• Abadi, M., and et al. , 2015: TensorFlow: Large-scale machine learning on heterogeneous distributed systems. TensorFlow, 19 pp., https://static.googleusercontent.com/media/research.google.com/en//pubs/archive/45166.pdf.

  • Adames, Á. F., and D. Kim, 2016: The MJO as a dispersive, convectively coupled moisture wave: Theory and observations. J. Atmos. Sci., 73, 913941, https://doi.org/10.1175/JAS-D-15-0170.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Ahmed, F., and J. D. Neelin, 2018: Reverse engineering the tropical precipitation–buoyancy relationship. J. Atmos. Sci., 75, 15871608, https://doi.org/10.1175/JAS-D-17-0333.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Aires, F., M. Schmitt, A. Chedin, and N. Scott, 1999: The “weight smoothing” regularization of MLP for Jacobian stabilization. IEEE Trans. Neural Network, 10, 15021510, https://doi.org/10.1109/72.809096.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Beucler, T., 2019: Interaction between water vapor, radiation and convection in the tropics. Ph.D. thesis, Massachusetts Institute of Technology, 251 pp.

  • Beucler, T., T. Cronin, and K. Emanuel, 2018: A linear response framework for radiative-convective instability. J. Adv. Model. Earth Syst., 10, 19241951, https://doi.org/10.1029/2018MS001280.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Brenowitz, N. D., 2019: Coarse-grained near-global aqua-planet simulation with computed dynamical tendencies. Zenodo, accessed 1 August 2020, https://doi.org/10.5281/zenodo.4075647.

    • Crossref
    • Export Citation
  • Brenowitz, N. D., and C. S. Bretherton, 2018: Prognostic validation of a neural network unified physics parameterization. Geophys. Res. Lett., 45, 62896298, https://doi.org/10.1029/2018GL078510.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Brenowitz, N. D., and C. S. Bretherton, 2019: Spatially extended tests of a neural network parametrization trained by coarse-graining. J. Adv. Model. Earth Syst., 11, 27282744, https://doi.org/10.1029/2019MS001711.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Brenowitz, N. D., and A. Kwa, 2020: nbren12/uwnet v0.9.0 (v0.9.0). Zenodo, https://doi.org/10.5281/ZENODO.4273210.

    • Crossref
    • Export Citation
  • Bretherton, C. S., M. E. Peters, and L. E. Back, 2004: Relationships between water vapor path and precipitation over the tropical oceans. J. Climate, 17, 15171528, https://doi.org/10.1175/1520-0442(2004)017<1517:RBWVPA>2.0.CO;2.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Bretherton, C. S., P. N. Blossey, and M. Khairoutdinov, 2005: An energy-balance analysis of deep convective self-aggregation above uniform SST. J. Atmos. Sci., 62, 42734292, https://doi.org/10.1175/JAS3614.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Chevallier, F., and J.-F. Mahfouf, 2001: Evaluation of the Jacobians of infrared radiation models for variational data assimilation. J. Appl. Meteor., 40, 14451461, https://doi.org/10.1175/1520-0450(2001)040<1445:EOTJOI>2.0.CO;2.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Chevallier, F., F. Chéruy, N. A. Scott, and A. Chédin, 1998: A neural network approach for a fast and accurate computation of a longwave radiative budget. J. Appl. Meteor., 37, 13851397, https://doi.org/10.1175/1520-0450(1998)037<1385:ANNAFA>2.0.CO;2.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Emanuel, K. A., 1994: Atmospheric Convection. Oxford University Press, 580 pp.

  • Gentine, P., M. Pritchard, S. Rasp, G. Reinaudi, and G. Yacalis, 2018: Could machine learning break the convection parameterization deadlock? Geophys. Res. Lett., 45, 57425751, https://doi.org/10.1029/2018GL078202.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Hayashi, Y., 1971: Instability of large-scale equatorial waves with a frequency-dependent CISK parameter. J. Meteor. Soc. Japan, 49, 5962, https://doi.org/10.2151/jmsj1965.49.1_59.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Herman, M. J., and Z. Kuang, 2013: Linear response functions of two convective parameterization schemes. J. Adv. Model. Earth Syst., 5, 510541, https://doi.org/10.1002/jame.20037.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • IPCC, 2014: Climate Change 2013: The Physical Science Basis. Cambridge University Press, 1535 pp., https://doi.org/10.1017/CBO9781107415324.

    • Search Google Scholar
    • Export Citation
  • Khairoutdinov, M. F., and D. A. Randall, 2001: A cloud resolving model as a cloud parameterization in the NCAR Community Climate System Model: Preliminary results. Geophys. Res. Lett., 28, 36173620, https://doi.org/10.1029/2001GL013552.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Khairoutdinov, M. F., and D. A. Randall, 2003: Cloud resolving modeling of the ARM summer 1997 IOP: Model formulation, results, uncertainties, and sensitivities. J. Atmos. Sci., 60, 607625, https://doi.org/10.1175/1520-0469(2003)060<0607:CRMOTA>2.0.CO;2.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Khairoutdinov, M. F., D. A. Randall, and C. DeMott, 2005: Simulations of the atmospheric general circulation using a cloud-resolving model as a superparameterization of physical processes. J. Atmos. Sci., 62, 21362154, https://doi.org/10.1175/JAS3453.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Khouider, B., and A. J. Majda, 2006: A simple multicloud parameterization for convectively coupled tropical waves. Part I: Linear analysis. J. Atmos. Sci., 63, 13081323, https://doi.org/10.1175/JAS3677.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Krasnopolsky, V. M., 2007: Reducing uncertainties in neural network Jacobians and improving accuracy of neural network emulations with NN ensemble approaches. Neural Network, 20, 454461, https://doi.org/10.1016/j.neunet.2007.04.008.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Krasnopolsky, V. M., M. S. Fox-Rabinovitz, and D. V. Chalikov, 2005: New approach to calculation of atmospheric model physics: Accurate and fast neural network emulation of longwave radiation in a climate model. Mon. Wea. Rev., 133, 13701383, https://doi.org/10.1175/MWR2923.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Krasnopolsky, V. M., M. S. Fox-Rabinovitz, and A. A. Belochitski, 2013: Using ensemble of neural networks to learn stochastic convection parameterizations for climate and numerical weather prediction models from data simulated by a cloud resolving model. Adv. Artif. Neural Syst., 2013, 485913, https://doi.org/10.1155/2013/485913.

    • Search Google Scholar
    • Export Citation
  • Kuang, Z., 2008: A moisture-stratiform instability for convectively coupled waves. J. Atmos. Sci., 65, 834854, https://doi.org/10.1175/2007JAS2444.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Kuang, Z., 2010: Linear response functions of a cumulus ensemble to temperature and moisture perturbations and implications for the dynamics of convectively coupled waves. J. Atmos. Sci., 67, 941962, https://doi.org/10.1175/2009JAS3260.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Kuang, Z., 2018: Linear stability of moist convecting atmospheres Part I: From linear response functions to a simple model and applications to convectively coupled waves. J. Atmos. Sci., 75, 28892907, https://doi.org/10.1175/JAS-D-18-0092.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Majda, A. J., and M. G. Shefter, 2001: Waves and instabilities for model tropical convective parameterizations. J. Atmos. Sci., 58, 896914, https://doi.org/10.1175/1520-0469(2001)058<0896:WAIFMT>2.0.CO;2.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • McGovern, A., R. Lagerquist, D. John Gagne, G. E. Jergensen, K. L. Elmore, C. R. Homeyer, and T. Smith, 2019: Making the black box more transparent: Understanding the physical implications of machine learning. Bull. Amer. Meteor. Soc., 100, 21752199, https://doi.org/10.1175/BAMS-D-18-0195.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Molnar, C., G. Casalicchio, and B. Bischl, 2018: IML: An R package for interpretable machine learning. J. Open Source Software, 3, 786, https://doi.org/10.21105/joss.00786.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Montavon, G., W. Samek, and K.-R. Müller, 2018: Methods for interpreting and understanding deep neural networks. Digital Signal Process., 73, 115, https://doi.org/10.1016/j.dsp.2017.10.011.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Neelin, J. D., O. Peters, and K. Hales, 2009: The transition to strong convection. J. Atmos. Sci., 66, 23672384, https://doi.org/10.1175/2009JAS2962.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • O’Gorman, P. A., and J. G. Dwyer, 2018: Using machine learning to parameterize moist convection: Potential for modeling of climate, climate change, and extreme events. J. Adv. Model. Earth Syst., 10, 25482563, https://doi.org/10.1029/2018MS001351.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Oueslati, B., and G. Bellon, 2015: The double ITCZ bias in CMIP5 models: Interaction between SST, large-scale circulation and precipitation. Climate Dyn., 44, 585607, https://doi.org/10.1007/s00382-015-2468-6.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Palmer, T. N., 2001: A nonlinear dynamical perspective on model error: A proposal for non-local stochastic-dynamic parametrization in weather and climate prediction models. Quart. J. Roy. Meteor. Soc., 127, 279304, https://doi.org/10.1002/qj.49712757202.

    • Search Google Scholar
    • Export Citation
  • Paszke, A., and et al. , 2019: Pytorch: An imperative style, high-performance deep learning library. Advances in Neural Information Processing Systems, H. Wallach et al., Eds., Curran Associates, Inc., 8024–8035, http://papers.neurips.cc/paper/9015-pytorch-an-imperative-style-high-performance-deep-learning-library.pdf.

  • Pritchard, M. S., C. S. Bretherton, and C. A. DeMott, 2014: Restricting 32–128 km horizontal scales hardly affects the MJO in the Superparameterized Community Atmosphere Model v. 3.0 but the number of cloud-resolving grid columns constrains vertical mixing. J. Adv. Model. Earth Syst., 6, 723739, https://doi.org/10.1002/2014MS000340.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Rasp, S., 2019: Coupled online learning as a way to tackle instabilities and biases in neural network parameterizations. arXiv, 1907.01351, https://arxiv.org/abs/1907.01351.

  • Rasp, S., M. S. Pritchard, and P. Gentine, 2018: Deep learning to represent subgrid processes in climate models. Proc. Natl. Acad. Sci. USA, 115, 96849689, https://doi.org/10.1073/pnas.1810286115.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Rushley, S. S., D. Kim, C. S. Bretherton, and M.-S. Ahn, 2018: Reexamining the nonlinear moisture-precipitation relationship over the tropical oceans. Geophys. Res. Lett., 45, 11331140, https://doi.org/10.1002/2017GL076296.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Samek, W., T. Wiegand, and K.-R. Müller, 2017: Explainable artificial intelligence: Understanding, visualizing and interpreting deep learning models. arXiv, 1708.08296, https://arxiv.org/abs/1708.08296.

  • Schneider, T., J. Teixeira, C. S. Bretherton, F. Brient, K. G. Pressel, C. Schär, and A. P. Siebesma, 2017: Climate goals and computing the future of clouds. Nat. Climate Change, 7, 35, https://doi.org/10.1038/nclimate3190.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Sobel, A., and E. Maloney, 2013: Moisture modes and the eastward propagation of the MJO. J. Atmos. Sci., 70, 187192, https://doi.org/10.1175/JAS-D-12-0189.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Toms, B. A., E. A. Barnes, and I. Ebert-Uphoff, 2019: Physically interpretable neural networks for the geosciences: Applications to earth system variability. arXiv, 1912.01752, https://arxiv.org/abs/1912.01752.

  • Wood, R., and C. S. Bretherton, 2006: On the relationship between stratiform low cloud cover and lower-tropospheric stability. J. Climate, 19, 64256432, https://doi.org/10.1175/JCLI3988.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
All Time Past Year Past 30 Days
Abstract Views 726 726 68
Full Text Views 279 279 30
PDF Downloads 301 301 28

Interpreting and Stabilizing Machine-Learning Parametrizations of Convection

View More View Less
  • 1 Vulcan, Inc., Seattle, Washington
  • | 2 Department of Earth System Science, University of California, Irvine, Irvine, California
  • | 3 Department of Earth and Environmental Engineering, Columbia University, New York, New York
  • | 4 Department of Applied Mathematics, University of Washington, Seattle, Washington
  • | 5 Department of Atmospheric Sciences, University of Washington, Seattle, Washington
© Get Permissions Rent on DeepDyve
Restricted access

Abstract

Neural networks are a promising technique for parameterizing subgrid-scale physics (e.g., moist atmospheric convection) in coarse-resolution climate models, but their lack of interpretability and reliability prevents widespread adoption. For instance, it is not fully understood why neural network parameterizations often cause dramatic instability when coupled to atmospheric fluid dynamics. This paper introduces tools for interpreting their behavior that are customized to the parameterization task. First, we assess the nonlinear sensitivity of a neural network to lower-tropospheric stability and the midtropospheric moisture, two widely studied controls of moist convection. Second, we couple the linearized response functions of these neural networks to simplified gravity wave dynamics, and analytically diagnose the corresponding phase speeds, growth rates, wavelengths, and spatial structures. To demonstrate their versatility, these techniques are tested on two sets of neural networks, one trained with a superparameterized version of the Community Atmosphere Model (SPCAM) and the second with a near-global cloud-resolving model (GCRM). Even though the SPCAM simulation has a warmer climate than the cloud-resolving model, both neural networks predict stronger heating/drying in moist and unstable environments, which is consistent with observations. Moreover, the spectral analysis can predict that instability occurs when GCMs are coupled to networks that support gravity waves that are unstable and have phase speeds larger than 5 m s−1. In contrast, standing unstable modes do not cause catastrophic instability. Using these tools, differences between the SPCAM-trained versus GCRM-trained neural networks are analyzed, and strategies to incrementally improve both of their coupled online performance unveiled.

Supplemental information related to this paper is available at the Journals Online website: https://doi.org/10.1175/JAS-D-20-0082.s1.

© 2020 American Meteorological Society. For information regarding reuse of this content and general copyright information, consult the AMS Copyright Policy (www.ametsoc.org/PUBSReuseLicenses).

Corresponding author: Noah D. Brenowitz, noahb@vulcan.com

Abstract

Neural networks are a promising technique for parameterizing subgrid-scale physics (e.g., moist atmospheric convection) in coarse-resolution climate models, but their lack of interpretability and reliability prevents widespread adoption. For instance, it is not fully understood why neural network parameterizations often cause dramatic instability when coupled to atmospheric fluid dynamics. This paper introduces tools for interpreting their behavior that are customized to the parameterization task. First, we assess the nonlinear sensitivity of a neural network to lower-tropospheric stability and the midtropospheric moisture, two widely studied controls of moist convection. Second, we couple the linearized response functions of these neural networks to simplified gravity wave dynamics, and analytically diagnose the corresponding phase speeds, growth rates, wavelengths, and spatial structures. To demonstrate their versatility, these techniques are tested on two sets of neural networks, one trained with a superparameterized version of the Community Atmosphere Model (SPCAM) and the second with a near-global cloud-resolving model (GCRM). Even though the SPCAM simulation has a warmer climate than the cloud-resolving model, both neural networks predict stronger heating/drying in moist and unstable environments, which is consistent with observations. Moreover, the spectral analysis can predict that instability occurs when GCMs are coupled to networks that support gravity waves that are unstable and have phase speeds larger than 5 m s−1. In contrast, standing unstable modes do not cause catastrophic instability. Using these tools, differences between the SPCAM-trained versus GCRM-trained neural networks are analyzed, and strategies to incrementally improve both of their coupled online performance unveiled.

Supplemental information related to this paper is available at the Journals Online website: https://doi.org/10.1175/JAS-D-20-0082.s1.

© 2020 American Meteorological Society. For information regarding reuse of this content and general copyright information, consult the AMS Copyright Policy (www.ametsoc.org/PUBSReuseLicenses).

Corresponding author: Noah D. Brenowitz, noahb@vulcan.com

Supplementary Materials

    • Supplemental Materials (ZIP 5.78 MB)
Save