Search NASA⌕ Search

SEARCH · Search NASA

Results for “value of data”

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 883 records · Page 49
All sourcesNASA NTRSDOE OSTI
NASA NTRS2000

Case Study Analyses of the SUCCESS DC-8 Scanning Lidar Database

Under project SUCCESS (Subsonic Aircraft Contrail and Cloud Effects Special Study) funded by the Atmospheric Effects of Aviation Program, SRI International (SRI) developed an angular scanning backscatter lidar for operation on the NASA DC-8 research aircraft and deployed the scanning lidar during the SUCCESS field campaign. The primary purpose of the lidar was to generate real-time video displays of clouds and contrails above, ahead of, and below the DC-8 as a means to help position the aircraft for optimum cloud and contrail sampling by onboard in situ sensors, and to help extend the geometrical domain of the in situ sampling records. A large, relatively complex lidar database was collected and several data examples were processed to illustrate the value of the lidar data for interpreting the other data records collected during SUCCESS. These data examples were used to develop a journal publication for the special SUCCESS Geophysical Research Letters issue. The data examples justified data analyses of a larger part of the DC-8 lidar database and is the objective of the current study. Efficient processing of the SUCCESS DC-8 scanning lidar database required substantial effort to enhance hardware and software components of the data system that was used for the initial analyses. MATLAB instructions are used to generate altitude and distance color-coded lidar displays corrected for effects introduced by aircraft pitch and forward movement during an angular scan time interval. Onboard in situ sensor atmospheric measurements are propagated to distances ahead of the DC-8 using recorded aircraft velocity so that they can be plotted on the lidar displays for comparison with lidar remotely observed aerosol distributions. Resulting lidar and in situ sensor polar scan displays over extended sampling intervals are integrated into a time series movie format for 36 case studies. Contrails and clouds were detected to ranges of 15 km by the forward-viewing angular scanning lidar and were progressively mapped as the aircraft approached and penetrated them. Near aircraft lidar observations were much better correlated with in situ sensor observations than lidar observations at greater distances ahead of the aircraft. The major cause of this difference was thought to be the about 2 deg. offset of the lidar viewing direction from the flight direction. Contrail spatial distributions were not of the quality obtainable from ground-based lidar observations. This results because contrails tend to become horizontally stratified, vertical distance between angular lidar observations increases with increased distance from the aircraft, and erratic aircraft motions during an angular scan. The most useful lidar observations were made with lidar viewing directions of vertically upward or vertically downward. These provided real-time information on aircraft altitudes to achieve optimum in situ cloud and contrail sampling. At sampling altitudes, the forward viewing angular scanning observations were useful for fine-tuning the aircraft altitude for cloud and contrail penetration. Best information on cloud and contrail properties were obtained from vertically directed lidar observations as the aircraft performed a series of upward and downward penetrations of contrails. This operational mode was especially well suited for lidar and radiometric evaluation of cloud and contrail optical and radiative properties. The vertical viewing lidar detected ice crystals thought to be precipitating from an aircraft contrail and their scavenging by a cirrus cloud layer. The lidar display indicates that the crystals are effective for increasing cirrus cloud density. Vertical angular scanning observations can evaluate the sharp decrease in lidar backscatter for small off-vertical viewing directions that result from horizontally aligned ice crystals and perhaps can provide additional information on crystal shapes. The about 2 deg. offset of the lidar viewing direction from the flight direction is thought to have greatly degraded the forward-viewing angular scanning observations and this mode of operation was not fully evaluated. However, the reasoning for this capability remains valid and the angular scan presentations collected during this program justifies modification of the lidar pod for true forward direction lidar viewing during future cloud and contrail studies.

Uthe, Edward E.↗
NASA NTRS2002

Maximum Likelihood Estimation of Spectra Information from Multiple Independent Astrophysics Data Sets

The Maximum Likelihood (ML) statistical theory required to estimate spectra information from an arbitrary number of astrophysics data sets produced by vastly different science instruments is developed in this paper. This theory and its successful implementation will facilitate the interpretation of spectral information from multiple astrophysics missions and thereby permit the derivation of superior spectral information based on the combination of data sets. The procedure is of significant value to both existing data sets and those to be produced by future astrophysics missions consisting of two or more detectors by allowing instrument developers to optimize each detector's design parameters through simulation studies in order to design and build complementary detectors that will maximize the precision with which the science objectives may be obtained. The benefits of this ML theory and its application is measured in terms of the reduction of the statistical errors (standard deviations) of the spectra information using the multiple data sets in concert as compared to the statistical errors of the spectra information when the data sets are considered separately, as well as any biases resulting from poor statistics in one or more of the individual data sets that might be reduced when the data sets are combined.

Howell, Leonard W., Jr.↗
NASA NTRS2022

A Robust Machine Learning Schema for Developing, Maintaining, and Disseminating Machine Learning Models

Recent advances in the development of machine learning (ML) algorithms have enabled the creation of predictive models that can improve decision making, decrease computational cost, and improve efficiency in a variety of fields. As an organization begins to develop and implement such models, the data used in the training, validation, and testing of ML models, the model parameters, and the use cases or limitations of the models must be properly stored to ensure models are both fully traceable and used correctly. In the context of predicting material behavior, advances in computationally intense, physics-based modeling of material behavior at various length scales and the emergence of Integrated Computational Materials Engineering (ICME) have driven the need for developing data-driven surrogate models of the physics-based simulation tools using ML techniques. Surrogate model development allows for accurate material behavior prediction at a fraction of the cost of its physics-based counterpart, allowing for multiscale simulations of real-world applications, further enabling the ability to design fit-for-purpose materials for a reasonable computational investment. However, training such models requires extensive data, and thus, effective data management is necessary to reach the full potential that ML can offer to material design and ICME. This paper proposes a generalized, robust schema that allows organizations to store both real (experimental) and virtual (simulation) data used to train ML models and the defining model parameters and architectures within the Granta MI Platform. The developed schema allows for various types of data inputs and outputs, including single point values, time-series data, and images that can be used in the prediction of material behavior, while following outlined best practices for effective data management. An effective schema for ML data and models can help prevent the recreation of virtual/real training data and surrogate models, help reduce the time to create new models similar to existing ones by offering a starting point in the hyperparameter determination stages, minimize resources devoted to verification and validation (V&V) and certification of models, and ensure that data and surrogate models are not misused due to full traceability of both the data and ML model. It also allows organizations access to models that have already been developed, such that they can be used in the design of new materials, enabling the overall goals of ICME.

Brandon L. Hearley↗
NASA NTRSDate not supplied

Predicting Fiber Failure of Plain Weave Fabric with Recursive Multiscale Micromechanics

Recent advances in the development of machine learning (ML) algorithms have enabled the creation of predictive models that can improve decision making, decrease computational cost, and improve efficiency in a variety of fields. As an organization begins to develop and implement such models, the data used in the training, validation, and testing of machine learning models, the model parameters, and the use cases or limitations of the models must be properly stored to ensure models are both fully traceable and used correctly. In the context of predicting material behavior, advances in computationally intense, physics-based, modeling of material behavior at various length scales, and the emergence of Integrated Computational Materials Engineering (ICME) have driven the need for developing data-driven surrogate models of the physics-based simulation tools using machine learning (ML) techniques. Surrogate model development allows for accurate material behavior prediction at a fraction of the cost of its physics-based counterpart, allowing for multiscale simulations of real-world applications, further enabling the ability to design fit-for-purpose materials for a reasonable computational investment. However, training such models requires extensive data, and thus effective data management is necessary to reach the full potential that ML can offer to material design and ICME. This paper proposes a generalized, robust schema that allows organizations to store both real (experimental) and virtual (simulation) data used to train machine learning models and the defining model parameters and architectures. The developed schema allows for various types of data inputs and outputs, including single point values, time-series data, and images that can be used in for various types of machine learning models while following outlined best practices for effective data management. An effective schema for machine learning data and models can help prevent the recreation of virtual/real training data and surrogate models, can help reduce the time to create new models similar to existing ones by offering a starting point in the hyperparameter determination stages, minimize resources devoted to verification and validation (V&V) and certification of models, and ensure that data and surrogate models are not misused due to full traceability of both the data and ML model. It also allows organizations access to models that have already been developed, such that they can be used in the design of new materials, enabling the overall goals of ICME.

Failure↗
NASA NTRSDate not supplied

The TOLNet 2.0 Website: How an API Can Promote Open Science and FAIR Principles

The Tropospheric Ozone Lidar Network (TOLNet) has generated over a decade of ozone vertical profile data products over North America. The science value of the TOLNet data has been demonstrated in numerous peer-reviewed publications on air quality and other ozone relevant research. To support the broad spectrum of data use, the TOLNet team launched a major effort to upgrade the web-based data repository aiming to enhance the data discoverability and to enable machine-to-machine data upload and download processes. Specifically, the TOLNet website included an application programming interface (API), which supports machine-to-machine data search and data download. The API also extracts selected variables from the files, which can be retrieved as JSON objects and used to create data displays without having to download or open the underlying files. The TOLNet science team members can also use the API for automated data upload, including a data file scanning feature to ensure data product integrity. To be presented will include a summary of key features of data repositories, an actual use case of machine-to-machine data access/use, as well as our journey to make TOLNet data more FAIR, i.e., more findable, accessible, interoperable, and (re)usable.

Crystal Gummo↗
DOE OSTI2015

Awe Bonner sphere measurements of the neutron leakage spectrum from the GODIVIA-IV critical assembly at NCERC - 14413

The US National Criticality Experiments Research Centre (NCERC) is home to several critical assembly machines that were relocated from Los Alamos National Laboratory in the last decade. The centre is sponsored by the Department of Energy's Nuclear Criticality Safety Program (NCSP), which has the mission to ensure the safe handling of fissile materials from a criticality perspective. Under the auspices of the NCSP, Lawrence Livermore National Laboratory (LLNL) is planning to host an international inter-comparison of nuclear accident dosimetry (NAD) using the Godiva-IV assembly at NCERC. The inter-comparison is motivated by a decline in the frequency of performance testing of NADs (both in the US, the UK and France) and the desire to modernize and harmonize dosimetry systems. A pre-requisite for the inter-comparison is the characterisation of the neutron and gamma-ray leakage spectra associated with the Godiva-IV assembly at NCERC. AWE provided its Bonner-sphere spectrometry capability to LLNL for two experimental campaigns, in November 2013 and May 2014, to measure the neutron leakage spectrum of the assembly in both a sub-critical 'steady-state' and a super-prompt-critical 'burst'. Nine points, at distances from 2 to 4 meters from the machine were measured with the Bonner-sphere spectrometer using both an active He-3 detector and a passive technique employing gold foils. This paper presents the spectra unfolded from the data and integral dose values, and compares these to previously published data. The limitations of the work and future plans are also discussed. (authors)

21 SPECIFIC NUCLEAR REACTORS AND ASSOCIATED PLANTS↗
NASA NTRS2010

Remote Characterization of Biomass Measurements: Case Study of Mangrove Forests

Accurately quantifying forest biomass is of crucial importance for climate change studies. By quantifying the amount of above and below ground biomass and consequently carbon stored in forest ecosystems, we are able to derive estimates of carbon sequestration, emission and storage and help close the carbon budget. Mangrove forests, in addition to providing habitat and nursery grounds for over 1300 animal species, are also an important sink of biomass. Although they only constitute about 3% of the total forested area globally, their carbon storage capacity -- in forested biomass and soil carbon -- is greater than that of tropical forests (Lucas et al, 2007). In addition, the amount of mangrove carbon -- in the form of litter and leaves exported into offshore areas is immense, resulting in over 10% of the ocean's dissolved organic carbon originating from mangroves (Dittmar et al, 2006) The measurement of forest above ground biomass is carried out on two major scales: on the plot scale, biomass can be measured using field measurements through allometric equation derivation and measurements of forest plots. On the larger scale, the field data are used to calibrate remotely sensed data to obtain stand-wide or even regional estimates of biomass. Currently, biomass can be calculated using average stand biomass values and optical data, such as aerial photography or satellite images (Landsat, Modis, Ikonos, SPOT, etc.). More recent studies have concentrated on deriving forest biomass values using radar (JERS, SIR-C, SRTM, Airsar) and/or lidar (ICEsat/GLAS, LVIS) active remote sensing to retrieve more accurate and detailed measurements of forest biomass. The implementation of a generation of new active sensors (UAVSar, DesdynI, Alos/Palsar, TerraX) has prompted the development of new tecm'liques of biomass estimation that use the combination of multiple sensors and datasets, to quantify past, current and future biomass stocks. Focusing on mangrove forest biomass estimation, this book chapter has 3 main objectives: a) To describe in detail the field methodologies used to derive accurate estimates of biomass in mangrove forests b) To explain how mangrove forest biomass can be measured using several remote sensing techniques and datasets c) To give a detailed explanation of the measurement challenges and errors that arise in each estimate of forest biomass

Fatoyinbo, Temilola E.↗
NASA NTRS1975

Extraction from flight data of longitudinal aerodynamic coefficients in maneuvering flight for F-8C aircraft

Flight-test data were used to extract the longitudinal aerodynamic parameters of the F-8C aircraft. The aircraft was trimmed in a steady turn at angles of attack of approximately 9 deg and 13 deg at Mach numbers of 0.7 and 0.8. The parameters extracted resulted in a good match to the flight data and the values obtained were reasonable. The values were further verified by comparing the period and time to damp to half-amplitude, as calculated by using the extracted parameter values, with the period and time to damp to half-amplitude actually measured from the flight data traces. These results show that for the set of data examined, a mathematical model using linear aerodynamics was adequate to describe the response motions at the test angles of attack.

Suit, W. T.↗
NASA NTRS1990

Aerodynamic parameters of High-Angle-of attack Research Vehicle (HARV) estimated from flight data

Aerodynamic parameters of the High-Angle-of-Attack Research Aircraft (HARV) were estimated from flight data at different values of the angle of attack between 10 degrees and 50 degrees. The main part of the data was obtained from small amplitude longitudinal and lateral maneuvers. A small number of large amplitude maneuvers was also used in the estimation. The measured data were first checked for their compatibility. It was found that the accuracy of air data was degraded by unexplained bias errors. Then, the data were analyzed by a stepwise regression method for obtaining a structure of aerodynamic model equations and least squares parameter estimates. Because of high data collinearity in several maneuvers, some of the longitudinal and all lateral maneuvers were reanalyzed by using two biased estimation techniques, the principal components regression and mixed estimation. The estimated parameters in the form of stability and control derivatives, and aerodynamic coefficients were plotted against the angle of attack and compared with the wind tunnel measurements. The influential parameters are, in general, estimated with acceptable accuracy and most of them are in agreement with wind tunnel results. The simulated responses of the aircraft showed good prediction capabilities of the resulting model.

Klein, Vladislav↗
NASA NTRS1990

Lightning observations from space shuttle

The experimental program of the Earth Sciences and Applications Division at NASA/MSFC includes development of the Lightning Imaging Sensor (LIS) for the NOAA Earth Observing System (EOS) Polar Platform. The research plan is to use existing lightning information to generate simulated data for the LIS experiment. Navigation algorithms were used to transform pixel locations to latitude and longitude values. The simulated data would then be used to test and develop algorithms for the analysis of LIS data. Individual frames of video imagery obtained from Space Shuttle Missions provide the raw data for the simulation. Individual video frames were digitized to get the pixel locations of lightning flashes. The pixel locations will be used to locate the geographical position of the event. Because of a lack of detailed knowledge of camera orientation with respect to the Space Shuttle, video scenes that contain identifiable city lights were chosen for analysis. A method for locating the payload bay camera axis was developed and tested. Two measurements are needed: the pixel location of the apparent horizon and a timed siting of a known location passing the principal line of the image. Individual video frames were navigated and lightning illuminated clouds were located on the map. Satisfactory agreement in location was achieved for cities and LLP lightning locations. Ground truth measurements were compared to satellite observations. A vertical lightning event was identified on the horizon. Very low frequency (VLF) transmission on this particular occassion shows a strong response to negative cloud to cloud flashes.

Boeck, William L.↗
NASA NTRS1989

Modelling Rift Valley fever (RVF) disease vector habitats using active and passive remote sensing systems

The NASA Ames Ecosystem Science and Technology Branch and the U.S. Army Medical Research Institute of Infectious Diseases are conducting research to detect Rift Valley fever (RVF) vector habitats in eastern Africa using active and passive remote-sensing. The normalized difference vegetation index (NDVI) calculated from Landsat TM and SPOT data is used to characterize the vegetation common to the Aedes mosquito. Relationships have been found between the highest NDVI and the 'dambo' habitat areas near Riuru, Kenya on both wet and dry data. High NDVI values, when combined with the vegetation classifications, are clearly related to the areas of vector habitats. SAR data have been proposed for use during the rainy season when optical systems are of minimal use and the short frequency and duration of the optimum RVF mosquito habitat conditions necessitate rapid evaluation of the vegetation/moisture conditions; only then can disease potential be stemmed and eradication efforts initiated.

Ambrosia, Vincent G.↗
DOE OSTI2018

Bayesian optimization of generalized data

Direct application of Bayes' theorem to generalized data yields a posterior probability distribution function (PDF) that is a product of a prior PDF of generalized data and a likelihood function, where generalized data consists of model parameters, measured data, and model defect data. The prior PDF of generalized data is defined by prior expectation values and a prior covariance matrix of generalized data that naturally includes covariance between any two components of generalized data. A set of constraints imposed on the posterior expectation values and covariances of generalized data via a given model is formally solved by the method of Lagrange multipliers. Posterior expectation values of the constraints and their covariance matrix are conventionally set to zero, leading to a likelihood function that is a Dirac delta function of the constraining equation. It is shown that setting constraints to values other than zero is analogous to introducing a model defect. Since posterior expectation values of any function of generalized data are integrals of that function over all generalized data weighted by the posterior PDF, all elements of generalized data may be viewed as nuisance parameters marginalized by this integration. One simple form of posterior PDF is obtained when the prior PDF and the likelihood function are normal PDFs. For linear models without a defect this PDF becomes equivalent to constrained least squares (CLS) method, that is, the χ2 minimization method.

Arbanas, Goran↗
DOE OSTI2020

Review of Landscaping Equipment Vibration Risk Exposure, LLNL Site 200 (Various Locations), Whole Body Vibration Seated and Standing Mowers

WBV (Whole-Body Vibration) for operators of seated and standing lawn mowers was conducted during landscape activities from April 2019 through August 2019 at various locations around LLNL Site 200. Mowing is generally performed by each landscaper for 2-3 days every other week with a maximum duration of 3-5 hours each day. On mowing days, other tasks are performed for the remainder of the day that do not expose the worker to WBV such as irrigation, blowing, or sweeping. The mowing terrain at LLNL is variable with typical reported conditions such as gopher holes, exposed tree roots, quick turns, normal and high growth, weeds, and mushy/wet ground that can result in jolts to the mower operator. The Larson Davis HVM200 seat pad with tri-axial accelerometer was used for monitoring workers. For the seated mowers, the HVM200 seat pad with tri-axial accelerometer was placed on the seat between the user and seat pan of the mower. For standing mowers, the seat pad with tri-axial accelerometer was wrapped with Dycem (for seat pad protection) and placed directly on top of the mower deck and underneath the end-user’s right foot. Monitoring for WBV was conducted while operators used different types, makes, and models of mowers. A total of sixteen samples were collected from the operators of seated mower and 6 samples were collected from operators of standing mower. This initial data sampling was part of a discovery phase to identify if WBV exposure met the 2016 ACGIH TLVs (Threshold Limit Values). This initial data identified that workers operating the mowers do not exceed either the TLV or Action Level (AL) for Weighted Acceleration Levels (A(8)) but typically exceed the Vibration Dose Value (VDV) lower boundary for the standing mowers.

59 BASIC BIOLOGICAL SCIENCES↗
NASA NTRS1993

Are existing magnetospheric models excessively stretched?

The distribution of the average north-south magnetic field component Bz in the vicinity of the neutral sheet has been investigated. This component is crucial for mappings between the nightside polar ionosphere and the equatorial magnetosphere. Data sets consisting of about 0.5 R(E) averages of magnetic field observations by the IMP/HEOS and ISEE spacecraft have been compared to the field predicted by the Tsyganenko (1987, 1989) models T87 and T89. It was found that both T87 and T89 underestimate Bz in the near tail region by as much as a factor of 2. Modified versions of the T87 model, incorporating plasma sheet warping, were obtained by fitting the model parameters via nonlinear least squares to the ISEE data set and yielded Bz values in agreement with the ISEE data. The study reveals an enormous scatter among the observed baseline values of Bz (on a time scale of 10-20 min), as well as intrinsic biases imposed by the mathematical structure of tail models, and these two factors (especially the first one) greatly limit the accuracy of model predictions of tail Bz.

Peredo, Mauricio↗
DOE OSTI2023

Quasar UV/X-ray relation luminosity distances are shorter than reverberation-measured radius–luminosity relation luminosity distances

ABSTRACT We use measurements of 59/58 quasars (QSOs), over a redshift range 0.0041 ≤ z ≤ 1.686, to do a comparative study of the radius–luminosity (R − L) and X-ray−UV luminosity (LX − LUV) relations and the implication of these relations for cosmological parameter estimation. By simultaneously determining R − L or LX − LUV relation parameters and cosmological parameters in six different cosmological models, we find that both R − L and LX − LUV relations are standardizable but provide only weak cosmological parameter constraints, with LX − LUV relation data favouring larger current non-relativistic matter density parameter Ωm0 values than R − L relation data and most other available data. We derive LX − LUV and R − L luminosity distances for each of the sources in the six cosmological models and find that LX − LUV relation luminosity distances are shorter than R − L relation luminosity distances as well as standard flat ΛCDM model luminosity distances. This explains why LX − LUV relation QSO data favour larger Ωm0 values than do R − L relation QSO data or most other cosmological measurements. While our sample size is small and only spans a small z range, these results indicate that more work is needed to determine whether the LX − LUV relation can be used as a cosmological probe.

Astronomy & Astrophysics↗
NASA NTRS1983

Geologic exploration: The contribution of LANDSAT-4 thematic mapper data

The major advantages of the TM data over that of MSS systems are increased spatial resolution and a greater number of narrow, strategically placed spectral bands. The 30 meter pixel size permits finer definition of ground features and improves reliability of the photointerpretation of geologic structure. The value of the spatial data increases relative to the value of the spectral data as soil and vegetation cover increase. In arid areas with good exposure, it is possible with careful digital processing and some inventive color compositing to produce enough spectral differentiation of rock types and thereby produce facsimiles of standard geologic maps with a minimum of field work or reference to existing maps. Hue-saturation value images are compared with geological maps of Death Valley, California, the Big Horn/Wind River Basin of Wyoming, the area around Cement, Oklahoma, and Detroit. False color composites of the Ontario region are also examined.

Everett, J. R.↗
DOE OSTI2026

REV-INR: Regularized Evidential Implicit Neural Representation for Uncertainty-Aware Volume Visualization

Applications of Implicit Neural Representations (INRs) have emerged as a promising deep learning approach for compactly representing large volumetric datasets. These models can act as surrogates for volume data, enabling efficient storage and on-demand reconstruction via model predictions. However, conventional deterministic INRs only provide value predictions without insights into the model’s prediction uncertainty or the impact of inherent noisiness in the data. This limitation can lead to unreliable data interpretation and visualization due to prediction inaccuracies in the reconstructed volume. Identifying erroneous results extracted from model-predicted data may be infeasible, as raw data may be unavailable due to its large size. To address this challenge, we introduce REV-INR, Regularized Evidential Implicit Neural Representation, which learns to predict data values accurately along with the associated coordinate-level data uncertainty and model uncertainty using only a single forward pass of the trained REV-INR during inference. By comprehensively comparing and contrasting REV-INR with existing well-established deep uncertainty estimation methods, we show that REV-INR achieves the best volume reconstruction quality with robust data (aleatoric) and model (epistemic) uncertainty estimates using the fastest inference time. Consequently, we demonstrate that REV-INR facilitates assessment of the reliability and trustworthiness of the extracted isosurfaces and volume visualization results, enabling analyses to be solely driven by model-predicted data.

Saklani, Shanu [Indian Institute of Technology, Ka↗
DOE OSTI2020

Radar Super Resolution using a Deep Convolutional Neural Network

Super-resolution involves synthetically increasing the resolution of gridded data beyond its native resolution. Typically, this is done using interpolation schemes, which estimate sub-grid scale values from neighboring data, and perform the same operation everywhere regardless of the large-scale context, or by requiring a network of radars with overlapping fields of view. Recently, significant progress has been made in single image super resolution using convolutional neural networks. Conceptually, a neural network may be able to learn relations between large scale precipitation features and the associated sub-pixel scale variability and outperform interpolation schemes. Here, we use a deep convolutional neural network to artificially enhance the resolution of NEXRAD PPI scans. The model is trained on 6-months of reflectivity observations from the Langley Hill WA (KLGX) radar, and we find that it substantially outperforms common interpolation schemes for x4 and x8 resolution increases based on several objective error and perceptual quality metrics.

radar, machine learning, super resolution, Remote ↗