Search NASA⌕ Search

SEARCH · Search NASA

Results for “Testing Procedure”

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

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

At least 1,747 records · Page 97
All sourcesNASA NTRSDOE OSTI
DOE OSTI2014

Use of proton beams with breast prostheses and tissue expanders

Since the early 2000s, a small but rapidly increasing number of patients with breast cancer have been treated with proton beams. Some of these patients have had breast prostheses or tissue expanders in place during their courses of treatment. Procedures must be implemented to plan the treatments of these patients. The density, kilovoltage x-ray computed tomography numbers (kVXCTNs), and proton relative linear stopping powers (pRLSPs) were calculated and measured for several test sample devices. The calculated and measured kVXCTNs of saline were 1% and 2.4% higher than the values for distilled water while the calculated RLSP for saline was within 0.2% of the value for distilled water. The measured kVXCTN and pRLSP of the silicone filling material for the test samples were approximately 1120 and 0.935, respectively. The conversion of kVXCTNs to pRLSPs by the treatment planning system standard tissue conversion function is adequate for saline-filled devices but for silicone-filled devices manual reassignment of the pRLSPs is required.

62 RADIOLOGY AND NUCLEAR MEDICINE↗
DOE OSTI2017

Learning an Eddy Viscosity Model Using Shrinkage and Bayesian Calibration: A Jet-in-Crossflow Case Study

In this paper, we demonstrate a statistical procedure for learning a high-order eddy viscosity model (EVM) from experimental data and using it to improve the predictive skill of a Reynolds-averaged Navier–Stokes (RANS) simulator. The method is tested in a three-dimensional (3D), transonic jet-in-crossflow (JIC) configuration. The process starts with a cubic eddy viscosity model (CEVM) developed for incompressible flows. It is fitted to limited experimental JIC data using shrinkage regression. The shrinkage process removes all the terms from the model, except an intercept, a linear term, and a quadratic one involving the square of the vorticity. The shrunk eddy viscosity model is implemented in an RANS simulator and calibrated, using vorticity measurements, to infer three parameters. The calibration is Bayesian and is solved using a Markov chain Monte Carlo (MCMC) method. A 3D probability density distribution for the inferred parameters is constructed, thus quantifying the uncertainty in the estimate. The phenomenal cost of using a 3D flow simulator inside an MCMC loop is mitigated by using surrogate models (“curve-fits”). A support vector machine classifier (SVMC) is used to impose our prior belief regarding parameter values, specifically to exclude nonphysical parameter combinations. The calibrated model is compared, in terms of its predictive skill, to simulations using uncalibrated linear and CEVMs. Finally, we find that the calibrated model, with one quadratic term, is more accurate than the uncalibrated simulator. The model is also checked at a flow condition at which the model was not calibrated.

42 ENGINEERING↗
DOE OSTI2013

Integrated Water Gas Shift Membrane Reactors Utilizing Novel, Non Precious Metal Mixed Matrix Membrane

Nanoparticles of zeolitic imidazolate frameworks and other related hybrid materials were prepared by modifying published synthesis procedures by introducing bases, changing stoichiometric ratios, or adjusting reaction conditions. These materials were stable at temperatures >300 °C and were compatible with the polymer matrices used to prepare mixed- matrix membranes (MMMs). MMMs tested at 300 °C exhibited a >30 fold increase in permeability, compared to those measured at 35 °C, while maintaining H 2 /CO 2 selectivity. Measurements at high pressure (up to 30 atm) and high temperature (up to 300 °C) resulted in an increase in gas flux across the membrane with retention of selectivity. No variations in permeability were observed at high pressures at either 35 or 300 °C. CO 2 -induced plasticization was not observed for Matrimid®, VTEC, and PBI polymers or their MMMs at 30 atm and 300 °C. Membrane surface modification by cross-linking with ethanol diamine resulted in an increase in H 2 /CO 2 selectivity at 35 °C. Spectrometric analysis showed that the cross-linking was effective to temperatures <150 °C. At higher temperatures, the cross-linked membranes exhibit a H 2 /CO 2 selectivity similar to the uncross-linked polymer. Performance of the polybenzimidazole (PBI) hollow fibers prepared at Santa Fe Science and Technology (SFST, Inc.) showed increased flux and selectivity at 300 °C, which is comparable to a flat PBI membrane. A water-gas shift reactor has been built and currently being optimized for testing under DOE conditions.

36 MATERIALS SCIENCE↗
DOE OSTI2021

Analysis of the MST-16 crimp process on cold 304L stainless steel

The purpose of this report is to assess the quality of the MST-16 crimp process through a series of cold experiments and metallographic analyses. Common practice in MST-16 uses a mechanical crimp on 304L stainless steel tubes to prevent radioactive contamination from escaping the tube interior for a variety of experiments. The crimped tubes must pass a leak check to demonstrate their radiological safety. Cold (i.e. non-radioactive samples) 304L tube samples were tested using two different approved crimping devices to assess the quality of the crimps. Each crimp was leak checked, followed by destructive metallographic analysis to determine the quality of both the crimp equipment and procedures used by MST-16 personnel.

42 ENGINEERING↗
DOE OSTI2022

Radiological Considerations Supporting the American Medical Isotope Producer Niowave

This report provides a summary of work performed by Savannah River National Laboratory (SRNL) during FY22 in support of Niowave, an American Medical Isotope Production (AMIP) facility. SRNL served in a technical support role, funded by NA-231, supporting Niowave in the beginning of non radiological functional testing of their airport facility in FY22. Through this agreement, SRNL provides Niowave access to subject matter experts (SMEs) for questions that may arise during the drafting of engineering plans, facility policies, and/or response procedures. This report summarizes discussions between SRNL and Niowave about guidance concerning issues with the floor of the hot cells, dispersible removable contamination (beta emitting fission product), hot cell operation specifically utilization of cameras or mirrors, and the process for removing contaminated waste from the hot cell. Lessons learned and other various resources provided to Niowave by SRNL during FY22 are referenced in this report. This work was funded through NNSA’s Office of Material Management and Minimization, Conversion Office (NA-231).

12 MANAGEMENT OF RADIOACTIVE AND NON-RADIOACTIVE W↗
DOE OSTI2021

Recovery and Enhanced Upgrading of Rare Earth Elements from Coal-Based Resources: Bioleaching and Precipitation

Rare earth elements (REEs) are of great importance to modern society and their reliable supply is a major concern of many industries that utilize them in metal alloys, semiconductors, electrical equipment, and defense equipment. REEs in the coal waste have been revealed to be an alternative resource for REEs production. In this study, the extraction, recovery, and upgrading of the REEs from coal waste has been realized with the bioleaching and precipitation processes. Reliable and sustainable acid and oxidant production from the oxidation of the pyrite with Acidithiobacillus ferrooxidans to generate acid for leaching were realized in this research. The acidified bioleaching solution was used to extract REEs from coal waste, with 13–14% yields for most REE elements (~72 h of leaching). However, recovery for longer duration tests was significant higher (varies from 40–60% for individual REEs). After extraction, precipitation and separation processes were designed with the aid of Visual Minteq calculations and modeling to concentrate the REEs. With the procedures designed in this research, a final REEs precipitate product containing 36.7% REEs was produced.

01 COAL, LIGNITE, AND PEAT↗
DOE OSTI2018

Development of Grade 91 inelastic model for incorporation in ASME Division 5

This report describes a unified viscoplastic model for Grade 91 suitable for use with the Section III, Division 5 design by inelastic analysis procedures in the ASME Boiler and Pressure Vessel Code. The report identifies several key features a successful design inelastic model must capture along with critical aspects of the response of Grade 91 for elevated temperature loading conditions typical of expected reactor operating conditions. An inelastic model for Code design must capture the average response of a material including heat-to-heat and product form variations. This work develops a method for calibrating such a model to a large experimental database using genetic algorithm optimization. Finally, the report compares the final model to several specialized experimental tests to validate that it will be suitable for use in high temperature structural design. The techniques developed here can be applied to the remaining Section III, Division 5, Class A materials to develop a nonmandatory appendix to the Code describing inelastic material models suitable for use with the design by inelastic analysis procedures.

21 SPECIFIC NUCLEAR REACTORS AND ASSOCIATED PLANTS↗
DOE OSTI2016

TU-H-CAMPUS-IeP3-02: Neurovascular 4D Parametric Imaging Using Co-Registration of Biplane DSA Sequences with 3D Vascular Geometry Obtained From Cone Beam CT

Purpose: To create 4D parametric images using biplane Digital Subtraction Angiography (DSA) sequences co-registered with the 3D vascular geometry obtained from Cone Beam-CT (CBCT). Methods: We investigated a method to derive multiple 4D Parametric Imaging (PI) maps using only one CBCT acquisition. During this procedure a 3D-DSA geometry is stored and used subsequently for all 4D images. Each time a biplane DSA is acquired, we calculate 2D parametric maps of Bolus Arrival Time (BAT), Mean Transit Time (MTT) and Time to Peak (TTP). Arterial segments which are nearly parallel with one of the biplane imaging planes in the 2D parametric maps are co-registered with the 3D geometry. The values in the remaining vascular network are found using spline interpolation since the points chosen for co-registration on the vasculature are discrete and remaining regions need to be interpolated. To evaluate the method we used a patient CT volume data set for 3D printing a neurovascular phantom containing a complete Circle of Willis. We connected the phantom to a flow loop with a peristaltic pump, simulating physiological flow conditions. Contrast media was injected with an automatic injector at 10 ml/sec. Images were acquired with a Toshiba Infinix C-arm and 4D parametric image maps of the vasculature were calculated. Results: 4D BAT, MTT, and TTP parametric image maps of the Circle of Willis were derived. We generated color-coded 3D geometries which avoided artifacts due to vessel overlap or foreshortening in the projection direction. Conclusion: The software was tested successfully and multiple 4D parametric images were obtained from biplane DSA sequences without the need to acquire additional 3D-DSA runs. This can benefit the patient by reducing the contrast media and the radiation dose normally associated with these procedures. Partial support from NIH Grant R01-EB002873 and Toshiba Medical Systems Corp.

60 APPLIED LIFE SCIENCES↗
DOE OSTI2012

Numerical Analysis of JNES Seismic Tests on Degraded Combined Piping System

Nuclear power plant safety under seismic conditions is an important consideration. The piping systems may have some defects caused by fatigue, stress corrosion cracking, etc., in aged plants. These cracks may not only affect the seismic response but also grow and break through causing loss of coolant. Therefore, an evaluation method needs to be developed to predict crack growth behavior under seismic excitation. This paper describes efforts conducted to analyze and better understand a series of degraded pipe tests under seismic loading that was conducted by Japan Nuclear Energy Safety Organization (JNES). A special 'cracked-pipe element' (CPE) concept, where the element represented the global moment-rotation response due to the crack, was developed. This approach was developed to significantly simplify the dynamic finite element analysis in fracture mechanics fields. In this paper, model validation was conducted by comparisons with a series of pipe tests with circumferential through-wall and surface cracks under different excitation conditions. These analyses showed that reasonably accurate predictions could be made using the abaqus connector element to model the complete transition of a circumferential surface crack to a through-wall crack under cyclic dynamic loading. The JNES primary loop recirculation piping test was analyzed in detail. This combined-component test had three crack locations and multiple applied simulated seismic block loadings. Comparisons were also made between the ABAQUS finite element (FE) analyses results to the measured displacements in the experiment. Good agreement was obtained, and it was confirmed that the simplified modeling is applicable to a seismic analysis for a cracked pipe on the basis of fracture mechanics. Pipe system leakage did occur in the JNES tests. The analytical predictions using the CPE approach did not predict leakage, suggesting that cyclic ductile tearing with large-scale plasticity was not the crack growth mode for the acceleration excitations considered here. Hence, the leakage was caused by low-cycle fatigue with small-scale yielding. The procedure used to make predictions of low-cycle fatigue crack growth with small-scale yielding was based on the Dowling {Delta}J procedure, which is an extension of linear-elastic fatigue crack growth methodology into the nonlinear plasticity region. The predicted moments from the CPE approach were used using a cycle-by-cycle crack growth procedure. The predictions compare quite well with the experimental measurements.

21 SPECIFIC NUCLEAR REACTORS AND ASSOCIATED PLANTS↗
DOE OSTI2011

Commissioning and benchmarking a 3D dosimetry system for clinical use

Purpose: A 3D dosimetry system is described which consists of two parts: a radiochromic plastic dosimeter PRESAGE (which responds to absorbed dose with a linear change in optical-density) and the Duke large-field-of-view optical-CT scanner (DLOS). The DLOS/PRESAGE system has recently been commissioned and benchmarked for clinical use and, in particular, for verification and commissioning of complex radiation treatments. Methods: DLOS commissioning involved determining the dynamic range, spatial resolution, noise, temporal, and other characteristics of the light source and imaging components. Benchmarking tests were performed on the combined DLOS/PRESAGE system to establish baseline dosimetric performance. The tests consisted of delivering simple radiation treatments to PRESAGE dosimeters, and comparing the measured 3D relative dose distributions with the known gold standard. The gold standard distribution was obtained from machine beam-data or the treatment planning system (TPS). All studies used standardized procedures to ensure consistency. Results: For commissioning, isotropic spatial resolution was submillimeter (MTF > 0.5 for frequencies of 1.5 lp/mm) and the dynamic range was {approx}60 dB. Flood field uniformity was within 10% and stable after 45 min of warm-up. Stray-light is small, due to telecentricity, but even the residual can be removed through deconvolution by a point-spread-function. For benchmarking, the mean 3D passing NDD (normalized dose distribution) rate (3%, 3mm, 5% dose threshold) over the benchmark data sets was 97.3% {+-} 0.6% (range 96%-98%), which is on par with other planar dosimeters used in external beam radiation therapy indicating excellent agreement. Noise was low at <2% of maximum dose (4-12 Gy) for 2 mm reconstructions. The telecentric design was critical to enabling fast imaging with minimal stray-light artifacts. Conclusions: This work presents the first comprehensive benchmarking of a 3D dosimetry system for clinical use. The DLOS/PRESAGE benchmark tests show consistently good agreement to simple known distributions. The system produces accurate isotropic 2 mm dose data over clinical volumes (e.g., 16 cm diameter phantoms, 12 cm height), in under 15 min. It represents a uniquely useful and versatile new tool for commissioning and verification of complex therapy treatments.

46 INSTRUMENTATION RELATED TO NUCLEAR SCIENCE AND ↗
DOE OSTI2014

Automated identification of elemental ions in macromolecular crystal structures

The solvent-picking procedure in phenix.refine has been extended and combined with Phaser anomalous substructure completion and analysis of coordination geometry to identify and place elemental ions. Many macromolecular model-building and refinement programs can automatically place solvent atoms in electron density at moderate-to-high resolution. This process frequently builds water molecules in place of elemental ions, the identification of which must be performed manually. The solvent-picking algorithms in phenix.refine have been extended to build common ions based on an analysis of the chemical environment as well as physical properties such as occupancy, B factor and anomalous scattering. The method is most effective for heavier elements such as calcium and zinc, for which a majority of sites can be placed with few false positives in a diverse test set of structures. At atomic resolution, it is observed that it can also be possible to identify tightly bound sodium and magnesium ions. A number of challenges that contribute to the difficulty of completely automating the process of structure completion are discussed.

75 CONDENSED MATTER PHYSICS, SUPERCONDUCTIVITY AND↗
DOE OSTI2019

A performance evaluation framework for building fault detection and diagnosis algorithms

Fault detection and diagnosis (FDD) algorithms for building systems and equipment represent one of the most active areas of research and commercial product development in the buildings industry. However, far more effort has gone into developing these algorithms than into assessing their performance. As a result, considerable uncertainties remain regarding the accuracy and effectiveness of both research-grade FDD algorithms and commercial products - a state of affairs that has hindered the broad adoption of FDD tools. This article presents a general, systematic framework for evaluating the performance of FDD algorithms. The article focuses on understanding the possible answers to two key questions: in the context of FDD algorithm evaluation, what defines a fault and what defines an evaluation input sample? The answers to these questions, together with appropriate performance metrics, may be used to fully specify evaluation procedures for FDD algorithms.

32 ENERGY CONSERVATION, CONSUMPTION, AND UTILIZATI↗
DOE OSTI2023

SULI Internship Deliverables (Flow Orifice Design: Measuring Degas Flow of Experiment Loops at the Advanced Test Reactor)

These deliverables present a design for a new orifice flow meter to be installed in three experiment loops at the Advanced Test Reactor. The current configuration of the experiment loop system has an orifice that is not properly sized to the low flow rates of the degas line of the pressurizer system, causing inaccurate flow measurements. Under consideration of flow conditions and system constraints, an iterative procedure was utilized to determine the optimal sizing for a new orifice. This orifice is designed to accurately and precisely measure the flow rate of the degas line, which will consequently improve control of the pressure and chemistry of the loop pressurizer system. The scope of this project includes recommendations for orifice sizing, material, flange type, face type, and installation location on the loop piping spools. The design is suitable for experiment loops 1D-N, 2B-SE, and 2D-SW and meets all necessary standards.

21 SPECIFIC NUCLEAR REACTORS AND ASSOCIATED PLANTS↗
DOE OSTI2015

Impact of Anatomical, Procedural, and Operator Skill Factors on the Success and Duration of Fluoroscopy-Guided Transjugular Intrahepatic Portosystemic Shunt

PurposeTo assess the impact of anatomical, procedural, and operator skill factors on the success and duration of fluoroscopy-guided transjugular intrahepatic portoystemic shunt following standard operating procedure (SOP).Material and MethodsDuring a 32-month period, 102 patients underwent transjugular intrahepatic portosystemic shunt creation (TIPS) by two interventional radiologists (IR) following our institutional SOP based on fluoroscopy guidance. Both demographic and procedural data were assessed. The duration of the intervention (D{sub Int}) and of the portal vein puncture (D{sub Punct}) was analyzed depending on the skill level of the IR as well as the anatomic or procedural factors.ResultsIn 99 of the 102 patients, successful TIPS without peri-procedural complications was performed. The mean D{sub Int} (IR1: 77 min; IR2: 51 min, P < 0.005) and the mean D{sub Punct} (IR1: 19 min; IR2: 13 min, P < 0.005) were significantly higher in TIPS performed by IR1 (with 2 years of clinical experience performing TIPS, n = 38) than by IR2 (>10 years of clinical experience performing TIPS, n = 61), (P < 0.005 both, Mann–Whitney U test). D{sub Int} showed a higher correlation with D{sub Punct} for IR2 (R{sup 2} = 0.63) than for IR1 (R{sup 2} = 0.13). There was no significant difference in the D{sub Punct} for both IRs with regard to the success of the wedged portography (P = 0.90), diameter of the portal vein (P = 0.60), central right portal vein length (P = 0.49), or liver function (MELD-Score before the TIPS procedure; P = 0.14).ConclusionTIPS following SOP is safe, fast, and reliable. The only significant factor for shorter D{sub Punct} and D{sub Int} was the clinical experience of the IR. Anatomic variability, successful portography, or liver function did not alter the duration or technical success of TIPS.

62 RADIOLOGY AND NUCLEAR MEDICINE↗
DOE OSTI2016

TU-D-209-04: How Useful Are Indirect Dose Metrics for Estimating Peak Skin Dose in Interventional Cardiology?

Purpose: The purpose of this study was to determine if a relationship between indirect dose metrics and PSD could be established for fluoroscopically-guided interventional cardiology procedures. Methods: PSD were measured directly using XR-RV3 radiochromic film for 94 consecutive fluoroscopically guided interventional cardiology procedures performed at two sites. Procedures were both diagnostic and therapeutic in nature. Radiation dose structured reports (RDSR) were collected for each procedure and used to calculate indirect estimates of PSD which were compared to the measured PSD. Reference air kerma (Ka,r) was also compared to the measured PSD. Pearson’s correlation coefficient was calculated for each metric and metrics were compared to measured PSD using a two-tailed t-test. Data were log transformed prior to statistical analysis. Results: Both Ka,r and the calculated PSD were closely correlated with measured PSD at each sites (Ka,r: 0.92 and 0.86, indirect PSD: 0.91 and 0.88). At one site, neither Ka,r nor indirect PSD was significantly different from the measured PSD (p = 0.22 and p=0.054, respectively), while at the second site both Ka,r and indirect PSD were significantly higher than measured PSD (p<0.0001 and p<0.0001, respectively). In almost all cases, both Ka,r and indirect PSD overestimated the true PSD. Conclusions: The use of a range of gantry angles and table positions, along with variation in procedural imaging requirements, limits the utility of indirect dose metrics for predicting PSD for interventional cardiology procedures. A. Kyle Jones and Alexander S. Pasciak are owners of Fluoroscopic Safety, LLC.

60 APPLIED LIFE SCIENCES↗
DOE OSTI2016

Stent Retriever Thrombectomy in Different Thrombus Locations of Anterior Cerebral Circulation

BackgroundMechanical thrombectomy (MT) is a safe and efficient treatment for acute ischemic stroke in patients with proximal anterior occlusion and large penumbra. We evaluated the technical and clinical success of MT in relation to the location of the occlusion (internal carotid artery, M1 and M2 segments of the middle cerebral artery).MethodsWe prospectively reviewed 130 patients of whom 105 met the inclusion criteria. Baseline clinical, procedural and imaging variables, technical outcome (TICI, thrombolysis in cerebral infarction), 24 h imaging outcome and three-month clinical outcome (mRS, modified Rankin Scale) were recorded. Differences between the groups were studied with statistical tests according to the type of the variable.ResultsThere were 37, 46 and 22 patients in the internal carotid artery (ICA), M1 and M2 groups, respectively. TICI 2b or 3 was achieved in 92 cases (88 %) with a non-significant trend towards a better recanalization outcome in the ICA and M1 groups. Overall, 57 of the 105 patients (55 %) experienced favorable clinical outcome (mRS ≤ 2) with no significant differences between the groups. Excellent outcome (mRS ≤ 1) was seen in 40 patients (39 %) and there proportionally more patients with excellent outcome in the ICA and M1 groups (ICA: 44 %, M1: 41 %, M2: 23 % of patients, p = 0.22).ConclusionsThere were no statistically significant differences in the technical or clinical outcomes between the different sites of occlusion (ICA, M1 or M2). There was a non-significant trend towards achieving excellent clinical outcome (3-month mRS ≤ 1) more often and better recanalization results in the two more proximal locations.

62 RADIOLOGY AND NUCLEAR MEDICINE↗
DOE OSTI2019

Computational Modeling of Fracture Toughness of Al-Si and Al-Zn-Mg-Cu Alloys with Detected Porosity

This article presents the results of an experimental and numerical analysis of the fracture resistance of cast aluminum alloys subjected to variable loads over time based on the fracture mechanics method. The subjects of the analysis were the AlSi9 and AlZn6MgCu alloys (usually used for plastic processing) in which the presence of porosities in the form of separate or dispersed voids was revealed experimentally. The material for the experimental procedures was obtained directly from a casting or from casting samples. Depending on the type of the defect detected, different numerical models were proposed. The results of the numerical calculations were verified experimentally. Static tensile tests, fatigue low-cycle tests, fracture mechanics tests, fatigue crack propagation tests, a microscopic examination of the sample fracture area, and an x-ray tomographic analysis were all performed. A correlation of the results between the numerical analysis and the experimental work has been found. On the basis of the proposed calculation models, the fatigue life of castings of the analyzed alloys can be predicted in the case of concentrated and dispersed defects.

36 MATERIALS SCIENCE↗
DOE OSTI2021

Gap-filling eddy covariance methane fluxes: Comparison of machine learning model predictions and uncertainties at FLUXNET-CH4 wetlands

Time series of methane fluxes measured by eddy-covariance require gap-filling to estimate annual emissions. Gap-filling methane fluxes is challenging because of high variability and complex responses to multiple drivers. To date, there is no widely established gap-filling standard for methane, with regards both to the best model algorithms and predictors. In this study, we address the need for standardization by synthesizing results of gap-filling methods applied at 17 wetland sites spanning boreal to tropical regions including all major wetlands classes and two rice paddies. We introduce new procedures for: 1) creating realistic artificial gap scenarios, 2) training and evaluating gap-filling models without overstating performance, and 3) predicting half-hourly methane fluxes and annual emissions with robust uncertainty estimates. We tested a conventional method (marginal distribution sampling) and four machine learning algorithms - penalized linear regression, artificial neural networks, random forests, and boosted decision trees - and four predictor sets, including temporal, meteorological, ecosystem carbon and energy flux, and soil predictors. We find that the conventional method can achieve similar median performance to the machine learning models but is worse than the best machine learning models and relatively insensitive to predictor choices. Of the machine learning models, decision tree algorithms performed the best in cross-validation experiments, even with a baseline predictor set, and artificial neural networks showed comparable performance when using all predictors. Soil temperature was frequently the most important predictor whilst water table depth was important at sites with substantial water table fluctuations, highlighting the value of data on soil conditions. Raw gap-filling uncertainties from the machine learning models were underestimated and we propose a method to calibrate uncertainties to observations. Finally, we gap-fill and provide summary evaluation metrics for all 81 sites in the FLUXNET-CH4 community dataset and publicly release the python code for model development, evaluation, and uncertainty estimation.

42 ENGINEERING↗