DOE News
    Doe Science news source
    The DOE Science News Source is a Newswise initiative to promote research news from the Office of Science of the DOE to the public and news media.
    • 2018-12-12 14:05:37
    • Article ID: 705358

    Cloud or no cloud, that is the question

    How deep learning is bringing automatic cloud detection to new heights

    • Credit: Adriel Kloppenburg on Unsplash

      Clouds come in all shapes and sizes, and now a deep learning model can help detect the finer details in cloud data.

     

    RICHLAND, Wash. — Kids lying on their backs in a grassy field might scan the clouds for images—perhaps a fluffy bunny here and a fiery dragon over there. Often, atmospheric scientists do the opposite—they search data images for the clouds as part of their research to understand Earth systems.

    Manually labeling data images pixel by pixel is time-consuming, so researchers rely on automatic processing techniques, such as cloud detection algorithms. But the algorithms’ output is not as exact as the scientists want it to be.

    Recently, researchers at the Department of Energy’s Pacific Northwest National Laboratory teamed up to find out if deep learning—a distinct subset of machine learning—can do a better job at identifying clouds in lidar data than the current physics-based algorithms. The answer: A clear “yes.” The new model is much closer to the answers scientists arrive at but in just a fraction of the time.

    Lidar is a remote sensing instrument that emits a pulsed laser and collects the return signal scattered back by cloud droplets or aerosols. This return signal provides information about the height and vertical structure of atmospheric features, such as clouds or smoke layers. Such data from ground-based lidars are an important part of global forecasting. 

    Earth scientist Donna Flynn noticed that, in some cases, what the algorithms detected as clouds in the lidar images did not match well with what her expert eye saw. The algorithms tend to overestimate the cloud boundaries.

    “The current algorithm identifies the clouds using broad brushstrokes,” says Flynn, a co-principal investigator on the project. “We need to more accurately determine the cloud’s true top and base and to distinguish multiple cloud layers.”

    Upgrade initiated

    Until recently, computing power limited artificial neural networks, a type of deep learning model, to a small number of computational layers. Now, with increased computing power available through supercomputing clusters, researchers can use more computations—each building off of the last one—in a series of layers. The more layers an artificial neural network has, the more powerful the deep learning network.

    Figuring out what those computations are is part of the model training. To start, the researchers need properly labeled lidar data images, or “ground truth” data, for the training and testing of the model.

    So, Flynn spent many long hours hand-labeling images pixel by pixel: cloud or no cloud. Her eye can distinguish the cloud boundaries and cloud versus an aerosol layer. She took 40 hours—the equivalent of a full work week—to label about 100 days of lidar data collected at the Southern Great Plains atmospheric observatory, part of DOE’s Atmospheric Radiation Measurement user facility, in Oklahoma.  

    Given how time and labor intensive the hand-labeling process is, PNNL computational scientist and co-principal investigator Erol Cromwell used learning methods that required minimal ground truth data.

    The model learns through self-feedback. It compares its own performance against hand-labeled results and adjusts its calculations accordingly, explains Cromwell. It cycles through these steps, improving each time through.

    Cromwell will be presenting the team’s findings at the Institute of Electrical and Electronics Engineers Winter Conference on Applications of Computer Vision in January.

    Goal achieved

    With the training, the deep learning model outperforms the current algorithms. The model’s precision is almost double and much closer to what a human expert would find—but in a fraction of the time.

    The next steps are to evaluate the model’s performance on lidar data collected at different locations and in different seasons. Initial tests on data from the ARM observatory at Oliktok Point in Alaska are promising.

    “An advantage of the deep learning model is transfer learning,” says Cromwell. “We can train the model further with data from Oliktok to make its performance more robust.”

    “Reducing sources of uncertainty in global model predictions is especially important to the atmospheric science community,” says Flynn. “With its improved precision, deep learning increases our confidence.”

    She adds: “Plus, it gives us more time to be outside looking at real clouds!”

    This research was supported by PNNL’s internal investment in Deep Learning for Scientific Discovery.

    • other-fb
    • other-tw
    X
    X
    X
    • Filters

    • × Clear Filters
    Study shows a much cheaper catalyst can generate hydrogen in a commercial device

    Study shows a much cheaper catalyst can generate hydrogen in a commercial device

    SLAC and Stanford researchers have shown for the first time that a cheap catalyst can split water and generate hydrogen gas for hours on end in the harsh environment of a commercial electrolyzer - a step toward large-scale hydrogen production for fuel, fertilizer and industry.

    Unlocking the Biochemical Treasure Chest Within Microbes

    Unlocking the Biochemical Treasure Chest Within Microbes

    An international team of scientists lead by the Joint Genome Institute has developed a genetic engineering tool that makes producing and analyzing microbial secondary metabolites - the basis for many important agricultural, industrial, and medical products - much easier than before, and could even lead to breakthroughs in biomanufacturing.

    Scientists Pinpoint Cause of Harmful Dendrites and Whiskers in Lithium Batteries

    Scientists Pinpoint Cause of Harmful Dendrites and Whiskers in Lithium Batteries

    Scientists have uncovered a root cause of the growth of needle-like structures--known as dendrites and whiskers--that plague lithium batteries, sometimes causing a short circuit or failure. The defects are a major factor holding back the batteries from broader widespread use and further improvement.

    Argonne and University of Illinois to form hydrogen fuel cell coalition

    Argonne and University of Illinois to form hydrogen fuel cell coalition

    Argonne and University of Illinois announce intent to form the Midwest Hydrogen and Fuel Cell Coalition.

    Six Degrees of Nuclear Separation

    Six Degrees of Nuclear Separation

    For the first time, Argonne scientists have printed 3D parts that pave the way to recycling up to 97 percent of the waste produced by nuclear reactors. From left to right: Peter Kozak, Andrew Breshears, M Alex Brown, co-authors of a recent Scientific Reports article detailing their breakthrough. (Image by Argonne National Laboratory.)

    Shaping nanoparticles for improved quantum information technology

    Shaping nanoparticles for improved quantum information technology

    Argonne researchers find that semiconductor nanoparticles in the shape of rings have attractive properties for quantum networking and computation.

    Science Snapshots - Waste to fuel, moire superlattices, mining cellphones for energy data

    Science Snapshots - Waste to fuel, moire superlattices, mining cellphones for energy data

    Science Snapshots - Waste to fuel, moire superlattices, mining cellphones for energy data

    New Electrolyte Stops Rapid Performance Decline of Next-Generation Lithium Battery

    New Electrolyte Stops Rapid Performance Decline of Next-Generation Lithium Battery

    Researchers at Argonne National Laboratory have designed and tested a new electrolyte composition that could greatly accelerate the adoption of the next generation of lithium-ion batteries.

    Light My Fire: How to Startup Fusion Devices Every Time

    Light My Fire: How to Startup Fusion Devices Every Time

    Researchers have constructed a framework for starting and raising a fusion plasma to temperatures rivaling the sun in hundreds of milliseconds.

    Atomic-level Imaging Could Offer Roadmap to Metals with New Properties

    Atomic-level Imaging Could Offer Roadmap to Metals with New Properties

    A team of researchers at the Georgia Institute of Technology has developed a new process that could help gain new insights into individual high-entropy alloys and help characterize their properties.


    • Filters

    • × Clear Filters

    Department of Energy Announces Private-Public Awards to Advance Fusion Energy Technology

    The U.S. Department of Energy (DOE) announced funding for 12 projects with private industry to enable collaboration with DOE national laboratories on overcoming challenges in fusion energy development. The awards are the first provided through the Innovation Network for Fusion Energy program (INFUSE).

    Denisov Leads High Energy Physics at Brookhaven

    Denisov Leads High Energy Physics at Brookhaven

    Dmitri Denisov, a leading physicist and spokesperson of the DZero experiment, has been named Deputy Associate Lab Director for High Energy Physics at the U.S. Department of Energy's (DOE) Brookhaven National Laboratory.

    Chemistry Postdoc Receives Battery500 Young Investigator Award

    Chemistry Postdoc Receives Battery500 Young Investigator Award

    Zulipiya Shadike, a postdoctoral fellow in the Chemistry Division at the U.S. Department of Energy's (DOE) Brookhaven National Laboratory, received a Young Investigator Award from the Battery500 Consortium, a DOE-sponsored consortium led by Pacific Northwest National Laboratory (PNNL) that aims to improve electric vehicle batteries.

    Two Brookhaven Lab Scientists Named Fellows of the American Physical Society

    Two Brookhaven Lab Scientists Named Fellows of the American Physical Society

    The American Physical Society (APS) has elected two scientists from Brookhaven National Laboratory as 2019 APS fellows.

    Versatile physics leader Stefan Gerhardt elected an APS fellow

    Versatile physics leader Stefan Gerhardt elected an APS fellow

    Profile of physicist Stefan Gerhardt who has been elected a 2019 fellow of the American Physical Society.

    PNNL, Sandia, and Georgia Tech Join Forces in AI Effort

    PNNL, Sandia, and Georgia Tech Join Forces in AI Effort

    Scientists from DOE's Pacific Northwest National Laboratory, DOE's Sandia National Laboratories, and the Georgia Institute of Technology will collaborate on solutions to some of the most challenging problems in AI today, thanks to $5.5 million in funding from DOE.

    Argonne Receives More Than $1 Million for Quantum Information Science

    Argonne Receives More Than $1 Million for Quantum Information Science

    Argonne scientists receive $1.19 million from DOE for quantum research.

    Department of Energy Announces $6.6 Million to Study Dark Matter

    The U.S. Department of Energy (DOE) announced $6.6 million for four new research awards to develop design concepts for dark matter search experiments.

    Department of Energy Office of Science and NNSA Award $3.5 Million for High Energy Density Plasma Research

    The Department of Energy's (DOE) Office of Science and DOE's National Nuclear Security Administration (NNSA) have awarded eight research grants totaling $3.5 million to support work related to High-Energy Density Laboratory Plasmas (HEDLP).

    Department of Energy Announces $21.4 Million for Quantum Information Science Research

    Department of Energy Announces $21.4 Million for Quantum Information Science Research

    The following news release was issued on Aug. 26, 2019 by the U.S. Department of Energy (DOE). It announces funding that DOE has awarded for research in quantum information science related to particle physics and fusion energy sciences. Scientists at DOE's Brookhaven National Laboratory are principal investigators on two of the 21 funded projects.


    • Filters

    • × Clear Filters
    Investigating Materials that Can Go the Distance in Fusion Reactors

    Investigating Materials that Can Go the Distance in Fusion Reactors

    Future fusion reactors will require materials that can withstand extreme operating conditions, including being bombarded by high-energy neutrons at high temperatures. Scientists recently irradiated titanium diboride (TiB2) in the High Flux Isotope Reactor (HFIR) to better understand the effects of fusion neutrons on performance.

    Better 3-D Imaging of Tumors in the Breast with Less Radiation

    Better 3-D Imaging of Tumors in the Breast with Less Radiation

    In breast cancer screening, an imaging technique based on nuclear medicine is currently being used as a successful secondary screening tool alongside mammography to improve the accuracy of the diagnosis. Now, a team is hoping to improve this imaging technique.

    Microbes are Metabolic Specialists

    Microbes are Metabolic Specialists

    Scientists can use genetic information to measure if microbes in the environment can perform specific ecological roles. Researchers recently analyzed the genomes of over 6,000 microbial species.

    Even Hard Materials Have Soft Spots

    Even Hard Materials Have Soft Spots

    The Achilles Heel of "metallic glasses" is that while they are strong materials--even stronger than conventional steels--they are also very brittle. The initial failures tend to be localized and catastrophic. This is due to their random amorphous (versus ordered crystalline) atomic structure. Computer simulations revealed that the structure is not completely random, however, and that there are some regions in the structure that are relatively weak. Defects nucleate more easily in these regions, which can lead to failure. This understanding of the mechanical properties has led to a strategy for making the material stronger and less brittle.

    2-D Atoms Do the Twist

    2-D Atoms Do the Twist

    In the study, scientists demonstrated, for the first time, an intrinsically rotating form of motion for the atoms in a crystal. The observations were on collective excitations of a single molecular layer of tungsten diselenide. Whether the rotation is clockwise or counter-clockwise depends on the wave's propagation direction.

    Location, Location, Location... How charge placement can control a self-assembled structure

    Location, Location, Location... How charge placement can control a self-assembled structure

    For years, scientists have formed polymers using the interaction of charges on molecular chains to determine the shape, geometry, and other properties. Now, a team achieved precise and predictable control of molecular chains by positioning charges. Their method leads to particles with reproducible sizes.

    Cracking in Harsh Environments Needs Stress and Corrosion, But Not at the Same Time

    Cracking in Harsh Environments Needs Stress and Corrosion, But Not at the Same Time

    Alloys (metals combining two or more metallic elements) are typically stronger and less susceptible to cracking than pure metals. Yet when alloys are subjected to stress and a harsh chemical environment, the alloy can fail. The reason? Cracks caused by corrosion.

    Simultaneous Clean and Repair

    Simultaneous Clean and Repair

    Scientists have developed a novel and efficient approach to surface cleaning, materials transport, and repair.

    Where Does Salt in the Amazon Air Come From?

    Where Does Salt in the Amazon Air Come From?

    Tiny particles of sodium salt float in the air over the pristine Amazon basin. Why? The only explanation before now has been that winds blow marine particles hundreds of miles inland from the Atlantic Ocean. An international team of scientists used chemical imaging and atmospheric models to prove otherwise.

    Testing the Toughness of Microbial Cell Walls

    Testing the Toughness of Microbial Cell Walls

    Microbial cells contain biological material that can be important for research or industrial use, such as DNA or proteins. Yet, reaching this cellular material can be a challenge.


    Spotlight





    Showing results

    0-4 Of 2215