scholarly journals Deep learning to represent subgrid processes in climate models

2018 ◽  
Vol 115 (39) ◽  
pp. 9684-9689 ◽  
Author(s):  
Stephan Rasp ◽  
Michael S. Pritchard ◽  
Pierre Gentine
Keyword(s):  
Neural Network ◽  
Deep Learning ◽  
General Circulation ◽  
Climate Models ◽  
Climate Model ◽  
Wave Spectrum ◽  
Computational Cost ◽  
Circulation Model ◽  
The Neural Network ◽  
Equatorial Wave

The representation of nonlinear subgrid processes, especially clouds, has been a major source of uncertainty in climate models for decades. Cloud-resolving models better represent many of these processes and can now be run globally but only for short-term simulations of at most a few years because of computational limitations. Here we demonstrate that deep learning can be used to capture many advantages of cloud-resolving modeling at a fraction of the computational cost. We train a deep neural network to represent all atmospheric subgrid processes in a climate model by learning from a multiscale model in which convection is treated explicitly. The trained neural network then replaces the traditional subgrid parameterizations in a global general circulation model in which it freely interacts with the resolved dynamics and the surface-flux scheme. The prognostic multiyear simulations are stable and closely reproduce not only the mean climate of the cloud-resolving simulation but also key aspects of variability, including precipitation extremes and the equatorial wave spectrum. Furthermore, the neural network approximately conserves energy despite not being explicitly instructed to. Finally, we show that the neural network parameterization generalizes to new surface forcing patterns but struggles to cope with temperatures far outside its training manifold. Our results show the feasibility of using deep learning for climate model parameterization. In a broader context, we anticipate that data-driven Earth system model development could play a key role in reducing climate prediction uncertainty in the coming decade.

scholarly journals Downscaling diurnal temperature over west and southwest Iran : A comparison of statistical downscaling approaches

Időjárás ◽  
2021 ◽  
Vol 125 (3) ◽  
pp. 491-511
Author(s):  
Mohammad Reza Poodineh
Keyword(s):  
Neural Network ◽  
Statistical Downscaling ◽  
General Circulation ◽  
Climate Model ◽  
Global Climate Model ◽  
Circulation Model ◽  
General Circulation Models ◽  
Baseline Period ◽  
Mountainous Regions ◽  
The Neural Network

This study aimed to forecast temperature variations in the western and southwestern part of Iran using a general circulation model and artificial neural networks (ANN). The data included mean diurnal temperatures from synoptic stations, National Centers for Environmental Prediction/ National Center for Atmospheric Research (NCEP/NCAR) reanalysis data, and outputs of a third-generation global climate model, the Hadley Centre Coupled Model version 3 (HadCM3), under A2 and B2 scenarios for the baseline period (1961–1990). The data of the first (1961–1975) and second 15 years (1976–1990) of the baseline period were used for model calibration and validation, respectively. Both models, however, produced reliable estimates at the plain stations with neither outperforming the other due to their negligible errors. However, the neural network results of mountain synoptic stations showed a lower error rate than the statistical downscaling model (SDSM) outputs. All in all, we can say that there was a larger amount of error in the outputs of the atmospheric general circulation models (AGCMs) in the mountainous regions. According to the outputs of the neural network and the AGCMs, temperatures at the studied stations were on the rise. In fact, this increase was more noticeable at the plain stations. This can be attributed to their proximity to the sea, to their latitude, and to the more intensive industrial activities (especially, extraction of petroleum and production of petroleum products) taking place near the plain stations.

scholarly journals A description of the FAMOUS (version XDBUA) climate model and control run

2008 ◽  
Vol 1 (1) ◽  
pp. 53-68 ◽  
Author(s):  
R. S. Smith ◽  
J. M. Gregory ◽  
A. Osprey
Keyword(s):  
General Circulation ◽  
High Performance ◽  
Climate Model ◽  
Computational Cost ◽  
Circulation Model ◽  
Cold Bias ◽  
Northern Latitudes ◽  
Climate Simulations ◽  
Atmosphere General Circulation Model ◽  
And Control

Abstract. FAMOUS is an ocean-atmosphere general circulation model of low resolution, capable of simulating approximately 120 years of model climate per wallclock day using current high performance computing facilities. It uses most of the same code as HadCM3, a widely used climate model of higher resolution and computational cost, and has been tuned to reproduce the same climate reasonably well. FAMOUS is useful for climate simulations where the computational cost makes the application of HadCM3 unfeasible, either because of the length of simulation or the size of the ensemble desired. We document a number of scientific and technical improvements to the original version of FAMOUS. These improvements include changes to the parameterisations of ozone and sea-ice which alleviate a significant cold bias from high northern latitudes and the upper troposphere, and the elimination of volume-averaged drifts in ocean tracers. A simple model of the marine carbon cycle has also been included. A particular goal of FAMOUS is to conduct millennial-scale paleoclimate simulations of Quaternary ice ages; to this end, a number of useful changes to the model infrastructure have been made.

scholarly journals The 1986–1989 ENSO cycle in a chemical climate model

2006 ◽  
Vol 6 (12) ◽  
pp. 4669-4685 ◽  
Author(s):  
S. Brönnimann ◽  
M. Schraner ◽  
B. Müller ◽  
A. Fischer ◽  
D. Brunner ◽  
...  
Keyword(s):  
North Atlantic ◽  
General Circulation ◽  
Climate Models ◽  
Climate Model ◽  
Circulation Model ◽  
Enso Cycle ◽  
Atlantic Sector ◽  
Ensemble Simulations ◽  
Chemical Climate ◽  
Middle Stratosphere

Abstract. A pronounced ENSO cycle occurred from 1986 to 1989, accompanied by distinct dynamical and chemical anomalies in the global troposphere and stratosphere. Reproducing these effects with current climate models not only provides a model test but also contributes to our still limited understanding of ENSO's effect on stratosphere-troposphere coupling. We performed several sets of ensemble simulations with a chemical climate model (SOCOL) forced with global sea surface temperatures. Results were compared with observations and with large-ensemble simulations performed with an atmospheric general circulation model (MRF9). We focus our analysis on the extratropical stratosphere and its coupling with the troposphere. In this context, the circulation over the North Atlantic sector is particularly important. Relative to the La Niña winter 1989, observations for the El Niño winter 1987 show a negative North Atlantic Oscillation index with corresponding changes in temperature and precipitation patterns, a weak polar vortex, a warm Arctic middle stratosphere, negative and positive total ozone anomalies in the tropics and at middle to high latitudes, respectively, as well as anomalous upward and poleward Eliassen-Palm (EP) flux in the midlatitude lower stratosphere. Most of the tropospheric features are well reproduced in the ensemble means in both models, though the amplitudes are underestimated. In the stratosphere, the SOCOL simulations compare well with observations with respect to zonal wind, temperature, EP flux, meridional mass streamfunction, and ozone, but magnitudes are underestimated in the middle stratosphere. With respect to the mechanisms relating ENSO to stratospheric circulation, the results suggest that both, upward and poleward components of anomalous EP flux are important for obtaining the stratospheric signal and that an increase in strength of the Brewer-Dobson circulation is part of that signal.

scholarly journals Prognostic parameterization of cloud ice with a single category in the aerosol-climate model ECHAM(v6.3.0)-HAM(v2.3)

2017 ◽  
Author(s):  
Remo Dietlicher ◽  
David Neubauer ◽  
Ulrike Lohmann
Keyword(s):  
General Circulation ◽  
Climate Models ◽  
Climate Model ◽  
Traditional Approach ◽  
Circulation Model ◽  
Cloud Microphysics ◽  
High Temporal Resolution ◽  
Conversion Rates ◽  
Single Category ◽  
Cloud Ice

Abstract. A new scheme for stratiform cloud microphysics has been implemented in the ECHAM6-HAM2 general circulation model. It features a widely used description of cloud water with two categories for cloud droplets and rain drops. The unique aspect of the scheme is the break with the traditional approach to describe cloud ice analogously. Here we parameterize cloud ice with a single, prognostic category as it has been done in regional models and most recently also in the global model CAM5. A single category does not rely on heuristic conversion rates from one category to another. At the same time it is conceptually easier and closer to first principles. This work shows that a single category is a viable approach to describe cloud ice in climate models. Prognostic representation of sedimentation is achieved by a nested approach for sub-stepping the microphysics scheme. This yields good results in terms of numerical stability and accuracy as compared to simulations with high temporal resolution. The improvement of the representation of cloud ice in ECHAM6-HAM2 is twofold. Not only are we getting rid of heuristic conversion rates but we also find that the prognostic treatment of sedimenting ice allows to unbiasedly represent the ice formation pathway from nucleation over growth by deposition and collisions to sedimentation.

Developing a data-driven ocean model

2020 ◽  
Author(s):  
Rachel Furner ◽  
Peter Haynes ◽  
Dan Jones ◽  
Dave Munday ◽  
Brooks Paige ◽  
...  
Keyword(s):  
Neural Network ◽  
Machine Learning ◽  
General Circulation Model ◽  
General Circulation ◽  
Climate Models ◽  
Circulation Model ◽  
Machine Learning Algorithms ◽  
Data Driven ◽  
Geophysical Research ◽  
Weather And Climate

<p>The recent boom in machine learning and data science has led to a number of new opportunities in the environmental sciences. In particular, climate models represent the best tools we have to predict, understand and potentially mitigate climate change, however these process-based models are incredibly complex and require huge amounts of high-performance computing resources. Machine learning offers opportunities to greatly improve the computational efficiency of these models.</p><p>Here we discuss our recent efforts to reduce the computational cost associated with running a process-based model of the physical ocean by developing an analogous data-driven model. We train statistical and machine learning algorithms using the outputs from a highly idealised sector configuration of general circulation model (MITgcm). Our aim is to develop an algorithm which is able to predict the future state of the general circulation model to a similar level of accuracy in a more computationally efficient manner.</p><p>We first develop a linear regression model to investigate the sensitivity of data-driven approaches to various inputs, e.g. temperature on different spatial and temporal scales, and meta-variables such as location information. Following this, we develop a neural network model to replicate the general circulation model, as in the work of Dueben and Bauer 2018, and Scher 2018.</p><p>We present a discussion on the sensitivity of data-driven models and preliminary results from the neural network based model.</p><p> </p><p><em>Dueben, P. D., & Bauer, P. (2018). Challenges and design choices for global weather and climate models based on machine learning. Geoscientific Model Development, 11(10), 3999-4009.</em></p><p><em>Scher, S. (2018). Toward Data‐Driven Weather and Climate Forecasting: Approximating a Simple General Circulation Model With Deep Learning. Geophysical Research Letters, 45(22), 12-616.</em></p>

scholarly journals Hydroclimate variability in Scandinavia over the last millennium – insights from a climate model–proxy data comparison

2017 ◽  
Vol 13 (12) ◽  
pp. 1831-1850 ◽  
Author(s):  
Kristina Seftigen ◽  
Hugues Goosse ◽  
Francois Klein ◽  
Deliang Chen
Keyword(s):  
General Circulation ◽  
Climate Models ◽  
Climate Model ◽  
Model Performance ◽  
Circulation Model ◽  
Last Millennium ◽  
Atmospheric Variability ◽  
Out Of Sample ◽  
Hydroclimate Variability ◽  
Transient Simulations

Abstract. The integration of climate proxy information with general circulation model (GCM) results offers considerable potential for deriving greater understanding of the mechanisms underlying climate variability, as well as unique opportunities for out-of-sample evaluations of model performance. In this study, we combine insights from a new tree-ring hydroclimate reconstruction from Scandinavia with projections from a suite of forced transient simulations of the last millennium and historical intervals from the CMIP5 and PMIP3 archives. Model simulations and proxy reconstruction data are found to broadly agree on the modes of atmospheric variability that produce droughts–pluvials in the region. Despite these dynamical similarities, large differences between simulated and reconstructed hydroclimate time series remain. We find that the GCM-simulated multi-decadal and/or longer hydroclimate variability is systematically smaller than the proxy-based estimates, whereas the dominance of GCM-simulated high-frequency components of variability is not reflected in the proxy record. Furthermore, the paleoclimate evidence indicates in-phase coherencies between regional hydroclimate and temperature on decadal timescales, i.e., sustained wet periods have often been concurrent with warm periods and vice versa. The CMIP5–PMIP3 archive suggests, however, out-of-phase coherencies between the two variables in the last millennium. The lack of adequate understanding of mechanisms linking temperature and moisture supply on longer timescales has serious implications for attribution and prediction of regional hydroclimate changes. Our findings stress the need for further paleoclimate data–model intercomparison efforts to expand our understanding of the dynamics of hydroclimate variability and change, to enhance our ability to evaluate climate models, and to provide a more comprehensive view of future drought and pluvial risks.

scholarly journals The BRIDGE HadCM3 family of climate models: HadCM3@Bristol v1.0

2017 ◽  
Vol 10 (10) ◽  
pp. 3715-3743 ◽  
Author(s):  
Paul J. Valdes ◽  
Edward Armstrong ◽  
Marcus P. S. Badger ◽  
Catherine D. Bradshaw ◽  
Fran Bragg ◽  
...  
Keyword(s):  
Climate Change ◽  
Ocean Circulation ◽  
Land Surface ◽  
General Circulation ◽  
Computational Models ◽  
Climate Models ◽  
Climate Model ◽  
Circulation Model ◽  
Good Representation ◽  
Comprehensive Overview

Abstract. Understanding natural and anthropogenic climate change processes involves using computational models that represent the main components of the Earth system: the atmosphere, ocean, sea ice, and land surface. These models have become increasingly computationally expensive as resolution is increased and more complex process representations are included. However, to gain robust insight into how climate may respond to a given forcing, and to meaningfully quantify the associated uncertainty, it is often required to use either or both ensemble approaches and very long integrations. For this reason, more computationally efficient models can be very valuable tools. Here we provide a comprehensive overview of the suite of climate models based around the HadCM3 coupled general circulation model. This model was developed at the UK Met Office and has been heavily used during the last 15 years for a range of future (and past) climate change studies, but has now been largely superseded for many scientific studies by more recently developed models. However, it continues to be extensively used by various institutions, including the BRIDGE (Bristol Research Initiative for the Dynamic Global Environment) research group at the University of Bristol, who have made modest adaptations to the base HadCM3 model over time. These adaptations mean that the original documentation is not entirely representative, and several other relatively undocumented configurations are in use. We therefore describe the key features of a number of configurations of the HadCM3 climate model family, which together make up HadCM3@Bristol version 1.0. In order to differentiate variants that have undergone development at BRIDGE, we have introduced the letter B into the model nomenclature. We include descriptions of the atmosphere-only model (HadAM3B), the coupled model with a low-resolution ocean (HadCM3BL), the high-resolution atmosphere-only model (HadAM3BH), and the regional model (HadRM3B). These also include three versions of the land surface scheme. By comparing with observational datasets, we show that these models produce a good representation of many aspects of the climate system, including the land and sea surface temperatures, precipitation, ocean circulation, and vegetation. This evaluation, combined with the relatively fast computational speed (up to 1000 times faster than some CMIP6 models), motivates continued development and scientific use of the HadCM3B family of coupled climate models, predominantly for quantifying uncertainty and for long multi-millennial-scale simulations.

scholarly journals Abrupt Circulation Responses to Tropical Upper-Tropospheric Warming in a Relatively Simple Stratosphere-Resolving AGCM

2012 ◽  
Vol 25 (12) ◽  
pp. 4097-4115 ◽  
Author(s):  
Shuguang Wang ◽  
Edwin P. Gerber ◽  
Lorenzo M. Polvani
Keyword(s):  
Climate Change ◽  
General Circulation ◽  
Climate Models ◽  
Climate Model ◽  
Polar Vortex ◽  
Circulation Model ◽  
Hadley Cell ◽  
Critical Threshold ◽  
Intergovernmental Panel ◽  
Upper Troposphere

Abstract The circulation response of the atmosphere to climate change–like thermal forcing is explored with a relatively simple, stratosphere-resolving general circulation model. The model is forced with highly idealized physics, but integrates the primitive equations at resolution comparable to comprehensive climate models. An imposed forcing mimics the warming induced by greenhouse gasses in the low-latitude upper troposphere. The forcing amplitude is progressively increased over a range comparable in magnitude to the warming projected by Intergovernmental Panel on Climate Change coupled climate model scenarios. For weak to moderate warming, the circulation response is remarkably similar to that found in comprehensive models: the Hadley cell widens and weakens, the tropospheric midlatitude jets shift poleward, and the Brewer–Dobson circulation (BDC) increases. However, when the warming of the tropical upper troposphere exceeds a critical threshold, ~5 K, an abrupt change of the atmospheric circulation is observed. In the troposphere the extratropical eddy-driven jet jumps poleward nearly 10°. In the stratosphere the polar vortex intensifies and the BDC weakens as the intraseasonal coupling between the troposphere and the stratosphere shuts down. The key result of this study is that an abrupt climate transition can be effected by changes in atmospheric dynamics alone, without need for the strong nonlinearities typically associated with physical parameterizations. It is verified that the abrupt climate shift reported here is not an artifact of the model’s resolution or numerics.

scholarly journals Tropical forests and the global carbon cycle: impacts of atmospheric carbon dioxide, climate change and rate of deforestation

2004 ◽  
Vol 359 (1443) ◽  
pp. 331-343 ◽  
Author(s):  
Wolfgang Cramer ◽  
Alberte Bondeau ◽  
Sibyll Schaphoff ◽  
Wolfgang Lucht ◽  
Benjamin Smith ◽  
...  
Keyword(s):  
Climate Change ◽  
Tropical Forests ◽  
General Circulation ◽  
Climate Models ◽  
Climate Model ◽  
Carbon Sink ◽  
Circulation Model ◽  
Tropical Deforestation ◽  
Dynamic Global Vegetation Model ◽  
Multiple Scenarios

The remaining carbon stocks in wet tropical forests are currently at risk because of anthropogenic deforestation, but also because of the possibility of release driven by climate change. To identify the relative roles of CO 2 increase, changing temperature and rainfall, and deforestation in the future, and the magnitude of their impact on atmospheric CO 2 concentrations, we have applied a dynamic global vegetation model, using multiple scenarios of tropical deforestation (extrapolated from two estimates of current rates) and multiple scenarios of changing climate (derived from four independent offline general circulation model simulations). Results show that deforestation will probably produce large losses of carbon, despite the uncertainty about the deforestation rates. Some climate models produce additional large fluxes due to increased drought stress caused by rising temperature and decreasing rainfall. One climate model, however, produces an additional carbon sink. Taken together, our estimates of additional carbon emissions during the twenty–first century, for all climate and deforestation scenarios, range from 101 to 367 Gt C, resulting in CO 2 concentration increases above background values between 29 and 129 p.p.m. An evaluation of the method indicates that better estimates of tropical carbon sources and sinks require improved assessments of current and future deforestation, and more consistent precipitation scenarios from climate models. Notwithstanding the uncertainties, continued tropical deforestation will most certainly play a very large role in the build–up of future greenhouse gas concentrations.

scholarly journals Hydroclimate variability in Scandinavia over the last millennium – insights from a climate model-proxy data comparison

2017 ◽  
Author(s):  
Kristina Seftigen ◽  
Hugues Goosse ◽  
Francois Klein ◽  
Deliang Chen
Keyword(s):  
General Circulation ◽  
Climate Models ◽  
Climate Model ◽  
Global Climate ◽  
Model Performance ◽  
Circulation Model ◽  
Global Climate Models ◽  
Last Millennium ◽  
Atmospheric Variability ◽  
Out Of Sample

Abstract. The integration of climate proxy information with General Circulation Model (GCM) results offers considerable potential for deriving greater understanding of the mechanisms underlying climate variability, as well as unique opportunities for out-of-sample evaluations of model performance. In this study, we combine insights from a new tree-ring hydroclimate reconstruction from Scandinavian with projections from a suite of forced transient simulations of the last millennium and historical intervals from the CMIP5 and PMIP3 archives. Model simulations and proxy reconstruction data are found to broadly agree on the modes of atmospheric variability that produces droughts/pluvials in the region. But despite these dynamical similarities, large differences between simulated and reconstructed hydroclimate time series remain. We find simulated interannual components of variability to be overestimated, while the multidecadal/longer timescale components generally are too weak. Specifically, summertime moisture variability and temperature are weakly negatively associated at inter-annual timescales but positively correlated at decadal timescales, revealed from observational and proxy evidences. On this background, the CMIP5/PMIP3 simulated timescale dependent relationship between regional precipitation and temperature is considerably biased, because the short-term negative association is overestimated, and the long-term relationship is significantly underestimated. The lack of adequate understanding for mechanisms linking temperature and moisture supply on longer timescales has important implication for future projections. Weak multidecadal variability in models also implies that inference about future persistent droughts and pluvials based on the latest generation global climate models will likely underestimate the true risk of these events.

Sign in / Sign up

Export Citation Format

Share Document