth’s energy budget, and their behavior is one of the largest uncertainties in future climate projections. Satellite observations should help in understanding cloud responses, but decades and petabytes of multispectral cloud imagery have to date received only limited use. This study describes a new analysis approach that reduces the dimensionality of satellite cloud observations by grouping them via a novel automated, unsupervised cloud classification technique based on a convolutional autoencoder, an artificial intelligence (AI) method good at identifying patterns in spatial data. Our technique combines a rotation-invariant autoencoder and hierarchical agglomerative clustering to generate cloud clusters that capture meaningful distinctions among cloud textures, using only raw multispectral imagery as input. Cloud classes are therefore defined based on spectral properties and spatial textures without reliance on location, time/season, derived physical properties, or pre-designated class definitions. We use this approach to generate a unique new cloud dataset, the AI-driven cloud classification atlas (AICCA), which clusters 22 years of ocean images from the Moderate Resolution Imaging Spectroradiometer (MODIS) on NASA’s Aqua and Terra instruments—198 million patches, each roughly 100 km × 100 km (128 × 128 pixels)—into 42 AI-generated cloud classes, a number determined via a newly-developed stability protocol that we use to maximize richness of information while ensuring stable groupings of patches. AICCA thereby translates 801 TB of satellite images into 54.2 GB of class labels and cloud top and optical properties, a reduction by a factor of 15,000. The 42 AICCA classes produce meaningful spatio-temporal and physical distinctions and capture a greater variety of cloud types than do the nine International Satellite Cloud Climatology Project (ISCCP) categories—for example, multiple textures in the stratocumulus decks along the West coasts of North and South America. We conclude that our methodology has explanatory power, capturing regionally unique cloud classes and providing rich but tractable information for global analysis. AICCA delivers the information from multi-spectral images in a compact form, enables data-driven diagnosis of patterns of cloud organization, provides insight into cloud evolution on timescales of hours to decades, and helps democratize climate research by facilitating access to core data.

97 MATHEMATICS AND COMPUTING↗