The computational future for climate and Earth system models: on the path to petaflop and beyond

The development of the climate and Earth system models has had a long history, starting with the building of individual atmospheric, ocean, sea ice, land vegetation, biogeochemical, glacial and ecological model components. The early researchers were much aware of the long-term goal of building the Earth system models that would go beyond what is usually included in the climate models by adding interactive biogeochemical interactions. In the early days, the progress was limited by computer capability, as well as by our knowledge of the physical and chemical processes. Over the last few decades, there has been much improved knowledge, better observations for validation and more powerful supercomputer systems that are increasingly meeting the new challenges of comprehensive models. Some of the climate model history will be presented, along with some of the successes and difficulties encountered with present-day supercomputer systems.

Download Full-text

CPMIP: measurements of real computational performance of Earth system models in CMIP6

Geoscientific Model Development ◽

10.5194/gmd-10-19-2017 ◽

2017 ◽

Vol 10 (1) ◽

pp. 19-34 ◽

Cited By ~ 14

Author(s):

Venkatramani Balaji ◽

Eric Maisonnave ◽

Niki Zadeh ◽

Bryan N. Lawrence ◽

Joachim Biercamp ◽

...

Keyword(s):

Climate Models ◽

Climate Model ◽

Initial Conditions ◽

Performance Model ◽

Earth System ◽

System Models ◽

Performance Engineering ◽

Computational Performance ◽

Sensitive Dependence ◽

Earth System Models

Abstract. A climate model represents a multitude of processes on a variety of timescales and space scales: a canonical example of multi-physics multi-scale modeling. The underlying climate system is physically characterized by sensitive dependence on initial conditions, and natural stochastic variability, so very long integrations are needed to extract signals of climate change. Algorithms generally possess weak scaling and can be I/O and/or memory-bound. Such weak-scaling, I/O, and memory-bound multi-physics codes present particular challenges to computational performance. Traditional metrics of computational efficiency such as performance counters and scaling curves do not tell us enough about real sustained performance from climate models on different machines. They also do not provide a satisfactory basis for comparative information across models. codes present particular challenges to computational performance. We introduce a set of metrics that can be used for the study of computational performance of climate (and Earth system) models. These measures do not require specialized software or specific hardware counters, and should be accessible to anyone. They are independent of platform and underlying parallel programming models. We show how these metrics can be used to measure actually attained performance of Earth system models on different machines, and identify the most fruitful areas of research and development for performance engineering. codes present particular challenges to computational performance. We present results for these measures for a diverse suite of models from several modeling centers, and propose to use these measures as a basis for a CPMIP, a computational performance model intercomparison project (MIP).

Download Full-text

CPMIP: Measurements of Real Computational Performance of Earth System Models

10.5194/gmd-2016-197 ◽

2016 ◽

Cited By ~ 1

Author(s):

V. Balaji ◽

E. Maisonnave ◽

N. Zadeh ◽

B. N. Lawrence ◽

J. Biercamp ◽

...

Keyword(s):

Climate Models ◽

Climate Model ◽

Initial Conditions ◽

Earth System ◽

System Models ◽

Performance Engineering ◽

Multi Scale ◽

Computational Performance ◽

Sensitive Dependence ◽

Earth System Models

Abstract. A climate model represents a multitude of processes on a variety of time and space scales; a canonical example of multi-physics multi-scale modeling. The underlying climate system is physically characterized by sensitive dependence on initial conditions, and natural stochastic variability, so very long integrations are needed to extract signals of climate change. Algorithms generally possess weak scaling and can be I/O and/or memory bound. Such weak-scaling, I/O and memory-bound multi-physics codes present particular challenges to computational performance. Traditional metrics of computational efficiency such as performance counters and scaling curves do not tell us enough about real sustained performance from climate models on different machines. They also do not provide a satisfactory basis for comparative information across models. We introduce a set of metrics that can be used for the study of computational performance of climate (and Earth System) models. These measures do not require specialized software or specific hardware counters, and should be accessible to anyone. They are independent of platform, and underlying parallel programming models. We show how these metrics can be used to measure actually attained performance of Earth system models on different machines, and identify the most fruitful areas of research and development for performance engineering. We present results for these measures for a diverse suite of models from several modeling centres, and propose to use these measures as a basis for a CPMIP, a computational performance MIP.

Download Full-text

Anticipating the computational performance of Earth System Models for pre-exascale systems

10.5194/egusphere-egu21-7888 ◽

2021 ◽

Author(s):

Xavier Yepes-Arbós ◽

Miguel Castrillo ◽

Mario C. Acosta ◽

Kim Serradell

Keyword(s):

Computational Efficiency ◽

Climate Model ◽

Global Climate ◽

Horizontal Resolution ◽

Limiting Factor ◽

Earth System ◽

Computational Point ◽

System Models ◽

Computational Performance ◽

Earth System Models

The increase in the capability of Earth System Models (ESMs) is strongly linked to the amount of computing power, given that the spatial resolution used for global climate experimentation is a limiting factor to correctly reproduce climate mean state and variability. However, higher spatial resolutions require new High Performance Computing (HPC) platforms, where the improvement of the computational efficiency of ESMs will be mandatory. In this context, porting a new ultra-high resolution configuration into a new and more powerful HPC cluster is a challenging task, involving technical expertise to deploy and improve the computational performance of such a novel configuration.To take advantage of this foreseeable landscape, the new EC-Earth 4 climate model is being developed by coupling OpenIFS 43R3 and NEMO 4 as atmosphere and ocean components respectively. An important effort has been made to improve the computational efficiency of this new EC-Earth version, such as extending the asynchronous I/O capabilities of the XIOS server to OpenIFS.&#160;In order to anticipate the computational behaviour of EC-Earth 4 for new pre-exascale machines such as the upcoming MareNostrum 5 of the Barcelona Supercomputing Center (BSC), OpenIFS and NEMO models are therefore benchmarked on a petascale machine (MareNostrum 4) to find potential computational bottlenecks introduced by new developments or to investigate if previous known performance limitations are solved. The outcome of this work can also be used to efficiently set up new ultra-high resolutions from a computational point of view, not only for EC-Earth, but also for other ESMs.Our benchmarking consists of large strong scaling tests (tens of thousands of cores) by running different output configurations, such as changing multiple XIOS parameters and number of 2D and 3D fields. These very large tests need a huge amount of computational resources (up to 2,595 nodes, 75 % of the supercomputer), so they require a special allocation that can be applied once a year.OpenIFS is evaluated with a 9 km global horizontal resolution (Tco1279) and using three different output data sets: no output, CMIP6-based fields and huge output volume (8.8 TB) to stress the I/O part. In addition, different XIOS parameters, XIOS resources, affinity, MPI-OpenMP hybridisation and MPI library are tested. Results suggest new features introduced in 43R3 do not represent a bottleneck in terms of performance as the model scales. The I/O scheme is also improved when outputting data through XIOS according to the scalability curve.NEMO is scaled using a 3 km global horizontal resolution (ORCA36) with and without the sea-ice module. As in OpenIFS, different I/O configurations are benchmarked, such as disabling model output, only enabling 2D fields, or either producing 3D variables on an hourly basis. XIOS is also scaled and tested with different parameters. While NEMO has good scalability during the most part of the exercise, a severe degradation is observed before the model uses 70% of the machine resources (2,546 nodes). The I/O overhead is moderate for the best XIOS configuration, but it demands many resources.

Download Full-text

Towards in-situ visualization integrated earth system models: RegESM 1.1 regional modelling system

10.5194/gmd-2018-179 ◽

2018 ◽

Cited By ~ 1

Author(s):

Ufuk Utku Turuncoglu

Keyword(s):

Data Analysis ◽

System Model ◽

Earth System ◽

System Models ◽

Modeling Systems ◽

Temporal And Spatial ◽

Model Components ◽

Earth System Models ◽

Modelling System

Abstract. The data volume being produced by regional and global multi-component earth system models are rapidly increasing due to the improved spatial and temporal resolution of the model components, sophistication of the used numerical models in terms of represented physical processes and their non-linear complex interactions. In particular, very short time steps have to be defined in multi-component and multi-scale non-hydrostatic modelling systems to represent the evolution of the fast-moving processes such as turbulence, extra-tropical cyclones, convective lines, jet streams, internal waves, vertical turbulent mixing and surface gravity waves. Consequently, the used small time steps cause extra computation and disk I/O overhead in the used modelling system even if today's most powerful high-performance computing and data storage systems are being considered. Analysis of the high volume of data from multiple earth system model components at different temporal and spatial resolution also poses a challenging problem to efficiently perform integrated data analysis of the massive amounts of data by relying on the conventional post-processing methods available today. This study basically aims to explore the feasibility and added value of integrating existing in-situ visualization and data analysis methods with the model coupling framework (ESMF) to increase interoperability between multi-component simulation code and data processing pipelines by providing easy to use, efficient, generic and standardized modeling environment for earth system science applications. The new data analysis approach enables simultaneous analysis of the vast amount of data produced by multi-component regional earth system models (atmosphere, ocean etc.) during the run process. The methodology aims to create an integrated modeling environment for analyzing fast-moving processes and their evolution in both time and space to support better understanding of the underplaying physical mechanisms. The state-of-art approach can also be used to solve common problems in earth system model development workflow such as designing new sub-grid scale parametrizations (convection, air–sea interaction etc.) that requires inspecting the integrated model behavior in a higher temporal and spatial scale during the run or supporting visual debugging of the multi-component modeling systems, which usually are not facilitated by existing model coupling libraries and modeling systems.

Download Full-text

An Overview of CMIP5 and the Experiment Design

Bulletin of the American Meteorological Society ◽

10.1175/bams-d-11-00094.1 ◽

2012 ◽

Vol 93 (4) ◽

pp. 485-498 ◽

Cited By ~ 8089

Author(s):

Karl E. Taylor ◽

Ronald J. Stouffer ◽

Gerald A. Meehl

Keyword(s):

Climate Change ◽

Climate Models ◽

Global Climate ◽

Experiment Design ◽

Global Climate Models ◽

Model Output ◽

Earth System ◽

Intergovernmental Panel ◽

System Models ◽

Earth System Models

The fifth phase of the Coupled Model Intercomparison Project (CMIP5) will produce a state-of-the- art multimodel dataset designed to advance our knowledge of climate variability and climate change. Researchers worldwide are analyzing the model output and will produce results likely to underlie the forthcoming Fifth Assessment Report by the Intergovernmental Panel on Climate Change. Unprecedented in scale and attracting interest from all major climate modeling groups, CMIP5 includes “long term” simulations of twentieth-century climate and projections for the twenty-first century and beyond. Conventional atmosphere–ocean global climate models and Earth system models of intermediate complexity are for the first time being joined by more recently developed Earth system models under an experiment design that allows both types of models to be compared to observations on an equal footing. Besides the longterm experiments, CMIP5 calls for an entirely new suite of “near term” simulations focusing on recent decades and the future to year 2035. These “decadal predictions” are initialized based on observations and will be used to explore the predictability of climate and to assess the forecast system's predictive skill. The CMIP5 experiment design also allows for participation of stand-alone atmospheric models and includes a variety of idealized experiments that will improve understanding of the range of model responses found in the more complex and realistic simulations. An exceptionally comprehensive set of model output is being collected and made freely available to researchers through an integrated but distributed data archive. For researchers unfamiliar with climate models, the limitations of the models and experiment design are described.

Download Full-text

Modular AWI-CM: An Earth System Model (ESM) prototype using the esm-interface library for a modular ESM coupling approach

10.5194/egusphere-egu2020-6990 ◽

2020 ◽

Author(s):

Nadine Wieters ◽

Dirk Barbi ◽

Luisa Cristini

Keyword(s):

Climate Model ◽

General Idea ◽

Second Step ◽

System Modelling ◽

Earth System ◽

First Results ◽

Coupling Approach ◽

Domain Models ◽

Model Components ◽

Earth System Models

Earth System Models (ESMs) are composed of different components, including submodels as well as whole domain models. Within such an ESM, these model components need to exchange information to account for the interactions between the different compartments. This exchange of data is the purpose of a &#8220;model coupler&#8221;.Within the Advanced Earth System Modelling Capacity (ESM) project, a goal is to develop a modular framework that allows for a flexible ESM configuration. One approach is to implement purpose build model couplers in a more modular way.For this purpose, we developed the esm-interface library, in consideration of the following objectives: (i) To obtain a more modular ESM, that allows model components and model couplers to be exchangeable; and (ii) to account for a more flexible coupling configuration of an ESM setup.As a first application of the esm-interface library, we implemented it into the AWI Climate Model (AWI-CM) [Sidorenko et al., 2015] as an interface between the model components and the model coupler (OASIS3-MCT; Valcke [2013]). In a second step, we extended the esm-interface library for a second coupler (YAC; Hanke et al. [2016]).In this presentation, we will discuss the general idea of the esm-interface library, it&#8217;s implementation in an ESM setup and show first results from the first modular prototype of AWI-CM.

Download Full-text

Coupling of a sediment diagenesis model (MEDUSA) and an Earth system model (CESM1.2): a contribution toward enhanced marine biogeochemical modelling and long-term climate simulations

10.5194/gmd-2019-231 ◽

2019 ◽

Author(s):

Takasumi Kurahashi-Nakamura ◽

André Paul ◽

Guy Munhoven ◽

Ute Merkel ◽

Michael Schulz

Keyword(s):

Climate Models ◽

Climate Model ◽

Coupled Model ◽

Earth System Model ◽

System Model ◽

Future Application ◽

Coupling Scheme ◽

Earth System ◽

Climate Simulations

Abstract. We developed a coupling scheme for the Community Earth System Model version 1.2 (CESM1.2) and the Model of Early Diagenesis in the Upper Sediment of Adjustable complexity (MEDUSA), and explored the effects of the coupling on solid components in the upper sediment and on bottom seawater chemistry by comparing the coupled model's behaviour with that of the uncoupled CESM having a simplified treatment of sediment processes. CESM is a fully-coupled atmosphere-ocean-sea ice-land model and its ocean component (the Parallel Ocean Program version 2, POP2) includes a biogeochemical component (BEC). MEDUSA was coupled to POP2 in an off-line manner so that each of the models ran separately and sequentially with regular exchanges of necessary boundary condition fields. This development was done with the ambitious aim of a future application for long-term (spanning a full glacial cycle; i.e., ~ 105 years) climate simulations with a state-of-the-art comprehensive climate model including the carbon cycle, and was motivated by the fact that until now such simulations have been done only with less-complex climate models. We found that the sediment-model coupling already had non-negligible immediate advantages for ocean biogeochemistry in millennial-time-scale simulations. First, the MEDUSA-coupled CESM outperformed the uncoupled CESM in reproducing an observation-based global distribution of sediment properties, especially for organic carbon and opal. Thus, the coupled model is expected to act as a better bridge between climate dynamics and sedimentary data, which will provide another measure of model performance. Second, in our experiments, the MEDUSA-coupled model and the uncoupled model had a difference of 0.2‰ or larger in terms of δ13C of bottom water over large areas, which implied potential significant model biases for bottom seawater chemical composition due to a different way of sediment treatment. Such a model bias would be a fundamental issue for paleo model–data comparison often relying on data derived from benthic foraminifera.

Download Full-text

Calibrating Simple Climate Models to Individual Earth System Models: Lessons Learned From Calibrating Hector

Earth and Space Science ◽

10.1029/2019ea000980 ◽

2020 ◽

Vol 7 (11) ◽

Cited By ~ 1

Author(s):

Kalyn Dorheim ◽

Robert Link ◽

Corinne Hartin ◽

Ben Kravitz ◽

Abigail Snyder

Keyword(s):

Climate Models ◽

Lessons Learned ◽

Earth System ◽

System Models ◽

Earth System Models

Download Full-text

The role of efficient climate models in the projection of future climate feedback and surface warming

10.5194/egusphere-egu21-852 ◽

2021 ◽

Author(s):

Philip Goodwin ◽

B.B. Cael

Keyword(s):

Climate Model ◽

Climate Feedback ◽

Model Output ◽

Earth System ◽

Surface Warming ◽

System Models ◽

Model Ensembles ◽

Computational Resources ◽

Earth System Models ◽

Global Mean

Projecting the global climate feedback and surface warming responses to anthropogenic forcing scenarios remains a key priority for climate science. Here, we explore possible roles for efficient climate model ensembles in contributing to quantitative projections of future global mean surface warming and climate feedback within model hierarchies. By comparing complex and efficient (sometimes termed &#8216;simple&#8217;) model output to data we: (1) explore potential Bayesian approaches to model ensemble generation; (2) ask what properties an efficient climate model should have to contribute to the generation of future warming and climate feedback projections; (3) present new projections from efficient model ensembles.&#160;Climate processes relevant to global surface warming and climate feedback act over at least 14 orders of magnitude in space and time; from cloud droplet collisions and photosynthesis up to the global mean temperature and carbon storage over the 21st century. Due to computational resources, even the most complex Earth system models only resolve around 3 orders of magnitude in horizontal space (from grid scale up to global scale) and 6 orders of magnitude in time (from a single timestep up to a century).&#160;Complex Earth system models must therefore contain a great many parameterisations (including specified functional forms of equations and their coefficient values) representing sub grid-scale and sub time-scale processes. We know that these parameterisations affect the quantitative model projections, because different complex models produce a range of historic and future projections. However, complex Earth system models are too computationally expensive to fully sample the plausible combinations of their own parameterisations, typically being able to realise only several tens of simulations.&#160;In contrast, efficient climate models are able to utilise computational resources to resolve their own plausible combinations of parameterisations, through the construction of very large model ensembles. However, this parameterisation resolution occurs at the expense of a much-reduced resolution of relevant climate processes. Since the relative simplicity of efficient model representations may not capture the required complexity of the climate system, the qualitative nature of their simulated projections may be too simplistic. For example, an efficient climate model may use a single climate feedback value for all time and for all sources of radiative forcing, when in complex models (and the real climate system) climate feedbacks may vary over time and may respond differently to, say, localised aerosol forcing than to well mixed greenhouse gases.&#160;By far the dominant quantitative projections of global mean surface warming in the scientific literature, as used in the Intergovernmental Panel on Climate Change Assessment Reports, derive from relatively small ensembles of complex climate model output. However, computational resources impose an inherent trade-off between model resolution of relevant climate processes (affecting the qualitative nature of the model framework) and model ensemble resolution of plausible parameterisations (affecting the quantitative exploration of projections within that model framework). This computationally imposed trade-off suggests there may be a significant role for efficient model output, within a hierarchy of model complexities, when generating future warming projections.

Download Full-text

Climate feedbacks in the Earth system and prospects for their evaluation

10.5194/esd-2018-84 ◽

2018 ◽

Author(s):

Christoph Heinze ◽

Veronika Eyring ◽

Pierre Friedlingstein ◽

Colin Jones ◽

Yves Balkanski ◽

...

Keyword(s):

Climate Models ◽

Climate Projections ◽

Earth System ◽

Climate Feedbacks ◽

Comprehensive Overview ◽

The Earth ◽

Cloud Dynamics ◽

Spatial And Temporal Heterogeneity ◽

Earth System Models

Abstract. Earth system models (ESMs) are key tools for providing climate projections under different scenarios of human-induced forcing. ESMs include a large number of additional processes and feedbacks such as biogeochemical cycles that traditional physical climate models do not consider. Yet, some processes such as cloud dynamics and ecosystem functional response still have fairly high uncertainties. In this article, we present an overview of climate feedbacks for Earth system components currently included in state-of-the-art ESMs and discuss the challenges to evaluate and quantify them. Uncertainties in feedback quantification arise from the interdependencies of biogeochemical matter fluxes and physical properties, the spatial and temporal heterogeneity of processes, and the lack of long-term continuous observational data to constrain them. We present an outlook for promising approaches that can help quantifying and constraining the large number of feedbacks in ESMs in the future. The target group for this article includes generalists with a background in natural sciences and an interest in climate change as well as experts working in interdisciplinary climate research (researchers, lecturers, and students). This study updates and significantly expands upon the last comprehensive overview of climate feedbacks in ESMs, which was produced 15 years ago (NRC, 2003).

Download Full-text