Engineering Papers⌕ Search

SEARCH · Engineering Papers

Results for “Spatiotemporal modeling”

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 19 records
All sourcesNASA NTRSDOE OSTI
DOE OSTI2024

Analyzing inference workloads for spatiotemporal modeling

Ensuring power grid resiliency, forecasting climate conditions, and optimization of transportation infrastructure are some of the many application areas where data is collected in both space and time. Spatiotemporal modeling is about modeling those patterns for forecasting future trends and carrying out critical decision-making by leveraging machine learning/deep learning. Once trained offline, field deployment of trained models for near real-time inference could be challenging because performance can vary significantly depending on the environment, available compute resources and tolerance to ambiguity in results. Users deploying spatiotemporal models for solving complex problems can benefit from analytical studies considering a plethora of system adaptations to understand the associated performance-quality trade-offs. To facilitate the co-design of next-generation hardware architectures for field deployment of trained models, it is critical to characterize the workloads of these deep learning (DL) applications during inference and assess their computational patterns at different levels of the execution stack. In this paper, we develop several variants of deep learning applications that use spatiotemporal data from dynamical systems. We study the associated computational patterns for inference workloads at different levels, considering relevant models (Long short-term Memory, Convolutional Neural Network and Spatio-Temporal Graph Convolution Network), DL frameworks (Tensorflow and PyTorch), precision (FP16, FP32, AMP, INT16 and INT8), inference runtime (ONNX and AI Template), post-training quantization (TensorRT) and platforms (Nvidia DGX A100 and Sambanova SN10 RDU). Overall, our findings indicate that although there is potential in mixed-precision models and post-training quantization for spatiotemporal modeling, extracting efficiency from contemporary GPU systems might be challenging. Instead, co-designing custom accelerators by leveraging optimized High Level Synthesis frameworks (such as SODA High-Level Synthesizer for customized FPGA/ASIC targets) can make workload-specific adjustments to enhance the efficiency.

97 MATHEMATICS AND COMPUTING↗
NASA NTRS2020

Advancing Methodologies for Applying Machine Learning and Evaluating Spatiotemporal Models of Fine Particulate Matter (PM 2.5 ) Using Satellite Data Over Large Regions

Reconstructing the distribution of fine particulate matter (PM 2.5 ) in space and time, even far from ground monitoring sites, is an important exposure science contribution to epidemiologic analyses of PM 2.5 health impacts. Flexible statistical methods for prediction have demonstrated the integration of satellite observations with other predictors, yet these algorithms are susceptible to overfitting the spatiotemporal structure of the training datasets. We present a new approach for predicting PM 2.5 using machine-learning methods and evaluating prediction models for the goal of making predictions where they were not previously available. We apply extreme gradient boosting (XGBoost) modeling to predict daily PM 2.5 on a 1 x 1 km 2 resolution for a 13 state region in the Northeastern USA for the years 2000–2015 using satellite-derived aerosol optical depth and implement a recursive feature selection to develop a parsimonious model. We demonstrate excellent predictions of withheld observations but also contrast an RMSE of 3.11 μg/m 3 in our spatial cross-validation withholding nearby sites versus an overfit RMSE of 2.10 μg/m 3 using a more conventional random ten-fold splitting of the dataset. As the field of exposure science moves forward with the use of advanced machine-learning approaches for spatiotemporal modeling of air pollutants, our results show the importance of addressing data leakage in training, overfitting to spatiotemporal structure, and the impact of the predominance of ground monitoring sites in dense urban sub-networks on model evaluation. The strengths of our resultant modeling approach for exposure in epidemiologic studies of PM 2.5 include improved efficiency, parsimony, and interpretability with robust validation while still accommodating complex spatiotemporal relationships.

air pollution↗
DOE OSTI2025

Intelligent Sampling of Extreme-Scale Turbulence Datasets for Accurate and Efficient Spatiotemporal Model Training

With the end of Moore’s law and Dennard scaling, efficient training increasingly requires rethinking data volume. Can we train better models with significantly less data via intelligent subsampling? To explore this, we develop SICKLE, a sparse intelligent curation framework for efficient learning, featuring a novel maximum entropy (MaxEnt) sampling approach, scalable training, and energy benchmarking. We compare MaxEnt with random and phase-space sampling on large direct numerical simulation (DNS) datasets of turbulence. Evaluating SICKLE at scale on Frontier, we show that subsampling as a preprocessing step can, in many cases, improve model accuracy and substantially lower energy consumption, with observed reductions of up to 38×.

Brewer, Wes [ORNL] (ORCID:0000000236393956)↗
DOE OSTI2025

Spatiotemporal Downscaling Model for Solar Irradiance Forecast Using Nearest-Neighbor Random Forest and Gaussian Process

Accurate solar photovoltaic (PV) capacity estimation requires high-resolution, site-specific solar irradiance data to account for localized variability. However, global datasets, such as the National Solar Radiation Database (NSRDB), provide regional averages that fail to capture the fine-scale fluctuations critical for large-scale grid integration. This limitation is particularly relevant in the context of increasing distributed energy resources (DERs) penetration, such as rooftop PV. Additionally, it is critical to the implementation of the U.S. Federal Energy Regulatory Commission (FERC) Order 2222, which facilitates DER participation in U.S. bulk power markets. To address this challenge, this study evaluates Nearest-Neighbor Random Forest (NNRF) and Nearest-Neighbor Gaussian Process (NNGP) models for spatiotemporal downscaling of global solar irradiance data. By leveraging historical irradiance and meteorological data, these models incorporate spatial, temporal, and feature-based correlations to enhance local irradiance predictions. The NNRF model, a machine-learning approach, prioritizes computational efficiency and predictive accuracy, while the NNGP model offers a level of interpretability and prediction uncertainty by numerically quantifying correlations and dependencies in the data. Model validation was conducted using day-ahead predictions. The results showed that the average Goodness of Fit (GoF) of the NNRF model of 90.61% across all eight sites outperformed the GoF of the NNGP of 85.88%. Additionally, the computational speed of NNRF was 2.5 times faster than the NNGP. Finally, the NNGP displayed polynomial scaling while the NNRF scaled linearly with increasing number of nearest neighbors. Additional validation of the model on five sites in Puerto Rico further confirmed the superiority of the NNRF model over the NNGP model. These findings highlight the robustness and computational efficiency of NNRF for large-scale solar irradiance downscaling, making it a strong candidate for improving PV capacity estimation and real-time electricity market integration for DERs.

Asiedu, Shadrack (ORCID:0009000646004826)↗
NASA NTRS2021

A Robust Hybrid Deep Learning Model for Spatiotemporal Image Fusion

Dense time-series remote sensing data with detailed spatial information are highly desired for the monitoring of dynamic earth systems. Due to the sensor tradeoff, most remote sensing systems cannot provide images with both high spatial and temporal resolutions. Spatiotemporal image fusion models provide a feasible solution to generate such a type of satellite imagery, yet existing fusion methods are limited in predicting rapid and/or transient phenological changes. Additionally, a systematic approach to assessing and understanding how varying levels of temporal phenological changes affect fusion results is lacking in spatiotemporal fusion research. The objective of this study is to develop an innovative hybrid deep learning model that can effectively and robustly fuse the satellite imagery of various spatial and temporal resolutions. The proposed model integrates two types of network models: super-resolution convolutional neural network (SRCNN) and long short-term memory (LSTM). SRCNN can enhance the coarse images by restoring degraded spatial details, while LSTM can learn and extract the temporal changing patterns from the time-series images. To systematically assess the effects of varying levels of phenological changes, we identify image phenological transition dates and design three temporal phenological change scenarios representing rapid, moderate, and minimal phenological changes. The hybrid deep learning model, alongside three benchmark fusion models, is assessed in different scenarios of phenological changes. Results indicate the hybrid deep learning model yields significantly better results when rapid or moderate phenological changes are present. It holds great potential in generating high-quality time-series datasets of both high spatial and temporal resolutions, which can further benefit terrestrial system dynamic studies. The innovative approach to understanding phenological changes’ effect will help us better comprehend the strengths and weaknesses of current and future fusion models.

spatiotemporal fusion↗
DOE OSTI2025

PaleoSTeHM v1.0: a modern, scalable spatiotemporal hierarchical modeling framework for paleo-environmental data

Abstract. Geological records of past environmental change provide crucial insights into long-term climate variability, trends, non-stationarity, and nonlinear feedback mechanisms. However, reconstructing spatiotemporal fields from these records is statistically challenging due to their sparse, indirect, and noisy nature. Here, we present PaleoSTeHM, a scalable and modern framework for spatiotemporal hierarchical modeling of paleo-environmental data. This framework enables the implementation of flexible statistical models that rigorously quantify spatial and temporal variability from geological data while clearly distinguishing measurement and inferential uncertainty from process variability. We illustrate its application by reconstructing temporal and spatiotemporal paleo-sea-level changes across multiple locations. Using various modeling and analysis choices, PaleoSTeHM demonstrates the impact of different methods on inference results and computational efficiency. Our results highlight the critical role of model selection in addressing specific paleo-environmental questions, showcasing the PaleoSTeHM framework's potential to enhance the robustness and transparency of paleo-environmental reconstructions.

58 GEOSCIENCES↗
DOE OSTI2026

MATEY: multiscale adaptive transformer models for spatiotemporal physical systems

Accurate representation of the multiscale features in spatiotemporal physical systems using vision transformer architectures requires extremely long, computationally prohibitive token sequences. To address this issue, we propose two novel adaptive tokenization schemes that dynamically adjust patch sizes based on local features: one ensures convergent behavior to uniform patch refinement, while the other offers better computational efficiency. Moreover, we present a set of spatiotemporal attention schemes, where the temporal or axial spatial dimensions are decoupled, to evaluate their baseline computational and data efficiencies and to determine whether adaptive tokenization can improve this performance. We assess the performance of the proposed multiscale adaptive model, MATEY, in a sequence of experiments. Compared to a full spatiotemporal attention scheme or a scheme that decouples only the temporal dimension, we find that fully decoupled axial attention is less efficient and expressive, requiring more training time and model parameters to achieve the same accuracy. The experiments on the adaptive tokenization schemes show that, compared to a uniformly refined model, the proposed schemes achieve comparable or improved accuracy at a much lower cost in the tested two-dimensional settings. While the asymptotic analysis suggests the potential for favorable scaling, empirical validation at substantially longer sequence lengths remains to be performed in future work. Finally, we demonstrate in two fine-tuning tasks featuring different physics that models pretrained on PDEBench data outperform the ones trained from scratch, especially in the low data regime with frozen attention.

adaptive tokenization↗
NASA NTRS1991

Modeling the spatiotemporal organization of velocity storage in the vestibuloocular reflex by optokinetic studies

1. A generalized three-dimensional state space model of visual vestibular interaction was developed. Matrix and dynamical system operators associated with inputs from the semicircular canals, otolith velocity estimator, and the visual system have been incorporated into the model, which focus on their relationship to the velocity storage integrator. 2. A relationship was postulated between the eigenvalues and the direction of the eigenvectors of the system matrix and the orientation of the spatial vertical. It was assumed that the system matrix for a tilted position was a composition of two linear transformations of the system matrix for the upright position. One transformation modifies the eigenvalues of the system matrix, whereas another rotates the eigenvectors of the system matrix. The pitch and roll eigenvectors rotate with the head, whereas the yaw axis eigenvector remains approximately spatially invariant. 3. Based on the three-dimensional model, a computational procedure was formulated to identify the eigenvalues and eigenvectors of the system matrix with the use of a modification of the marquardt algorithm. With the use of data obtained from a monkey, it was shown that the three-dimensional behavior of velocity storage cannot be predicted solely in terms of its time constants, i.e., the inverse of its eigenvalues. With the use of the same eigenvalues the data could either be fit or not fit, depending on the eigenvector directions. Therefore, it is necessary to specify eigenvector directions when characterizing velocity storage in three dimensions. 4. Parameters found with the use of the Marquardt algorithm were incorporated into the model. Diagonal matrices in a head coordinate frame were introduced for coupling the visual system to the integrator and to the direct optokinetic pathway. Simulations of optokinetic nystagmus (OKN) and optokinetic after-nystagmus (OKAN) were run. The model predicted the behavior of yaw and pitch OKN and OKAN when the animal is upright. It also predicted the cross-coupling in the side down position. The trajectories in velocity space were also accurately simulated. 5. One of the predictions of the model is that when the stimulus direction is along an eigenvector, the trajectory in velocity space is a straight line. Using the "spectral width" of the residuals from a straight line sequence during OKAN, we developed a methodology to estimate how close the OKAN decay was to an eigenvector trajectory. 6. Thus we have developed a model-based approach for studying and interpreting the response characteristics of velocity storage in three dimensions.(ABSTRACT TRUNCATED AT 400 WORDS).

Non-NASA Center↗
DOE OSTI2025

Machine learning and process-based modeling of spatiotemporal changes in active layer thickness across Alaska

Permafrost degradation poses a growing threat to infrastructure stability and ecosystem resilience in the rapidly warming Arctic. We investigated the spatiotemporal dynamics of active layer thickness (ALT) across Alaska by integrating field observations, environmental datasets, a physically based Stefan model, and machine learning (ML) techniques. Using weather projections from the Coupled Model Intercomparison Project Phase 6 under two Shared Socioeconomic Pathways (SSP 2-4.5 and SSP 5-8.5), we assessed ALT sensitivity to projected future weather conditions. The random forest (RF) model outperformed the Stefan approach in predicting ALT on the training dataset (R² = 0.84 vs. 0.53) but demonstrated lower generalizability on the test dataset (R² = 0.24 vs. 0.54). The root mean square error (RMSE) for the RF model for training and testing ranged from 14 to 22 cm, compared to 17 and 18 cm for the Stefan model. Variable importance analysis revealed that mean annual temperature and slope angle were the strongest predictors of ALT, accounting for 19% and 18% of the variance, respectively, followed by sediment transport index (14%) and stream power index (11%). Comparative analysis of baseline ALT predictions showed the Stefan model tended to project a thicker active layer (mean ± SD: 65 ± 16 cm), compared to the RF model (mean ± SD: 59 ± 8.8) cm). Both models indicated a latitudinal gradient in ALT, with shallower depths at higher latitudes. Projected ALT increases by 2100 were estimated at 3.3 ± 2.2 cm under SSP 2-4.5 and 5.9 ± 4.0 cm under SSP 5-8.5 for the ML model, whereas the Stefan model projected substantially larger increases of 13 ± 2.6 cm (SSP 2-4.5) and 28 ± 4.4 cm (SSP5-8.5). Spatial analysis showed the greatest ALT increases in northern Alaska, with relatively smaller changes in southern regions. These findings highlight the complex, multifactorial nature of ALT dynamics and the value of hybrid modeling approaches. As rising temperatures accelerate permafrost thaw, changes in ALT can disrupt ecosystems, damage infrastructures, and enhance the release of stored soil carbon, highlighting the urgent need for improved predictive capabilities to inform adaptation strategies in the Arctic.

Climate sciences↗
DOE OSTI2026

Estimating CO 2 fluxes through integrating spatial and temporal input layers via deep learning algorithms

Background Accurate estimation of net ecosystem exchange of CO 2 fluxes (Fc) is essential for understanding carbon cycle processes and assessing ecosystem carbon budgets. However, conventional modeling approaches often emphasize temporal dynamics while overlooking the pronounced spatial heterogeneity within the footprint of eddy covariance (EC) towers, potentially limiting predictive accuracy and interpretability of Fc estimates. To address this challenge, we developed a spatiotemporal model that integrates high-resolution footprint-weighted spatial information with sequential environmental drivers. Results The integrated model combines a deeper graph convolutional network to characterize fine-scale spatial variability within EC footprints and a gated recurrent unit network to capture temporal dependencies in biophysical conditions. Using multi-year flux tower observations, remote sensing vegetation indices and footprint modeling, we evaluate the proposed method across three land cover types. This spatiotemporal model consistently outperforms temporal-only and spatial-only baselines, achieving the highest overall accuracy (R 2 = 0.9569) and the lowest RMSE (1.8128 μmol m −2 s −1 ) and MAE (1.1939 μmol m −2 s −1 ). Performance gains are particularly evident in ecosystems with strong vegetation heterogeneity, where spatial structure substantially modulates Fc variability. Conclusions This study demonstrates the importance of joint modeling spatial heterogeneity and temporal dynamics for improving Fc estimation and provides a robust method for advancing footprint-based Fc estimates across diverse ecosystems, supporting refined assessments of terrestrial carbon fluxes, and enhancing scientific foundations for carbon studies.

CO2 flux estimate↗
NASA NTRS2021

Spatiotemporal Associations Between Social Vulnerability, Environmental Measurements, and COVID-19 in the Conterminous United States

This study summarizes the results from fitting a Bayesian hierarchical spatiotemporal model to coronavirus disease 2019 (COVID-19) cases and deaths at the county level in the United States for the year 2020. Two models were created, one for cases and one for deaths, utilizing a scaled Besag, York, Mollié model with Type I spatial-temporal interaction. Each model accounts for 16 social vulnerability and 7 environmental variables as fixed effects. The spatial pattern between COVID-19 cases and deaths is significantly different in many ways. The spatiotemporal trend of the pandemic in the United States illustrates a shift out of many of the major metropolitan areas into the United States Southeast and Southwest during the summer months and into the upper Midwest beginning in autumn. Analysis of the major social vulnerability predictors of COVID-19 infection and death found that counties with higher percentages of those not having a high school diploma, having non-White status and being Age 65 and over to be significant. Among the environmental variables, above ground level temperature had the strongest effect on relative risk to both cases and deaths. Hot and cold spots, areas of statistically significant high and low COVID-19 cases and deaths respectively, derived from the convolutional spatial effect show that areas with a high probability of above average relative risk have significantly higher Social Vulnerability Index composite scores. The same analysis utilizing the spatiotemporal interaction term exemplifies a more complex relationship between social vulnerability, environmental measurements, COVID-19 cases, and COVID-19 deaths.

spatial epidemiology↗
DOE OSTI2025

Inferring Plant Acclimation and Improving Model Generalizability With Differentiable Physics‐Informed Machine Learning of Photosynthesis

Net photosynthesis (A N ) is a key component of the global carbon cycle influencing climate feedback over decadal scales. Although plant acclimation to environmental changes can modify A N , traditional vegetation models in Earth system models (ESMs) often rely on plant functional type (PFT)-specific parameterizations or simplified acclimation assumptions limiting generalizability across time, space, and PFTs. In this study, we developed a differentiable photosynthesis model to learn the environmental dependencies of V c,max25 (maximum carboxylation rate at 25°C, representing photosynthetic capacity), as this genre of hybrid physics-informed machine learning can seamlessly train neural networks and process-based equations together. Compared to PFT-specific parameterization of V c,max25 , learning the environment dependencies of key photosynthetic parameters improved model spatiotemporal generalizability. Applying environmental acclimation to V c,max25 led to substantial variations in global mean A N indicating the need to address acclimation in ESMs. The model effectively captured multivariate observations (V c,max25 , A N , and stomatal conductance (g s )) simultaneously with multivariate constraints, improving generalization across space and PFTs. It also learned sensible acclimation relationships of V c,max25 to different environmental conditions. The model explained more than 54%, 57%, and 62% of the variance of A N , g s , and V c,max25 , respectively, presenting a first global-scale spatial test benchmark of A N and g s . These results highlight the potential for differentiable modeling to enhance process-based modules in ESMs and effectively leverage information from large, multivariate data sets.

54 ENVIRONMENTAL SCIENCES↗
DOE OSTI2024

STFM: Accurate Spatio-Temporal Fusion Model for Weather Forecasting

Meteorological prediction is crucial for various sectors, including agriculture, navigation, daily life, disaster prevention, and scientific research. However, traditional numerical weather prediction (NWP) models are constrained by their high computational resource requirements, while the accuracy of deep learning models remains suboptimal. In response to these challenges, we propose a novel deep learning-based model, the Spatiotemporal Fusion Model (STFM), designed to enhance the accuracy of meteorological predictions. Our model leverages Fifth-Generation ECMWF Reanalysis (ERA5) data and introduces two key components: a spatiotemporal encoder module and a spatiotemporal fusion module. The spatiotemporal encoder integrates the strengths of convolutional neural networks (CNNs) and recurrent neural networks (RNNs), effectively capturing both spatial and temporal dependencies. Meanwhile, the spatiotemporal fusion module employs a dual attention mechanism, decomposing spatial attention into global static attention and channel dynamic attention. This approach ensures comprehensive extraction of spatial features from meteorological data. The combination of these modules significantly improves prediction performance. Experimental results demonstrate that STFM excels in extracting spatiotemporal features from reanalysis data, yielding predictions that closely align with observed values. In comparative studies, STFM outperformed other models, achieving a 7% improvement in ground and high-altitude temperature predictions, a 5% enhancement in the prediction of the u/v components of 10 m wind speed, and an increase in the accuracy of potential height and relative humidity predictions by 3% and 1%, respectively. This enhanced performance highlights STFM’s potential to advance the accuracy and reliability of meteorological forecasting.

54 ENVIRONMENTAL SCIENCES↗
DOE OSTI2026

Scaling kinetic Monte-Carlo simulations of grain growth with combined convolutional and graph neural networks

Graph neural networks (GNN) have emerged as a promising machine learning method for microstructure simulations such as grain growth. However, accurate modeling of realistic grain boundary networks requires large simulation cells, which GNN has difficulty scaling up to. To alleviate the computational costs and memory footprint of GNN, we suggest a hybrid architecture combining a convolutional neural network (CNN) based bijective autoencoder to compress the spatial dimensions, and a GNN that evolves the microstructure in the latent space of reduced spatial sizes. Our results demonstrate that the new design significantly reduces computational costs with using fewer message passing layer (from 12 down to 3) compared with GNN alone. The reduction in computational cost becomes more pronounced as the spatial size increases, indicating strong computational scalability. For the largest mesh evaluated (160 3 ), our method reduces memory usage and runtime in inference by 117× and 115×, respectively, compared with GNN-only baseline. More importantly, it shows higher accuracy and stronger spatiotemporal capability than the GNN-only baseline, especially in long-term testing. Such combination of scalability and accuracy is essential for simulating realistic material microstructures over extended time scales. The improvements can be attributed to the bijective autoencoder’s ability to compress information losslessly from spatial domain into a high dimensional feature space, thereby producing more expressive latent features for the GNN to learn from, while also contributing its own spatiotemporal modeling capability. Training data are generated from stochastic grain growth simulations, providing realistic variability for learning robust microstructure evolution. Comprehensive system validation confirms that the model is accurate, robust, and scalable.

36 MATERIALS SCIENCE↗
DOE OSTI2025

Subsurface heatwaves in lakes

Abstract Lake heatwaves (extreme hot water events) can substantially disrupt aquatic ecosystems. Although surface heatwaves are well studied, their vertical structures within lakes remain largely unexplored. Here we analyse the characteristics of subsurface lake heatwaves (extreme hot events occurring below the surface) using a spatiotemporal modelling framework. Our findings reveal that subsurface heatwaves are frequent, often longer lasting but less intense than surface events. Deep-water heatwaves (bottom heatwaves) have increased in frequency (7.2 days decade −1 ), duration (2.1 days decade −1 ) and intensity (0.2 °C days decade −1 ) over the past 40 years. Moreover, vertically compounding heatwaves, where extreme heat occurs simultaneously at the surface and bottom, have risen by 3.3 days decade −1 . By the end of the century, changes in heatwave patterns, particularly under high emissions, are projected to intensify. These findings highlight the need for subsurface monitoring to fully understand and predict the ecological impacts of lake heatwaves.

Environmental Sciences & Ecology↗
NASA NTRS1991

A space-time neural network

Introduced here is a novel technique which adds the dimension of time to the well known back propagation neural network algorithm. Cited here are several reasons why the inclusion of automated spatial and temporal associations are crucial to effective systems modeling. An overview of other works which also model spatiotemporal dynamics is furnished. A detailed description is given of the processes necessary to implement the space-time network algorithm. Several demonstrations that illustrate the capabilities and performance of this new architecture are given.

Villarreal, James A.↗