Physically regularized machine learning emulators of aerosol activation

Abstract. The activation of aerosol into cloud droplets is an important step in the formation of clouds and strongly influences the radiative budget of the Earth. Explicitly simulating aerosol activation in Earth system models is challenging due to the computational complexity required to resolve the necessary chemical and physical processes and their interactions. As such, various parameterizations have been developed to approximate these details at reduced computational cost and accuracy. Here, we explore how machine learning emulators can be used to bridge this gap in computational cost and parameterization accuracy. We evaluate a set of emulators of a detailed cloud parcel model using physically regularized machine learning regression techniques. We find that the emulators can reproduce the parcel model at higher accuracy than many existing parameterizations. Furthermore, physical regularization tends to improve emulator accuracy, most significantly when emulating very low activation fractions. This work demonstrates the value of physical constraints in machine learning model development and enables the implementation of improved hybrid physical and machine learning models of aerosol activation into next-generation Earth system models.

[1]  Noah D. Brenowitz,et al.  Prognostic Validation of a Neural Network Unified Physics Parameterization , 2018, Geophysical Research Letters.

[2]  J. Seinfeld,et al.  Atmospheric Chemistry and Physics: From Air Pollution to Climate Change , 1997 .

[3]  Peter D. Ivatt,et al.  Improving the prediction of an atmospheric chemistry transport model using gradient-boosted regression trees , 2020 .

[4]  Trevor Hastie,et al.  Regularization Paths for Generalized Linear Models via Coordinate Descent. , 2010, Journal of statistical software.

[5]  Ville Kolehmainen,et al.  Correction of approximation errors with Random Forests applied to modelling of cloud droplet formation , 2013 .

[6]  Tianqi Chen,et al.  XGBoost: A Scalable Tree Boosting System , 2016, KDD.

[7]  S. Twomey.  The Influence of Pollution on the Shortwave Albedo of Clouds , 1977 .

[8]  Markus Reichstein,et al.  Physics‐Constrained Machine Learning of Evapotranspiration , 2019, Geophysical Research Letters.

[9]  J. Mülmenstädt,et al.  Bounding Global Aerosol Radiative Forcing of Climate Change , 2020, Reviews of geophysics.

[10]  Pierre Baldi,et al.  Enforcing Analytic Constraints in Neural Networks Emulating Physical Systems. , 2019, Physical review letters.

[11]  Athanasios Nenes,et al.  Continued development of a cloud droplet formation parameterization for global climate models , 2005 .

[12]  I. Mammarella,et al.  A Deep Learning Parameterization for Ozone Dry Deposition Velocities , 2019, Geophysical Research Letters.

[13]  C. Bretherton,et al.  Improving our fundamental understanding of the role of aerosol−cloud interactions in the climate system , 2016, Proceedings of the National Academy of Sciences.

[14]  S. Twomey,et al.  The nuclei of natural cloud formation part II: The supersaturation in natural clouds and the variation of cloud droplet concentration , 1959 .

[15]  P. Nowack,et al.  Using machine learning to build temperature-based ozone parameterizations for climate sensitivity simulations , 2018, Environmental Research Letters.

[16]  H. Cai,et al.  Evaluation of SVM, ELM and four tree-based ensemble models for predicting daily reference evapotranspiration using limited meteorological data in different climates of China , 2018, Agricultural and Forest Meteorology.

[17]  Robert Tibshirani,et al.  The Elements of Statistical Learning: Data Mining, Inference, and Prediction , 2001, Springer Series in Statistics.

[18]  B. Albrecht.  Aerosols, Cloud Microphysics, and Fractional Cloudiness , 1989, Science.

[19]  R. Carey.  Atmospheric Science: An Introductory Survey , 1978 .

[20]  S. Twomey.  Pollution and the Planetary Albedo , 1974 .

[21]  M. Christensen,et al.  Aerosols enhance cloud lifetime and brightness along the stratus-to-cumulus transition , 2020, Proceedings of the National Academy of Sciences.

[22]  Joaquim R. R. A. Martins,et al.  A Python surrogate modeling framework with derivatives , 2019, Adv. Eng. Softw..

[23]  C. Bretherton,et al.  Combining Emergent Constraints for Climate Sensitivity , 2020 .

[24]  B. Stevens,et al.  Bounding aerosol radiative forcing of climate change , 2019 .

[25]  Mikhail Ovchinnikov,et al.  Droplet nucleation: Physically‐based parameterizations and comparative evaluation , 2011 .

[26]  S. Ghan,et al.  A parameterization of aerosol activation: 2. Multiple aerosol types , 2000 .

[27]  Pierre Gentine,et al.  Could Machine Learning Break the Convection Parameterization Deadlock? , 2018, Geophysical Research Letters.

[28]  V. Ramaswamy,et al.  A New Parameterization of Cloud Droplet Activation Applicable to General Circulation Models , 2006 .

[29]  Pierre Gentine,et al.  Deep learning to represent subgrid processes in climate models , 2018, Proceedings of the National Academy of Sciences.

[30]  Paris Perdikaris,et al.  Physics-informed neural networks: A deep learning framework for solving forward and inverse problems involving nonlinear partial differential equations , 2019, J. Comput. Phys..

[31]  M. J. Evans,et al.  Improving the prediction of an atmospheric chemistry transport model using gradient-boosted regression trees , 2019, Atmospheric Chemistry and Physics.

[32]  Joachim Denzler,et al.  Deep learning and process understanding for data-driven Earth system science , 2019, Nature.

[33]  Sam J. Silva,et al.  Development of a reduced-complexity plant canopy physics surrogate model for use in chemical transport models: a case study with GEOS-Chem v12.3.0 , 2020, Geoscientific Model Development.

[34]  Sam J. Silva,et al.  Exploring the Constraints on Simulated Aerosol Sources and Transport Across the North Atlantic With Island‐Based Sun Photometers , 2020, Earth and space science.

[35]  Chien Wang,et al.  Metamodeling of Droplet Activation for Global Climate Models , 2016 .