Search NASA⌕ Search

SEARCH · Search NASA

Results for “Scientific data interpolation”

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 37 records · Page 2

RACORO aerosol data processing

The RACORO aerosol data (cloud condensation nuclei (CCN), condensation nuclei (CN) and aerosol size distributions) need further processing to be useful for model evaluation (e.g., GCM droplet nucleation parameterizations) and other investigations. These tasks include: (1) Identification and flagging of 'splash' contaminated Twin Otter aerosol data. (2) Calculation of actual supersaturation (SS) values in the two CCN columns flown on the Twin Otter. (3) Interpolation of CCN spectra from SGP and Twin Otter to 0.2% SS. (4) Process data for spatial variability studies. (5) Provide calculated light scattering from measured aerosol size distributions. Below we first briefly describe the measurements and then describe the results of several data processing tasks that which have been completed, paving the way for the scientific analyses for which the campaign was designed. The end result of this research will be several aerosol data sets which can be used to achieve some of the goals of the RACORO mission including the enhanced understanding of cloud-aerosol interactions and improved cloud simulations in climate models.

54 ENVIRONMENTAL SCIENCES↗

Feature-Identification and Data-Compression Software

A report discusses the continuing development of Windows Interface for Nominal Displacement Selection (WINDS), a computer program for automated analysis of images of the Sun and planets acquired by scientific instruments aboard spacecraft. WINDS is intended to afford capabilities for identification of features, measurement of displacements and velocities, analysis of terrain and of atmospheres, and synthesis of animation sequences of images of terrains and atmospheres from small sets of samples by use of velocity-based interpolation. A major element of WINDS will be a nonlinear correlator capable of tracking small features in complex image sequences. For dynamic image sequences, the correlator will enable compression of data by factors >100. In processing image data, WINDS will take account of such factors as texture in image data, rotation of features during measurement intervals, effects of viewing and solar illumination angles, and vertical structures of atmospheres. WINDS will also take account of positions, aiming directions, and fields of view of cameras to determine three-dimensional feature structures by use of triangulation and stereoscopic analysis techniques.

Jong, Eric De↗

Processing TES Level-2 Data

TES Level 2 Subsystem is a set of computer programs that performs functions complementary to those of the program summarized in the immediately preceding article. TES Level-2 data pertain to retrieved species (or temperature) profiles, and errors thereof. Geolocation, quality, and other data (e.g., surface characteristics for nadir observations) are also included. The subsystem processes gridded meteorological information and extracts parameters that can be interpolated to the appropriate latitude, longitude, and pressure level based on the date and time. Radiances are simulated using the aforementioned meteorological information for initial guesses, and spectroscopic-parameter tables are generated. At each step of the retrieval, a nonlinear-least-squares- solving routine is run over multiple iterations, retrieving a subset of atmospheric constituents, and error analysis is performed. Scientific TES Level-2 data products are written in a format known as Hierarchical Data Format Earth Observing System 5 (HDF-EOS 5) for public distribution.

Poosti, Sassaneh↗

Advances in Kriging-Based Autonomous X-Ray Scattering Experiments

Autonomous experimentation is an emerging paradigm for scientific discovery, wherein measurement instruments are augmented with decision-making algorithms, allowing them to autonomously explore parameter spaces of interest. We have recently demonstrated a generalized approach to autonomous experimental control, based on generating a surrogate model to interpolate experimental data, and a corresponding uncertainty model, which are computed using a Gaussian process regression known as ordinary Kriging (OK). We demonstrated the successful application of this method to exploring materials science problems using x-ray scattering measurements at a synchrotron beamline. Here, we report several improvements to this methodology that overcome limitations of traditional Kriging methods. The variogram underlying OK is global and thus insensitive to local data variation. We augment the Kriging variance with model-based measures, for instance providing local sensitivity by including the gradient of the surrogate model. As with most statistical regression methods, OK minimizes the number of measurements required to achieve a particular model quality. However, in practice this may not be the most stringent experimental constraint; e.g. the goal may instead be to minimize experiment duration or material usage. We define an adaptive cost function, allowing the autonomous method to balance information gain against measured experimental cost. We provide synthetic and experimental demonstrations, validating that this improved algorithm yields more efficient autonomous data collection.

36 MATERIALS SCIENCE↗

Developing Information Power Grid Based Algorithms and Software

This was an exploratory study to enhance our understanding of problems involved in developing large scale applications in a heterogeneous distributed environment. It is likely that the large scale applications of the future will be built by coupling specialized computational modules together. For example, efforts now exist to couple ocean and atmospheric prediction codes to simulate a more complete climate system. These two applications differ in many respects. They have different grids, the data is in different unit systems and the algorithms for inte,-rating in time are different. In addition the code for each application is likely to have been developed on different architectures and tend to have poor performance when run on an architecture for which the code was not designed, if it runs at all. Architectural differences may also induce differences in data representation which effect precision and convergence criteria as well as data transfer issues. In order to couple such dissimilar codes some form of translation must be present. This translation should be able to handle interpolation from one grid to another as well as construction of the correct data field in the correct units from available data. Even if a code is to be developed from scratch, a modular approach will likely be followed in that standard scientific packages will be used to do the more mundane tasks such as linear algebra or Fourier transform operations. This approach allows the developers to concentrate on their science rather than becoming experts in linear algebra or signal processing. Problems associated with this development approach include difficulties associated with data extraction and translation from one module to another, module performance on different nodal architectures, and others. In addition to these data and software issues there exists operational issues such as platform stability and resource management.

Dongarra, Jack↗

The Application of Jason-1 Measurements to Estimate the Global Near Surface Ocean Circulation for Climate Research

The scientific objective of this research program was to utilize drifter, Jason-1 altimeter data and a variety of wind data for the determination of time mean and time variable wind driven surface currents of the global ocean. To accomplish this task has required the interpolation of 6-hourly winds on drifter tracks and the computation of the wind coherent motions of the drifters. These calculations showed that the Ekman current model proposed by Ralph and Niiler for the tropical Pacific was valid for all the oceans south of 40N latitude. Improvements to RN99 model were computed and poster presentations of the results were given in several ocean science venues, including the November 2004 GODAY meeting in St. Petersburg, FL.

Niiler, Peran P.↗

Spline Trajectory Algorithm Development: Bezier Curve Control Point Generation for UAVs

A greater need for sophisticated autonomous piloting systems has risen in direct correlation with the ubiquity of Unmanned Aerial Vehicle (UAV) technology. Whether surveying unknown or unexplored areas of the world, collecting scientific data from regions in which humans are typically incapable of entering, locating lost or wanted persons, or delivering emergency supplies, an unmanned vehicle moving in close proximity to people and other vehicles, should fly smoothly and predictably. The mathematical application of spline interpolation can play an important role in autopilots' on-board trajectory planning. Spline interpolation allows for the connection of Three-Dimensional Euclidean Space coordinates through a continuous set of smooth curves. This paper explores the motivation, application, and methodology used to compute the spline control points, which shape the curves in such a way that the autopilot trajectory is able to meet vehicle-dynamics limitations. The spline algorithms developed used to generate these curves supply autopilots with the information necessary to compute vehicle paths through a set of coordinate waypoints.

Howell, Lauren R.↗

A kinetic-based regularization method for data science applications

We propose a physics-based regularization technique for function learning, inspired by statistical mechanics. By drawing an analogy between optimizing the parameters of an interpolator and minimizing the energy of a system, we introduce corrections that impose constraints on the lower-order moments of the data distribution. This minimizes the discrepancy between the discrete and continuum representations of the data, in turn allowing to access more favorable energy landscapes, thus improving the accuracy of the interpolator. Our approach improves performance in both interpolation and regression tasks, even in high-dimensional spaces. Unlike traditional methods, it does not require empirical parameter tuning, making it particularly effective for handling noisy data. We also show that thanks to its local nature, the method offers computational and memory efficiency advantages over Radial Basis Function interpolators, especially for large datasets.

97 MATHEMATICS AND COMPUTING↗

Validating a large geophysical data set: Experiences with satellite-derived cloud parameters

We are validating the global cloud parameters derived from the satellite-borne HIRS2 and MSU atmospheric sounding instrument measurements, and are using the analysis of these data as one prototype for studying large geophysical data sets in general. The HIRS2/MSU data set contains a total of 40 physical parameters, filling 25 MB/day; raw HIRS2/MSU data are available for a period exceeding 10 years. Validation involves developing a quantitative sense for the physical meaning of the derived parameters over the range of environmental conditions sampled. This is accomplished by comparing the spatial and temporal distributions of the derived quantities with similar measurements made using other techniques, and with model results. The data handling needed for this work is possible only with the help of a suite of interactive graphical and numerical analysis tools. Level 3 (gridded) data is the common form in which large data sets of this type are distributed for scientific analysis. We find that Level 3 data is inadequate for the data comparisons required for validation. Level 2 data (individual measurements in geophysical units) is needed. A sampling problem arises when individual measurements, which are not uniformly distributed in space or time, are used for the comparisons. Standard 'interpolation' methods involve fitting the measurements for each data set to surfaces, which are then compared. We are experimenting with formal criteria for selecting geographical regions, based upon the spatial frequency and variability of measurements, that allow us to quantify the uncertainty due to sampling. As part of this project, we are also dealing with ways to keep track of constraints placed on the output by assumptions made in the computer code. The need to work with Level 2 data introduces a number of other data handling issues, such as accessing data files across machine types, meeting large data storage requirements, accessing other validated data sets, processing speed and throughput for interactive graphical work, and problems relating to graphical interfaces.

Kahn, Ralph↗

MPEX AI Digital Twins

All magnetically confined plasma fusion power plant concepts (Tokamak, Spherical Tokamak, Stellarator, Mirror, ...) must exhaust the heat and plasma from the core confinement region to the material walls. The primary channel for this exhaust is through a plasma divertor which directs plasma along open magnetic field lines to a material target. The Material Plasma Exposure eXperiment (MPEX) illustrated in Figure 1, is a high-power, steady-state linear plasma device designed to produce the plasma material interaction (PMI) conditions of the divertor of future magnetic confinement fusion power plants: energy flux 20MW/m 2 , ion fluence 1031/m 2 , pulse duration 106 sec. These goals of plasma exposure in MPEX are well beyond those achieved in magnetic fusion experimental devices. Successfully achieving these high power steady state conditions for long pulses requires operational control of the heating and particle sources and the plasma flux to the walls and target. The MPEX AI Hot Spot Controller, proposed in this project, will help achieve the operational milestones of MPEX. The MPEX device will begin commissioning at the end of FY26. A smaller proto-MPEX was operated for 14,666 plasma discharges and will resume operation in September of 2025 as proto-MPEX-lite, with reduced capability, to test a new window for the Helicon plasma source. The proto-MPEX data has undergone surrogate modeling with machine learning methods (R. Archibald, 2022 IEEE International Conference on Big Data). This proto-MPEX data will be used to begin development of the AI digital twins described in this white paper. The scientific mission of MPEX is to qualify materials of different composition for use in the high energy and plasma flux conditions of a fusion power plant. The materials exposed in MPEX will in some cases be exposed to high neutron fluxes at other ORNL facilities to measure the changes to their PMI properties. The targets exposed in MPEX will be transported under vacuum to a Surface Analysis Station (SAS). The SAS will be equipped with the following diagnostics: Focused Ion Beam (FIB) for trench milling, 100-400 angstrom resolution scanning electron microscope (SEM), surface mapping x-ray spectrometer, high resolution camera, and a future upgrade to a laser induced breakdown spectroscopy quadruple mass spectrometer (LIBS-QMS). The MPEX experiments will generate diverse pre- and post-exposure measurement data of detailed material properties down to the crystal grain level in 3D for post-exposure assessment of PMI damage (e.g. cracking, melting, erosion and redeposition of the material). Physics models for the PMI, and how the material composition and manufacturing impact its performance under high energy plasma exposure, need to be validated with MPEX data to guide the selection of new candidate materials. Our vision for the MPEX AI Digital Twins project is to supply experimental and physics model simulation data to train Artificial Intelligence (AI) models for data processing, analysis, operational control, PMI and materials simulation to maximize the scientific output of the MPEX device. Ultimately, an AI digital twin of MPEX material assessment metrics for tested and synthetic material types with simulated PMI will be trained by the AI Modeling Teams on the experimental and physics simulation data submitted to the American Science Cloud by this project. A purely empirical search for the best material is inefficient given the finite number of samples that can be tested on MPEX. In order to expand the material properties database for training the MPEX Material Assessment AI Digital Twin, and to gain physics understanding of the PMI processes, physics models of the material properties and PMI processes are required. The physics simulations provide detailed simulation data, like impact angles for plasma ions, sputtering yields, transport of the ionized sputtered target material in the plasma, and redeposition locations. This simulation data expands the measurement data for deeper physics understanding. The experimental data is essential to validate the PMI and material structure simulation models. The validated models can then be used to generate new simulation data of MPEX material assessments for synthetic material compositions that have not been exposed in MPEX. These predictive simulations, plus the whole experimental dataset, will be used to train the MPEX Material Assessment AI Digital Twin allowing a rapid generative AI search for new materials with reduced PMI damage by interpolating the domain of the training set. These new optimum materials can be simulated with the physics codes and/or tested in MPEX. The ability of AI neural networks to interpolate multi-dimensional parameter spaces and generate virtual data is exploited for a more efficient search for optimum materials. The advent of the Transformational AI Models Consortium (TAIMC) is an opportunity to engage with state of the art private and public AI developers to achieve the goals of the AI digital twins and AI accelerated physics models proposed in this project. Our partners at ORNL from the Advance Scientific Computing Research (ASCR) organization will collaborate in accelerating the integrated plasma material interaction simulation framework. This simulation framework will provide a platform for generating simulation data across a range of physical fidelities, including hybrid methods that produce multi-fidelity results. This data will be leveraged for AI model development, both for generation of surrogates and the automation of simulation campaigns. A part of the research below will include collaborative efforts with the TAIMC to (i) adapt data storage approaches to ensure AI-readiness, (ii) provide a protypical exemplar to inform and exercise constructed workflows, and (iii) generate and share data, using the TAIMC unified AI data standard, for foundational models that will be trained from multiple sources across the DOE complex. We will also collaborate with the TAIMC, as well as the planned AI modeling teams, to develop approaches for reducing the cost of data generation. These include tailored multi-fidelity approaches as well as fine-tuning strategies to augment general, large-scale foundational models.

70 PLASMA PHYSICS AND FUSION TECHNOLOGY↗

Physically Motivated Deep Learning to Superresolve and Cross Calibrate Solar Magnetograms

Abstract Superresolution (SR) aims to increase the resolution of images by recovering detail. Compared to standard interpolation, deep learning-based approaches learn features and their relationships to leverage prior knowledge of what low-resolution patterns look like in higher resolution. Deep neural networks can also perform image cross-calibration by learning the systematic properties of the target images. While SR for natural images aims to create perceptually convincing results, SR of scientific data requires careful quantitative evaluation. In this work, we demonstrate that deep learning can increase the resolution and calibrate solar imagers belonging to different instrumental generations. We convert solar magnetic field images taken by the Michelson Doppler Imager (resolution ∼2″ pixel −1 ; space based) and the Global Oscillation Network Group (resolution ∼2.″5 pixel −1 ; ground based) to the characteristics of the Helioseismic and Magnetic Imager (resolution ∼0.″5 pixel −1 ; space based). We also establish a set of performance measurements to benchmark deep-learning-based SR and calibration for scientific applications.

Muñoz-Jaramillo, Andrés (ORCID:0000000247160840)↗

VTK-m User's Guide (V.1.8)

Extreme scale systems require a new programming model and a fundamental change in how we design algorithms. To address these issues we created VTK-m: the visualization toolkit for multi-/many-core architectures. VTK-m supports a number of algorithms and the ability to design further algorithms through a top-down design with an emphasis on extreme parallelism. VTK-m also provides support for finding and building links across topologies, making it possible to perform operations that determine manifold surfaces, interpolate generated values, and find adjacencies. Although VTK-m provides a simplified high-level interface for programming, its template-based code removes the overhead of abstraction. VTK-m simplifies the development of parallel scientific visualization algorithms by providing a framework of supporting functionality that allows developers to focus on visualization operations. Additionally, VTK-m provides data abstractions not provided by other libraries that make code written in VTK-m more versatile.

97 MATHEMATICS AND COMPUTING↗

NASA Soil Moisture Active Passive Mission Status and Science Highlights

The Soil Moisture Active Passive (SMAP) observatory was launched January 31, 2015, and its L-band radiometer and radar instruments became operational during April 2015. This paper provides a summary of the quality assessment of its baseline soil moisture and freeze/thaw products as well as an overview of new products. The first new product explores the Backus Gilbert optimum interpolation based on the oversampling characteristics of the SMAP radiometer. The second one investigates the disaggregation of the SMAP radiometer data using the European Space Agency's Sentinel-1 C-band synthetic aperture radar (SAR) data to obtain soil moisture products at about 1 to 3 km resolution. In addition, SMAPs L-band data have been found useful for many scientific applications, including depictions of water cycles, vegetation opacity, ocean surface salinity and hurricane ocean surface wind mapping. Highlights of these new applications will be provided.The SMAP soil moisture, freeze/taw state and SSSprovide a synergistic view of water cycle. For example, Fig.7 illustrates the transition of freeze/thaw state, change of soilmoisture near the pole and SSS in the Arctic Ocean fromApril to October in 2015 and 2016. In April, most parts ofAlaska, Canada, and Siberia remained frozen. Melt onsetstarted in May. Alaska, Canada, and a big part of Siberia havebecome thawed at the end of May; some freshwater dischargecould be found near the mouth of Mackenzie in 2016, but notin 2015. The soil moisture appeared to be higher in the Oband Yenisei river basins in Siberia in 2015. As a result,freshwater discharge was more widespread in the Kara Seanear the mouths of both rivers in June 2015 than in 2016. TheNorth America and Siberia have become completely thawedin July. After June, the freshwater discharge from other riversinto the Arctic, indicated by blue, also became visible. Thefreeze-up started in September and the high latitude regionsin North America and Eurasia became frozen. Comparing thespread of freshwater in August 2015 and 2016 suggests thatthere was more discharge from Ob and Yenisei in 2015,which appeared to correspond to a higher soil moisturecontent in the Ob and Yenisei basins. In contrast, Mackenzieappeared to have more discharge in September 2016.

soil moisture↗

Experimenting with the GMAO 4D Data Assimilation

The Global Modeling and Assimilation Office (GMAO) has been working to promote its prototype four-dimensional variational (4DVAR) system to a version that can be exercised at operationally desirable configurations. Beyond a general circulation model (GeM) and an analysis system, traditional 4DV AR requires availability of tangent linear (TL) and adjoint (AD) models of the corresponding GeM. The GMAO prototype 4DVAR uses the finite-volume-based GEOS GeM and the Grid-point Statistical Interpolation (GSI) system for the first two, and TL and AD models derived ITom an early version of the finite-volume hydrodynamics that is scientifically equivalent to the present GEOS nonlinear GeM but computationally rather outdated. Specifically, the TL and AD models hydrodynamics uses a simple (I-dimensional) latitudinal MPI domain decomposition, which has consequent low scalability and prevents the prototype 4DV AR ITom being used in realistic applications. In the near future, GMAO will be upgrading its operational GEOS GCM (and assimilation system) to use a cubed-sphere-based hydrodynamics. This versions of the dynamics scales to thousands of processes and has led to a decision to re-derive the TL and AD models for this more modern dynamics, thus taking advantage of a two-dimensional MPI decomposition and improved scalability properties. With the aid of the Transformation of Algorithms in FORTRAN (l'AF) automatic adjoint generation tool and some hand-coding, a version of the cubed-sphere-based TL and AD models, with a simplified vertical diffusion scheme, is now available, enabling multiple configurations of standard implementations of 4DV AR in GEOS. Concurrent to this development, collaboration with the National Centers for Environmental Prediction (NCEP) and the Earth System Research Laboratory (ESRL) has allowed GMAO to implement a hybrid-ensemble capability within the GEOS data assimilation system. Both 3Dand 4D-ensemble capabilities are presently available thus allowing GMAO to now evaluate the performance and benefit of various ensemble and variational assimilation strategies. This presentation will cover the most recent developments taking place at GMAO and show results from various comparisons from traditional techniques to more recent ensemble-based ones.

Todling, R.↗

Multidimensional Distributional Neural Network Output Demonstrated in Super‐Resolution of Surface Wind Speed

Accurate quantification of uncertainty in neural network predictions remains a central challenge for scientific applications involving high-dimensional, correlated data. While existing methods capture either aleatoric or epistemic uncertainty, few offer closed-form, multidimensional distributions that preserve spatial correlation while remaining computationally tractable. In this work, we present a framework for training neural networks with a multidimensional Gaussian loss, generating a closed-form predictive distribution over outputs informed by non-identically distributed training data. Our approach captures aleatoric uncertainty by iteratively estimating the means and covariance matrices, and is demonstrated on a super-resolution example out-of-training-sample. We leverage a Fourier representation of the covariance matrix to stabilize network training and preserve spatial correlation. We introduce a novel regularization strategy—referred to as information sharing—that interpolates between image-specific and global covariance estimates, enabling convergence of the super-resolution downscaling network trained on image-specific distributional loss functions. This framework allows for efficient sampling, explicit correlation modeling, and extensions to more complex distribution families all without disrupting prediction performance. We demonstrate the method on a surface wind speed downscaling task and discuss its broader applicability to uncertainty-aware prediction in scientific models.

17 WIND ENERGY↗

A Kriging-Based Approach to Autonomous Experimentation with Applications to X-Ray Scattering

Abstract Modern scientific instruments are acquiring data at ever-increasing rates, leading to an exponential increase in the size of data sets. Taking full advantage of these acquisition rates will require corresponding advancements in the speed and efficiency of data analytics and experimental control. A significant step forward would come from automatic decision-making methods that enable scientific instruments to autonomously explore scientific problems—that is, to intelligently explore parameter spaces without human intervention, selecting high-value measurements to perform based on the continually growing experimental data set. Here, we develop such an autonomous decision-making algorithm that is physics-agnostic, generalizable, and operates in an abstract multi-dimensional parameter space. Our approach relies on constructing a surrogate model that fits and interpolates the available experimental data, and is continuously refined as more data is gathered. The distribution and correlation of the data is used to generate a corresponding uncertainty across the surrogate model. By suggesting follow-up measurements in regions of greatest uncertainty, the algorithm maximally increases knowledge with each added measurement. This procedure is applied repeatedly, with the algorithm iteratively reducing model error and thus efficiently sampling the parameter space with each new measurement that it requests. We validate the method using synthetic data, demonstrating that it converges to faithful replica of test functions more rapidly than competing methods, and demonstrate the viability of the approach in an experimental context by using it to direct autonomous small-angle (SAXS) and grazing-incidence small-angle (GISAXS) x-ray scattering experiments.

36 MATERIALS SCIENCE↗

CERES Fast Longwave And SHortwave Radiative Flux (FLASHFlux): Research to Operation

The Clouds and Earth’s Radiant Energy System (CERES) project provides satellite-based observations of how Earth’s energy flows are varying in time and space and how clouds and aerosols are affecting the Earth’s radiation budget. Nominally, CERES data products require months of validation and calibration before releasing a climate quality data. The Fast Longwave and Shortwave Radiative Flux (FLASHFlux) data product was developed as to provide key radiative flux data products within a week of observations for scientific, applied science research and educational usage. FLASHFlux achieves this by using simplified calibration process and optimized CERES data fusion production system, an operation meteorology product from Global Modeling and Assimilation Office(GMAO), and surface parameterized radiative flux models. The CERES FLASHFlux provides two data products: 1) an hourly Level 2 Single Scanner Footprint (SSF) data separately for Aqua and Terra observations, and 2) a daily Level 3 Time Interpolated and Spatially Averaged (TISA) 1ox1o gridded data that combines Aqua and Terra observations.The current operational Version 3C is being transitioned to Version4A. We compare the FLASHFlux Version4A fluxes using the specialized calibration to Version3C and to the climate quality CERES Edition4A TOA and surface fluxes over time for Terra and Aqua to determine the uncertainties of the calibration process. In addition, we also compare the impact of the fluxes by using the near real-time GMAO GEOS FP-IT instead of GMAO G5-CERES meteorology data.We also compare both SSF overpass products (for Terra and Aqua) and TISA radiative flux products to a set of surface measurement sites that are globally distributed and assess agreement.Lastly, we introduce the first CERES FF SSF data products from the CERES instrument on board NOAA-20.

P Sawaengphokhai↗

Deep-freeze graph training for latent learning

Scientific and engineering advances are primarily driven by multi-tier conceptual constructs and conditional theoretical frameworks. The theories allow predictions of hypothetical system responses, given a set of approximate conditions (ranges of applicability) imposed on latent parameters that cannot be measured directly. Learning to estimate the latent variables (Latent Learning) helps to pinpoint the anticipated range-edge anomalies and improves the confidence in interpretation, interpolation and extrapolation of limited experimental data. Due to high dimensionality and extreme non-linearity of the materials science problems, very large datasets are typically required for conventional data-driven model development. The vital experimental data collection, particularly on microstructural phases, is very challenging, which makes it difficult to compile a high-quality database. Incorporation of the domain knowledge into the computational graph structure, initialization and optimization processes presents a viable mechanism for developing accurate models, with limited datasets. Furthermore, this study successfully utilized the approach to build the Deep Freeze Graph (DeepFreG) by mapping known causality relationships and by digitizing empirical domain knowledge for Latent Learning (LL), with specific applications in materials science.

36 MATERIALS SCIENCE↗