Engineering Papers⌕ Search

SEARCH · Engineering Papers

Results for “sampling variability”

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 1,477 records · Page 82
All sourcesNASA NTRSDOE OSTI
DOE OSTI2012

Characterization and evaluation of caustic wash tank and solvent hold tank samples from MCU from august to september 2011

During processing of Salt Batches 3 and 4 in the Modular Caustic-Side Solvent Extraction Unit (MCU), the decontamination efficiency for cesium declined from historical values and from expectations based on laboratory testing. This report documents efforts to analyze samples of solvent and process solutions from MCU in an attempt to understand the cause of the reduced performance and to recommend mitigations. CWT Solutions from MCU from the time period of variable decontamination factor (DF) performance which covers from April 2011 to September 2011 (during processing of Salt Batch 4) were examined for impurities using chromatography and spectroscopy. The results indicate that impurities were found to be of two types: aromatic containing impurities most likely from Modifier degradation and aliphatic type impurities most likely from Isopar{reg_sign} L and tri-n-octylamine (TOA) degradation. Caustic washing the Solvent Hold Tank (SHT) solution with 1M NaOH improved its extraction ability as determined from {sup 22}Na uptake tests. Evidence from this work showed that pH variance in the aqueous solutions within the range of 1M nitric acid to 1.91M NaOH that contacted the solvent samples does not influence the analytical determination of the TOA concentration by GC-MS.

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

Extreme metrics from large ensembles: investigating the effects of ensemble size on their estimates

Abstract. We consider the problem of estimating the ensemble sizes required to characterize the forced component and the internal variability of a number of extreme metrics. While we exploit existing large ensembles, our perspective is that of a modeling center wanting to estimate a priori such sizes on the basis of an existing small ensemble (we assume the availability of only five members here). We therefore ask if such a small-size ensemble is sufficient to estimate accurately the population variance (i.e., the ensemble internal variability) and then apply a well-established formula that quantifies the expected error in the estimation of the population mean (i.e., the forced component) as a function of the sample size n, here taken to mean the ensemble size. We find that indeed we can anticipate errors in the estimation of the forced component for temperature and precipitation extremes as a function of n by plugging into the formula an estimate of the population variance derived on the basis of five members. For a range of spatial and temporal scales, forcing levels (we use simulations under Representative Concentration Pathway 8.5) and two models considered here as our proof of concept, it appears that an ensemble size of 20 or 25 members can provide estimates of the forced component for the extreme metrics considered that remain within small absolute and percentage errors. Additional members beyond 20 or 25 add only marginal precision to the estimate, and this remains true when statistical inference through extreme value analysis is used. We then ask about the ensemble size required to estimate the ensemble variance (a measure of internal variability) along the length of the simulation and – importantly – about the ensemble size required to detect significant changes in such variance along the simulation with increased external forcings. Using the F test, we find that estimates on the basis of only 5 or 10 ensemble members accurately represent the full ensemble variance even when the analysis is conducted at the grid-point scale. The detection of changes in the variance when comparing different times along the simulation, especially for the precipitation-based metrics, requires larger sizes but not larger than 15 or 20 members. While we recognize that there will always exist applications and metric definitions requiring larger statistical power and therefore ensemble sizes, our results suggest that for a wide range of analysis targets and scales an effective estimate of both forced component and internal variability can be achieved with sizes below 30 members. This invites consideration of the possibility of exploring additional sources of uncertainty, such as physics parameter settings, when designing ensemble simulations.

54 ENVIRONMENTAL SCIENCES↗
DOE OSTI2016

Trans-Pacific transport and evolution of aerosols: Evaluation of quasi-global WRF-Chem simulation with multiple observations

A fully coupled meteorology-chemistry model (WRF-Chem, the Weather Research and Forecasting model coupled with chemistry) has been configured to conduct quasi-global simulation for 5 years (2010–2014) and evaluated with multiple observation data sets for the first time. The evaluation focuses on the simulation over the trans-Pacific transport region using various reanalysis and observational data sets for meteorological fields and aerosol properties. The simulation generally captures the overall spatial and seasonal variability of satellite retrieved aerosol optical depth (AOD) and absorbing AOD (AAOD) over the Pacific that is determined by the outflow of pollutants and dust and the emissions of marine aerosols. The assessment of simulated extinction Ångström exponent (EAE) indicates that the model generally reproduces the variability of aerosol size distributions as seen by satellites. In addition, the vertical profile of aerosol extinction and its seasonality over the Pacific are also well simulated. The difference between the simulation and satellite retrievals can be mainly attributed to model biases in estimating marine aerosol emissions as well as the satellite sampling and retrieval uncertainties. Compared with the surface measurements over the western USA, the model reasonably simulates the observed magnitude and seasonality of dust, sulfate, and nitrate surface concentrations, but significantly underestimates the peak surface concentrations of carbonaceous aerosol likely due to model biases in the spatial and temporal variability of biomass burning emissions and secondary organic aerosol (SOA) production. A sensitivity simulation shows that the trans-Pacific transported dust, sulfate, and nitrate can make significant contribution to surface concentrations over the rural areas of the western USA, while the peaks of carbonaceous aerosol surface concentrations are dominated by the North American emissions. Both the retrievals and simulation show small interannual variability of aerosol characteristics for 2010–2014 averaged over three Pacific sub-regions. Furthermore, the evaluation in this study demonstrates that the WRF-Chem quasi-global simulation can be used for investigating trans-Pacific transport of aerosols and providing reasonable inflow chemical boundaries for the western USA, allowing one to further understand the impact of transported pollutants on the regional air quality and climate with high-resolution nested regional modeling.

54 ENVIRONMENTAL SCIENCES↗
DOE OSTI2020

Optimal Eco-Driving Control of Connected and Autonomous Vehicles Through Signalized Intersections

This article focuses on the speed planning problem for connected and automated vehicles (CAVs) communicating to traffic lights. The uncertainty of traffic signal timing for signalized intersections on the road is considered. The eco-driving problem is formulated as a data-driven chance-constrained robust optimization problem. Effective red-light duration (ERD) is defined as a random variable, and describes the feasible passing time through the signalized intersections. Usually, the true probability distribution for ERD is unknown. Consequently, a data-driven approach is adopted to formulate chance constraints based on empirical sample data. This incorporates robustness into the eco-driving control problem with respect to uncertain signal timing. Dynamic programming (DP) is employed to solve the optimization problem. The simulation results demonstrate that the proposed method can generate optimal speed reference trajectories with 40% less vehicle fuel consumption, while maintaining the arrival time at a similar level compared to a modified intelligent driver model (IDM). The proposed control approach significantly improves the controller's robustness in the face of uncertain signal timing, without requiring to know the distribution of the random variable a priori.

33 ADVANCED PROPULSION SYSTEMS↗
DOE OSTI2016

Retrospective analysis of pulse oximeter alarm settings in an intensive care unit patient population

Background: The cacophony of alerts and alarms in a hospital produced by medical devices results in alarm fatigue. The pulse oximeter is one of the most common sources of alarms. One of the ways to reduce alarm rates is to adjust alarm settings at the bedside. This study is aimed to retrospectively examine individual pulse oximeter alarm settings on alarm rates and inter- and intra- patient variability. Methods: Nine hundred sixty-two previously collected intensive care unit (ICU) patient records were obtained from the Multiparameter Intelligent Monitoring in Intensive Care II Database (Beth Israel Deaconess Medical Center, Boston, MA). Inclusion criteria included patient records that contained SpO 2 trend data sampled at 1 Hz for at least 1 h and a matching clinical record. SpO 2 alarm rates were simulated by applying a range of thresholds (84, 86, 88, and 90 %) and delay times (10 to 60 s) to the SpO 2 data. Patient records with at least 12 h of SpO 2 data were examined for the variability in alarm rate over time. Results: Decreasing SpO 2 thresholds and increasing delay times resulted in decreased alarm rates. A limited number of patient records accounted for most alarms, and this number increased as alarm settings loosened (the top 10 % of patient records were responsible for 57.4 % of all alarms at an SpO 2 threshold of 90 % and 15 s delay and 81.6 % at an SpO 2 threshold of 84 % and 45 s delay). Alarm rates were not consistent over time for individual patients with periods of high and low alarms for all alarm settings. Conclusion: Pulse oximeter SpO 2 alarm rates are variable between patients and over time, and the alarm rate and the extent of inter- and intra-patient variability can be affected by the alarm settings. Personalized alarm settings for a patient’s current status may help to reduce alarm fatigue for nurses.

59 BASIC BIOLOGICAL SCIENCES↗
DOE OSTI2016

ERRATIC FLARING OF BL LAC IN 2012–2013: MULTIWAVELENGTH OBSERVATIONS

BL Lac, the eponymous blazar, flared to historically high levels at millimeter, infrared, X-ray, and gamma-ray wavelengths in 2012. We present observations made with Herschel, Swift, NuSTAR, Fermi, the Submillimeter Array, CARMA, and the VLBA in 2012–2013, including three months with nearly daily sampling at several wavebands. We have also conducted an intensive campaign of 30 hr with every-orbit observations by Swift and NuSTAR, accompanied by Herschel, and Fermi observations. The source was highly variable at all bands. Time lags, correlations between bands, and the changing shapes of the spectral energy distributions can be explained by synchrotron radiation and inverse Compton emission from nonthermal seed photons originating from within the jet. The passage of four new superluminal very long baseline interferometry knots through the core and two stationary knots about 4 pc downstream accompanied the high flaring in 2012–2013. The seed photons for inverse Compton scattering may arise from the stationary knots and from a Mach disk near the core where relatively slow-moving plasma generates intense nonthermal radiation. The 95 spectral energy distributions obtained on consecutive days form the most densely sampled, broad wavelength coverage for any blazar. The observed spectral energy distributions and multi-waveband light curves are similar to simulated spectral energy distributions and light curves generated with a model in which turbulent plasma crosses a conical shock with a Mach disk.

79 ASTRONOMY AND ASTROPHYSICS↗
DOE OSTI2022

The Time Domain Spectroscopic Survey: Changing-look Quasar Candidates from Multi-epoch Spectroscopy in SDSS-IV

Active galactic nuclei (AGN) can vary significantly in their rest-frame optical/UV continuum emission, and with strong associated changes in broad line emission, on much shorter timescales than predicted by standard models of accretion disks around supermassive black holes. Most such changing-look or changing-state AGN—and at higher luminosities, changing-look quasars (CLQs)—have been found via spectroscopic follow-up of known quasars showing strong photometric variability. The Time Domain Spectroscopic Survey of the Sloan Digital Sky Survey IV (SDSS-IV) includes repeat spectroscopy of large numbers of previously known quasars, many selected irrespective of photometric variability, and with spectral epochs separated by months to decades. Our visual examination of these repeat spectra for strong broad line variability yielded 61 newly discovered CLQ candidates. We quantitatively compare spectral epochs to measure changes in continuum and Hβ broad line emission, finding 19 CLQs, of which 15 are newly recognized. The parent sample includes only broad line quasars, so our study tends to find objects that have dimmed, i.e., turn-off CLQs. However, we nevertheless find four turn-on CLQs that meet our criteria, albeit with broad lines in both dim and bright states. We study the response of Hβ and Mg ii emission lines to continuum changes. The Eddington ratios of CLQs are low, and/or their Hβ broad line width is large relative to the overall quasar population. Repeat quasar spectroscopy in the upcoming SDSS-V black hole Mapper program will reveal significant numbers of CLQs, enhancing our understanding of the frequency and duty cycle of such strong variability, and the physics and dynamics of the phenomenon.

79 ASTRONOMY AND ASTROPHYSICS↗
DOE OSTI2022

Resolve the Clear-Sky Continuous Diurnal Cycle of High-Resolution ECOSTRESS Evapotranspiration and Land Surface Temperature

Although numerous evapotranspiration (ET) estimates have been developed from remote sensing measurements, a diurnally resolved product with high spatial resolution (<100 m) is still lacking, which is critically needed for agricultural and ecological monitoring. The ECOsystem Spaceborne Thermal Radiometer Experiment on Space Station (ECOSTRESS) provides, for the first time, ET at 70 m resolution with diurnal samplings, enabling a step toward filling this data gap. This study developed the first framework to resolve the full diurnal cycle of ET and Land Surface Temperature (LST) from sporadic ECOSTRESS measurements. We first constructed 70 m diurnal LST utilizing a diurnal temperature cycle (DTC) model that fused ECOSTRESS and Geostationary Operational Environmental Satellite (GOES) LST. Next, we derived 70 m diurnal ET from the diurnally resolved LST, along with ancillary meteorological and reflectance data sets, using the Priestley-Taylor Jet Propulsion Laboratory (PT-JPL) algorithm. This framework was tested in two regions: A relatively homogeneous semiarid grassland landscape (centered at the US-Seg Ameriflux site) and a heterogeneous humid cropland landscape (centered at US-ARM). Our diurnally resolved LST and ET successfully reproduced the spatial patterns in the native ECOSTRESS measurements during overpasses for both landscapes (LST: r > 0.97, RMSE ~2°C; ET: r > 0.99, and RMSE = 1–7 W m -2 ). The constructed time series also well captured the temporal variability of the in situ measurements (especially under clear-sky conditions) at both US-Seg and US-ARM, albeit performance varying with availability and quality of ECOSTRESS samples. This study sets the stage for testing and applying our framework to broader climates, biomes, and landscapes toward eventually generating diurnally resolved 70 m global operational LST/ET products to enhance ecological/agricultural applications.

54 ENVIRONMENTAL SCIENCES↗
DOE OSTI2022

Peak Rain Rate Sensitivity to Observed Cloud Condensation Nuclei and Turbulence in Continental Warm Shallow Clouds During CACTI

Abstract Warm clouds strongly affect Earth's energy budget but remain imperfectly represented in climate models, partly due to the complexity and covariability of relevant processes influencing warm rain. This work presents a detailed analysis of different factors affecting rain rate peak intensity (RR) in continental warm clouds. Clouds were identified with vertically pointing radar and lidar observations and categorized via a temperature‐based cloud type classification algorithm from which warm clouds were isolated. Observations and retrievals of liquid water path (LWP), cloud condensation nuclei concentration (N CCN ), cloud depth, and cloud duration of more than 3,000 separate warm clouds sampled during the Cloud, Aerosol, and Complex Terrain Interactions (CACTI) field campaign are analyzed in this work. Multiple linear regression (MLR) and random forest (RF) models are applied to assess the relative impact of these variables on RR. Overall, RR tends to increase as cloud depth, LWP, and cloud duration increase, or N CCN decreases. Cloud depth affects RR the most while N CCN impacts it the least. When considering over 170 warm clouds observed at least 1 hr in which in‐cloud turbulence is retrieved, the effect of N CCN on RR remains most likely suppressive, but it is not significant at a 75% level for MLR and is highly uncertain for RF. The impact of in‐cloud turbulence depends on the moment and location it is sampled. Cloud base turbulence around the time of RR suppresses RR, while cloud top turbulence effects are inconclusive. Possible difficulties in isolating robust CCN and turbulence effects on RR are discussed.

54 ENVIRONMENTAL SCIENCES↗
DOE OSTI2015

Investigating wind turbine impacts on near-wake flow using profiling Lidar data and large-eddy simulations with an actuator disk model

In this study, wind turbine impacts on the atmospheric flow are investigated using data from the Crop Wind Energy Experiment (CWEX-11) and large-eddy simulations (LESs) utilizing a generalized actuator disk (GAD) wind turbine model. CWEX-11 employed velocity-azimuth display (VAD) data from two Doppler lidar systems to sample vertical profiles of flow parameters across the rotor depth both upstream and in the wake of an operating 1.5 MW wind turbine. Lidar and surface observations obtained during four days of July 2011 are analyzed to characterize the turbine impacts on wind speed and flow variability, and to examine the sensitivity of these changes to atmospheric stability. Significant velocity deficits (VD) are observed at the downstream location during both convective and stable portions of four diurnal cycles, with large, sustained deficits occurring during stable conditions. Variances of the streamwise velocity component, σ u , likewise show large increases downstream during both stable and unstable conditions, with stable conditions supporting sustained small increases of σ u , while convective conditions featured both larger magnitudes and increased variability, due to the large coherent structures in the background flow. Two representative case studies, one stable and one convective, are simulated using LES with a GAD model at 6 m resolution to evaluate the compatibility of the simulation framework with validation using vertically profiling lidar data in the near wake region. Virtual lidars were employed to sample the simulated flow field in a manner consistent with the VAD technique. Simulations reasonably reproduced aggregated wake VD characteristics, albeit with smaller magnitudes than observed, while σu values in the wake are more significantly underestimated. The results illuminate the limitations of using a GAD in combination with coarse model resolution in the simulation of near wake physics, and validation thereof using VAD data.

17 WIND ENERGY↗
DOE OSTI2020

Chapter 14: Machine Learning of Combustion LES Models from Reacting Direct Numerical Simulation

In this chapter we demonstrate how supervised deep learning techniques can be used to construct models for the filtered progress variable source term necessary for large eddy simulation (LES). The source data for the model is a direct numerical simulation (DNS) of a reacting flow in a low swirl burner configuration. Filtered quantities taken from the DNS data are used to train a deep neural network (DNN)-based model. An efficient data sampling strategy was devised to ensure that a uniform representation of all the states observed in the filtered DNS data are equally present in the training dataset. A-priori testing of the DNN-based model highlights the representative power of DNN to accurately reproduce the filtered reaction progress variable source term over a range of scales and various flame regimes as seen in an industrial burner.

combustion LES models↗
DOE OSTI2018

A stochastic downscaling approach for generating high-frequency solar irradiance scenarios

Solar power is increasingly cost viable with solar photovoltaic (PV) installations becoming commonplace. PV planning and operational studies, however, require high-frequency solar irradiance scenarios to understand potential electric grid impacts due to the variability and uncertainty of the underlying solar resource. Existing remote sensing solar data products are often available over large spatial domains, but are limited in temporal resolution. For example, the global horizontal irradiance (GHI) component contained within the National Solar Radiation Database (NSRDB) is available at time resolution of 30 min on an approximately four-kilometer grid. In contrast, substantial solar variability is present at finer time scales and this article describes an algorithm to stochastically generate one-minute GHI from widely available sub-hourly NSRDB. A generalized linear modeling (GLM) framework is proposed, which includes non-Gaussian mixtures, and extends the literature involving the synthesis of GHI data. The model is trained on a set of sample locations around Oregon, USA, and validated across the USA using both the Surface Radiation Budget Network (SURFRAD) dataset and Solar Radiation Monitoring Laboratory (SRML) network. Simulated ensembles show good coverage properties and temporal correlation structure. The resulting downscaled ensembles allow for understanding the unpredictable variability inherent in GHI at locations without direct measurements. Here, future work can leverage the algorithm as part of a stochastic optimization of electric grid operations with high-penetration PV systems.

14 SOLAR ENERGY↗
DOE OSTI2024

A comparison of model validation approaches for echo state networks using climate model replicates

As global temperatures continue to rise, climate mitigation strategies such as stratospheric aerosol injections (SAI) are increasingly discussed, but the downstream effects of these strategies are not well understood. As such, there is interest in developing statistical methods to quantify the evolution of climate variable relationships during the time period surrounding an SAI. Feature importance applied to echo state network (ESN) models has been proposed as a way to understand the effects of SAI using a data-driven model. This approach depends on the ESN fitting the data well. If not, the feature importance may place importance on features that are not representative of the underlying relationships. Typically, time series prediction models such as ESNs are assessed using out-of-sample performance metrics that divide the times series into separate training and testing sets. However, this model assessment approach is geared towards forecasting applications and not scenarios such as the motivating SAI example where the objective is using a data driven model to capture variable relationships. Here, in this paper, we demonstrate a novel use of climate model replicates to investigate the applicability of the commonly used repeated hold-out model assessment approach for the SAI application. Simulations of an SAI are generated using a simplified climate model, and different initialization conditions are used to provide independent training and testing sets containing the same SAI event. The climate model replicates enable out-of-sample measures of model performance, which are compared to the single time series hold-out validation approach. For our case study, it is found that the repeated hold-out sample performance is comparable, but conservative, to the replicate out-of-sample performance when the training set contains enough time after the aerosol injection.

54 ENVIRONMENTAL SCIENCES↗
DOE OSTI2017

Targeting tumor cells based on Phosphodiesterase 3A expression

Highlights: • PDE3A mRNA overexpression correlates to PDE3A protein overexpression. • High levels of PDE3A protein can be detected in clinical samples using IHC. • Primary cells from subgroup of patients with high PDE3A respond to PDE3 inhibitors. • PDE3A is overexpressed in subsets of patient derived samples from various tumors. • Gastrointestinal stromal tumors show the highest expression of PDE3A. We and others have previously reported a correlation between high phosphodiesterase 3 A (PDE3A) expression and selective sensitivity to phosphodiesterase (PDE) inhibitors. This indicates that PDE3A could serve both as a drug target and a biomarker of sensitivity to PDE3 inhibition. In this report, we explored publicly available mRNA gene expression data to identify cell lines with different PDE3A expression. Cell lines with high PDE3A expression showed marked in vitro sensitivity to PDE inhibitors zardaverine and quazinone, when compared with those having low PDE3A expression. Immunofluorescence and immunohistochemical stainings were in agreement with PDE3A mRNA expression, providing suitable alternatives for biomarker analysis of clinical tissue specimens. Moreover, we here demonstrate that tumor cells from patients with ovarian carcinoma show great variability in PDE3A protein expression and that level of PDE3A expression is correlated with sensitivity to PDE inhibition. Finally, we demonstrate that PDE3A is highly expressed in subsets of patient tumor cell samples from different solid cancer diagnoses and expressed at exceptional levels in gastrointestinal stromal tumor (GIST) specimens. Importantly, vulnerability to PDE3 inhibitors has recently been associated with co-expression of PDE3A and Schlafen family member 12 (SLFN12). We here demonstrate that high expression of PDE3A in clinical specimens, at least on the mRNA level, seems to be frequently associated with high SLFN12 expression. In conclusion, PDE3A seems to be both a promising biomarker and drug target for individualized drug treatment of various cancers.

60 APPLIED LIFE SCIENCES↗
DOE OSTI2018

Transmission in situ and operando high temperature X-ray powder diffraction in variable gaseous environments

In this paper, we describe a device for time-resolved synchrotron-based in situ and operando X-ray powder diffraction measurements at elevated temperatures under controllable gaseous environments. The respective gaseous sample environment is realized via a gas-tight capillary-in-capillary design, where the gas flow is achieved through an open-end 0.5 mm capillary located inside a 0.7 mm capillary filled with a sample powder. Thermal mass flow controllers provide appropriate gas flows and computer-controlled on-the-fly gas mixing capabilities. The capillary system is centered inside an infrared heated, proportional integral differential-controlled capillary furnace allowing access to temperatures up to 1000 °C.

36 MATERIALS SCIENCE↗
DOE OSTI2013

A retrospective comparison of smart prep and test bolus multi-detector CT pulmonary angiography protocols

Optimal arterial opacification is crucial in imaging the pulmonary arteries using computed tomography (CT). This poses the challenge of precisely timing data acquisition to coincide with the transit of the contrast bolus through the pulmonary vasculature. The aim of this quality assurance exercise was to investigate if a change in CT pulmonary angiography (CTPA) scanning protocol resulted in improved opacification of the pulmonary arteries. Comparison was made between the smart prep protocol (SPP) and the test bolus protocol (TBP) for opacification in the pulmonary trunk. A total of 160 CTPA examinations (80 using each protocol) performed between January 2010 and February 2011 were assessed retrospectively. CT attenuation coefficients were measured in Hounsfield Units (HU) using regions of interest at the level of the pulmonary trunk. The average pixel value, standard deviation (SD), maximum, and minimum were recorded. For each of these variables a mean value was then calculated and compared for these two CTPA protocols. Minimum opacification of 200 HU was achieved in 98% of the TBP sample but only 90% of the SPP sample. The average CT attenuation over the pulmonary trunk for the SPP was 329 (SD = ±21) HU, whereas for the TBP it was 396 (SD = ±22) HU (P = 0.0017). The TBP also recorded higher maximum (P = 0.0024) and minimum (P = 0.0039) levels of opacification. This study has found that a TBP resulted in significantly better opacification of the pulmonary trunk than the SPP.

62 RADIOLOGY AND NUCLEAR MEDICINE↗
DOE OSTI2024

In situ aerobic methane oxidation rates in a stratified lake

Abstract Microbial aerobic methane oxidation is an important sink for aquatic methane worldwide. Despite its importance to global methane fluxes, few aerobic methane oxidation rates have been obtained in freshwater or marine environments without imposing changes to the microbial community through use of ex situ methods. A novel in situ incubation method for continuous time‐series measurements was used in Jordan Lake, North Carolina, during 2020–2021, to determine reaction kinetics for aerobic methane oxidation rates across a wide range of naturally varying methane (55–1833 nM) and dissolved oxygen (DO; 28–366 μM) concentrations and temperatures (17–30°C). Methane oxidation began immediately at the start of each of 21 incubations and methane oxidation rates were 1 st order with respect to methane. The data density allowed for accurate calculation of 1 st ‐order rate constants,k, that ranged from 0.018 to 0.462 h −1 (R 2 > 0.967). Addition of ammonium (20–45 μM) to natural concentrations ranging from 0.057 to 2.4 μM did not change aerobic methane oxidation rate kinetics, suggesting that the natural population of aerobic methane oxidizers in this eutrophic lake was not nitrogen limited. Values ofkinversely correlated most strongly with initial DO concentrations (R 2 = 0.82) rather than temperature. Values forkincreased with Julian day throughout our sampling period, suggesting seasonal influences on methane oxidation via responses to geochemical changes or shifts in microbial community abundance and composition. These experiments demonstrate a high variability in the enzymatic capacity for 1 st ‐order methane oxidation rates in this eutrophic lake that is tightly and inversely coupled to oxygen concentrations. Measurements of in situ aerobic methane oxidation rate constants allow for the direct quantification and modeling of the microbial community's capacity for methane oxidation over a wide range of natural methane concentrations.

Marine & Freshwater Biology↗
DOE OSTI2025

Karhunen–Loève deep learning method for surrogate modeling and approximate Bayesian parameter estimation

We evaluate the performance of the Karhunen-Loève Deep Neural Network (KL-DNN) framework for surrogate modeling and approximate Bayesian parameter estimation in partial differential equation models. In the surrogate model, the Karhunen-Loève (KL) expansions are used for the dimensionality reduction of the number of unknown parameters and variables, and a deep neural network is employed to relate the reduced space of parameters to that of the state variables. The KL-DNN surrogate model is used to formulate a maximum-a-posteriori-like least-squares problem, which is randomized to draw samples of the posterior distribution of the parameters. We test the proposed framework for a hypothetical unconfined aquifer via comparison with the forward MODFLOW and inverse PEST++ iterative ensemble smoother (IES) solutions as well as the state-of-the-art Fourier neural operator (FNO) and deep operator networks (DeepONets) operator learning surrogate models. Our results show that the KL-DNN surrogate model outperforms FNO and DeepONet for forward predictions. For solving inverse problems, the randomized algorithm provides the same or more accurate Bayesian predictions of the parameters than IES as evidenced by the higher log-predictive probability of both the estimated parameter field and the forecast hydraulic head. The posterior mean obtained from the randomized algorithm is closer to the reference parameter field than that obtained with FNO as the maximum a posteriori estimate.

Approximate Bayesian inference↗