An Evaluation of the Software System Dependency of a Global Atmospheric Model

Hong, Song‐You; Koo, Myung-Seo; Jang, Jihyeon; Kim, Jung-Eun Esther; Park, Hoon; Joh, Minsu; Kang, Ju-Hyung; Oh, Tae-Jin

doi:10.1175/mwr-d-12-00352.1

Cited by 32 publications

(15 citation statements)

References 14 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…In summer (Fig. 2), there is a very large spread of model responses with some RCMs predicting a widespread cooling from forestation (CCLM-TERRA and RCA), a widespread warming (RegCM-CLM4.5, REMO-iMOVE and the WRF Davies (1976) Davies (1976) Davies (1976) Davies (1976) Turbulence and planetary boundary layer scheme Level 2.5 closure for turbulent kinetic energy as prognostic variable (Mellor and Yamada, 1982) Level 2.5 closure for turbulent kinetic energy as prognostic variable (Mellor and Yamada, 1982) Level 2.5 closure for turbulent kinetic energy as prognostic variable (Mellor and Yamada, 1982) (Vogelezang and Holtslag, 1996) (Iacono et al, 2008) RRTMG scheme (Iacono et al, 2008) RRTMG scheme (Iacono et al, 2008) Tiedtke (1996) for cumulus convec-tion (Tiedtke, 1989) with modifications after Nordeng (1994) Grell and Freitas (2014) for cumulus convection and Global/Regional Integrated Model system (GRIMs) Scheme (Hong et al, 2013) for shallow convection (Kain, 2004); no shallow convection (Kain, 2004); no shallow convection Microphysics scheme one-moment cloud micro-physics scheme (Seifert and Beheng, 2001) one-moment cloud microphysics scheme (Seifert and Beheng, 2001) one-moment cloud microphysics scheme (Seifert and Beheng, 2001) values from tables models) or a mixed response (CCLM-VEG3D and CCLM-CLM4.5). Overall this highlights the strong seasonal contrasts in the temperature effect of forestation and the larger uncertainties associated with the summer response.…”

Section: Temperature Responsementioning

confidence: 99%

Biogeophysical impacts of forestation in Europe: first results from the LUCAS (Land Use and Climate Across Scales) regional climate model intercomparison

et al. 2020

View full text Add to dashboard Cite

Abstract. The Land Use and Climate Across Scales Flagship Pilot Study (LUCAS FPS) is a coordinated community effort to improve the integration of land use change (LUC) in regional climate models (RCMs) and to quantify the biogeophysical effects of LUC on local to regional climate in Europe. In the first phase of LUCAS, nine RCMs are used to explore the biogeophysical impacts of re-/afforestation over Europe: two idealized experiments representing respectively a non-forested and a maximally forested Europe are compared in order to quantify spatial and temporal variations in the regional climate sensitivity to forestation. We find some robust features in the simulated response to forestation. In particular, all models indicate a year-round decrease in surface albedo, which is most pronounced in winter and spring at high latitudes. This results in a winter warming effect, with values ranging from +0.2 to +1 K on average over Scandinavia depending on models. However, there are also a number of strongly diverging responses. For instance, there is no agreement on the sign of temperature changes in summer with some RCMs predicting a widespread cooling from forestation (well below −2 K in most regions), a widespread warming (around +2 K or above in most regions) or a mixed response. A large part of the inter-model spread is attributed to the representation of land processes. In particular, differences in the partitioning of sensible and latent heat are identified as a key source of uncertainty in summer. Atmospheric processes, such as changes in incoming radiation due to cloud cover feedbacks, also influence the simulated response in most seasons. In conclusion, the multi-model approach we use here has the potential to deliver more robust and reliable information to stakeholders involved in land use planning, as compared to results based on single models. However, given the contradictory responses identified, our results also show that there are still fundamental uncertainties that need to be tackled to better anticipate the possible intended or unintended consequences of LUC on regional climates.

show abstract

Section: Temperature Responsementioning

confidence: 99%

Biogeophysical impacts of forestation in Europe: first results from the LUCAS (Land Use and Climate Across Scales) regional climate model intercomparison

et al. 2020

View full text Add to dashboard Cite

show abstract

“…By chance, we found that even tiny numerical precision errors due to the use of different computational architecture can lead to large differences in TC development, which is also found in Hong et al . []. Using exactly the same initial conditions and model setups, the SH5 ensemble set was run on two supercomputers at the Texas Advanced Computing Center (TACC): Ranger (Figure a), which has since been retired and replaced by Stampede (Figure b).…”

Section: Intrinsic Versus Practical Limits Of Predictabilitymentioning

confidence: 99%

Effects of vertical wind shear on the predictability of tropical cyclones: Practical versus intrinsic limit

Tao

Zhang

2015

J Adv Model Earth Syst

View full text Add to dashboard Cite

The effects of environmental shear on the dynamics and predictability of tropical cyclones (TCs) are further explored through a series of cloud-permitting ensemble sensitivity experiments with small, random initial condition perturbations on the low-level moisture fields. As an expansion of earlier studies, it is found that larger the shear magnitude, less predictable the TCs, especially the onset time of the rapid intensification (RI), until the shear is too large for the TC formation. Systematic differences amongst the ensemble members begin to arise right after the initial burst of moist convection associated with the incipient vortex. This randomness inherent in moist convection first changes the TC vortex structure subtly, but the location and strength of subsequent moist convection are greatly influential to the precession and alignment of the TC vortex as well as the RI onset time. Additional ensemble sensitivity experiments with different magnitude random perturbations to the mean environmental shear (6 m s 21 ) show that when the standard deviation of the random shear perturbations among different ensemble members is as small as 0.5 m s 21 , the difference in shear magnitude overwhelms the randomness of moist convection in influencing the TC development and rapid intensification (indicative of limited practical predictability). However, for the ensemble with standard deviation of 0.1 m s 21 in random shear perturbations, the uncertainty in TC onset timing is comparable to the ensemble that is perturbed only by small random moisture conditions in the initial moisture field (indicative of the limit in intrinsic predictability).

show abstract

“…They all came to the same conclusion that changes in behavior induced by hardware or software differences were not negligible compared to other sources of error such as uncertainty in model parameters or initial conditions. These conclusions were found to hold from weather (Thomas et al, 2002) to seasonal (Hong et al, 2013) and even climate change (Knight et al, 2007) time scales.…”

Section: Origins Of Non-replicabilitymentioning

confidence: 76%

Replicability of the EC-Earth3 Earth System Model under a change in computing environment

Massonnet¹,

Ménégoz²,

Acosta³

et al. 2019

Preprint

View full text Add to dashboard Cite

Abstract. Most Earth System Models (ESMs) are running under different high-performance computing (HPC) environments. This has several advantages, from allowing different groups to work with the same tool in parallel to leveraging the burden of ensemble climate simulations but also offering alternative solutions in case of shutdown (expected or not) of any of the environments. However, for obvious scientific reasons, it is critical to ensure that ESMs provide identical results under changes in computing environment. While strict bit-for-bit reproducibility is not always guaranteed with ESMs, it is desirable that results obtained under one computing environment are at least statistically indistinguishable from those obtained under another environment, which we term a replicability condition following the metrology nomenclature. Here, we develop a protocol to assess the replicability of the EC-Earth ESM. Using two versions of EC-Earth, we present one case of non-replicability and one case of replicability. The non-replicable case occurs with the older version of the model and likely finds its origin in the treatment of river runoffs along Antarctic coasts. By contrast, the more recent version of the model provides replicable results. The methodology presented here has been adopted as a standard test by the EC-Earth consortium (27 institutions in Europe) to evaluate the replicability of any new model version across platforms, including for CMIP6 experiments. To a larger extent, it can be used to assess whether other ESMs can safely be ported from one HPC environment to another for studying climate-related questions. Our results and experience with this work suggest that the default assumption should be that ESMs are not replicable under changes in the HPC environment, until proven otherwise.

show abstract

An Evaluation of the Software System Dependency of a Global Atmospheric Model

Cited by 32 publications

References 14 publications

Biogeophysical impacts of forestation in Europe: first results from the LUCAS (Land Use and Climate Across Scales) regional climate model intercomparison

Biogeophysical impacts of forestation in Europe: first results from the LUCAS (Land Use and Climate Across Scales) regional climate model intercomparison

Effects of vertical wind shear on the predictability of tropical cyclones: Practical versus intrinsic limit

Replicability of the EC-Earth3 Earth System Model under a change in computing environment

Contact Info

Product

Resources

About