Engineering Papers⌕ Search

SEARCH · Engineering Papers

Results for “collocated”

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 397 records · Page 22
All sourcesNASA NTRSDOE OSTI
DOE OSTI2023

Joint Optimization for Transport and Bucket Loading Phases of Automated Wheel Loaders

This article investigates optimization of fuel-efficiency and productivity for automated wheel loaders. A control-oriented model for both the transport phase and bucket loading phase is proposed. Here, the vehicle model includes an automatic gear shift schedule that can be incorporated into the optimization problem. Based on the model, the multistage optimization problem is formulated to simultaneously consider all phases of a short cycle with physical constraints. Cycle time and fuel efficiency are used as the weighted performance indexes in a multiobjective cost function. Bucket fill factor is included as a constraint during the bucket loading phase. A nonlinear programming problem is created with collocation using MATLAB and CasADi. The optimization solver IPOPT solves the problem to obtain the optimal state and control trajectories, which can be used as a reference for automated wheel loaders or even as a driver advisory for human-driven wheel loaders.

42 ENGINEERING↗
DOE OSTI2021

Geostatistical Mapping of Salinity Conditioned on Borehole Logs, Montebello Oil Field, California

We present a geostatistics-based stochastic salinity estimation framework for the Montebello Oil Field that capitalizes on available total dissolved solids (TDS) data from groundwater samples as well as electrical resistivity (ER) data from borehole logging. Data from TDS samples (n = 4924) was coded into an indicator framework based on falling below four selected thresholds (500, 1000, 3000, and 10,000 mg/L). Collocated TDS-ER data from the surrounding groundwater basin were then employed to produce a kernel density estimator to establish conditional probabilities for ER data (n = 8 boreholes) falling below the selected TDS thresholds within the Montebello Oil Field area. Directional variograms were estimated from these indicator coded data, and 500 TDS realizations from conditional indicator simulation were generated for the subsurface region above the Montebello Oil Field reservoir. Simulations were summarized as 3D maps of median TDS, most likely salinity class, and probability for exceeding each of the specified TDS thresholds. Results suggested TDS was below 500 mg/L in most of the study area, with a trend toward higher values (500 to 1000 mg/L) to the southwest; consistent with the average regional groundwater flow direction. Discrete localized zones of TDS greater than 1000 mg/L were observed, with one of these zones in the greater than 10,000 mg/L range; however, these areas were not prevalent. The probabilistic approach used here is adaptable and is readily modified to include additional data and types and can be employed in time-lapse salinity modeling through Bayesian updating.

54 ENVIRONMENTAL SCIENCES↗
DOE OSTI2021

Whole-lithosphere shear during oblique rifting

Processes controlling the formation of continental whole-lithosphere shear zones are debated, but their existence requires that the lithosphere is mechanically coupled from base to top. We document the formation of a dextral, whole-lithosphere shear zone in the Death Valley region (DVR), southwest United States. Dextral deflections of depth gradients in the lithosphere-asthenosphere boundary and Moho are stacked vertically, defining a 20–50-km-wide, lower lithospheric shear zone with ~60 km of shear. These deflections underlie an upper-crustal fault zone that accrued ~60 km of dextral slip since ca. 8–7 Ma, when we infer that whole-lithosphere shear began. This dextral offset is less than net dextral offset on the upper-crustal fault zone (~90 km, ca. 13–0 Ma) and total upper-crustal extension (~250 km, ca. 16–0 Ma). We show that, before ca. 8–7 Ma, weak middle crust decoupled upper-crustal deformation from deformation in the lower crust and mantle lithosphere. Between 16 and 7 Ma, detachment slip thinned, uplifted, cooled, and thus strengthened the middle crust, which is exposed in metamorphic core complexes collocated with the whole-lithosphere shear zone. Midcrustal strengthening coupled the layered lithosphere vertically and therefore enabled whole-lithosphere dextral shear. Where thick crust exists (as in pre–16 Ma DVR), midcrustal strengthening is probably a necessary condition for whole-lithosphere shear.

58 GEOSCIENCES↗
DOE OSTI2020

Low-Memory, Discrete Ordinates, Discontinuous Galerkin Methods for Radiative Transport

The discrete ordinates discontinuous Galerkin (SN-DG) method is a well-established and practical approach for solving the radiative transport equation. In this paper, we study a low-memory variation of the upwind SN-DG method. The proposed method uses a smaller finite element space that is constructed by coupling spatial unknowns across collocation angles, thereby yielding an approximation with fewer degrees of freedom than the standard method. Like the original SN-DG method, the low-memory variation still preserves the asymptotic diffusion limit and maintains the characteristic structure needed for mesh sweeping algorithms. While we observe second-order convergence in the scattering dominated, diffusive regime, the low-memory method is in general only first-order accurate. To address this issue, we use upwind reconstruction to recover second-order accuracy. Finally, for both methods, numerical procedures based on upwind sweeps are proposed to reduce the system dimension in the underlying Krylov solver strategy.

97 MATHEMATICS AND COMPUTING↗
DOE OSTI2024

S-OPT: A Points Selection Algorithm for Hyper-Reduction in Reduced Order Models

While projection-based reduced order models can reduce the dimension of full order solutions, the resulting reduced models may still contain terms that scale with the full order dimension. Hyper-reduction techniques are sampling-based methods that further reduce this computational complexity by approximating such terms with a much smaller dimension. The goal of this work is to introduce the points selection algorithm developed by Shin and Xiu as a hyper-reduction method. The selection algorithm was originally proposed as a stochastic collocation method for uncertainty quantification. Since the algorithm aims at maximizing a quantity $\mathcal{S}$ that measures both the column orthogonality and the determinant, we refer to the algorithm as S-OPT. Numerical examples are provided to demonstrate the performance of S-OPT and to compare its performance with a gappy proper orthogonal decomposition (POD) algorithm. Here, we found that using the S-OPT algorithm is shown to predict the full order solutions with higher accuracy than gappy POD especially when the number of sampling points is small, although we note that S-OPT shows slow asymptotic convergence with respect to the number of samples for some applications, e.g., Lagrangian hydrodynamics.

97 MATHEMATICS AND COMPUTING↗
DOE OSTI2023

Efficient Reformulation of Linear and Nonlinear Solid-Phase Diffusion in Lithium-ion Battery Models using Symmetric Polynomials: Mass Conservation and Computational Efficiency

Lithium-ion batteries are typically modeled using porous electrode theory coupled with various transport and reaction mechanisms, along with suitable discretization or approximations for the solid-phase diffusion equation. The solid-phase diffusion equation represents the main computational burden for typical pseudo-2-dimensional (p2D) models since these equations in the pseudo r -dimension must be solved at each point in the computational grid. This substantially increases the complexity of the model as well as the computational time. Traditional approaches towards simplifying solid-phase diffusion possess certain significant limitations, especially in modeling emerging electrode materials which involve phase changes and variable diffusivities. A computationally efficient representation for solid-phase diffusion is discussed in this paper based on symmetric polynomials using Orthogonal Collocation and Galerkin formulation (weak form). A systematic approach is provided to increase the accuracy of the approximation (p form in finite element methods) to enable efficient simulation with a minimal number of semi-discretized equations, ensuring mass conservation even for non-linear diffusion problems involving variable diffusivities. These methods are then demonstrated by incorporation into the full p2D model, illustrating their advantages in simulating high C-rates and short-time dynamic operation of Lithium-ion batteries.

25 ENERGY STORAGE↗
DOE OSTI2024

Development of a Full-Scale Connected U-Net for Reflectivity Inpainting in Spaceborne Radar Blind Zones

CloudSat’s Cloud Profiling Radar is a valuable tool for remotely monitoring high-latitude snowfall, but its ability to observe hydrometeor activity near the Earth’s surface is limited by a radar blind zone caused by ground clutter contamination. This study presents the development of a deeply supervised U-Net-style convolutional neural network to predict cold season reflectivity profiles within the blind zone at two Arctic locations. The network learns to predict the presence and intensity of near-surface hydrometeors by coupling latent features encoded in blind zone-aloft clouds with additional context from collocated atmospheric state variables (i.e., temperature, specific humidity, and wind speed). Results show that the U-Net predictions outperform traditional linear extrapolation methods, with low mean absolute error, a 38% higher Sørensen–Dice coefficient, and vertical reflectivity distributions 60% closer to observed values. The U-Net is also able to detect the presence of near-surface cloud with a critical success index (CSI) of 72% and cases of shallow cumuliform snowfall and virga with 18% higher CSI values compared to linear methods. An explainability analysis shows that reflectivity information throughout the scene, especially at cloud edges and at the 1.2-km blind zone threshold, along with atmospheric state variables near the tropopause, are the most significant contributors to model skill. This surface-trained generative inpainting technique has the potential to enhance current and future remote sensing precipitation missions by providing a better understanding of the nonlinear relationship between blind zone reflectivity values and the surrounding atmospheric state.

54 ENVIRONMENTAL SCIENCES↗
DOE OSTI2021

Observations of Aerosol, Cloud, Turbulence, and Radiation Properties at the Top of the Marine Boundary Layer over the Eastern North Atlantic Ocean: The ACORES Campaign

We report on the Azores Stratocumulus Measurements of Radiation, Turbulence and Aerosols (ACORES) campaign, which took place around Graciosa and Pico Islands/Azores in July 2017. The main objective was to investigate the vertical distribution of aerosol particles, stratocumulus microphysical and radiative properties, and turbulence parameters in the eastern North Atlantic. The vertical exchange of mass, momentum, and energy between the free troposphere (FT) and the cloudy marine boundary layer (MBL) was explored over a range of scales from submeters to kilometers. To cover these spatial scales with appropriate measurements, helicopter-borne observations with unprecedented high resolution were realized using the Airborne Cloud Turbulence Observation System (ACTOS) and Spectral Modular Airborne Radiation Measurement System–Helicopter-Borne Observations (SMART-HELIOS) instrumental payloads. Here, the helicopter-borne observations were combined with ground-based aerosol measurements collected at two continuously running field stations on Pico Mountain (2,225 m above sea level, in the FT), and at the Atmospheric Radiation Measurement (ARM) station on Graciosa (at sea level). First findings from the ACORES observations we are discussing in the paper are as follows: (i) we have observed a high variability of the turbulent cloud-top structure on horizontal scales below 100 m with local temperature gradients of up to 4 K over less than 1 m vertical distance, (ii) we have collected strictly collocated radiation measurements supporting the relevance of small-scale processes by revealing significant inhomogeneities in cloud-top brightness temperature to scales well below 100 m, and (iii) we have concluded that aerosol properties are completely different in the MBL and FT with often-complex stratification and frequently observed burst-like new particle formation.

54 ENVIRONMENTAL SCIENCES↗
DOE OSTI2021

Aerosol–Ice Formation Closure: A Southern Great Plains Field Campaign

Prediction of ice formation in clouds presents one of the grand challenges in the atmospheric sciences. Immersion freezing initiated by ice-nucleating particles (INPs) is the dominant pathway of primary ice crystal formation in mixed-phase clouds, where supercooled water droplets and ice crystals coexist, with important implications for the hydrological cycle and climate. However, derivation of INP number concentrations from an ambient aerosol population in cloud-resolving and climate models remains highly uncertain. We conducted an aerosol–ice formation closure pilot study using a field-observational approach to evaluate the predictive capability of immersion freezing INPs. The closure study relies on collocated measurements of the ambient size-resolved and single-particle composition and INP number concentrations. The acquired particle data serve as input in several immersion freezing parameterizations, which are employed in cloud-resolving and climate models, for prediction of INP number concentrations. We discuss in detail one closure case study in which a front passed through the measurement site, resulting in a change of ambient particle and INP populations. We achieved closure in some circumstances within uncertainties, but we emphasize the need for freezing parameterization of potentially missing INP types and evaluation of the choice of parameterization to be employed. Overall, this closure pilot study aims to assess the level of parameter details and measurement strategies needed to achieve aerosol–ice formation closure. Furthermore, the closure approach is designed to accurately guide immersion freezing schemes in models, and ultimately identify the leading causes for climate model bias in INP predictions.

58 GEOSCIENCES↗
DOE OSTI2022

The Response of the Polar Vortex to Tropospheric Temperature Eddies in an Idealized General Circulation Model

Abstract A dry-core idealized general circulation model with a stratospheric polar vortex in the Northern Hemisphere is run with a combination of simplified topography and imposed tropospheric temperature perturbations, each located in the Northern Hemisphere with a zonal wavenumber of 1. The phase difference between the imposed temperature wave and the topography is varied to understand what effect this has on the occurrence of polar vortex displacements. Geometric moments are used to identify the centroid of the polar vortex for the purposes of classifying whether or not the polar vortex is displaced. Displacements of the polar vortex are a response to increased tropospheric wave activity. Compared to a model run with only topography, the likelihood of the polar vortex being displaced increases when the warm region is located west of the topography peak, and decreases when the cold region is west of the topography peak. This response from the polar vortex is due to the modulation of vertically propagating wave activity by the temperature forcing. When the southerly winds on the western side of the topographically forced anticyclone are collocated with warm- or cold-temperature forcing, the vertical wave activity flux in the troposphere becomes more positive or negative, respectively. This is in line with recent reanalysis studies that showed that anomalous warming west of the surface pressure high, in the climatological standing wave, precedes polar vortex disturbances.

54 ENVIRONMENTAL SCIENCES↗
DOE OSTI2024

Droplet Growth or Evaporation Does Not Buffer the Variability in Supersaturation in Clean Clouds

Abstract Water vapor supersaturation in clouds is a random variable that drives activation and growth of cloud droplets. The Pi Convection–Cloud Chamber generates a turbulent cloud with a microphysical steady state that can be varied from clean to polluted by adjusting the aerosol injection rate. The supersaturation distribution and its moments, e.g., mean and variance, are investigated for varying cloud microphysical conditions. High-speed and collocated Eulerian measurements of temperature and water vapor concentration are combined to obtain the temporally resolved supersaturation distribution. This allows quantification of the contributions of variances and covariances between water vapor and temperature. Results are consistent with expectations for a convection chamber, with strong correlation between water vapor and temperature; departures from ideal behavior can be explained as resulting from dry regions on the warm boundary, analogous to entrainment. The saturation ratio distribution is measured under conditions that show monotonic increase of liquid water content and decrease of mean droplet diameter with increasing aerosol injection rate. The change in liquid water content is proportional to the change in water vapor concentration between no-cloud and cloudy conditions. Variability in the supersaturation remains even after cloud droplets are formed, and no significant buffering is observed. Results are interpreted in terms of a cloud microphysical Damköhler number (Da), under conditions corresponding to , i.e., the slow-microphysics regime. This implies that clouds with very clean regions, such that is satisfied, will experience supersaturation fluctuations without them being buffered by cloud droplet growth. Significance Statement The saturation ratio (humidity) in clouds controls the growth rate and formation of cloud droplets. When air in a turbulent cloud mixes, the humidity varies in space and time throughout the cloud. This is important because it means cloud droplets experience different growth histories, thereby resulting in broader size distributions. It is often assumed that growth and evaporation of cloud droplets buffers out some of the humidity variations. Measuring these variations has been difficult, especially in the field. The purpose of this study is to measure the saturation ratio distribution in clouds with a range of conditions. We measure the in-cloud saturation ratio using a convection cloud chamber with clean to polluted cloud properties. We found in clouds with low concentrations of droplets that the variations in the saturation ratio are not suppressed.

54 ENVIRONMENTAL SCIENCES↗
DOE OSTI2020

The Modulation of Gulf Stream Influence on the Troposphere by the Eddy-Driven Jet

This study suggests that the Gulf Stream influence on the wintertime North Atlantic troposphere is most pronounced when the eddy-driven jet (EDJ) is farthest south and better collocated with the Gulf Stream. Using the reanalysis dataset NCEP-CFSR for December–February 1979–2009, the daily EDJ latitude is separated into three regimes (northern, central, and southern). It is found that the average trajectory of atmospheric fronts covaries with EDJ latitude. In the southern EDJ regime (~19% of the time), the frequency of near-surface atmospheric fronts that pass across the Gulf Stream is maximized. Analysis suggests that this leads to significant strengthening in near-surface atmospheric frontal convergence resulting from strong air–sea sensible heat flux gradients (due to strong temperature gradients in the atmosphere and ocean). In recent studies, it was shown that the pronounced band of time-mean near-surface wind convergence across the Gulf Stream is set by atmospheric fronts. Here, it is shown that an even smaller subset of atmospheric fronts—those associated with a southern EDJ—primarily sets the time mean, due to enhanced Gulf Stream air–sea interaction. Furthermore, statistically significant anomalies in vertical velocity extending well above the boundary layer are identified in association with changes in EDJ latitude. These anomalies are particularly strong for a southern EDJ and are spatially consistent with increases in near-surface atmospheric frontal convergence over the Gulf Stream. These results imply that much of the Gulf Stream influence on the time-mean atmosphere is modulated on synoptic time scales, and enhanced when the EDJ is farthest south.

54 ENVIRONMENTAL SCIENCES↗
DOE OSTI2020

Columnar Vertical Profile (CVP) Methodology for Validating Polarimetric Radar Retrievals in Ice Using In Situ Aircraft Measurements

Abstract A novel way to process polarimetric radar data collected via plan position indicator (PPI) scans and display those data in a time–height format is introduced. The columnar vertical profile (CVP) methodology uses radar data collected via multiple elevation scans, limited to data within a set region in range and azimuth relative to the radar, to create vertical profiles of polarimetric radar data representative of that limited region in space. This technique is compared to others existing in the literature, and various applications are discussed. Polarimetric ice microphysical retrievals are performed on CVPs created within the stratiform rain region of two mesoscale convective systems sampled during two field campaigns, where CVPs follow the track of research aircraft. Aircraft in situ data are collocated to microphysical retrieval data, and the accuracy of these retrievals is tested against other retrieval techniques in the literature.

Engineering↗
DOE OSTI2025

Estimating Ice Water Content for Winter Storms from Millimeter-Wavelength Radar Measurements Using a Synthesis of Polarimetric and Dual-Frequency Radar Observations

The potential of millimeter-wavelength radar-based ice water content (IWC) estimation is demonstrated using a Ka-band Scanning Polarimetric Radar (KASPR) for the U.S. northeast coast winter storms. Two IWC relations for Ka-band polarimetric radar measurements are proposed: one that uses a combination of the radar reflectivity Z and the estimated total number concentration of snow particles N t and the other based on the joint use of Z, specific differential phase K DP , and the degree of riming f rim . A key element of the algorithms is to obtain the “Rayleigh-equivalent” value of Z measured at the Ka band, i.e., the corresponding Z at a longer radar wavelength for which Rayleigh scattering takes place. This is achieved via polarimetric retrieval of the mean volume diameter D m and incorporating the relationship between the dual-wavelength ratio DWR S/Ka and D m . Those techniques allow for retrievals from single millimeter-wavelength radar measurements and do not necessarily require the dual-wavelength ratio (DWR) measurements, if the DWR–D m relation and Rayleigh assumption for Ka-band K DP are valid. Comparison between the quasivertical profile product obtained from KASPR and the columnar vertical profile product generated from the nearby WSR-88D S-band radar measurements demonstrates that the DWR S/Ka can be estimated from the two close radars without the need for collocated radar beams and synchronized antenna scanning and can be used for determining the Rayleigh-equivalent value of Z. As a result, the performance of the suggested techniques is evaluated for seven winter storms using surface disdrometer and snow accumulation measurements.

Cloud retrieval↗
DOE OSTI2021

A Bimodal Diagnostic Cloud Fraction Parameterization. Part I: Motivating Analysis and Scheme Description

Cloud fraction parameterizations are beneficial to regional, convection-permitting numerical weather prediction. For its operational regional midlatitude forecasts, the Met Office uses a diagnostic cloud fraction scheme that relies on a unimodal, symmetric subgrid saturation-departure distribution. This scheme has been shown before to underestimate cloud cover and hence an empirically based bias correction is used operationally to improve performance. This first of a series of two papers proposes a new diagnostic cloud scheme as a more physically based alternative to the operational bias correction. The new cloud scheme identifies entrainment zones associated with strong temperature inversions. Additionally, for model grid boxes located in this entrainment zone, collocated moist and dry Gaussian modes are used to represent the subgrid conditions. The mean and width of the Gaussian modes, inferred from the turbulent characteristics, are then used to diagnose cloud water content and cloud fraction. It is shown that the new scheme diagnoses enhanced cloud cover for a given gridbox mean humidity, similar to the current operational approach. It does so, however, in a physically meaningful way. Using observed aircraft data and ground-based retrievals over the southern Great Plains in the United States, it is shown that the new scheme improves the relation between cloud fraction, relative humidity, and liquid water content. An emergent property of the scheme is its ability to infer skewed and bimodal distributions from the large-scale state that qualitatively compare well against observations. A detailed evaluation and resolution sensitivity study will follow in Part II.

54 ENVIRONMENTAL SCIENCES↗
DOE OSTI2022

Deep Convection Initiation, Growth, and Environments in the Complex Terrain of Central Argentina during CACTI

This study characterizes the wide range of deep convective cloud life cycles and their relationships with ambient environments observed during the Cloud, Aerosol, and Complex Terrain Interactions (CACTI) field campaign near the Sierras de Córdoba (SDC) range in central Argentina. We develop a novel convective cell tracking database for the entire field campaign using C-band polarimetric radar observations. The cell tracking database includes timing, location, area, depth, merge/split information, microphysical properties, collocated satellite-retrieved cloud properties, and sounding-derived environmental conditions. Results show that the SDC exerts a strong control on convection initiation (CI) and growth. CI preferentially occurs east of the SDC ridge during the afternoon, and cells often undergo upscale growth through the evening as they travel eastward toward the plains. Larger and more intense cells tend to occur in more unstable and humid low-level environments, and surface-based cells are stronger than elevated cells. Midtropospheric relative humidity and vertical wind shear also jointly affect the size and depth of the cells. Rapid cell area growth rates exhibit dependence on both large environmental wind shear and low-level moisture. Evolution of convective cell macro- and microphysical properties are strongly influenced by convective available potential energy and low-level humidity, as well as the presence of other cells in their vicinity. This cell tracking database demonstrates a framework that ties measurements from various platforms centering around convective life cycles to facilitate process understanding of factors that control convective evolution.

54 ENVIRONMENTAL SCIENCES↗
DOE OSTI2024

Temporal and spatial characterization of a thermogenic, fault-controlled gas hydrate system, Woolsey Mound, Gulf of Mexico

Woolsey Mound, located at Mississippi Canyon Lease Block 118 (MC118), is the site of the Gulf of Mexico hydrate research consortium’s seafloor observatory, where gas hydrates outcrop at the seafloor. The presence of gas hydrates in the mound is confirmed directly by coring and indirectly by 3D seismic reflection data. Craters, pockmarks, chemosynthetic communities, and authigenic carbonates populate the seafloor at Woolsey Mound. Each crater is characterized by a network of shallow crestal faults that connect the hydrate mound to the underlying allochthonous salt body. We characterize the temporal and spatial evolution of gas hydrates at Woolsey Mound under natural perturbations using four collocated 3D seismic reflection data sets that span over 14 years. Data acquisition differences embedded in the data sets arising from variation in geometry, sample rate, and phase are minimized using the “cross-equalization” method. Our results indicate that hydrate formation and dissociation vary temporally and spatially in close connection to the shallow crestal faults. Evidence of gas hydrate dissociation is observed over a period of three years (2000–2003), where major dissociation occurred along the southern portion of the crestal fault in the southeast crater. The dissociation is less prominent in the southwest crater. Evidence of methane venting is observed between 2000 and 2010, which is mostly concentrated in the southeast crater. The residual amplitude anomalies observed between 2000 and 2014 in the mound are mostly positive, implying that the methane venting had increased significantly. The positive anomalies are correlated with the methane seepage recorded in 2011. Our results indicate the evolution of a fault-controlled gas hydrate system in the northern Gulf of Mexico, which would aid in assessing its impact on the seafloor.

Geochemistry & Geophysics↗
DOE OSTI2022

Profile Images and Annotations for Vehicle Re-identification Algorithms (PRIMAVERA)

This dataset contains 636,246 profile images of vehicles representing 13,963 unique vehicles. The data was collected by a set of roadside sensors over the course of three years. Each time a vehicle passed by one of the sensors, a series of images was collected. The images were processed to detect and localize each vehicle, and a license plate reader collocated with the sensor was used to provide a unique ID for the vehicle. Actual license plate numbers have been obfuscated by replacing with an arbitrary numerical ID for each vehicle. After localizing the vehicle in each image, the original RGB image was rotated, scaled, and shifted to produce a new RGB image of size 234x234 pixels such that the outermost two wheels are located at predetermined pixel locations in the image. In this way, all vehicle images are aligned to one another. This registration process occasionally results in a portion of certain vehicles being cutoff at the edges of the image. The dataset has been partitioned into two sets called training and validation. The two partitions no common vehicles, i.e., a vehicle present in one partition is guaranteed not to be present in the other. In this way, an algorithm can be validated against a set of new vehicles that were not seen during the training process. The training set contains 543,926 images from 64,440 vehicle passes representing 11,918 unique vehicles, while the validation set contains 92,320 images from 10,991 vehicle passes representing 2,045 unique vehicles. Vehicle images are organized by directories corresponding to unique vehicles. The file naming scheme is as follows: veh_{vehID}_tr_{passID}_{frameID}_{elevation}_{timeofday}.jpg where {vehID} is the vehicle ID (unique across the entire dataset), {passID} is an identifier for each tracked vehicle pass (unique across the entire dataset), {frameID} is the index of the frame within the given vehicle pass starting at 0, {elevation} is a two-letter string indicating whether the sensor was elevated (el) or at ground-level (gl), and {timeofday} is a two-letter string indicating whether the image was captured during daytime (dt) or nighttime (nt).

image↗