Engineering Papers⌕ Search

SEARCH · Engineering Papers

Results for “Datasets”

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 1,441 records · Page 80

Leveraging large language models to address data scarcity in machine learning for graphene synthesis

Machine learning in experimental materials science faces significant challenges due to the scarcity of data, which are costly and time-consuming to generate, particularly when relying on in-house experiments. Literature data mining offers a potential solution but introduces issues like mixed data quality, inconsistent formats, and non-uniform reporting of synthesis parameters, resulting in partially missing and heterogeneous features across the dataset. Here, we propose data imputation and feature engineering methods that employ pre-trained large language models (LLMs) to enhance machine learning performance on scarce, heterogeneous datasets, demonstrated on graphene CVD synthesis data and the ML-HydPARK hydrogen storage dataset. GPT models perform data imputation via tailored prompting and semantic normalization of inconsistently reported features through embeddings, for example, to harmonize the complex nomenclature of CVD substrates. Beyond yielding more diverse and richer feature representations than traditional methods such as K-nearest neighbors (KNN) and Multivariate Imputation by Chained Equations (MICE), LLM-based data imputation is evaluated against dataset characteristics and prompting strategies. We vary the level of autonomy granted to the LLM, from generic prompting that leverages pre-trained knowledge for autonomous data generation to data-informed prompting that constrains outputs using target-specific information, and demonstrate which level of autonomy yields superior imputation performance across datasets and feature types. The proposed data engineering methods markedly improve downstream performance; for example, in graphene layer number classification using a support vector machine (SVM), binary accuracy increases from 39% to 65% and ternary accuracy from 52% to 72%. Fine-tuning experiments on both datasets show that combining our proposed LLM-based data imputation and feature encoding methods with numerical machine learning predictors outperforms standalone fine-tuned LLM predictors in data-scarce settings. The proposed strategies emphasize data enhancement techniques rather than refining learning architectures or regularizing loss functions, offering a broadly applicable framework for improving machine learning performance on scarce, inhomogeneous datasets.

Chemical vapor deposition↗

GriddingMachine, a database and software for Earth system modeling at global and regional scales

Land and Earth system modeling is moving towards more explicit biophysical representations, requiring increasing variety of datasets for initialization and benchmarking. However, researchers often have difficulties in identifying and integrating non-standardized datasets from various sources. We aim towards a standardized database and one-stop distribution method of global datasets. Here, we present the GriddingMachine as (1) a database of global-scale datasets commonly used to parameterize or benchmark the models, from plant traits to vegetation indices and geophysical information and (2) a cross-platform open source software to download and request a subset of datasets with only a few lines of code. The GriddingMachine datasets can be accessed either manually through traditional HTTP, or automatically using modern programming languages including Julia, Matlab, Octave, Python, and R. The GriddingMachine collections can be used for any land and Earth modeling framework and ecological research at the regional and global scales, and the number of datasets will continue to grow to meet the increasing needs of research communities.

58 GEOSCIENCES↗

Assessing Spatial Representativeness of Global Flux Tower Eddy-Covariance Measurements Using Data from FLUXNET2015

Large datasets of carbon dioxide, energy, and water fluxes were measured with the eddy-covariance (EC) technique, such as FLUXNET2015. These datasets are widely used to validate remote-sensing products and benchmark models. One of the major challenges in utilizing EC-flux data is determining the spatial extent to which measurements taken at individual EC towers reflect model-grid or remote sensing pixels. To minimize the potential biases caused by the footprint-to-target area mismatch, it is important to use flux datasets with awareness of the footprint. This study analyze the spatial representativeness of global EC measurements based on the open-source FLUXNET2015 data, using the published flux footprint model (SAFE-f). The calculated annual cumulative footprint climatology (ACFC) was overlaid on land cover and vegetation index maps to create a spatial representativeness dataset of global flux towers. The dataset includes the following components: (1) the ACFC contour (ACFCC) data and areas representing 50%, 60%, 70%, and 80% ACFCC of each site, (2) the proportion of each land cover type weighted by the 80% ACFC (ACFCW), (3) the semivariogram calculated using Normalized Difference Vegetation Index (NDVI) considering the 80% ACFCW, and (4) the sensor location bias (SLB) between the 80% ACFCW and designated areas (e.g. 80% ACFCC and window sizes) proxied by NDVI. Finally, we conducted a comprehensive evaluation of the representativeness of each site from three aspects: (1) the underlying surface cover, (2) the semivariogram, and (3) the SLB between 80% ACFCW and 80% ACFCC, and categorized them into 3 levels. The goal of creating this dataset is to provide data quality guidance for international researchers to effectively utilize the FLUXNET2015 dataset in the future.

54 ENVIRONMENTAL SCIENCES↗

Lack of clear standards and usable comparisons of downscaled climate projections pose a roadblock for US climate discovery and adaptation

Abstract The release of global climate projections coupled with the demand for local-resolution climate-forced meteorology has prompted many research groups to downscale these projections using various statistical, dynamical, and current machine learning techniques. Such downscaled datasets are being used to plan infrastructure and other community needs over the coming decades. Faced with roughly a dozen available US downscaled datasets, many practitioners ask, ‘What are the relevant differences between datasets?’ This work highlights the difficulty of comparing downscaled datasets and illustrates ways in which datasets differ even when using identical climate model input data. We show that substantial variability in precipitation projections arises from downscaling alone and that the downscaled dataset agreement varies depending on global climate projection. This analysis emphasizes the need for greater coordination and movement toward rigorous benchmarking of downscaling strategies within the downscaling research community, à la the land-modeling community, to better quantify downscaling dataset differences, strengths, and weaknesses for practitioners.

Hartke, Samantha H. (ORCID:0000000202394723)↗

Unsupervised atomic data mining via multi-kernel graph autoencoders for machine learning force fields

Constructing a chemically diverse dataset while avoiding sampling bias is critical to training efficient and generalizable force fields. However, in computational chemistry and materials science, many common dataset generation techniques are prone to oversampling regions of the potential energy surface. Furthermore, these regions can be difficult to identify and isolate from each other or may not align well with human intuition, making it challenging to systematically remove bias in the dataset. While traditional clustering and pruning (down-sampling) approaches can be useful for this, they can often lead to information loss or a failure to properly identify distinct regions of the potential energy surface due to difficulties associated with the high dimensionality of atomic descriptors. In this work, we introduce the Multi-kernel Edge Attention-based Graph Autoencoder (MEAGraph) model, an unsupervised approach for analyzing atomic datasets. MEAGraph combines multiple linear kernel transformations with attention-based message passing to capture geometric sensitivity and enable effective dataset pruning without relying on labels or extensive training. Demonstrated applications on niobium, tantalum, and iron datasets show that MEAGraph efficiently groups similar atomic environments, allowing for the use of basic pruning techniques for removing sampling bias. This approach provides an effective method for representation learning and clustering that can be used for data analysis, outlier detection, and dataset optimization.

Materials science↗

Characterizing Quantum Classifier Utility in Natural Language Processing Workflows

Quantum Natural Language Processing (QNLP) develops natural language processing (NLP) models for deployment on quantum computers. We explore feature and data prototype selection techniques to address challenges posed by encoding high dimensional features. Our study builds quantum circuit classifiers that includes classical feature pre-processing, quantum embedding and quantum model training. The quantum models are built on 4 or 6 qubits and the quantum neural network (QNN) uses the established bricklayer design. We compare the dependence of model performance (in terms of accuracy and F1 scores) on feature length, embedding gates and parameterized unitary design. We compare the performance of quantum machine learning models to classical convolution neural network model (CNN) on binary and multi-class classification tasks using two datasets of synthetic features and labels. The first is the ECP-CANDLE P3B3 dataset a corpus of synthetically generated cancer pathology reports. The second dataset is extracted from well-known benchmark dataset (MADELON) - features are generated with a combination of informative, repeated and uninformative features. Both datasets are used for binary classification and multi-class classification with 3 classes. We observe robust, accurate performance from all models on the binary classification tasks, but multiclass classification is a challenge for the quantum models-there is a notable decrease in accuracy when using 3 classes. Overall the performance is comparable in terms of recall and accuracy between QNNs and CNNs, even with large datasets. These results provide a point of comparison between quantum and classical models on real-world datasets.

Hamilton, Kathleen↗

Enhancing Electron Microscopy Image Classification Using Data Augmentation

Manual labeling for machine learning tasks such as image classification is tedious and labor-intensive; as a result, scientific datasets suitable for deep learning applications are scarce and limited. While data augmentation techniques have shown promise for extending image datasets, very little work has been done to understand the impact of combining multiple augmentation methods sequentially or the limits of their effectiveness when combined. Our work addresses this gap by examining how standard and combinatorial data augmentation affects the performance of machine learning models when trained on small datasets for label classification tasks. For our analysis, we generate single, double and quadruple-augmented datasets for a microscopy image classification task using six standard augmentation methods, and compare the resultant improvements observed in binary classification accuracy with three standard image classification models (DenseNet169, MobileNetV2, ResNet101V2). Our experiments show a non-monotonic relationship between the number of simultaneous augmentation methods and classification accuracy, indicating that there is a trade-off between the degree of augmentation and the model performance. These findings suggest that the optimal number of augmentation methods will vary by domain and use case. We also find that the order in which augmentation methods are applied to a limited dataset matters when combining augmentation schemes, with our use case showing performance differences up to 2.6% when the augmentation order is reversed for double-augmented datasets. Our work offers insights to the limits of data augmentation when working on image classification tasks with limited datasets.

Welsman, Jordan A↗

Cumulative sum quality control for calibrated breast density measurements

Purpose: Breast density is a significant breast cancer risk factor. Although various methods are used to estimate breast density, there is no standard measurement for this important factor. The authors are developing a breast density standardization method for use in full field digital mammography (FFDM). The approach calibrates for interpatient acquisition technique differences. The calibration produces a normalized breast density pixel value scale. The method relies on first generating a baseline (BL) calibration dataset, which required extensive phantom imaging. Standardizing prospective mammograms with calibration data generated in the past could introduce unanticipated error in the standardized output if the calibration dataset is no longer valid. Methods: Sample points from the BL calibration dataset were imaged approximately biweekly over an extended timeframe. These serial samples were used to evaluate the BL dataset reproducibility and quantify the serial calibration accuracy. The cumulative sum (Cusum) quality control method was used to evaluate the serial sampling. Results: There is considerable drift in the serial sample points from the BL calibration dataset that is x-ray beam dependent. Systematic deviation from the BL dataset caused significant calibration errors. This system drift was not captured with routine system quality control measures. Cusum analysis indicated that the drift is a sign of system wear and eventual x-ray tube failure. Conclusions: The BL calibration dataset must be monitored and periodically updated, when necessary, to account for sustained system variations to maintain the calibration accuracy.

62 RADIOLOGY AND NUCLEAR MEDICINE↗

Dosimetric comparison of treatment plans based on free breathing, maximum, and average intensity projection CTs for lung cancer SBRT

Purpose: To determine whether there is a CT dataset may be more favorable for planning and dose calculation by comparing dosimetric characteristics between treatment plans calculated using free breathing (FB), maximum and average intensity projection (MIP and AIP, respectively) CTs for lung cancer patients receiving stereotactic body radiation therapy (SBRT). Methods: Twenty lung cancer SBRT patients, treated on a linac with 2.5 mm width multileaf-collimator (MLC), were analyzed retrospectively. Both FB helical and four-dimensional CT scans were acquired for each patient. Internal target volume (ITV) was delineated based on MIP CTs and modified based on both ten-phase datasets and FB CTs. Planning target volume (PTV) was then determined by adding additional setup margin to ITV. The PTVs and beams in the optimized treatment plan based on FB CTs were copied to MIP and AIP CTs, with the same isocenters, MLC patterns and monitor units. Mean effective depth (MED) of beams, and some dosimetric parameters for both PTVs and most important organ at risk (OAR), lung minus PTV, were compared between any two datasets using two-tail paired t test. Results: The MEDs in FB and AIP plans were similar but significantly smaller (Ps < 0.001) than that in MIP plans. Minimum dose, mean dose, dose covering at least 90% and 95% of PTVs in MIP plans were slightly higher than two other plans (Ps < 0.008). The absolute volume of lung minus PTV receiving greater than 5, 10, and 20 Gy in MIP plans were significantly smaller than those in both FB and AIP plans (Ps < 0.008). Conformity index for FB plans showed a small but statistically significantly higher. Conclusions: Dosimetric characteristics of AIP plans are similar to those of FB plans. Slightly better target volume coverage and significantly lower low-dose region ({<=}30 Gy) in lung was observed in MIP plans. The decrease in low-dose region in lung was mainly caused by the change of lung volume contoured on two datasets rather than the differences of dose distribution between AIP and MIP plans. Compare with AIP datasets, FB datasets were more prone to significant image artifacts and MIP datasets may overestimate or underestimate the target volume when the target is closer to the denser tissue, so AIP seems favorable for planning and dose calculation for lung SBRT.

61 RADIATION PROTECTION AND DOSIMETRY↗

Deep machine learning-assisted multiphoton microscopy to reduce light exposure and expedite imaging

Two-photon excitation fluorescence (2PEF) allows imaging of tissue up to about one millimeter in thickness. Typically, reducing fluorescence excitation exposure reduces the quality of the image. However, using deep learning super resolution techniques, these low-resolution images can be converted to high-resolution images. This work explores improving human tissue imaging by applying deep learning to maximize image quality while reducing fluorescence excitation exposure. We analyze two methods: a method based on U-Net, and a patch-based regression method. Both methods are evaluated on a skin dataset and an eye dataset. The eye dataset includes 1200 paired high power and low power images of retinal organoids. The skin dataset contains multiple frames of each sample of human skin. High-resolution images were formed by averaging 70 frames for each sample and low-resolution images were formed by averaging the first 7 and 15 frames for each sample. The skin dataset includes 550 images for each of the resolution levels. We track two measures of performance for the two methods: mean squared error (MSE) and structural similarity index measure (SSIM). For the eye dataset, the patches method achieves an average MSE of 27,611 compared to 146,855 for the U-Net method, and an average SSIM of 0.636 compared to 0.607 for the U-Net method. For the skin dataset, the patches method achieves an average MSE of 3.768 compared to 4.032 for the U-Net method, and an average SSIM of 0.824 compared to 0.783 for the U-Net method. Despite better performance on image quality, the patches method is worse than the U-Net method when comparing the speed of prediction, taking 303 seconds to predict one image compared to less than one second for the U-Net method.

62 RADIOLOGY AND NUCLEAR MEDICINE↗

Utilizing novel diversity estimators to quantify multiple dimensions of microbial biodiversity across domains

Background: Microbial ecologists often employ methods from classical community ecology to analyze microbial community diversity. However, these methods have limitations because microbial communities differ from macro-organismal communities in key ways. This study sought to quantify microbial diversity using methods that are better suited for data spanning multiple domains of life and dimensions of diversity. Diversity profiles are one novel, promising way to analyze microbial datasets. Diversity profiles encompass many other indices, provide effective numbers of diversity (mathematical generalizations of previous indices that better convey the magnitude of differences in diversity), and can incorporate taxa similarity information. To explore whether these profiles change interpretations of microbial datasets, diversity profiles were calculated for four microbial datasets from different environments spanning all domains of life as well as viruses. Both similarity-based profiles that incorporated phylogenetic relatedness and naïve (not similarity-based) profiles were calculated. Simulated datasets were used to examine the robustness of diversity profiles to varying phylogenetic topology and community composition. Results: Diversity profiles provided insights into microbial datasets that were not detectable with classical univariate diversity metrics. For all datasets analyzed, there were key distinctions between calculations that incorporated phylogenetic diversity as a measure of taxa similarity and naïve calculations. The profiles also provided information about the effects of rare species on diversity calculations. Additionally, diversity profiles were used to examine thousands of simulated microbial communities, showing that similarity-based and naïve diversity profiles only agreed approximately 50% of the time in their classification of which sample was most diverse. This is a strong argument for incorporating similarity information and calculating diversity with a range of emphases on rare and abundant species when quantifying microbial community diversity. Conclusions: For many datasets, diversity profiles provided a different view of microbial community diversity compared to analyses that did not take into account taxa similarity information, effective diversity, or multiple diversity metrics. These findings are a valuable contribution to data analysis methodology in microbial ecology.

54 ENVIRONMENTAL SCIENCES↗

SMC 2021 Data Challenge: Analyzing Resource Utilization and User Behavior on Titan Supercomputer

Resource utilization statistics of submitted jobs on a supercomputer can help us understand how users from various scientific domains use HPC platforms and better design a job scheduler. We explore to generate insight regarding workload distribution and usage pattern domains from job scheduler trace, GPU failure information, and project-specific information collected from Titan supercomputer. Furthermore, we want to know how the scheduler performance varies over time and how the users' scheduling behavior changes following a system failure. These observations have the potential to provide valuable insight, which is helpful to prepare for system failures. These practices will help us develop and apply novel machine learning algorithms in understanding system behavior, requirement, and better scheduling of HPC systems. There are two datasets, RUR and GPU: RUR dataset is the job scheduler traces collected from the Titan supercomputer from 01/01/2015 to 07/31/2019 (2015.csv - 2019.csv). These were collected using resource Utilization Report (RUR), a Cray-developed resource-usage data collection and reporting system. It contains the usage information of its critical resources (CPU, Memory, GPU, and I/O) of each running job on Titan during that period (https://ieeexplore.ieee.org/abstract/document/8891001). It includes ProjectAreas as additional information, every job is associated with a project ID. TheProjectAreas.csv dataset provides a mapping of the project ID to its domain science. GPU dataset has information regarding GPU failure on Titan. There have been some hardware-related issues in the GPUs in Titan that caused some GPUs to fail, sometimes irrecoverably during some job runs. This dataset provides information regarding these failures during the execution of the submitted jobs. GPUs on Titan are uniquely identified by a serial number (SN), and they are installed in a location. A GPU can be installed in a location, then removed from that location following a failure, and then re-installed in a different location after fixing the problem. If the failure can't be recovered, the GPU might be removed entirely from Titan. There are two prominent types of failures that resulted in the removal of GPUs from Titan: Double Bit Error (DBE) and Out of the Bus (OTB). The dataset (gc_full.csv) has seven attributes, we provided a short description of these attributes in the ReadMe file. To learn more about this dataset, please refer to the git repository https://github.com/olcf/TitanGPULife and the related publication (https://ieeexplore.ieee.org/abstract/document/9355319).

42 ENGINEERING↗

ORNL_AISD-Ex: Quantum chemical prediction of UV/Vis absorption spectra for over 10 million organic molecules

We performed calculations of electronic excitation energies and associated oscillator strengths based on the time-dependent density-functional tight-binding (TD-DFTB) method [1]. The SMILES (Simplified molecular-input line-entry system) strings of the molecules from the AISD HOMO-LUMO database [2] were converted to a 3D atomistic structure and stored in a PDB file after preliminary geometry optimization using the Merck Molecular Force Field (MMFF94) in RDKit [3,4]. The primary information stored in the PDB file archive consists of Cartesian coordinates for each atom of the molecule in their 3D location in space, along with summary information about the structure, sequence, and experiment. We then performed molecular geometry optimization using the density-functional tight-binding (DFTB) method [5] in the electronic ground state, followed by single-point excited states calculations, as described below. We note that, since RDKit employs a random choice for the generation of molecular conformers, the molecular geometries obtained in this dataset could be different from the ones that were generated when the AISD HOMO-LUMO dataset was generated. The computed excitation energies and associated oscillator strengths can be converted to predict UV/Vis absorption spectra, where excitation energies correspond to absorption peak positions, and oscillator strengths are a good measure of the probability of absorption of visible or UV light in transitions between electronic ground and excited states. The conversion of SMILES strings to 3D Cartesian coordinates of fully DFTB-optimized molecules was successful for 10,502,904 out of 10,502,917 molecules. For these molecules, both geometry optimizations and excited states calculations were successful. The DFTB calculations did not complete for 13 molecules of the original AISD HOMO-LUMO dataset. We still provide information about the geometry of these molecules. The molecules are diverse for chemical compositions (which span 5 non-hydrogen elements: oxygen, carbon, nitrogen, fluorine, sulfur) and molecular size (the smallest molecule contains 5 non-hydrogen atoms, and the largest molecule contains 71 non-hydrogen atoms). The DFTB method [5] is an approximation to density functional theory (DFT), utilizing a minimal basis set in conjunction with a two-center approximation to the electronic Hamiltonian and overlap matrix elements. The DFTB total energy is the sum of an electronic and a repulsive energy contribution, and their calculation requires optimized electronic parameters and diatomic repulsive potential energy functions. All DFTB calculations were performed using the DFTB+ code [6] (version 21.2) and the wrapper for DFTB+ in the Atomic Simulation Environment (ASE) (version 3.22.1) [7], which performed an internal conversion of Cartesian coordinates from PDB to the .gen file format. For the geometry optimizations on the electronic ground state potential energy surface of the molecules, we have chosen the third-order DFTB (DFTB3) method [5c] and employed the matching 3ob set of electronic parameters and repulsive potentials [8]. The empirical γ-damping for hydrogen bond correction, and Grimme's D3 empirical dispersion correction with Becke-Johnson damping (D3(BJ)) [9] dispersion correction was included to improve the description of non-covalent interactions. For excited states single-point energy calculations, we employed the TD-DFTB method in conjunction with the DFTB2 method [5b] and the matching mio [5b,10] and halorg [11] parameter sets. We opted to request the simultaneous calculation of 50 excited states for singlet transition to investigate sufficient number of excited states, based on linear response theory using the Casida equation [Ref: T. A. Niehaus, S. Suhai, F. Della Sala, P Lugli, M. Elstner, G. Seifert, and Th. Frauenheim. Tight-binding approach to time-dependent density-functional response theory. Phys. Rev. B, 63:085108, 2001] and the ARPACK diagonalizer [R. B. Lehoucq, D. C. Sorensen, and C. Yang. Arpack users guide: Solution of large-scale eigenvalue problems by implicitly restarted arnoldi methods, 1997. 46, 51]. The dataset contains 1001 tar.gz files. Tar files are named as “ornl_aisd_ex_1.tar.gz†through “ornl_aisd_ex_1000.tar.gzâ€. Additionally, the 13 failed molecules are in “ornl_aisd_ex_unprocessed.tar.gzâ€. Except for the tar files listed below, each tar file contains 10,500 molecules. Tar files numbered 34, 121, 128, 352, 360, 429, 495, 509, 518, 627, 676, 668, and 862 contain 10,499 molecules each. The last tar file numbered 1000 contains 13,417 molecules. The total size of the uncompressed dataset is over 283 Gigabytes. The code for calculating the electronic excitation energies and statistical analysis of the dataset is provided at the following GitLab repository: https://github.com/ORNL/Analysis-of-Large-Scale-Molecular-Datasets-with-Python Calculating the UV spectrum of a molecule requires performing 3 main operations: 1. Converting the smiles string representation of a molecule into a geometric structure where each atom is assigned XYZ coordinates. The geometric structure is written to the file smiles.pdb. 2. Using smiles.pdb to compute the relaxed geometry of the molecule, which corresponds with the position of the atoms at the position of equilibrium at the ground state. This generates the files band.out, detailed.out, and geo_end.gen. 3. Using geo_end.gen to calculate the UV spectrum of the molecule which is written into the file EXC.DAT. Every molecule in the dataset has its own directory. The files contained in each molecule directory are as follows: 1. geo_end.gen 2. detailed.out 3. band.out 4. EXC.DAT 5. smiles.pdb REFERENCES [1] Niehaus, T. A.; Suhai, S.; Della Salla, F.; Lugli, P.; Elstner, M.; Seifert, G.; Frauenheim, Th. Tight-binding approach to time-dependent density-functional response theory. Phys. Rev. B, 2001, 63, 085108/1-9. [2] Blanchard, A.; Gounley, J.; Metha, K.; Yoo, P.; Irle, S. AISD HOMO-LUMO. DOI: 10.13139/ORNLNCCS/1869409 [3] RDKit: Cheminformatics and Machine Learning Software. 2013, [http://www.rdkit.org] [4] Tosco, P.; Stiefl, N. and Landrum, G. Bringing the MMFF force field to the RDKit: implementation and validation. J Cheminform. 2014, 6, 1–4. [5] a) Porezag, D.; Frauenheim, T.; Kohler, T.; Seifert, G.; Kaschner, Construction of tight-binding-like potentials on the basis of density-functional theory: Application to carbon, R. Phys. Rev. B 1995, 51, 12947-12957; b) Elstner, M.; Porezag, D.; Jungnickel, G.; Elsner, J.; Haugk, M.; Frauenheim, Th.; Suhai, S.; Seifert, G.; Phys. Rev. B 1998, 58, 7260-7268; c) Gaus, M.; Cui, Q.; Elstner, M. DFTB3: Extension of the Self-Consistent-Charge Density-Functional Tight-Binding Method (SCC-DFTB), J. Chem. Theory Comput. 2011, 7, 931-948; d) Cui, Q.; Elstner, M. Density functional tight binding: values of semi-empirical methods in an ab initio era, Phys. Chem. Chem. Phys. 2014, 16, 14368-14377. [6] Hourahine, B. et al. DFTB+, a software package for efficient approximate density functional theory based atomistic simulations, J. Chem. Phys. 2020, 152, 124101/1-19. [7] Larsen, A. H. et al. The atomic simulation environment—a Python library for working with atoms. J. Phys.: Cond. Matter 2017, 29, 273002. [8] Kubillus, M.; Kubar, T.; Gaus, M.; Rezac, J.; Elstner, M. Parameterization of the DFTB3 Method for Br, Ca, Cl, F, I, K, and Na in Organic and Biological Systems, J. Chem. Theory Comput. 2015, 11, 332-342. [9] Brandenburg, J. G.; Grimme, S. Accurate Modeling of Organic Molecular Crystals by Dispersion-Corrected Density Functional Tight Binding (DFTB), J. Phys. Chem. Lett. 2014, 5, 1785−1789. [10] a) Niehaus, T. A.; Elstner, M.; Frauenheim, Th.; Suhai, S. Application of an approximate density-functional method to sulfur containing compounds. J. Mol. Struct.: THEOCHEM 2001, 541, 185-94; b) Elstner, M.; Hobza, P.; Frauenheim, Th.; Suhai, S.; Kaxiras, E. Hydrogen bonding and stacking interactions of nucleic acid base pairs: A density-functional-theory based treatment. J. Chem. Phys. 2001, 114, 5149-55. [11] Kubar, T.; Bodrog, Z.; Gaus, M.; Köhler, C.; Aradi, B.; Frauenheim, Th.; Elstner, M. Parametrization of the SCC-DFTB Method for Halogens. J. Chem. Theory Comput. 2013, 9, 2939-49.

36 MATERIALS SCIENCE↗

HydraGNN_Predictive_GFM_2024 - Ensemble of predictive graph foundation models for ground state atomistic materials modeling

We provide the ensemble of fifteen pre-trained graph foundation models (GFMs) for atomistic materials modeling applications. Each one of the fifteen GFMs has been trained on five open-source datasets that (once aggregated) amount to over 154 million atomistic structures, which cover over two-thirds of the natural elements of the periodic table and that comprises a broad set of organic and inorganic compounds. This vast set of atomistic structures comprises ground state configurations that are dynamically stable (i.e., equilibrated structures with atomic forces approximately close to zero values) as well as dynamically unstable structures (i.e., non-equilibrium structures with non-negligible non-zero values of atomic forces). The ensemble of datasets aggregated does NOT include excited states. The datasets have been curated to remove atomistic structures with spectral norm of the force tensor above 100 eV/angstrom. Moreover, a linear term of the energy was computed for each dataset using a linear regression model that uses the chemical concentration of each natural element as regressor. The linear term predicted by the linear regression model has been subtracted from each original energy value to perform a re-alignment of the energy values across different electronic structures approximation theories performed to generate the diverse multi-source, multi-fidelity datasets. The folder "ADIOS_files" contains the set of pre-processed datasets in Adaptable I/O System (ADIOS) format (https://www.exascaleproject.org/research-project/adios/) that have been used for the development and training of GFMs in this work. The "ADIOS_files" directory contains 6 sub-directories named as follows: - ANI1x-v3.bp - MPTrj-v3.bp - OC2020-20M-v3.bp - OC2020-v3.bp - OC2022-v3.bp - qm7x-v3.bp Each sub-directory contains the pre-processed datasets converted in Adaptable I/O System (ADIOS) format (https://www.exascaleproject.org/research-project/adios/) that have been used to the development, training, and performance testing of the ensemble go predictive graph foundation models. Each GFM was developed using HydraGNN (https://github.com/ORNL/HydraGNN) as underlying graph neural network (GNN) architecture. The multi-task learning (MTL) capability of HydraGNN was used to simultaneously train the GFMs on labeled values for direct predictions of energy (a total system property of an atomistic structure that measures the chemical stability) and atomic forces (an atomic level property of an atomistic structure that measures the dynamical stability). The hyper parameters of the GFM have been tuned using scalable hyperparameter optimization (HPO) algorithms implemented in the software DeepHyper (https://github.com/deephyper/deephyper). The pre-training of each HPO trial was performed using distributed data parallelism (DDP) to scale the training across 128 compute nodes of the exascale OLCF supercomputer Frontier. Each HPO trial was trained only for 10 epochs and an early stopping was performed to avoid wasting significant computational resources on GNN architectures that were clearly underperforming. For each HPO trial, the 'omnistat' tool developed by (AMD Research - Advanced Micro Device) was used to measure the total energy consumption in kWh. The ensemble of GFMs was obtained by selecting the fifteen best performing HPO trials. Four models have been selected for their clear advantage in accuracy, and these are the GFMs with IDs 229, 156, 147, 260. Additional eleven models have been selected based on judicious balance between accuracy and energy consumption needed for training, and these are the GFMs with IDs 165, 78, 137, 1, 175, 171, 181, 67, 179, 167, 351. Each selected GFM of the ensemble was continued to cumulate a total of at most 30 epochs. In some cases, the total number of epochs actually performed was les than 30 due to two combined factors: (1) the size of the GFM (i.e., the number of model parameters to train) and (2) the total wall-clock time for which the computational resources could be allocated on OLCF-Frontier. The "Ensemble_of_models" directory contains 15 sub-directories named as follows: - gfm_0.229 - gfm_0.156 - gfm_0.147 - gfm_0.260 - gfm_0.165 - gfm_0.78 - gfm_0.137 - gfm_0.1 - gfm_0.175 - gfm_0.171 - gfm_0.181 - gfm_0.67 - gfm_0.179 - gfm_0.167 - gfm_0.351 Each one of these sub-directories refers to one of the fifteen HPO trials that have been selected to continue the pre-training with at most 30 epochs. With each sub-directory associated with a specific HPO trial, the following files can be found: - config.json: file for argument parsing to develop and train an HydraGNN architecture - gfm_0.ID_epoch_N.pk: file with model parameters for HPO ID trial after N epochs of training The ensemble of fifteen GFM architectures was used for (1) ensemble averaging to stabilize the predictions of energy and atomic forces after pre-training for post-processing analysis and (2) ensemble uncertainty quantification (UQ). The code used to develop, pre-train, and load the pre-trained models for post-processing analysis is available on the ORNL-GitHub at the following link: https://github.com/ORNL/HydraGNN/tree/Predictive_GFM_2024

36 MATERIALS SCIENCE↗

Medium- and Heavy-Duty Truck Duty Cycles

This dataset provides second-by-second duty cycle data for Class 6 and Class 8 diesel trucks in Texas, including key vehicle metrics, engine-related data, and GPS data (excluding GPS latitude and longitude to ensure confidentiality). The data were collected via tablets installed on the trucks and organized into daily datasets, each associated with a unique vehicle ID and date. There are 12 daily datasets for Class 6 diesel trucks (three unique vehicle IDs) and 43 daily datasets for Class 8 diesel trucks (six unique vehicle IDs). The units associated with each column are included in the name. The engine performance data include columns such as engine speed, engine percent torque, and engine fuel rate. Road grade (%/100) was estimated using the GPS altitude and wheel-based vehicle speed, which is used as an input for FASTSim. Cumulative distance was also calculated using the wheel-based vehicle speed. Additional columns include: - Engine Speed (RPM): Removed inaccurate readings and used to calculate angular velocity (radians/second). - Torque (N·m): Calculated using engine percent torque, nominal friction percent torque, and engine reference torque values (those columns were removed from dataset), then normalized to express as torque (%). - Flywheel Power (%): Calculated using the angular velocity and torque (in kW), then normalized as a percentage of the maximum value. - Engine Fuel Rate (%) and Torque (%): Both metrics were normalized by dividing by their respective maximum values within each dataset to express them as percentages. The datasets were analyzed to assess the energy impact of various driving behaviors, simulate energy efficiency, and recommend optimal routes for diesel trucks using NLR’s tool called RouteE. For driver coaching, factors like speed and acceleration limits were considered, and idle periods were reduced (assuming the engine was off during idling) to adjust each drive cycle. These adjusted drive cycles were then simulated in FASTSim to evaluate their effect on fleet energy consumption and estimate potential energy savings. The original cycles are available for download on this page ![image](CoVaR_Image_for_Data_Page_Kenworth_Truck.jpg)

1Hz↗

Continuous snow depth and temperature measurements from dense network of above-ground distributed temperature profiling systems from 2021-09-23 to 2024-08-23, Seward Peninsula, Alaska

The dataset contains temperature measurements from distributed temperature profiling (DTP) systems (Dafflon et al., 2022; Wielandt et al., 2022; Wang et al., 2024a; Fiolleau et al., 2024) deployed vertically above the ground surface at a large number of locations from 2021 to 2024. The research is designed to improve understanding of the local heterogeneity in snow depth and snow thermal insulation dynamics, as well as their interactions in a discontinuous permafrost region (Wang et al., 2025). The DTP systems were deployed at 96 locations in a watershed along the Nome-Teller road at mile marker 27 (T27) and at 54 locations on a hillslope along the Kougarok road at mile marker 64 (K64) in the Seward Peninsula, Alaska. The probe location information is stored in Probe_locations_T27.csv and Probe_locations_K64.csv. Temperature measurements were recorded at 15-minute intervals using high-precision digital sensors (accuracy: ±0.1°C, resolution: 0.0078°C). The temperature probes, either 1.4 m or 1.6 m long, contain sensors spaced every 5 cm or 10 cm along their length. The temperature data are stored in compressed files following the format: DTP_snow_air_temperature_(site)_(start)_(end).zip, where site is either T27 or K64, and start and end represent the time series period. Within each ZIP file, individual CSV files are named by probe ID and contain temperature records at different heights above the ground surface.This dataset also includes derived snow depth time series over three snow seasons, estimated from temperature measurements. Snow depth was estimated by identifying the consecutive sensor pair that exhibited the largest drop in high-frequency temperature fluctuations (detailed in the methods). These data are stored in: Snow_depths_flags_(site)_(start)_(end).csv, which includes snow depth time series and corresponding quality flags (defined in the methods) from different probes. Additionally, the dataset includes derived metrics and supporting measurements at selected locations over two snow seasons, contributing to the manuscript of Wang et al., 2025. These locations were chosen based on the availability of high-quality snow depth time series during both seasons. The additional data include: (1) Air temperature proxies measured from the top sensors on the pole when they were not buried by snow, stored in Air_temperature_proxies_(site)_(start)_(end).csv (2) Ground interface temperature, recorded at 3 cm above the ground, stored in Ground_interface_temperature_(site)_(start)_(end).csv (3) Site characteristics, including vegetation height, elevation, and the topographic position index (TPI) within a 50 m radius, stored in Selected_probe_locations_gps_vegheight_tpi_elevation_(site).csv. These metrics were derived from 1 m resolution summer LiDAR-based digital elevation models and digital surface models from Singhania et al., 2023, DOI:10.5440/1832016. Metadata files include data descriptions (_dd.csv) for tabular data. All included files are listed and described in xxxx_flmd.csv.This dataset is an updated version of a previous archive (Wang et al., 2024b, DOI: 10.15485/2475020), incorporating multiple seasons and improved snow depth estimation. Please note that due to large amount of information present in this dataset, many specificities associated with the acquisition of snow temperature, air temperature proxy and estimation of snow depth, and the future archiving of additional datasets on the soil temperature, thaw depth and soil characteristics at these locations, the author would welcome being contacted by people planning to use this dataset.The Next-Generation Ecosystem Experiments: Arctic (NGEE Arctic), was a research effort to reduce uncertainty in Earth System Models by developing a predictive understanding of carbon-rich Arctic ecosystems and feedbacks to climate. NGEE Arctic was supported by the Department of Energy's Office of Biological and Environmental Research.The NGEE Arctic project had two field research sites: 1) located within the Arctic polygonal tundra coastal region on the Barrow Environmental Observatory (BEO) and the North Slope near Utqiagvik (Barrow), Alaska and 2) multiple areas on the discontinuous permafrost region of the Seward Peninsula north of Nome, Alaska.Through observations, experiments, and synthesis with existing datasets, NGEE Arctic provided an enhanced knowledge base for multi-scale modeling and contributed to improved process representation at global pan-Arctic scales within the Department of Energy's Earth system Model (the Energy Exascale Earth System Model, or E3SM), and specifically within the E3SM Land Model component (ELM).

54 ENVIRONMENTAL SCIENCES↗

Soil biogeochemical properties and metrics of tree-mycorrhizal dominance for a 25-Ha forest in South Central Indiana, USA.

This data package contains a dataset used in the papers “Seeing the forest for all the trees: Mycorrhizal-associated nutrient economies are modulated by stem density and the synchrony between overstory and understory communities” and “Mycorrhizal associations of tree species influence soil nitrogen dynamics via effects on soil acid–base chemistry”. Four csv files are included along with a dataset. The dataset features chemical soil properties for a single sampling campaign within the 25 Ha Lilly-Dickey Woods Smithsonian Forest Global Earth Observatory (ForestGEO) plot in South Central Indiana, USA (ldw_dat_raw.csv). Also included are separate files focused on pH (pH_data.csv), carbon and nitrogen (CN_data.csv), and nitrification rates (Nitrification_data.csv). These variables are commonly associated with the tree-mycorrhizal dominance of forest stands. In these data subsets, each soil variable was matched to a 10 meter radius neighborhood wherein metrics of tree-mycorrhizal dominance (basal area, stem count, importance value, etc.) were calculated. Models between these soil variables and dominance metrics were used to investigate how different assessments of mycorrhizal associated nutrient economies (MANE) capture these relationships. This research was performed as a part of the Smithsonian ForestGEO project. This data package can be used to explore spatial variability in soil chemistry within a mature hardwood forest, or it can be combined with the included tree data, other fine-scale spatial information, or other tree inventory data for the site to evaluate how soil chemistry varies with tree community composition or edaphic or topographic properties.

Craig, Matthew [ORNL] (ORCID:0000000288907920)↗

Water Observations of Flow/No-Flow for the East-Taylor Watershed, Colorado (June-July 2025 and 2026)

This dataset provides multi-year, ground-truth visual observations of surface water flow/no-flow conditions within the East-Taylor Watershed, Colorado, collected during June and July of 2025 and 2026. In June and July 2025, on-the-ground visual observations of flow/no-flow were collected as part of the Watershed Function Scientific Focus Area (SFA) and Rocky Mountain Biological Laboratory (RMBL) Colorado Headwaters Ecological Spectroscopy Study (CHESS) campaign (further details are provided within the CHESS Project Description). We obtained 377 water observations of flow/no-flow within the East-Taylor Watershed, Colorado. These ground-truth observations were collected to validate classification maps from remote sensing data and model results within the East-Taylor Watershed. In 2025, flow/no-flow measurements were collected using a field-based app for the CHESS Campaign (Zerion iForm). Within the field app, a water observation form was created to collect coordinates and metadata about the observation. Information collected for the water observation points included information about visually-assessed streamflow presence/absence (standard question obtained from Colorado State University’s StreamTracker project), flow estimate, stream or ponded area width, canopy cover, manganese films, iron seeps, and beaver activity. For 2025 water observations, this dataset contains: (1) a data file with the water observations and coordinates (2025_Water_Observations.csv); (2) a Keyhole Markup Language Zipped (KMZ) with the water observation locations and metadata (2025_Water_Observations_Locations.kmz); (3) photos (.jpg and .jpeg) of the water observation points, organized by location, contained within 2025_Water_Observations_FieldPhotographs.zip file; and (4) water observation protocols and figures (2025_Water_Observation_Protocols.pdf). In June and July 2026, on-the-ground visual observations of flow/no-flow were collected as part of the Watershed Function SFA project. We obtained 365 water observations of flow/no-flow within the East-Taylor Watershed, Colorado. The 2026 observations focused on collecting repeat measurements at the 2025 flow/no-flow observation locations conducted as part of the CHESS campaign. These ground-truth observations were collected to understand differences in flow/no-flow in 2026, given the unprecedented 2026 drought in Colorado. In 2026, flow/no-flow measurements were collected using ArcGIS (Geographic Information System) Survey123. Within the field app, a water observation form was created to collect coordinates and metadata about the observation. Information collected for the water observation points included repeat information from the 2025 water observation effort, including visually-assessed streamflow presence/absence (standard question obtained from Colorado State University’s StreamTracker project), flow estimate, stream or ponded area width, canopy cover, manganese films, iron seeps, beaver activity, and a new metadata component of estimated stream depth (for select locations). For 2026 water observations, this dataset contains: (1) a data file with the water observations and coordinates (2026_Water_Observations.csv); (2) a Keyhole Markup Language Zipped (KMZ) with the water observation locations and metadata (2026_Water_Observations_Locations.kmz); (3) photos (.jpg) of the water observation points, organized by location, contained within 2026_Water_Observations_FieldPhotographs.zip file; and (4) water observation protocols and figures (2026_Water_Observation_Protocols.pdf). For 2025 and 2026 water observations, this dataset contains: (1) a location metadata file (locations.csv); (6) a file-level metadata (flmd.csv) file that lists each file contained in the dataset with associated metadata; and (7) a data dictionary (dd.csv) file that contains column/row headers used throughout the files along with a definition, units, and data type. CHESS Project Description: The Colorado Headwaters Ecological Spectroscopy Study (CHESS) comprised a multi-week airborne remote sensing and field observation campaign in the Upper Gunnison Basin, Colorado, conducted in June and July of 2025. Airborne remote sensing was conducted by the National Ecological Observatory Network Airborne Observation Platform (NEON AOP), concurrent with a field campaign run by the Rocky Mountain Biological Laboratory (RMBL), the Lawrence Berkeley National Laboratory (LBNL) and SLAC National Accelerator Laboratory Watershed Function Science Focus Area (SFA), and NASA-JPL (Jet Propulsion Laboratory) Earth Surface Mineral Dust Source Investigation (EMIT) program. Between June 10 and July 18, 2025, the NEON AOP flight team collected high-resolution aerial imaging spectroscopy and Light Detection and Ranging (LiDAR) data over three domains: the Upper East River (CRBU), Almont Triangle (ALMO), and the Upper Taylor Basin (UPTA). In coordination with the flights, a field campaign acquired ground-truth observations, including observations of vegetation composition, foliar traits, forest demography, and subsurface properties in 18 core sampling areas within the domains. Additional surface water observations were taken at over 380 point locations. All CHESS campaign datasets can be found within the CHESS ESS-DIVE data portal: https://data.ess-dive.lbl.gov/portals/chess. This work was supported by the Watershed Function Science Focus Area at Lawrence Berkeley National Laboratory funded by the US Department of Energy, Office of Science, Biological and Environmental Research under Contract No. DE-AC02-05CH11231. 2026-09-02: This dataset was updated to include 2026 water observation measurements. The 2025 observation files were also updated to ensure a consistent file naming convention across water observation years.

2018 NEON and 2025 CHESS Campaigns↗