Search Results

You are looking at 1 - 10 of 11 items for :

  • Boundary currents x
  • The 1st NOAA Workshop on Leveraging AI in the Exploitation of Satellite Earth Observations & Numerical Weather Prediction x
  • Refine by Access: All Content x
Clear All
Noah D. Brenowitz, Tom Beucler, Michael Pritchard, and Christopher S. Bretherton

parameterization in a causal way (humidity affects precipitation) when the true causality is likely reversed. On the other hand, the instabilities in SPCAM do not appear to be sensitive to this causal ambiguity and are not yet fully understood, but sensitivities to hyperparameter tuning are suggestive. Regardless of its origin, for NNs, the numerical stability problem is catastrophic because current architectures can predict unbounded heating and moistening rates once they depart the envelope of the training

Restricted access
Ryan Lagerquist, Amy McGovern, Cameron R. Homeyer, David John Gagne II, and Travis Smith

tornado-modeling and postprocessing methods, the latter of which combine multisource data into explicit tornado predictions ( Karstens et al. 2018 ). Much work in this area falls under the Warn-on-Forecast initiative (WoF; Stensrud et al. 2009 , 2013 ). The main goal of WoF is to shift the current warning paradigm from extrapolation based on current observations (warn on detection) to use of short-range CAM simulations. This effort includes creating explicit probabilistic tornado forecasts at 0–1-h

Free access
Christina Kumler-Bonfanti, Jebb Stewart, David Hall, and Mark Govett

observational data are selected to be used in NWP models, and an even smaller fraction of that data are actually assimilated into the models ( Weingroff 2014 ). GOES-16 and GOES-17 produce over 100 times as much data as the previous GOES missions, with high potential value for NWP. With the current system, the amount of data far exceeds the computing time available to process it and instead, simple data thinning techniques are applied and the majority of the data are discarded. In contrast, targeted

Restricted access
Andrew E. Mercer, Alexandria D. Grimes, and Kimberly M. Wood

episodes meet this threshold ( Kaplan et al. 2010 ). Additionally, the exact physical processes governing RI remain poorly understood ( Wang and Wu 2004 ; Grimes and Mercer 2014 ), an issue compounded by the relative lack of boundary layer observations within the TC environment and heavy reliance on global operational dynamic forecast models to fill these observational gaps. Recent work has improved our understanding of processes governing the RI of Atlantic Ocean TCs. The probability of RI increases

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

standpoint of precipitation, having significant impacts on human activities, are also the areas that have the least amount of data to constrain estimates of the current atmospheric state. One approach is radiance assimilation (RA), which has the advantage of being physically based, making it simpler to interpret. Okamoto et al. (2019) , Honda et al. (2018a , b ), and Sawada et al. (2019) tested assimilation of Himawari-8 water vapor absorption bands, finding improvements for heavy rain cases

Open access
Hanoi Medina, Di Tian, Fabio R. Marin, and Giovanni B. Chirico

to spatial inconsistencies at the boundaries between tiles ( Hamill and Whitaker 2006 ; Hamill et al. 2006 ). However, this is not an issue present in this study. Leave-one-out cross validation are carried out by excluding the current year from the list of potential analogs. For a detailed description and theoretical basis of the analog method, the readers can refer to Hamill and Whitaker (2006) . 2) Logistic regression method In the logistic regression (LR) method a nonlinear function is

Full access
Sid-Ahmed Boukabara, Vladimir Krasnopolsky, Jebb Q. Stewart, Eric S. Maddy, Narges Shahroudi, and Ross N. Hoffman

foreseeable future, artificial general intelligence will not be available, and ML will require some degree of human expertise, intuition, and intervention to succeed. The steps in applying the general ML approach—identifying the problem, designing or selecting the ML architecture, selecting and normalizing inputs and outputs, preparing training sets, selecting a training algorithm and its parameters, making decisions about sufficient approximation accuracy, and validating the resulting ML model—currently

Free access
Eric D. Loken, Adam J. Clark, Amy McGovern, Montgomery Flora, and Kent Knopfmeier

), they can undersample the forecast probability density function (PDF; e.g., Schwartz et al. 2010 , 2014 ; Roberts et al. 2019 ), potentially leading to degraded reliability and underdispersion, especially in the absence of neighborhood evaluation or postprocessing methods ( Schwartz et al. 2014 ). Indeed, most CAMs and CAEs are currently underdispersive (e.g., Romine et al. 2014 ). One method to increase CAE spread is to increase the diversity of the ensemble membership, which can be achieved by

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

areas where significant progress has been made recently in the science of numerical weather prediction, including forecasting extreme weather events, and in exploiting satellite data. We then present a few potential directions that AI applications in Earth science may take in the future. We extend and update the perspective of Boukabara et al. (2019b) to include current activities, and expected future trends, based on presentations and discussion from the first National Oceanic and Atmospheric

Open access
Imme Ebert-Uphoff and Kyle Hilburn

function (aka loss function ) that continuously measures the NN’s performance during training, such as the mean square error of predictions generated for the training samples by the current NN. All NN weights are assigned random values at first. Then the loss function is minimized iteratively using gradient descent, i.e., the gradient of the loss function is calculated with respect to the NN weights and the NN weights are adjusted accordingly. This step is known as back propagation and is repeated

Full access