Search Results

You are looking at 1 - 5 of 5 items for :

  • The 1st NOAA Workshop on Leveraging AI in the Exploitation of Satellite Earth Observations & Numerical Weather Prediction x
  • All content x
Clear All
Ryan Lagerquist, Amy McGovern, Cameron R. Homeyer, David John Gagne II, and Travis Smith

. 2016 ) and reduces the amount of preprocessing needed, relative to traditional ML methods. CNNs have been used in atmospheric science to estimate sea ice concentration ( Wang et al. 2016 ) and tropical-cyclone intensity ( Wimmers et al. 2019 ) from satellite images, detect extreme-weather patterns in model output ( Racah et al. 2017 ; Kurth et al. 2018 ; Lagerquist et al. 2019 ), replace subgrid-scale parameterizations in numerical models ( Bolton and Zanna 2019 ), and improve the understanding

Free access
Noah D. Brenowitz, Tom Beucler, Michael Pritchard, and Christopher S. Bretherton

-global aquaplanet simulation performed with the System for Atmospheric Modeling (SAM), version 6.10 ( Khairoutdinov and Randall 2003 ). This simulation is run in a channel configuration (from 46°S to 46°N) with a horizontal grid spacing of 4 km and 34 vertical levels of varying thickness, over a zonally symmetric ocean surface with a sea surface temperature of 300.15 K at the equator and 278.15 K at the poleward boundaries. These GCRM training data consist of 80 days of instantaneous three-dimensional fields

Restricted access
Kyle A. Hilburn, Imme Ebert-Uphoff, and Steven D. Miller

efficiency effects related to the production of cloud ice ( Rutledge et al. 2020 ), and also possibly to the unsteady nature of updrafts. Instead spatial variability contains more reliable information content, supplementing missing information at very high optical depths, and is especially useful at night. While there is spatial variability in GLM detection efficiency ( Marchand et al. 2019 ), our CNN is more sensitive to the presence of lightning rather than the magnitude of lightning activity, which

Open access
Amy McGovern, Ryan Lagerquist, David John Gagne II, G. Eli Jergensen, Kimberly L. Elmore, Cameron R. Homeyer, and Travis Smith

this method provides one plot for each predictor, which can be overwhelming with hundreds of predictors. Related MIV methods attempt to explain the model’s prediction for an individual example. Individual conditional expectation (ICE; Goldstein et al. 2015 ) is the PDP for a specific example. The ICE plot (which can be shown on the same axes as the PDP) identifies clusters of model behavior, which are regions of the predictor space where the model treats examples similarly. Another method is

Full access
Sid-Ahmed Boukabara, Vladimir Krasnopolsky, Stephen G. Penny, Jebb Q. Stewart, Amy McGovern, David Hall, John E. Ten Hoeve, Jason Hickey, Hung-Lung Allen Huang, John K. Williams, Kayo Ide, Philippe Tissot, Sue Ellen Haupt, Kenneth S. Casey, Nikunj Oza, Alan J. Geer, Eric S. Maddy, and Ross N. Hoffman

.5194/gmd-12-4261-2019 Wang , L. , K. A. Scott , and D. A. Clausi , 2017 : Sea ice concentration estimation during freeze-up from SAR imagery using a convolutional neural network . Remote Sens . , 9 , 408 , . 10.3390/rs9050408 Wang , Y.-J. , and C.-T. Lin , 1998 : Runge-Kutta neural network for identification of dynamical systems in high accuracy . IEEE Trans. Neural Network , 9 , 294 – 307 , . 10

Open access