Engineering Papers⌕ Search

SEARCH · Engineering Papers

Results for “time domain 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 523 records · Page 29
All sourcesNASA NTRSDOE OSTI
DOE OSTI2024

Application of Modified Meshgraphnets for Subsurface Prediction during CO2 Sequestration

In the face of the increasingly dire consequences of anthropogenic climate change, capturing and storing carbon dioxide is paramount. However, several impediments exist to the safe and effective subsurface storage of CO2, such as cost of transport, identification of suitable sites for subsurface storage, and assessment of long-term risk from storage in subsurface aquifers. Accurate subsurface modeling is necessary to ensure that CO2 storage is both safe and effective. Still, such modeling has traditionally required either substantial time and computational power (numerical simulation) or a substantial amount of pre-existing data for training (machine learning models). Additionally, these models lack flexibility in dealing with both changes in discretization of the input data and generalizability beyond the data on which they are trained. In order to address these issues, this research applies graph neural networks (GNNs) to predict subsurface saturation and pressure during CO₂ injection in a model of the Illinois Basin-Decatur Project (IBDP). GNNs provide a flexible, intuitive method for representing and manipulating complex unstructured data, which is often found in many practical domain problems such as fluid flow and subsurface characterization. These unstructured grids are easily represented in GNNs by representing spatially-localized features such as permeability, porosity, saturation, and pressure as nodes in a graph and relationships between these properties as edges connecting these nodes. This research applies a specific GNN model called MeshGraphNets (MGN) to model the change in CO2 saturation and pressure over a 50-month time period (36 months of injection, 14 months post-injection). The MGN model leverages a message passing process that allows the network to learn both the spatial and temporal dynamics of this system simultaneously. Additionally, training on a limited dataset (64 realizations, 20 time points each) resulted in a high degree of accuracy in saturation prediction both within the same timeframe as the training (20 months, 0.039 average RMSE) and when projecting out to the end of injection (36 months, 0.053 average RMSE). Temporal predictions such as those generated by MGNs and other similar models are prone to accumulated error over time; in order to address this, a multi-step rollout (MSR) training process was applied to calculate training loss. This method mimics the forward prediction during inference by “rolling out” multiple time points in a single training step using the previous prediction as input to the MGN model. By calculating the loss several time steps forward from the current prediction, the model is forced to find a more stable state over time. Application of MSR to the MGN model resulted in an average 15% reduction in inference error over time during forward prediction. This study showcases the immense potential of GNNs as a game-changing methodology for predicting pressure and saturation evolution in CCS projects, ultimately paving the way for more sustainable and effective carbon storage solutions. Presentation prepared for the 2024 AiChE Annual Meeting, October 27 to November 1 2024, San Diego, CA.

Holcomb, Paul↗
DOE OSTI2022

SANS characterization of time dependent, slow molecular exchange in an SDS micellar system

Here, we have investigated the molecular exchange of sodium dodecyl sulfate (SDS) micelles in aqueous solution by time-resolved small angle neutron scattering (TR-SANS) measurements as a function of the surfactant and salt concentration. Starting with deuterated (d-SDS) and protonated (h-SDS) SDS micelles, surfactant exchange across the micelles leads to a randomized distribution of d-SDS and h-SDS within each micelle. By employing the contrast matching technique, we have studied this randomization process which is a direct measure of the molecular exchange of this system. Our results show that the randomization of the pure h-SDS and d-SDS micelles occurs in two steps: first, an almost instantaneous drop in the scattering intensity is observed where ~80% of the micelles are randomized (contrast matched). After this, micelle randomization progresses slowly spanning over ~100 hours. Importantly, we show that the kinetics in the second step are dominated by the formation of domains rich in either h-SDS, d-SDS and randomized (50 : 50 h-SDS : d-SDS). The slow exchange step is modeled via a phenomenological approach by drawing analogy to the Langmuir adsorption theory. Finally, the effects of the surfactant and salt concentrations on the instantaneous, and the time dependent randomization of SDS micelles are discussed.

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗
DOE OSTI2024

Connected Traffic Signal Coordination Optimization Framework through Network-Wide Adaptive Linear Quadratic Regulator–Based Control Strategy

Traffic congestion in metropolitan areas causes several significant challenges, such as longer travel times, decreased productivity, increased fuel consumption and vehicle emissions, and even severe injuries during crashes. Traffic signal control is a management approach to reduce traffic congestion and allocate the appropriate right of way for safety and mobility efficiency, both in temporal and spatial domains. Here, this study proposes a network-wide adaptive signal control coordination optimization framework based on the linear quadratic regulator algorithm. The traffic flow conditions driven by signal control inputs are formulated based on their network-wide state-space representation. After modeling traffic control regulation constraints, an adaptive linear quadratic regulator algorithm is designed to maximize the network-wide total throughput under the current conditions. Optimal signal control split time durations for multiple intersections in the network are derived by solving the algebraic Riccati equation. Furthermore, the recursive least square parameter estimation method is employed to quantify dynamic traffic condition changes. To verify the effectiveness of this proposed signal control framework, both simulation and real-world experimental tests are conducted for multiple intersections in downtown Chattanooga, Tennessee, United States. In preparation for real-world experimental tests, pipelines for real-time data processing implementation and historical traffic flow data analysis are conducted. The test results demonstrate that the proposed control framework achieves a decrease in travel time by up to 19.4%, total time spent (TTS) by up to 11.9%, and relative queue balance (RQB) by up to 15.6%. The research findings indicate that the proposed signal control framework can be generalized to handle large scale signal control optimization network-wide.

97 MATHEMATICS AND COMPUTING↗
DOE OSTI2024

Synthesis of historical reservoir operations from 1980 to 2020 for the evaluation of reservoir representation in large-scale hydrologic models

Abstract. All the major river systems in the contiguous United States (CONUS) (and many in the world) are impacted by dams, yet reservoir operations remain difficult to quantify and model due to a lack of data. Reservoir operation data are often inaccessible or distributed across many local operating agencies, making the acquisition and processing of data records quite time-consuming. As a result, large-scale models often rely on simple parameterizations for assumed reservoir operations and have a very limited ability to evaluate how well these approaches match actual historical operations. Here, we use the first national dataset of historical reservoir operations in the CONUS domain, ResOpsUS, to analyze reservoir storage trends and operations in more than 600 major reservoirs across the US. Our results show clear regional differences in reservoir operations. In the eastern US, which is dominated by flood control storage, we see storage peaks in the winter months with sharper decreases in the operational range (i.e., the difference between monthly maximum and minimum storage) in the summer. In the more arid western US where storage is predominantly for irrigation, we find that storage peaks during the spring and summer with increases in the operational range during the summer months. The Lower Colorado region is an outlier because its seasonal storage dynamics more closely mirrored those of flood control basins, yet the region is classified as arid, and most reservoirs have irrigation uses. Consistent with previous studies, we show that average annual reservoir storage has decreased over the past 40 years, although our analyses show a much smaller decrease than previous work. The reservoir operation characterizations presented here can be used directly for development or evaluation of reservoir operations and their derived parameters in large-scale models. We also evaluate how well historical operations match common assumptions that are often applied in large-scale reservoir parameterizations. For example, we find that 100 dams have maximum storage values greater than the reported reservoir capacity from the Global Reservoirs and Dams database (GRanD). Finally, we show that operational ranges have been increasing over time in more arid regions and decreasing in more humid regions, pointing to the need for operating policies which are not solely based on static values.

Geology↗
DOE OSTI2025

Transformers and Long Short-Term Memory Transfer Learning for GenIV Reactor Temperature Time Series Forecasting

Automated monitoring of the coolant temperature can enable autonomous operation of generation IV reactors (GenIV), thus reducing their operating and maintenance costs. Automation can be accomplished with machine learning (ML) models trained on historical sensor data. However, the performance of ML usually depends on the availability of large amount of training data, which is difficult to obtain for GenIV, as this technology is still under development. We propose the use of transfer learning (TL), which involves utilizing knowledge across different domains, to compensate for this lack of training data. TL can be used to create pre-trained ML models with data from small-scale research facilities, which can then be fine-tuned to monitor GenIV reactors. In this work, we develop pre-trained Transformer and long short-term memory (LSTM) networks by training them on temperature measurements from thermal hydraulic flow loops operating with water and Galinstan fluids at room temperature at Argonne National Laboratory. The pre-trained models are then fine-tuned and re-trained with minimal additional data to perform predictions of the time series of high temperature measurements obtained from the Engineering Test Unit (ETU) at Kairos Power. The performance of the LSTM and Transformer networks is investigated by varying the size of the lookback window and forecast horizon. The results of this study show that LSTM networks have lower prediction errors than Transformers, but LSTM errors increase more rapidly with increasing lookback window size and forecast horizon compared to the Transformer errors.

LSTM↗
DOE OSTI2022

Cumulant methods for electron-phonon problems. I. Perturbative expansions

Here in this work, we investigate the ability of the cumulant expansion (CE) to capture one-particle spectral information in electron-phonon coupled systems at both zero and finite temperatures. In particular, we present a comprehensive study of the second- and fourth-order CEs for the one-dimensional Holstein model as compared with numerically exact methods. We investigate both finite sized systems as well as the approach to the thermodynamic limit, drawing distinctions, and connections between the behavior of systems in and away from the thermodynamic limit that enable a greater understanding of the ability of the CE to capture real-frequency information across the full range of wave vectors. We find that for zero electronic momentum, the spectral function is well described by the second-order CE at low and high temperatures. However, for nonzero electronic momenta, the CE is only accurate at high temperature. We analyze the fourth-order cumulant and find that while it improves the description of the short-time dynamics encoded in the one-particle Green's function, it can introduce divergences in the time domain as well as unphysical negative spectral weight in the spectral function. When well-behaved, the fourth-order CE does provide notable accurate corrections to the second-order CE. Finally, we use our results to comment on the use of the CE as a tool for calculating transport behavior in the realistic ab initio modeling of materials.

36 MATERIALS SCIENCE↗
DOE OSTI2020

Physics-guided machine learning approaches to predict the ideal stability properties of fusion plasmas

One of the biggest challenges to achieve the goal of producing fusion energy in tokamak devices is the necessity of avoiding disruptions of the plasma current due to instabilities. The disruption event characterization and forecasting (DECAF) framework has been developed in this purpose, integrating physics models of many causal events that can lead to a disruption. Two different machine learning approaches are proposed to improve the ideal magnetohydrodynamic (MHD) no-wall limit component of the kinetic stability model included in DECAF. First, a random forest regressor (RFR), was adopted to reproduce the DCON computed change in plasma potential energy without wall effects for a large database of equilibria from the national spherical torus experiment (NSTX). This tree-based method provides an analysis of the importance of each input feature, giving an insight into the underlying physics phenomena. Secondly, a fully-connected neural network has been trained on sets of calculations with the DCON code, to get an improved closed form equation of the no-wall β limit as a function of the relevant plasma parameters indicated by the RFR. The neural network has been guided by physics theory of ideal MHD in its extension outside the domain of the NSTX experimental data. The estimated value has been incorporated into the DECAF kinetic stability model and tested against a set of experimentally stable and unstable discharges. Moreover, the neural network results were used to simulate a real-time stability assessment using only quantities available in real-time. Finally, the portability of the model was investigated, showing encouraging results by testing the NSTX-trained algorithm on the mega ampere spherical tokamak (MAST).

70 PLASMA PHYSICS AND FUSION TECHNOLOGY↗
DOE OSTI2024

Multiwavelength study of OT 081: broadband modelling of a transitional blazar

ABSTRACT OT 081 is a well-known, luminous blazar that is remarkably variable in many energy bands. We present the first broadband study of the source, which includes very high energy (VHE, $E\gt $ 100 GeV) $\gamma$-ray data taken by the MAGIC (Major Atmospheric Gamma-ray Imaging Cherenkov telescopes) and H.E.S.S. (High Energy Stereoscopic System) imaging Cherenkov telescopes. The discovery of VHE $\gamma$-ray emission happened during a high state of $\gamma$-ray activity in July 2016, observed by many instruments from radio to VHE $\gamma$-rays. We identify four states of activity of the source, one of which includes VHE $\gamma$-ray emission. Variability in the VHE domain is found on daily time-scales. The intrinsic VHE spectrum can be described by a power law with index $3.27\pm 0.44_{\rm stat}\pm 0.15_{\rm sys}$ (MAGIC) and $3.39\pm 0.58_{\rm stat}\pm 0.64_{\rm sys}$ (H.E.S.S.) in the energy range of 55–300 and 120–500 GeV, respectively. The broadband emission cannot be successfully reproduced by a simple one-zone synchrotron self-Compton model. Instead, an additional external Compton component is required. We test a lepto-hadronic model that reproduces the data set well and a proton-synchrotron-dominated model that requires an extreme proton luminosity. Emission models that are able to successfully represent the data place the emitting region well outside of the broad-line region to a location at which the radiative environment is dominated by the infrared thermal radiation field of the dusty torus. In the scenario described by this flaring activity, the source appears to be a flat spectrum radio quasar (FSRQ), in contrast with past categorizations. This suggests that the source can be considered to be a transitional blazar, intermediate between BL Lac and FSRQ objects.

Abe, H.↗
DOE OSTI2022

Observability of negative capacitance of a ferroelectric film: Theoretical predictions

Here, we theoretically explore mechanisms that can potentially give rise to the steady-state and transient negative capacitance in a uniaxial ferroelectric film stabilized by a dielectric layer. The analytical expressions for the steady-state capacitance of a single-domain and polydomain states are derived within the Landau-Ginzburg-Devonshire approach and used to study the state stability vs the domain splitting as a function of dielectric layer thickness. Analytical expressions for the critical thickness of the dielectric layer, polarization amplitude, equilibrium domain period, and susceptibility are obtained and corroborated by finite element modeling. Furthermore we explore the possible effects of nonlinear screening by two types of screening charges at the ferroelectric-dielectric interface and show that if at least one of the screening charges is very slow, the total polarization dynamics can exhibit complex time- and voltage-dependent behaviors that can be interpreted as an observable negative capacitance. In this setting, the transient negative capacitance effect is accompanied by almost zero dielectric susceptibility in a wide voltage range and low frequencies. These results may help to elucidate the observation of the transient negative capacitance in thin ferroelectric films, and identify materials systems that can give rise to the behavior.

75 CONDENSED MATTER PHYSICS, SUPERCONDUCTIVITY AND↗
DOE OSTI2026

A Physics-Aligned Multi-Domain Machine Learning Framework for Time-Localised Diagnosis of Power Electronics Faults

This paper presents a physics-aligned framework for fault diagnosis in multi-phase power-electronic systems using cycle-synchronous windowing and multi-domain features derived from Fourier, wavelet, and Hilbert–Huang representations. While both logistic regression and multilayer perceptron (MLP) models achieve perfect performance under standard evaluation, blind unseen testing reveals a critical failure in a baseline MLP. This is shown to arise from model selection based on validation accuracy. Using validation-loss-based selection restores correct unseen performance and improves confidence. Feature ablation shows that Fourier and wavelet features dominate, while computational analysis indicates that feature extraction, particularly HHT, governs runtime.

Kumar, Praveen [ORNL] (ORCID:0000000291877857)↗
DOE OSTI2022

Mason–Weaver theory: Revised and extended for a semi-infinite domain

Mason and Weaver developed equations to describe small particles settling under the influence of gravity and Brownian motion, including the limiting case for an infinitely deep suspension. We encountered this common convection–diffusion equation and no-flux boundary conditions in a model for dynamics of adsorbed polymers in dead end pores of a depolymerization catalyst. Close examination reveals that the Mason–Weaver solution is not correct for the infinite domain with a non-uniform initial condition. In this paper, we obtain the time dependent Green’s function for the no flux boundary condition and also for a more general reactive boundary condition. We demonstrate how the results provide solutions, via superposition, which provide solutions for several boundary conditions and all initial conditions.

Chen, Ziqiu↗
DOE OSTI2021

Correlating Time-Resolved Pressure Measurements With Rim Sealing Effectiveness for Real-Time Turbine Health Monitoring

Purge flow is bled from the upstream compressor and supplied to the under-platform region to prevent hot main gas path ingress that damages vulnerable under-platform hardware components. A majority of turbine rim seal research has sought to identify methods of improving sealing technologies and understanding the physical mechanisms that drive ingress. While these studies directly support the design and analysis of advanced rim seal geometries and purge flow systems, the studies are limited in their applicability to real-time monitoring required for condition-based operation and maintenance. As operational hours increase for in-service engines, this lack of rim seal performance feedback results in progressive degradation of sealing effectiveness, thereby leading to reduced hardware life. To address this need for rim seal performance monitoring, the present study utilizes measurements from a one-stage turbine research facility operating with true-scale engine hardware at engine-relevant conditions. Time-resolved pressure measurements collected from the rim seal region are regressed with sealing effectiveness through the use of common machine learning techniques to provide real-time feedback of sealing effectiveness. Two modelling approaches are presented that use a single sensor to predict sealing effectiveness accurately over a range of two turbine operating conditions. Results show that an initial purely data-driven model can be further improved using domain knowledge of relevant turbine operations, which yields sealing effectiveness predictions within three percent of measured values.

Compressors↗
DOE OSTI2025

Knowledge Oriented Graph Unified Transformer (KOGUT) v0.1

KOGUT — Knowledge Oriented Graph Unified Transformer KOGUT implements the Relational Graph Transformer (RelGT) architecture for knowledge graph link prediction in biological domains, with a primary focus on microbial growth media prediction. While the original RelGT (arXiv:2505.10960) targets relational tables, time series, and multi-table databases, KOGUT adapts this architecture for heterogeneous biological knowledge graphs, providing first-in-class AI predictive models for microbial cultivation. Key Adaptations Beyond Original RelGT: - Knowledge Graph Focus: Applied to biological KGs with semantic node types (taxa, chemicals, media, phenotypes, environments) versus generic relational database tables, trained on the KG-Microbe knowledge graph (1.3M entities, 2.9M edges, 24 relation types). - Multimodal Node Encoding: Integrates node labels, categories, descriptions, and synonyms from KG metadata through learned embedding layers—adapting relational column features to graph node attributes with textual semantics. - Extended K-Hop Subgraph Strategy: Optimized neighborhood sampling (3-hop default, configurable up to 200 nodes) tuned for sparse biological networks, building on the original local-global attention framework with biological relation preservation. - Biolink Predicate Preservation: Type-specific transformations for 24 biological edge semantics (occurs_in, consumes, produces, has_phenotype, subclass_of) beyond standard relational foreign keys, enabling multi-relation link prediction. - Inductive Learning Support: Enables zero-shot predictions for novel taxa through feature-based embeddings (temperature, oxygen requirements, gram stain, cell shape), extending the original transductive relational benchmark scope to uncultured microorganisms. CheapSOTA Performance Optimizations (This Distribution): - VQ-EMA Centroid Attention: Vector quantization with exponential moving average for improved global context modeling (+5-10% MRR improvement). - HDF5 Precomputed Data Loading: One-time preprocessing of k-hop subgraphs to eliminate redundant graph traversals (2-5× training speedup). - Distributed Data Parallel Training: Multi-GPU support for scaling to larger knowledge graphs (tested on 4× NVIDIA A100 GPUs at NERSC Perlmutter). - Mixed Precision Training: Automatic mixed precision (AMP) for memory efficiency and faster training. Advantages Over Standard Knowledge Graph Embedding Models: Combines RelGT's proven multi-element tokenization (features, type, hop, structure) with graph-native biological representations, enabling interpretable link prediction across heterogeneous entities that standard embedding models (TransE, RotatE, ComplEx) and table-based transformers cannot directly model. Achieves near-perfect performance on microbial growth media prediction (MRR: 0.9966, Precision@1: 0.9932, Hit@10: 1.0000) while maintaining explainability through attention-based reasoning over biological pathways. Training Data: - KG-Microbe merged knowledge graph: 1,379,337 nodes, 2,960,472 edges - 24 biological relation types including taxonomic hierarchies, metabolic interactions, phenotype associations, and environmental relationships - Primary prediction task: Growth media suitability for microbial taxa (biolink:occurs_in, 50K edges) - Multi-relation capability: Predicts links for any of the 24 relation types, including chemical consumption/production, phenotype associations, and taxonomic classification Citation: Original RelGT Architecture: Dwivedi et al., "Relational Graph Transformer", arXiv:2505.10960, 2025 KOGUT Implementation: Knowledge Oriented Graph Unified Transformer for Microbial Growth Media Prediction Developed at Lawrence Berkeley National Laboratory (LBNL) Trained on NERSC Perlmutter supercomputer

Joachimiak, Marcin [Lawrence Berkeley National Lab↗
DOE OSTI2022

Criticality in the Shallow-to-Deep Transition of Simulated Tropical Marine Convection

Idealized simulations of tropical, marine convection depict shallow, nonprecipitating cumuli located beneath the 0°C level transitioning into cumulonimbi that reach up to 12 km and higher. The timing of the transition was only weakly related to environmental stability, and 13 of the 15 simulations run with 5 different lapse-rate profiles had rain develop at nearly the same time after model start. The key quantity that apparently controlled deep convective formation was vertical acceleration inside cloudy updrafts between cloud base and the 0°C level. Below a critical value of updraft vertical acceleration, little rainfall occurred. Just as the domain-mean updraft acceleration reached the critical value, the first convection quickly grew to past 12 km altitude. Then, as acceleration increased above the critical value, rain rate averaged in the model domain increased quickly over about a 3-h-long period. The specific value of the critical updraft acceleration depended on how updrafts were defined and in what layer the acceleration was averaged; however, regardless of how criticality was defined, a robust relationship between domain-mean updraft vertical acceleration and rain rate occurred. Positive acceleration of updrafts below the 0°C level was present below 2.75 km and was largest in the 500 m above cloud base. However, the maximum difference between updraft and environmental temperatures occurred between 2 and 3 km. The domain-mean Archimedean buoyancy of updrafts relative to some reference state was a poor predictor for domain-mean rain rate. In conclusion, the exact value of the critical updraft acceleration likely depends on numerous other factors that were not investigated.

54 ENVIRONMENTAL SCIENCES↗
DOE OSTI2021

SULI Research Report Paper

Hybrid-CMOS (hCMOS) nanosecond x-ray imagers are a powerful timeresolved HED (high-energy-density) diagnostic due to their sensitivity to visible light, x-rays, and charged particles. However, significant pseudosinusoidal background oscillations have been observed in these sensors when exposed to an early incident excitation, a phenomenon that has been experimentally linked the sensor’s bond wire inductance. We analyze the oscillations both in the time domain and spatially to account for variations on the pixel level. Both theoretical techniques and analysis of experimental data have shown that the oscillations are a global sensor phenomenon. In addition, we effectively describe the oscillations with a physics-based model, giving further insight into the processes affecting them. This technique allows the hCMOS user to subtract out the oscillations from their measurements and improve accuracy.

46 INSTRUMENTATION RELATED TO NUCLEAR SCIENCE AND ↗
DOE OSTI2025

Low-energy electrodynamics and a hidden Fermi liquid in the heavy-fermion compound CeCoIn 5

We present time-domain THz spectroscopy of thin films of the heavy-fermion superconductor CeCoIn 5 . Below the ≈40 K Kondo coherence temperature, a narrow Drude-like peak forms, as a result of the 𝑓-orbital–conduction-electron hybridization and the formation of the heavy-fermion state. The complex optical conductivity is analyzed through a Drude model and extended Drude model analysis. Via the extended Drude model analysis, we measure the frequency-dependent scattering rate (1/𝜏) and effective mass (𝑚*/𝑚 𝑏 ). This scattering rate shows a linear dependence on temperature, which matches the dependence of the resistivity as expected. Nevertheless, the width of the low-frequency Drude peak itself that is set by the renormalized quasiparticle scattering rate (1/𝜏*=𝑚 𝑏 /𝑚*⁢𝜏) shows a 𝑇 2 dependence. This is the scattering rate that characterizes the relaxation time of the renormalized quasiparticles. In conclusion, this gives evidence for a Fermi liquid state, which in conventional transport experiments is hidden by the strong temperature dependent mass.

75 CONDENSED MATTER PHYSICS, SUPERCONDUCTIVITY AND↗
DOE OSTI2023

Field evaluation of semi‐automated moisture estimation from geophysics using machine learning

Geophysical methods can provide three-dimensional (3D), spatially continuous estimates of soil moisture. However, point-to-point comparisons of geophysical properties to measure soil moisture data are frequently unsatisfactory, resulting in geophysics being used for qualitative purposes only. This is because (1) geophysics requires models that relate geophysical signals to soil moisture, (2) geophysical methods have potential uncertainties resulting from smoothing and artifacts introduced from processing and inversion, and (3) results from multiple geophysical methods are not easily combined within a single soil moisture estimation framework. To investigate these potential limitations, an irrigation experiment was performed wherein soil moisture was monitored through time, and several surface geophysical datasets indirectly sensitive to soil moisture were collected before and after irrigation: ground penetrating radar, electrical resistivity tomography (ERT), and frequency domain electromagnetics (FDEM). Data were exported in both raw and processed form, and then snapped to a common 3D grid to facilitate moisture prediction by standard calibration techniques, multivariate regression, and machine learning. A combination of inverted ERT data, raw FDEM, and inverted FDEM data was most informative for predicting soil moisture using a random regression forest model (one-thousand 60/40 training/test cross-validation folds produced root mean squared errors ranging from 0.025–0.046 cm 3 /cm 3 ). This cross-validated model was further supported by a separate evaluation using a test set from a physically separate portion of the study area. Machine learning was conducive to a semi-automated model-selection process that could be used for other sites and datasets to locally improve accuracy.

54 ENVIRONMENTAL SCIENCES↗