### Abstract

The poor representation of tropical convection in general circulation models (GCMs) is believed to be responsible for much of the uncertainty in the predictions of weather and climate in the tropics. The stochastic multicloud model (SMCM) was recently developed by Khouider et al. (Commun Math Sci 8(1):187–216, 2010) to represent the missing variability in GCMs due to unresolved features of organized tropical convection. The SMCM is based on three cloud types (congestus, deep and stratiform), and transitions between these cloud types are formalized in terms of probability rules that are functions of the large-scale environment convective state and a set of seven arbitrary cloud timescale parameters. Here, a statistical inference method based on the Bayesian paradigm is applied to estimate these key cloud timescales from the Giga-LES dataset, a 24-h large-eddy simulation (LES) of deep tropical convection (Khairoutdinov et al. in J Adv Model Earth Syst 1(12), 2009) over a domain comparable to a GCM gridbox. A sequential learning strategy is used where the Giga-LES domain is partitioned into a few subdomains, and atmospheric time series obtained on each subdomain are used to train the Bayesian procedure incrementally. Convergence of the marginal posterior densities for all seven parameters is demonstrated for two different grid partitions, and sensitivity tests to other model parameters are also presented. A single column model simulation using the SMCM parameterization with the Giga-LES inferred parameters reproduces many important statistical features of the Giga-LES run, without any further tuning. In particular it exhibits intermittent dynamical behavior in both the stochastic cloud fractions and the large scale dynamics, with periods of dry phases followed by a coherent sequence of congestus, deep, and stratiform convection, varying on timescales of a few hours consistent with the Giga-LES time series. The chaotic variations of the cloud area fractions were captured fairly well both qualitatively and quantitatively demonstrating the stochastic nature of convection in the Giga-LES simulation.

Original language | English (US) |
---|---|

Pages (from-to) | 1845-1861 |

Number of pages | 17 |

Journal | Climate Dynamics |

Volume | 47 |

Issue number | 5-6 |

DOIs | |

State | Published - Sep 1 2016 |

### Fingerprint

### Keywords

- Bayesian inference
- General circulation models
- Giga-LES
- Large matrix exponential
- Markov Chain Monte Carlo
- Parallel and high performance computing
- Parameter estimation
- Stochastic cumulus parameterization
- Stochastic multicloud model
- Tropical convection

### ASJC Scopus subject areas

- Atmospheric Science

### Cite this

*Climate Dynamics*,

*47*(5-6), 1845-1861. https://doi.org/10.1007/s00382-015-2936-z

**Stochasticity of convection in Giga-LES data.** / De La Chevrotière, Michèle; Khouider, Boualem; Majda, Andrew J.

Research output: Contribution to journal › Article

*Climate Dynamics*, vol. 47, no. 5-6, pp. 1845-1861. https://doi.org/10.1007/s00382-015-2936-z

}

TY - JOUR

T1 - Stochasticity of convection in Giga-LES data

AU - De La Chevrotière, Michèle

AU - Khouider, Boualem

AU - Majda, Andrew J.

PY - 2016/9/1

Y1 - 2016/9/1

N2 - The poor representation of tropical convection in general circulation models (GCMs) is believed to be responsible for much of the uncertainty in the predictions of weather and climate in the tropics. The stochastic multicloud model (SMCM) was recently developed by Khouider et al. (Commun Math Sci 8(1):187–216, 2010) to represent the missing variability in GCMs due to unresolved features of organized tropical convection. The SMCM is based on three cloud types (congestus, deep and stratiform), and transitions between these cloud types are formalized in terms of probability rules that are functions of the large-scale environment convective state and a set of seven arbitrary cloud timescale parameters. Here, a statistical inference method based on the Bayesian paradigm is applied to estimate these key cloud timescales from the Giga-LES dataset, a 24-h large-eddy simulation (LES) of deep tropical convection (Khairoutdinov et al. in J Adv Model Earth Syst 1(12), 2009) over a domain comparable to a GCM gridbox. A sequential learning strategy is used where the Giga-LES domain is partitioned into a few subdomains, and atmospheric time series obtained on each subdomain are used to train the Bayesian procedure incrementally. Convergence of the marginal posterior densities for all seven parameters is demonstrated for two different grid partitions, and sensitivity tests to other model parameters are also presented. A single column model simulation using the SMCM parameterization with the Giga-LES inferred parameters reproduces many important statistical features of the Giga-LES run, without any further tuning. In particular it exhibits intermittent dynamical behavior in both the stochastic cloud fractions and the large scale dynamics, with periods of dry phases followed by a coherent sequence of congestus, deep, and stratiform convection, varying on timescales of a few hours consistent with the Giga-LES time series. The chaotic variations of the cloud area fractions were captured fairly well both qualitatively and quantitatively demonstrating the stochastic nature of convection in the Giga-LES simulation.

AB - The poor representation of tropical convection in general circulation models (GCMs) is believed to be responsible for much of the uncertainty in the predictions of weather and climate in the tropics. The stochastic multicloud model (SMCM) was recently developed by Khouider et al. (Commun Math Sci 8(1):187–216, 2010) to represent the missing variability in GCMs due to unresolved features of organized tropical convection. The SMCM is based on three cloud types (congestus, deep and stratiform), and transitions between these cloud types are formalized in terms of probability rules that are functions of the large-scale environment convective state and a set of seven arbitrary cloud timescale parameters. Here, a statistical inference method based on the Bayesian paradigm is applied to estimate these key cloud timescales from the Giga-LES dataset, a 24-h large-eddy simulation (LES) of deep tropical convection (Khairoutdinov et al. in J Adv Model Earth Syst 1(12), 2009) over a domain comparable to a GCM gridbox. A sequential learning strategy is used where the Giga-LES domain is partitioned into a few subdomains, and atmospheric time series obtained on each subdomain are used to train the Bayesian procedure incrementally. Convergence of the marginal posterior densities for all seven parameters is demonstrated for two different grid partitions, and sensitivity tests to other model parameters are also presented. A single column model simulation using the SMCM parameterization with the Giga-LES inferred parameters reproduces many important statistical features of the Giga-LES run, without any further tuning. In particular it exhibits intermittent dynamical behavior in both the stochastic cloud fractions and the large scale dynamics, with periods of dry phases followed by a coherent sequence of congestus, deep, and stratiform convection, varying on timescales of a few hours consistent with the Giga-LES time series. The chaotic variations of the cloud area fractions were captured fairly well both qualitatively and quantitatively demonstrating the stochastic nature of convection in the Giga-LES simulation.

KW - Bayesian inference

KW - General circulation models

KW - Giga-LES

KW - Large matrix exponential

KW - Markov Chain Monte Carlo

KW - Parallel and high performance computing

KW - Parameter estimation

KW - Stochastic cumulus parameterization

KW - Stochastic multicloud model

KW - Tropical convection

UR - http://www.scopus.com/inward/record.url?scp=84983260392&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84983260392&partnerID=8YFLogxK

U2 - 10.1007/s00382-015-2936-z

DO - 10.1007/s00382-015-2936-z

M3 - Article

AN - SCOPUS:84983260392

VL - 47

SP - 1845

EP - 1861

JO - Climate Dynamics

JF - Climate Dynamics

SN - 0930-7575

IS - 5-6

ER -