Engineering Papers⌕ Search

SEARCH · Engineering Papers

Results for “RESOLUTION”

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 469 records · Page 26
All sourcesNASA NTRSDOE OSTI
DOE OSTI2015

THE INFLUENCE OF SPATIAL RESOLUTION ON NONLINEAR FORCE-FREE MODELING

The nonlinear force-free field (NLFFF) model is often used to describe the solar coronal magnetic field, however a series of earlier studies revealed difficulties in the numerical solution of the model in application to photospheric boundary data. We investigate the sensitivity of the modeling to the spatial resolution of the boundary data, by applying multiple codes that numerically solve the NLFFF model to a sequence of vector magnetogram data at different resolutions, prepared from a single Hinode/Solar Optical Telescope Spectro-Polarimeter scan of NOAA Active Region 10978 on 2007 December 13. We analyze the resulting energies and relative magnetic helicities, employ a Helmholtz decomposition to characterize divergence errors, and quantify changes made by the codes to the vector magnetogram boundary data in order to be compatible with the force-free model. This study shows that NLFFF modeling results depend quantitatively on the spatial resolution of the input boundary data, and that using more highly resolved boundary data yields more self-consistent results. The free energies of the resulting solutions generally trend higher with increasing resolution, while relative magnetic helicity values vary significantly between resolutions for all methods. All methods require changing the horizontal components, and for some methods also the vertical components, of the vector magnetogram boundary field in excess of nominal uncertainties in the data. The solutions produced by the various methods are significantly different at each resolution level. We continue to recommend verifying agreement between the modeled field lines and corresponding coronal loop images before any NLFFF model is used in a scientific setting.

79 ASTRONOMY AND ASTROPHYSICS↗
DOE OSTI2014

Deformable elastic network refinement for low-resolution macromolecular crystallography

An overview of applications of the deformable elastic network (DEN) refinement method is presented together with recommendations for its optimal usage. Crystals of membrane proteins and protein complexes often diffract to low resolution owing to their intrinsic molecular flexibility, heterogeneity or the mosaic spread of micro-domains. At low resolution, the building and refinement of atomic models is a more challenging task. The deformable elastic network (DEN) refinement method developed previously has been instrumental in the determinion of several structures at low resolution. Here, DEN refinement is reviewed, recommendations for its optimal usage are provided and its limitations are discussed. Representative examples of the application of DEN refinement to challenging cases of refinement at low resolution are presented. These cases include soluble as well as membrane proteins determined at limiting resolutions ranging from 3 to 7 Å. Potential extensions of the DEN refinement technique and future perspectives for the interpretation of low-resolution crystal structures are also discussed.

75 CONDENSED MATTER PHYSICS, SUPERCONDUCTIVITY AND↗
DOE OSTI2018

New capabilities in high-resolution neutron Larmor diffraction at ORNL

Using superconducting magnetic Wollaston prisms, high-resolution neutron Larmor diffraction has been implemented at the High-Flux Isotope Reactor of Oak Ridge National Laboratory (ORNL), Tennesse, USA. This technique allows the inverse relationship between the achievable diffraction resolution and the usable neutron flux to be overcome. Instead of employing physically tilted radio-frequency spin flippers, the method uses magnetic Wollaston prisms which are electromagnetically tuned by changing the field configurations in the device. As implemented, this method can be used to measure lattice-spacing changes induced, for example, by thermal expansion or strain with a resolution of Δd/d ≃ 10 –6 , and the splitting of sharp Bragg peaks with a resolution of Δd/d = 3 × 10 –4 . The resolution for discerning a change in the profile of a Bragg peak is Δd/d < 10 –5 . This is a remarkable degree of precision for a neutron diffractometer as compact as the one used in this implementation. Higher precision could be obtained by implementing this technique in an instrument with a larger footprint. In conclusion, the availability of this technique will provide an alternative when standard neutron diffraction methods fail and will greatly benefit the scientific communities that require high-resolution diffraction measurements.

72 PHYSICS OF ELEMENTARY PARTICLES AND FIELDS↗
DOE OSTI2023

High-Resolution Synthetic Solar Irradiance Sequence Generation: An LSTM-Based Generative Adversarial Network

The rapid growth of renewable energy resources penetration is bringing more challenges to power system planning and operation. Relevant renewable energy integration studies, such as the capability and dynamic performance of inverter-based resources' primary frequency response and fast frequency response, require high-resolution renewable generation output data that are representative of renewable energy resources. This paper focuses on creating synthetic but realistic solar irradiance data and proposes a long short-term memory-based generative adversarial network to generate high-resolution (second-level) solar irradiance sequences from low-resolution (minute-level) measurements. Combined with a classifier to recognize the solar irradiance patterns, the proposed model is trained using multi-loss functions to accurately capture the temporal correlations among both high-resolution and low-resolution sequences. Verification of the proposed approach is performed on the data set of the Oahu Solar Measurement Grid collected through the National Renewable Energy Laboratory. The results of the case studies demonstrate the proposed approach's capability to capture the statistical characteristics of different solar irradiance patterns and to generate high-quality synthetic solar irradiance sequences in high resolution.

dynamic scheduling↗
DOE OSTI2020

Time Resolution Measurements of EJ-232Q With Single- and Dual-Sided Readouts

In view of its importance to applications in high energy physics, nuclear security, and time-of-flight PET, the time resolution of 0.5% benzophenone-quenched EJ-232Q plastic scintillators of various lengths was studied using a fast waveform digitizer and fast PMTs. The scintillator sizes were 5 mm × 5 mm × 10, 25, 50, 75, or 100 mm. First, the timing measurements were performed by using a single PMT reading out one end. The degradation in time resolution with increasing scintillator length was studied. The irradiation geometry was also observed to have a non-negligible impact on the time resolution due to the variation of depth-of-interaction positions, especially for the scintillators with a high aspect ratio. Second, two PMTs were employed for a dual-sided readout. This readout together with the mean time method was demonstrated to be an effective approach to improve the timing performance since the depth-of-interaction-related effect was considerably mitigated. A time resolution of 79 ± 1 ps (full-width at half-maximum) has been achieved for the 5 × 5 × 10 mm 3 scintillator. The 5 × 5 × 100 mm 3 scintillator can still provide a time resolution as good as 172 ± 1 ps. Here, the energy range used for each was between 340 keVee and 450 keVee. In addition, the time information from individual PMTs in a dual-sided readout was able to show a time resolution almost identical to that determined with the mean-time-based technique after correcting for the depth-of-interaction variation for the 100 mm long scintillator.

72 PHYSICS OF ELEMENTARY PARTICLES AND FIELDS↗
DOE OSTI2014

High resolution Cerenkov light imaging of induced positron distribution in proton therapy

Purpose: In proton therapy, imaging of the positron distribution produced by fragmentation during or soon after proton irradiation is a useful method to monitor the proton range. Although positron emission tomography (PET) is typically used for this imaging, its spatial resolution is limited. Cerenkov light imaging is a new molecular imaging technology that detects the visible photons that are produced from high-speed electrons using a high sensitivity optical camera. Because its inherent spatial resolution is much higher than PET, the authors can measure more precise information of the proton-induced positron distribution with Cerenkov light imaging technology. For this purpose, they conducted Cerenkov light imaging of induced positron distribution in proton therapy. Methods: First, the authors evaluated the spatial resolution of our Cerenkov light imaging system with a {sup 22}Na point source for the actual imaging setup. Then the transparent acrylic phantoms (100 × 100 × 100 mm{sup 3}) were irradiated with two different proton energies using a spot scanning proton therapy system. Cerenkov light imaging of each phantom was conducted using a high sensitivity electron multiplied charge coupled device (EM-CCD) camera. Results: The Cerenkov light’s spatial resolution for the setup was 0.76 ± 0.6 mm FWHM. They obtained high resolution Cerenkov light images of the positron distributions in the phantoms for two different proton energies and made fused images of the reference images and the Cerenkov light images. The depths of the positron distribution in the phantoms from the Cerenkov light images were almost identical to the simulation results. The decay curves derived from the region-of-interests (ROIs) set on the Cerenkov light images revealed that Cerenkov light images can be used for estimating the half-life of the radionuclide components of positrons. Conclusions: High resolution Cerenkov light imaging of proton-induced positron distribution was possible. The authors conclude that Cerenkov light imaging of proton-induced positron is promising for proton therapy.

46 INSTRUMENTATION RELATED TO NUCLEAR SCIENCE AND ↗
DOE OSTI2015

SU-E-J-27: Shifting Multiple EPID Imager Layers to Improve Image Quality and Resolution in MV CBCT

Purpose: Vertical stacking of four conventional EPID layers can improve DQE for MV-CBCT applications. We hypothesize that shifting each layer laterally by half a pixel relative to the layer above, will improve the contrast-to-noise ratio (CNR) and image resolution. Methods: For CNR assessment, a 20 cm diameter digital phantom with 8 inserts is created. The attenuation coefficient of the phantom is similar to lung at the average energy of a 6 MV photon beam. The inserts have attenuations 1, 2…8 times of lung. One of the inserts is close to soft tissue, resembling the case of a tumor in lung. For resolution assessment, a digital phantom featuring a bar pattern is created. The phantom has an attenuation coefficient similar to soft tissue and the bars have an attenuation coefficient of calcium sulfate. A 2 MeV photon beam is attenuated through these phantoms and hits each of the four stacked detector layers. Each successive layer is shifted by half a pixel in the x only, y only, and x and y (combined) directions, respectively. Blurring and statistical noise are added to the projections. Projections from one, two, three and four layers are used for reconstruction. CNR and image resolution are evaluated and compared. Results: When projections from multiple layers are combined for reconstruction, CNR increases with the number of layers involved. CNR in reconstructions from two, three and four layers are 1.4, 1.7 and 1.99 times that from one layer. The resolution from the shifted four layer detector is also improved from a single layer. In a comparison between one layer versus four layers in this preliminary study, the resolution from four shifted layers is at least 20% better. Conclusion: Layer-shifting in a stacked EPID imager design enhances resolution as well as CNR for half scan MV-CBCT. The project described was supported, in part, by a grant from Varian Medical Systems, Inc., and Award No. R01CA188446-01 from the National Cancer Institute. The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Cancer Institute or the National Institutes of Health.

60 APPLIED LIFE SCIENCES↗
DOE OSTI2016

SU-F-T-559: High-Resolution Scintillating Fiber Array for In-Vivo Real-Time SRS and SBRT Patient QA

Purpose: A high-resolution scintillating fiber detector was built for in-vivo real-time patient specific quality assurance (QA). The detector is designed for stereotactic body radiotherapy (SBRT) and stereotactic radiosurgery (SRS) to monitor treatment delivery and detect real-time deviations from planned dose to increase patient safety and treatment accuracy. Methods: The detector consists of two high-density scintillating fiber arrays layered to form an X-Y grid which can be attached to the accessory tray of a medical linac for SBRT and cone SRS treatment QA. Fiber arrays consist of 128 scintillating fibers embedded within a precision-machined, high-transmission polymer substrate with 0.8mm pitch. The fibers are coupled on both ends to high-sensitivity photodetectors and the output is recorded through a high-speed analog-to-digital converter to capture the linac pulse sequence as treatment delivery progresses. The detector has a software controlled 360 degree rotational system to capture angular beam projections for high-resolution beam profile reconstruction. Results: The detector was validated using SRS cone sizes from 6mm to 34mm and MLC defined field sizes from 5×5mm2 to 100×100mm2. The detector output response is linear with dose and is dose rate independent. Each field can be reconstructed accurately with a spatial resolution of 0.8mm and the current beam output is displayed every 50msec. Dosimetric errors of 1% with respect to the treatment plan can be identified and clinically significant deviations from the expected treatment can be displayed in real-time to alert the therapists. Conclusion: The high resolution detector is capable of reconstructing beam profiles in real-time with submillimeter resolution and 1% dose resolution. This system has the ability to project in-vivo both spatial and dosimetric errors during SBRT and SRS treatments when only a non-clinically significant fraction of the intended dose was delivered. The device has the potential to establish new standards for in-vivo patient specific QA.

60 APPLIED LIFE SCIENCES↗
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 ↗
DOE OSTI2019

Retrieving Temperatures and Abundances of Exoplanet Atmospheres with High-resolution Cross-correlation Spectroscopy

High-resolution spectroscopy (R ≥ 25,000) has recently emerged as one of the leading methods for detecting atomic and molecular species in the atmospheres of exoplanets. However, it has so far been lacking a robust method for extracting quantitative constraints on the temperature structure and molecular/atomic abundances. In this work, we present a novel Bayesian atmospheric retrieval framework applicable to high-resolution cross-correlation spectroscopy (HRCCS) that relies on the cross-correlation between data and models to extract the planetary spectral signal. We successfully test the framework on simulated data and show that it can correctly determine Bayesian credibility intervals on atmospheric temperatures and abundances, allowing for a quantitative exploration of the inherent degeneracies. Furthermore, our new framework permits us to trivially combine and explore the synergies between HRCCS and low-resolution spectroscopy to maximally leverage the information contained within each. This framework also allows us to quantitatively assess the impact of molecular line opacities at high resolution. We apply the framework to VLT CRIRES K-band spectra of HD 209458 b and HD 189733 b and retrieve abundant carbon monoxide but subsolar abundances for water, which are largely invariant under different model assumptions. This confirms previous analysis of these data sets, but is possibly at odds with detections of H{sub 2}O at different wavelengths and spectral resolutions. The framework presented here is the first step toward a true synergy between space observatories and ground-based high-resolution observations.

79 ASTRONOMY AND ASTROPHYSICS↗
DOE OSTI2017

Prospects for Measuring Abundances of >20 Elements with Low-resolution Stellar Spectra

Understanding the evolution of the Milky Way calls for the precise abundance determination of many elements in many stars. A common perception is that deriving more than a few elemental abundances ([Fe/H], [ α /Fe], perhaps [C/H], [N/H]) requires medium-to-high spectral resolution, R ≳ 10,000, mostly to overcome the effects of line blending. In a recent work, we presented an efficient and practical way to model the full stellar spectrum, even when fitting a large number of stellar labels simultaneously. In this paper, we quantify to what precision the abundances of many different elements can be recovered, as a function of spectroscopic resolution and wavelength range. In the limit of perfect spectral models and spectral normalization, we show that the precision of elemental abundances is nearly independent of resolution, for a fixed exposure time and number of detector pixels; low-resolution spectra simply afford much higher S/N per pixel and generally larger wavelength range in a single setting. We also show that estimates of most stellar labels are not strongly correlated with one another once R ≳ 1000. Modest errors in the line-spread function, as well as small radial velocity errors, do not affect these conclusions, and data-driven models indicate that spectral (continuum) normalization can be achieved well enough in practice. These results, to be confirmed with an analysis of observed low-resolution data, open up new possibilities for the design of large spectroscopic stellar surveys and for the reanalysis of archival low-resolution data sets.

79 ASTRONOMY AND ASTROPHYSICS↗
NASA NTRS2011

Analysis of Ultra High Resolution Sea Surface Temperature Level 4 Datasets

Sea surface temperature (SST) studies are often focused on improving accuracy, or understanding and quantifying uncertainties in the measurement, as SST is a leading indicator of climate change and represents the longest time series of any ocean variable observed from space. Over the past several decades SST has been studied with the use of satellite data. This allows a larger area to be studied with much more frequent measurements being taken than direct measurements collected aboard ship or buoys. The Group for High Resolution Sea Surface Temperature (GHRSST) is an international project that distributes satellite derived sea surface temperatures (SST) data from multiple platforms and sensors. The goal of the project is to distribute these SSTs for operational uses such as ocean model assimilation and decision support applications, as well as support fundamental SST research and climate studies. Examples of near real time applications include hurricane and fisheries studies and numerical weather forecasting. The JPL group has produced a new 1 km daily global Level 4 SST product, the Multiscale Ultrahigh Resolution (MUR), that blends SST data from 3 distinct NASA radiometers: the Moderate Resolution Imaging Spectroradiometer (MODIS), the Advanced Very High Resolution Radiometer (AVHRR), and the Advanced Microwave Scanning Radiometer ? Earth Observing System(AMSRE). This new product requires further validation and accuracy assessment, especially in coastal regions.We examined the accuracy of the new MUR SST product by comparing the high resolution version and a lower resolution version that has been smoothed to 19 km (but still gridded to 1 km). Both versions were compared to the same data set of in situ buoy temperature measurements with a focus on study regions of the oceans surrounding North and Central America as well as two smaller regions around the Gulf Stream and California coast. Ocean fronts exhibit high temperature gradients (Roden, 1976), and thus satellite data of SST can be used in the detection of these fronts. In this case, accuracy is less of a concern because the primary focus is on the spatial derivative of SST. We calculated the gradients for both versions of the MUR data set and did statistical comparisons focusing on the same regions.

sea surface temperature SST)↗
NASA NTRSDate not supplied

Super-Resolution from Space: Using MERRA-2 and MAIAC Satellite Imagery to Produce Daily Continuous 1 km PM2.5 Estimates

PM2.5 measurements from ground stations are the gold standard when available, but the expense and coverage of such stations limits widespread monitoring. Having accurate PM2.5 estimates outside the range of these stations is important for monitoring this crucial aspect of air quality. The goal of this project is to produce daily 1 km continuous PM2.5 estimates for the contiguous US relying primarily on satellite-derived data sources. This is important because models based on such data can be more easily expanded outside the study area and produce global estimates as well. The temporal availability of such data products is often weekly/daily, unlike land-use products with are often available at a yearly or worse temporal resolution. To achieve our goal, we use a couple of different deep neural network architectures to produce PM2.5 measurements at 10 km and 1 km resolution. We use two model architectures, a UNET-like model and a GAN-based model. We train both models using MERRA-2 data and MAIAC AOD data scaled to 10 km and 1 km for the two different prediction resolutions. MERRA-2 imagery is data rich with a wide range of geospatial variables at 50 km and has long historical availability (beginning in 1980). We’re also using higher spatial resolution MAIAC data at 1 km to provide finer resolution spatial context. This essentially leverages the spatial resolution of MAIAC data and the “wider” information of MERRA-2 data to predict PM2.5. For the target data we’re using a modeled 1 km PM2.5 dataset produced by Harvard to pre-train our models and then fine-tune our models using ground station measurements. Not only are our results comparable with the performance of the Harvard dataset, but can be generalized to any area or time where MERRA-2 and MAIAC data is available.

satellite imagery↗
DOE OSTI2015

Estimation of position resolution for DOI-PET detector using diameter 0.2 mm WLS fibers

We have been developing sub mm resolution and $ 1 million DOI-PET detector using wavelength shifting fibers (WLSF), scintillation crystals of plate shape and SiPM (MPPC: HAMAMATSU K. K.). Conventional design of DOI-PET detector is obtained about mm{sup 3} of resolution by using some blocks detecting gamma-ray in mm 3 voxel. It requires the production cost of $ a few ten million or more for high technique of processing crystal and a lot of number of photo-devices, and this technology is reaching the limit of the resolution. Both higher resolution and lower cost of DOI-PET detector production is challenging for PET diagnosis population. We propose two type of detector. One is a whole body PET system, and the other for brain or small animal. Both PET system consist 6 blocks. the former consist of 6 layers 300 mm x 300 mm x 4 mm crystal plate. The latter consist 16 crystal layers, 4 x 4 crystal array. The size of crystal plate is 40 mm x 40 mm x 1 mm.The WLSF sheets connect to upper and lower plane. The whole PET systems connect 8 SiPMs are bonded on each side. For the brain PET, 9 WLSF fibers are bond on the each side. The expected position resolution maybe less than 1 mm at the former. We have estimation experimental performance the system using {sup 22}Na radioactive source. The collection efficiency of WLSF (R-3) sheet was achieved 10% with GAGG at 511 keV. The relation between reconstruction position and incident position is obtained linearity and achieved the resolution of 0.7 mm FWHM for x-axis of DOI by readout WLSF. (authors)

46 INSTRUMENTATION RELATED TO NUCLEAR SCIENCE AND ↗
DOE OSTI2015

Estimation of position resolution for DOI-PET detector using diameter 0.2 mm WLS fibers [ANIMMA--2015-IO-x5]

We have been developing a submillimeter resolution and low-cost DOI-PET detector using wavelength shifting fibers (WLSF), scintillating crystal plates and MPPCs (Hamamatsu Photonics). Conventional design of DOI-PET detectors had approximately mm{sup 3} of resolution by using some scintillating blocks with a volume of 1 mm{sup 3}, which detects gamma-ray. They are expensive due to difficulties in processing scintillating crystals and a large number of photo-detectors, and these technologies are likely to reach the limit of the resolution. Development of a lower cost DOI-PET detector with higher resolution is challenging to popularize the PET diagnosis. We propose two type of PET detector. One is a whole body PET system, and the other is a PET system for brain or small animals. Each PET system consists 6 blocks. The former consists of 6 layers of crystal plates with 300 mm x 300 mm x 4 mm. The latter consists of 16 crystal layers, forming 4 x 4 crystal arrays. The size of the crystal plate is 40 mm x 40 mm x 1 mm. Wavelength shifting fiber (WLSF) sheets are attached to above and up and down side of crystal planes. The whole PET system has 8 MPPCs attached on each side. For the brain PET detector, 9 WLSF fibers are attached on the each side. The expected position resolution would be less than 1 mm at the former system. We have performed an experimental performance estimation for the system component using {sup 22}Na radioactive source. We achieved a collection efficiency of 10% using the WLSF sheet and Ce:Gd{sub 3}(Al,Ga){sub 5}O{sub 12} (GAGG) crystals at 511 keV. The linear relationship between reconstruction position and incident position was obtained, and a resolution of 0.7 mm (FWHM) for x-axis of DOI by the WLSF readout was achieved. (authors)

46 INSTRUMENTATION RELATED TO NUCLEAR SCIENCE AND ↗
DOE OSTI2015

Resolution‐dependent behavior of subgrid‐scale vertical transport in the Z hang‐ M c F arlane convection parameterization

Abstract To better understand the behavior of quasi‐equilibrium‐based convection parameterizations at higher resolution, we use a diagnostic framework to examine the resolution‐dependence of subgrid‐scale vertical transport of moist static energy as parameterized by the Zhang‐McFarlane convection parameterization (ZM). Grid‐scale input to ZM is supplied by coarsening output from cloud‐resolving model (CRM) simulations onto subdomains ranging in size from 8 × 8 to 256 × 256 km 2 . Then the ZM‐based parameterization of vertical transport of moist static energy for scales smaller than the subdomain size ( ) are compared to those directly calculated from the CRM simulations ( ) for different subdomain sizes. The ensemble mean decreases by more than half as the subdomain size decreases from 128 to 8 km across while decreases with subdomain size only for strong convection cases and increases for weaker cases. The resolution dependence of is determined by the positive‐definite grid‐scale tendency of convective available potential energy (CAPE) in the convective quasi‐equilibrium (QE) closure. Further analysis shows the actual grid‐scale tendency of CAPE (before taking the positive definite value) and behave very similarly as the subdomain size changes because they are both tied to grid‐scale advective tendencies. We can improve the resolution dependence of significantly by averaging the grid‐scale tendency of CAPE over an appropriately large area surrounding each subdomain before taking its positive definite value. Even though the ensemble mean decreases with increasing resolution, its variability increases dramatically. cannot capture such increase in the variability, suggesting the need for stochastic treatment of convection at relatively high spatial resolution (8 or 16 km).

54 ENVIRONMENTAL SCIENCES↗
DOE OSTI2016

An evaluation of the variable-resolution CESM for modeling California's climate

In this paper, the recently developed variable-resolution option within the Community Earth System Model (VR-CESM) is assessed for long-term regional climate modeling of California at 0.25° (~ 28 km) and 0.125° (~ 14 km) horizontal resolutions. The mean climatology of near-surface temperature and precipitation is analyzed and contrasted with reanalysis, gridded observational data sets, and a traditional regional climate model (RCM)—the Weather Research and Forecasting (WRF) model. Statistical metrics for model evaluation and tests for differential significance have been extensively applied. VR-CESM tended to produce a warmer summer (by about 1–3°C) and overestimated overall winter precipitation (about 25%–35%) compared to reference data sets when sea surface temperatures were prescribed. Increasing resolution from 0.25° to 0.125° did not produce a statistically significant improvement in the model results. By comparison, the analogous WRF climatology (constrained laterally and at the sea surface by ERA-Interim reanalysis) was ~1–3°C colder than the reference data sets, underestimated precipitation by ~20%–30% at 27 km resolution, and overestimated precipitation by ~ 65–85% at 9 km. Overall, VR-CESM produced comparable statistical biases to WRF in key climatological quantities. Moreover, this assessment highlights the value of variable-resolution global climate models (VRGCMs) in capturing fine-scale atmospheric processes, projecting future regional climate, and addressing the computational expense of uniform-resolution global climate models.

54 ENVIRONMENTAL SCIENCES↗
DOE OSTI2020

Super resolution for root imaging

Premise High‐resolution cameras are very helpful for plant phenotyping as their images enable tasks such as target vs. background discrimination and the measurement and analysis of fine above‐ground plant attributes. However, the acquisition of high‐resolution images of plant roots is more challenging than above‐ground data collection. An effective super‐resolution (SR) algorithm is therefore needed for overcoming the resolution limitations of sensors, reducing storage space requirements, and boosting the performance of subsequent analyses. Methods We propose an SR framework for enhancing images of plant roots using convolutional neural networks. We compare three alternatives for training the SR model: (i) training with non‐plant‐root images, (ii) training with plant‐root images, and (iii) pretraining the model with non‐plant‐root images and fine‐tuning with plant‐root images. The architectures of the SR models were based on two state‐of‐the‐art deep learning approaches: a fast SR convolutional neural network and an SR generative adversarial network. Results In our experiments, we observed that the SR models improved the quality of low‐resolution images of plant roots in an unseen data set in terms of the signal‐to‐noise ratio. We used a collection of publicly available data sets to demonstrate that the SR models outperform the basic bicubic interpolation, even when trained with non‐root data sets. Discussion The incorporation of a deep learning–based SR model in the imaging process enhances the quality of low‐resolution images of plant roots. We demonstrate that SR preprocessing boosts the performance of a machine learning system trained to separate plant roots from their background. Our segmentation experiments also show that high performance on this task can be achieved independently of the signal‐to‐noise ratio. We therefore conclude that the quality of the image enhancement depends on the desired application.

Ruiz‐Munoz, Jose F.↗