CL5.1.2

CL5.1 EDI
Emulators and simple climate models: development and applications 

Modelling the Earth system with state-of-the-art Earth System models is computationally expensive. Therefore, emulators, also known as simple or reduced complexity climate models, are useful as they are able to produce individual climate projections with reduced computational resources. They play a key role in the climate assessments of the IPCC’s Working Group 1 and Working Group 3 reports. Emulators also play a role in coupling climate models to Integrated Assessment models and exploring potential effects of missing feedbacks in the development of socioeconomic and emissions scenarios.

The lower computational burden of emulators allows for a multitude of possibilities. Large probabilistic model ensembles can represent uncertainty in climate projections that is difficult to quantify in a full-complexity ESM, usually through the variation in a number of parameters that can be tuned to alter the emulator behaviour. Emulators may have physical, statistical, or machine learning elements, and can range in complexity from spreadsheet-based projections of global-mean temperature from prescribed radiative forcing through to simplified variants of operational ESMs.

We invite presentations on all aspects of the development and application of simple climate and geophysical models, including but not limited to:
1. the development of emulators, results, and case studies
2. the role of emulators in integrated assessment and scenario generation
3. best practices in tuning and calibration of emulators to observations and full-complexity Earth System models
4. Use of emulators to better understand complex model differences and behaviours
5. Strategies for emulating internal variability, teleconnections, regional climate, climate extremes, tipping points and missing processes in Earth System models (e.g. permafrost)
6. models that focus on one particular complex aspect of the Earth system (for example land and ocean carbon cycles, aerosol-climate interactions or the cryosphere)
7. uses of emulators in outreach, education and policymaking

Public information:

This session explores the utility of simple climate and geophysical models for process-based and global-level understanding. Simplified models use physical or statistical methods to emulate processes in the Earth system at higher computational efficiency, allowing for uncertainty analysis with large ensembles. In many cases, emulators have lower code overheads and a tractable number of equations, lowering the barrier to entry for Earth System modelling.

Convener: Christopher SmithECSECS | Co-conveners: Zebedee R. NichollsECSECS, Kalyn DorheimECSECS, Ben Sanderson, Bjorn H. Samset
Presentations
| Fri, 27 May, 10:20–11:50 (CEST), 13:20–14:05 (CEST)
 
Room 0.49/50
Public information:

This session explores the utility of simple climate and geophysical models for process-based and global-level understanding. Simplified models use physical or statistical methods to emulate processes in the Earth system at higher computational efficiency, allowing for uncertainty analysis with large ensembles. In many cases, emulators have lower code overheads and a tractable number of equations, lowering the barrier to entry for Earth System modelling.

Presentations: Fri, 27 May | Room 0.49/50

Chairpersons: Christopher Smith, Zebedee R. Nicholls, Bjorn H. Samset
10:20–10:26
|
EGU22-3961
|
ECS
|
Highlight
|
Virtual presentation
Duncan Watson-Parris, Yuhan Rao, Dirk Olivié, Øyvind Seland, Peer Nowack, Gustau Camps-Valls, Philip Stier, Shahine Bouabid, Maura Dewey, Emilie Fons, Jessenia Gonzalez, Paula Harder, Kai Jeggle, Julien Lenhardt, Peter Manshausen, Maria Novitasari, Lucile Ricard, and Carla Roesch

Exploration of future emissions scenarios mostly relies on one-dimensional impulse response models, or simple pattern scaling approaches to approximate the physical climate response to a given scenario. Such approaches are unable to reliably predict climate variables which respond non-linearly to emissions or forcing (such as precipitation) and must rely on heavily simplified representations of e.g., aerosol, neglecting important spatial dependencies.

Here we present ClimateBench - a benchmark dataset based on a suite of CMIP, AerChemMIP and DAMIP simulations performed by NorESM2, and a set of baseline machine learning models that emulate its response to a variety of forcers. These surrogate models can skilfully predict annual mean global distributions of temperature, diurnal temperature range and precipitation (including extreme precipitation) given a wide range of emissions and concentrations of carbon dioxide, methane and spatially resolved aerosol. We discuss the accuracy and interpretability of these emulators and consider their robustness to physical constraints such as total energy conservation. Future opportunities incorporating such physical constraints directly in the machine learning models and using the emulators for detection and attribution studies are also discussed. This opens a wide range of opportunities to improve prediction, consistency and mathematical tractability.

We hope that by defining a clear baseline with appropriate metrics and providing a variety of baseline models we can bring the power of modern machine learning techniques to bear on the important problem of efficiently and robustly sampling future climates.

How to cite: Watson-Parris, D., Rao, Y., Olivié, D., Seland, Ø., Nowack, P., Camps-Valls, G., Stier, P., Bouabid, S., Dewey, M., Fons, E., Gonzalez, J., Harder, P., Jeggle, K., Lenhardt, J., Manshausen, P., Novitasari, M., Ricard, L., and Roesch, C.: ClimateBench: A benchmark for data-driven climate projections, EGU General Assembly 2022, Vienna, Austria, 23–27 May 2022, EGU22-3961, https://doi.org/10.5194/egusphere-egu22-3961, 2022.

10:26–10:32
|
EGU22-7553
|
ECS
|
Highlight
|
On-site presentation
Fiona Turner and Tamsin Edwards

Changes in the cryosphere are the leading component of global sea level rise. There is great uncertainty in what these changes will look like in the coming centuries, partly due to the unknown effects of the climate and ice sheet models used to model these changes. Modelling these contributions are necessary to understand how coastal communities and low-lying states will be affected by climate change; in order to do this, and to quantify the inherent uncertainties to make more informed estimates, statistical methods are required.

 

Here we describe our work building on Edwards et al. (2021) in the use of Gaussian process emulators to predict the land ice contribution to future sea level rise. Rather than building an emulator of an ensemble of ice sheet models, we emulate each model individually, allowing us to better understand the inherent biases and internal variability within each model. We then compare our combined estimates with our previous results to test how treating each model individually affects our predictions. 

 

We predict changes for different Shared Socioeconomic Pathways (SSPs), to investigate how different future levels of greenhouse emissions will affect sea level rise this century. We also explore differences in sensitivity of the models to different inputs, building a range of sea level predictions. In particular, sensitivity to the basal melt parameter in Antarctica has a significant effect on the upper tail of our distributions; further analysis of other inputs will also be explored. 

How to cite: Turner, F. and Edwards, T.: Predicting the land ice contribution to sea level rise with Gaussian process emulation, EGU General Assembly 2022, Vienna, Austria, 23–27 May 2022, EGU22-7553, https://doi.org/10.5194/egusphere-egu22-7553, 2022.

10:32–10:38
|
EGU22-6304
|
ECS
|
On-site presentation
Sandy Avrutin, Philip Goodwin, Ivan Haigh, and Robert Nicholls

Sea level rise is a major result of climate change that threatens coastal communities and has the potential to incur annual costs by 2100 of $11-95 billion in flood damages alone, assuming a global mean sea level rise of 25-123 cm (Hinkel et al. 2014). Projecting sea level rise as temperatures rise is therefore crucial for policy and decision-making.

The two methods currently used to project future sea level change are process-based modelling and semi-empirical modelling. Process-based models rely on combining outputs from coupled atmosphere/ocean models for each component of sea level rise. Semi-empirical models calculate sea level as an integrated response to either warming or radiative forcing, using parameters constrained from past observations.

Historically, there is little agreement in sea-level projections between these two methods (Orlić and Pasarić, 2013). One potential source of the discrepancies is uncertainty in land ice response to warming; although nonlinearities exist within processes affecting this response, most existing semi-empirical models treat the relationship between warming and ice-melt as linear.

Non-linear processes in sea level rise may have not yet affected the observational record (such as tipping points as future warming crosses some threshold) or may have already occurred (such as non-linear effects that apply across all levels of warming, or for which the threshold has been passed). Here, we examine the effect on semi-empirical projections of sea level rise of nonlinearities that have already affected the observed sea level record, by adding a nonlinear term to the relationship between warming and the rate of sea level rise within a large ensemble of historically constrained efficient earth systems model simulations.

Projections reach a median sea level rise of 0.47m by 2100 following SSP245, and 0.77m by 2100 following SSP585. Preliminary results suggest that nonlinear interactions in each ensemble member can be sublinear, superlinear or 0, with a mainly symmetrical distribution – although there are high-end, low-probability superlinear interactions up to 3x greater than low-end sublinear. Thus, we find that observation-consistent nonlinear interactions in the model configuration lead to insignificant differences in sea level rise by 2300 over the entire ensemble. However, it is key to note that nonlinear interactions that have not yet occurred but that may occur in the future, are not considered – these will lead to an increased projection of sea level rise by 2300 if not earlier (e.g. DeConto and Pollard, 2016).

References

  • Hinkel, J. et al. Coastal flood damage and adaptation costs under 21st century sea-level rise. Proc. Natl. Acad. Sci. U. S. A. 111, 3292–3297 (2014).
  • Kopp, R. E. et al. Probabilistic 21st and 22nd century sea‐level projections at a global network of tide‐gauge sites. Earth’s Future. 2, 383–406 (2014).
  • Jevrejeva, S., Moore, J. C. & Grinsted, A. How will sea level respond to changes in natural and anthropogenic forcings by 2100? Geophys. Res. Lett. 37, 1–5 (2010).
  • Orlić, M. & Pasarić, Z. Semi-empirical versus process-based sea-level projections for the twenty-first century. Nat. Clim. Chang. 3, 735–738 (2013).

How to cite: Avrutin, S., Goodwin, P., Haigh, I., and Nicholls, R.: Observation-Consistent Nonlinear Ice Interactions in an Efficient Earth Systems Model and their Implications for Sea-Level Projections, EGU General Assembly 2022, Vienna, Austria, 23–27 May 2022, EGU22-6304, https://doi.org/10.5194/egusphere-egu22-6304, 2022.

10:38–10:44
|
EGU22-11514
|
On-site presentation
Emmanuele Russo, Jonathan Buzan, Guillaume Jouvet, Denis Cohen, and Christoph C. Raible

Glacier modelling of the Alpine region during past ice ages has received increasing attention in the recent years. Considering the complexity of the Alpine topography, high spatial resolution climate information is required for running glacier models over the area. However, continuous climatic signal at resolution higher than 10 km and covering several hundred thousand of years cannot be directly derived using dynamical climate models. Alternative strategies must be considered.

Here, a climate emulator providing monthly temperature and precipitation over the Alpine region, with a horizontal resolution of 2x2 km and covering the last 400’000 years at intervals of 100 years, is presented. The emulatorcombines a dynamical modelling chain of Earth  System Models (ESMs) and a Regional Climate Model (RCM) with different statistical modelling methods. The dynamical modelling chain delivers climate information at highest accuracy based upon physical prognostic equations for specific time slices. The subsequent statistical modelling uses these time slices as physically consistent boundaries and estimates the climate conditions in between them, thus generatinga long-term climate evolution.

A total of 19 climate model experiments are conducted for different time-slices of the considered study period, including also sensitivity tests with changes in the ice sheet height of the Northern Hemisphere and land cover type. Starting from a simple linear regression, a series of different statistical approaches is tested for building the emulator. An evaluation of the different versions is then conducted against one of the RCM time-slice experiments, left out in turns from the training set of the statistical model.

Results show robust skills of the emulator in the representation of temperature, whose changes are mainly driven by smooth variations in the seasonal pattern of insolation at different time-steps, already using a simple linear regression. For precipitation, non-linearities associated to changes in the large-scale atmospheric circulation seem to dominate, making the use of more complex statistical approaches more appropriate. Additional evaluation tests conducted using glacier modelling driven by the outputs of the developed emulator confirm its potential for reconstructing the ice extent over the Alpine region during the last ice ages.

How to cite: Russo, E., Buzan, J., Jouvet, G., Cohen, D., and Raible, C. C.: A regional climate emulator to estimate glacial-interglacial changes over the Alps, EGU General Assembly 2022, Vienna, Austria, 23–27 May 2022, EGU22-11514, https://doi.org/10.5194/egusphere-egu22-11514, 2022.

10:44–10:50
|
EGU22-11069
|
ECS
|
On-site presentation
Matthew Kasoar, Carlo Corsaro, and Apostolos Voulgarakis

The Absolute Global Temperature change Potential (AGTP) and Absolute Global Precipitation change Potential (AGPP) are widely used climate change indices.  They can be applied quickly and easily to estimate the global mean temperature and precipitation responses to a pulse emission of a long-lived climate pollutant at a given time horizon, making them invaluable policy-relevant metrics.  They can also be extended to short-lived climate pollutants - where a sustained emission is more useful to consider than a pulse emission - by using their time-integrated forms (iAGTP and iAGPP).

However, these metrics are only useful when taking a global-average perspective, and do not allow us to account for the regional nature of either emissions or their climate response.  Although long-lived greenhouse gases induce a relatively homogeneous radiative forcing (RF) which is not sensitive to emission location, nonetheless due to transport of heat there is not a one-to-one correspondence between the RF in a region and the local temperature response.  Moreover when considering short-lived pollutants such as aerosols, the region of emission is potentially critical because the short lifetime of such pollutants results in an inhomogeneous distribution of RF.  Therefore, for both long-lived and short-lived pollutants the AGTP/AGPP (or iAGTP/iAGPP) are not adequate when looking at climate responses on a regional scale, even though this would be the most relevant when evaluating different policy scenarios or climate change impacts.

Here, we combine the results of simulations from the Precipitation Driver Response Model Intercomparison Project (PDRMIP) where emissions (or concentrations) of multiple long- and short-lived climate pollutants were perturbed globally in nine different climate models, with the results of simulations using the HadGEM3 model where sulfate aerosol emissions are perturbed one at a time in several key geopolitical regions: the United States, Europe, India, East Asia, or the whole Northern Hemisphere Mid-Latitudes.  We use these results to adapt the (i)AGTP/(i)AGPP to the case where both the emission and the response are regional.  Data from the regional HadGEM3 simulations allow us to estimate normalised regional forcing-response relationships for aerosols, whilst the PDRMIP multi-model means and ensemble spread are used to derive estimates of radiative efficiency for both long- and short-lived pollutants and their corresponding uncertainties, as well as the regional climate sensitivities for long-lived pollutants.

Finally, using these regional temperature and precipitation change potentials, we produce a simple model in Python which allows the user to specify arbitrary combinations of different future emission scenarios for different pollutants from different regions, allowing rapid projections of the regional climate responses to diverse emissions policies.

How to cite: Kasoar, M., Corsaro, C., and Voulgarakis, A.: Metrics for Regional Climate Responses to Regional Pollutant Emissions, EGU General Assembly 2022, Vienna, Austria, 23–27 May 2022, EGU22-11069, https://doi.org/10.5194/egusphere-egu22-11069, 2022.

10:50–10:56
|
EGU22-12999
|
Presentation form not yet defined
Boris Faybishenko, Bhavna Arora, Dipankar Dwivedi, and Eoin Brodie

A statistical framework to assess the long-term climatic water balance changes includes the following phases of the data analysis and predictions: (1) Preparation of daily, monthly, and yearly averaged time series of meteorological parameters (temperature, relative humidity, precipitation, wind speed, etc.), and an evaluation of the temporal structural breakpoints (breakthroughs) of meteorological parameters trends, (2) calculations of potential evapotranspiration, aridity index, actual evapotranspiration (ET), Standard Precipitation Index (SPI), and Standard Precipitation-Evapotranspiration Index (SPEI), as well as an evaluation of breakthroughs of their trends, (3) climatic zonation based on the application of the hierarchical k-means and Principal Component Analysis (PCA) clustering of temporal trends of ET and SPEI for the periods before and after the breakthroughs, and (4) simple forecasting hierarchical time series for different forecasting situations.

The statistical framework was applied to 17 locations at the East River watershed for the period from 1966 to 2020. Structural changes of time trends of measured and calculated water balance parameters are used to determine the time of abrupt climatic changes and breakthroughs. Calculations of the evapotranspiration are conducted using the Budyko model, with the potential/reference evapotranspiration (ETo) calculated using the Penman-Monteith (PM) equation. The results of calculations of ETo based on the PM model were compared to the ETo calculated using the Thornthwaite and Hargreaves equations. The results of the hierarchical clustering using ET and SPEI are illustrated using the tree dendrograms and the PCA plots of clusters of the studied sites for the periods of before and after the breakthroughs. A significant shift in the cluster arrangements for the time periods before and after the temporal structural breakpoints indicate that zonation patterns are driven by dynamic climatic processes, which are variable through time, and the watershed zonation requires periodic re-evaluation. Examples of time series forecasting are also shown.

How to cite: Faybishenko, B., Arora, B., Dwivedi, D., and Brodie, E.: A statistical framework to assess time trend and predict near-future climatic conditions, EGU General Assembly 2022, Vienna, Austria, 23–27 May 2022, EGU22-12999, https://doi.org/10.5194/egusphere-egu22-12999, 2022.

10:56–11:05
11:05–11:11
|
EGU22-11049
|
ECS
|
Highlight
|
Presentation form not yet defined
Sarah Schöngart, Quentin Lejeune, Carl-Friedrich Schleußner, Sonia Seneviratne, Lukas Gudmundsson, and Lea Beusch

Earth System Models (ESMs) are essential for understanding the dynamics of our climate system, but their computational costs make nuanced investigations of future climatic conditions difficult. Using statistical techniques, computationally efficient tools known as emulators, which mimic ESM simulations, can be built. Emulators allow to (i) project the regional climate change for a broad variety of emission scenarios and to (ii) thoroughly sample the uncertainty space associated with natural variability as well as structural model uncertainties. Both tasks would be computationally infeasible with actual ESMs. In this contribution, we introduce a probabilistic, bivariate ESM emulation framework that produces joint monthly spatial fields of temperature and precipitation for a given global mean temperature trajectory. This contribution adds to the existing modular MESMER framework developed by Beusch et al. (2020). The building blocks of the new emulator are: (i) A module for approximating the annual global mean temperature trajectory from ESM output. This module is adapted from the existing MESMER framework. (ii) A module capturing the deterministic local response of monthly temperature and precipitation to global mean temperature. The response function is assumed to be linear with coefficients fitted independently for each month, grid-cell and variable. (iii) A module capturing the residual variability, that follows a probabilistic, non-parametric approach to reproduce spatial and temporal variance, covariance and cross-covariance structures of both variables. The emulator is trained and tested on ESM ensembles generated during CMIP6. Near-term development steps include the quantification of inter-ESM differences through the trained parameters and the coupling of the emulator to the simple climate model MAGICC (Meinshausen et al., 2020) to explore the emission scenario space.

 

Beusch, Lea, Lukas Gudmundsson, and Sonia I. Seneviratne. "Emulating Earth system model temperatures with MESMER: from global mean temperature trajectories to grid-point-level realizations on land." Earth System Dynamics 11.1 (2020): 139-159.

Meinshausen, Malte, et al. "The shared socio-economic pathway (SSP) greenhouse gas concentrations and their extensions to 2500." Geoscientific Model Development 13.8 (2020): 3571-3605.

 

How to cite: Schöngart, S., Lejeune, Q., Schleußner, C.-F., Seneviratne, S., Gudmundsson, L., and Beusch, L.: A spatially explicit approach for joint temperature-precipitation emulation of Earth System Model simulation, EGU General Assembly 2022, Vienna, Austria, 23–27 May 2022, EGU22-11049, https://doi.org/10.5194/egusphere-egu22-11049, 2022.

11:11–11:17
|
EGU22-5034
|
ECS
|
Presentation form not yet defined
Yann Quilcaille, Lukas Gudmundsson, Lea Beusch, Mathias Hauser, and Sonia Seneviratne

Emulators of Earth System Models (ESMs) are complementary to ESMs, in that they provide climate information with reduced computational costs. However, climate extremes, one of the most impactful consequences of climate change, remain challenging to emulate in all of their aspects. Here, we propose a method for the emulation of local annual maximum temperatures, with a focus on reproducing essential statistical properties and correlation in space and time. The Modular Earth System Model Emulator for Regional eXtremes (MESMER-X) is based on sampling from the generalized extreme value distribution and we show quantitatively that the resulting emulations of annual maximum temperature can reproduce the temporal evolution and spatial patterns of the underlying ESM simulations. Given the general design of the emulator and the good performances for annual maximum temperatures, the proposed methodology can be applied to other indicators of climate extremes, illustrated here with an indicator of fire weather.

How to cite: Quilcaille, Y., Gudmundsson, L., Beusch, L., Hauser, M., and Seneviratne, S.: Emulating spatially resolved annual maximum temperatures of Earth system models using MESMER-X , EGU General Assembly 2022, Vienna, Austria, 23–27 May 2022, EGU22-5034, https://doi.org/10.5194/egusphere-egu22-5034, 2022.

11:17–11:23
|
EGU22-12390
|
ECS
|
Virtual presentation
Marit Sandstad, Ragnhild Bieltvedt Skeie, and Bjørn Hallvard Samset

The Cicero Simple Climate Model (CICERO-SCM) is an energy balance model originally developed around 20 years ago, in Fortran, that has since been in continuous use and subject to minor revisions to keep up with updated best estimates in the science. It was recently used as one of a suite of emulators linking Working Groups 1 and 3 of the IPCC 6th Assessment Report. In this presentation, we outline the model and its key features and components and show its native projections of future climate following the SSP Pathways and its performance as an IPCC AR6 emulator. We also present a python port of the model that will shortly be made publicly available.

 

For AR6, CICERO-SCM was tuned to reproduce the surface temperature evolution assessed by Working Group 1, as well as a range of other parameters. For probabilistic uncertainty estimation, we built on the method of (Skeie et al. 2018) where a large set of observation based prior assumptions on ocean heat content and temperature change was ran through the model to create a large consistent set of parameters. In the AR6 process, this set of parameters was used as an initial pool of useable parameter sets. From there the AR6 statistical distributions of current temperature, ECS and aerosol forcing were used to create a parameter subset. As part of this effort, a python wrapper was developed and integrated into the openscm-runner framework, to go between the formats and setups of the AR6 inputs, and the setup expected by the Fortran based binary.

CICERO-SCM has recently been ported to python, and is currently being tested for public, open-source release. The base version will include tunable parameters and the possibility for running user generated scenarios and will form the basis for a number of planned extensions – notably regarding short-lived climate forcers and the interaction of anthropogenic climate change with natural variability.

How to cite: Sandstad, M., Skeie, R. B., and Samset, B. H.: The updated CICERO Simple Climate Model – an open-source emulator contribution to the AR6 process, EGU General Assembly 2022, Vienna, Austria, 23–27 May 2022, EGU22-12390, https://doi.org/10.5194/egusphere-egu22-12390, 2022.

11:23–11:29
|
EGU22-9163
|
ECS
|
Virtual presentation
Maybritt Schillinger, Beatrice Ellerhoff, Kira Rehfeld, and Robert Scheichl

Reliable climate projections in face of global warming require a firm and detailed understanding of climate variability. Variations in climate can be externally-forced, for example by anthropogenic emissions, or internally-generated, for example from chaotic atmosphere and ocean dynamics. To investigate the climatic response to radiative forcing, a common concept is the equilibrium climate sensitivity (ECS). Many studies estimate the ECS by fitting simple energy balance models (EBMs) to observational data. This approach has benefitted from advances in numerical analysis and statistics, enabling a fully Bayesian analysis. Via Bayes theorem, it quantifies the probability of certain climate parameters given observations, for example of surface temperature. To this end, it combines the goodness of the model fit with assumptions on measurement errors and climate variability as well as prior information. Here, we analyse and discuss Bayesian inference of climate parameters such as ECS from global mean temperatures using multibox EBMs. We therefore present an R package which relies on the Markov Chain Monte Carlo algorithm and includes an extension of the one-box model with a time-dependent feedback parameter. Using measurements from the instrumental period as well as temperature reconstructions and model data from the last millennium, we validate and demonstrate the package. We find that the two-box model performs significantly better in fitting the observations than the one-box model, and generates 21st century projections that agree more closely with AR5 estimates. Further, we evaluate the robustness of the estimate against uncertainties in temperature and forcing data through synthetic experiments. To this end, we quantify how estimation errors depend on the strength of noise in temperature data and compare the influence of dating and amplitude uncertainties in forcing reconstructions. In summary, we provide an effective tool for Bayesian estimation of climate parameters and elaborate its potential for studying the response to external forcing.  

How to cite: Schillinger, M., Ellerhoff, B., Rehfeld, K., and Scheichl, R.: Bayesian Inference of Climate Parameters Using Multibox EBMs, EGU General Assembly 2022, Vienna, Austria, 23–27 May 2022, EGU22-9163, https://doi.org/10.5194/egusphere-egu22-9163, 2022.

11:29–11:35
|
EGU22-6346
|
ECS
|
On-site presentation
Mikkel Bennedsen, Eric Hillebrand, and Siem Jan Koopman

In this paper, we propose a new, fully statistical, reduced complexity climate model. The starting point for our model is a number of physical equations for the global climate system, which we show how to cast in non-linear state-space form. The resulting model incorporates measurement errors, capturing the fact that observations of physical quantities might be contaminated by error, as well as internal stochastic error processes, capturing the fact that the physical equations used are approximations to the true underlying  climate system. The state-space formulation allows for statistical estimation of the parameters in the model, using the method of maximum likelihood, as well as filtering and smoothing of latent quantities in the model, such as ocean and surface temperatures. Further, the explicit statistical formulation of the model allows for conducting a number of useful analyses, such as the estimation of parameter uncertainty, model selection, and probabilistic scenario analysis. 

 

By considering a range of different scenarios for greenhouse gas emissions, we set up simulation studies that can be used to investigate  the effect that a given scenario has on parameter estimates. We find substantial differences in the performance of the estimation procedure, depending on the precise scenario considered. These investigations can help decide what kind of data are best suited for estimating/calibrating the parameters of reduced complexity climate models, e.g. to what extend the historical data record can be used to reliably estimate parameters and/or which CMIP experiments are best suited for calibrating such models.

 

Using a data set of historical observations from 1959-2020, we estimate the model and report key parameter estimates and associated standard errors. A likelihood ratio test sheds light on the most appropriate equation for converting the atmospheric concentration of carbon dioxide (GtC) into forcings (W/m2). We then use the estimated model and assumptions on future greenhouse gas emissions to project global mean surface temperature out to the year 2100. The statistical nature of the model allows us to attach uncertainty bands to the projections, as well as quantify how much of the uncertainty is "aleatoric" (uncertainty arising from the internal variability of the climate system) and how much is "epistemic" (uncertainty arising from unknown model parameters). We find that epistemic uncertainty is by far the most important contributor to the uncertainty on the projected future global temperature increase.

How to cite: Bennedsen, M., Hillebrand, E., and Koopman, S. J.: A New Statistical Reduced Complexity Climate Model, EGU General Assembly 2022, Vienna, Austria, 23–27 May 2022, EGU22-6346, https://doi.org/10.5194/egusphere-egu22-6346, 2022.

11:35–11:41
|
EGU22-5538
|
ECS
|
Highlight
|
On-site presentation
Susanne Baur, Benjamin Sanderson, Roland Séférian, and Laurent Terray

The Earth system response to climate forcers can be broken down to multiple timescales, with the land surface responding within a few years to a change in forcing while the deep ocean layers have only fully equilibrated after several hundreds to thousands of years. In this work we assume that there is a number of distinct timescales represented in the thermal response to pulse injections of different climate forcers in the Coupled Model Intercomparison Project Phase 6 (CMIP6) Earth System Models (ESMs), which can be estimated by fitting a sum of decaying exponential responses  to a set of non-noisy Empirical Orthogonal Functions of each model. Using these exponential decay functions and a regression-based pattern scaling approach we are able to emulate the gridded transient surface temperature response to an input forcing timeseries. We determine that for the abrupt-4xCO2 experiment the thermal response in most CMIP6 ESMs can be represented by a similar set of timescales, but early results suggest diverse spatial warming patterns. This work introduces the concept that the evolving spatial patterns associated with the thermal response on different timescales for pulse injections of different climate forcers can be simply and accurately emulated and ultimately be used to predict transient simulations.

How to cite: Baur, S., Sanderson, B., Séférian, R., and Terray, L.: Thermal response timescales and associated spatial patterns in ESMs to pulse injections of climate forcers, EGU General Assembly 2022, Vienna, Austria, 23–27 May 2022, EGU22-5538, https://doi.org/10.5194/egusphere-egu22-5538, 2022.

11:41–11:50
Lunch break
Chairpersons: Zebedee R. Nicholls, Bjorn H. Samset, Christopher Smith
13:20–13:26
|
EGU22-9739
|
ECS
|
Highlight
|
On-site presentation
Gaurav Ganti, Robert J. Brecha, Robin D. Lamboll, Zebedee Nicholls, Bill Hare, Jared Lewis, Malte Meinshausen, Michiel Schaeffer, Christopher J. Smith, and Matthew J. Gidden

Scientifically rigorous guidance to policy makers on mitigation options for meeting the Paris Agreement long-term temperature goal requires an evaluation of long-term global-warming implications of greenhouse gas emissions pathways. Here, we present a uniform and transparent methodology to evaluate the climate outcome, and hence the Paris Agreement consistency of influential institutional emission scenarios from the grey literature, including those from the International Energy Agency1,2, BP3, and Shell4. We first identify challenges to performing such an assessment and then proceed to outline a sequence of steps to address these challenges by harmonizing5 all emissions to a consistent base-year (2010), extending all pathways to 2100, and filling in missing emission species6. We employ two simple climate models, MAGICC7 and FaIR8,9 to assess peak and end-of-century temperatures, and find that few published scenarios that claim to be compatible with the Paris Agreement are so.

 

References

 1. International Energy Agency. World Energy Outlook 2020. (2020).

2. International Energy Agency. Net Zero by 2050 - A Roadmap for the Global Energy Sector. (2021).

3. BP. Global Energy Outlook 2020. (2020).

4. Shell. The Energy Transformation Scenarios. (2021) 

5. Gidden, M. J. et al. A methodology and implementation of automated emissions harmonization for use in Integrated Assessment Models. Environ. Model. Softw. 105, 187–200 (2018)

6. Lamboll, R. D., Nicholls, Z. R. J., Kikstra, J. S., Meinshausen, M. & Rogelj, J. Silicone v1.0.0 : an open-source Python package for inferring missing emissions data for climate change research. Geosci. Model Dev 13, 5259–5275 (2020)

7. Meinshausen, M., Raper, S. C. B. & Wigley, T. M. L. Emulating coupled atmosphere-ocean and carbon cycle models with a simpler model, MAGICC6 - Part 1: Model description and calibration. Atmos. Chem. Phys. 11, 1417–1456 (2011).

8. Smith, C. J. et al. FAIR v1.3: A simple emissions-based impulse response and carbon cycle model. Geosci. Model Dev. 11, 2273–2297 (2018)

9. Millar, J. R., Nicholls, Z. R., Friedlingstein, P. & Allen, M. R. A modified impulse-response representation of the global near-surface air temperature and atmospheric concentration response to carbon dioxide emissions. Atmos. Chem. Phys. 17, 7213–7228 (2017).

How to cite: Ganti, G., J. Brecha, R., D. Lamboll, R., Nicholls, Z., Hare, B., Lewis, J., Meinshausen, M., Schaeffer, M., J. Smith, C., and J. Gidden, M.: Assessing the consistency of institutional pathways with the Paris Agreement , EGU General Assembly 2022, Vienna, Austria, 23–27 May 2022, EGU22-9739, https://doi.org/10.5194/egusphere-egu22-9739, 2022.

13:26–13:32
|
EGU22-1075
|
ECS
|
Highlight
|
On-site presentation
Thomas Bossy, Thomas Gasser, Philippe Ciais, Katsumasa Tanaka, and Franck Lecocq

The physical reality of the Earth system implies that there are clear conditions to respect the Paris agreement, or to limit any climate impact below a certain level. To be policy relevant, these conditions should be expressed in terms of emission targets, such as peaking date, budget, or annual value of global CO2 emissions. They have been explored by the IPCC using integrated assessment models. However, past work has focused on bottom-up scenarios, and on temperature as the only metric evaluating climate impacts, even though not all impacts are linearly related to it.

Here, we show that for these emission targets, across thousands of scenarios we have generated, there are points of no return after which limiting a given climate impact becomes geophysically infeasible. In addition to the Paris Agreement objectives (consisting of a 1.5 °C temperature target above pre-industrial (PI) era possibly overshot by no more than 0.5°C), we investigate three other climate targets:  ocean acidification, sea-level elevation rate, and Arctic sea-ice melting. We use a newly developed model called PathFinder; a reduced-form carbon-climate model that also emulates the three global climate impacts we investigate. The model is calibrated through Bayesian inference, using outputs from the state-of-the-art CMIP6 models as prior parameters, and the latest IPCC assessment and observations of the Earth system as constraints. This advanced calibration is enabled by the model’s capacity of using temperature and atmospheric CO2 concentration as inputs (instead of anthropogenic emissions and non-CO2 radiative forcing).

Thanks to this backward approach, we demonstrate that, for every emission target considered, the combination of climate impact targets is non-linear. While the Paris Agreement insists on the importance of reaching a carbon neutral world in 2050, our results show that global CO2 emissions must peak before 2030 but do not have to reach net-zero to keep all targets reachable with at least 50% chances. We also highlight the inevitable role of geoengineering technologies in reaching the Paris Agreement, as chances to keep it reachable goes from at least 69% if SRM or CDR are available to 10% if none of them is.

How to cite: Bossy, T., Gasser, T., Ciais, P., Tanaka, K., and Lecocq, F.: Points of no return to respect the Paris Agreement, EGU General Assembly 2022, Vienna, Austria, 23–27 May 2022, EGU22-1075, https://doi.org/10.5194/egusphere-egu22-1075, 2022.

13:32–13:38
|
EGU22-12530
|
On-site presentation
Thomas Gasser

Simple climate models (SCMs) are most often composed of ad hoc parametric laws that emulate the behaviour of more complex Earth system models (ESMs). The emulation allows investigating experiments or scenarios that would be too costly to compute with ESMs. However, the “SCM” denomination refers to a fairly broad range of models whose complexity can go from a couple of boxes that only emulate one part of the climate system (e.g. a global temperature impulse response function) to hundreds or thousands of boxes representing the different cycles of greenhouse gases and induced climate change (e.g. the compact Earth system model OSCAR). Simpler models are easier and faster to solve, but they may not adequately represent physical processes. Therefore, finding the “simplest but not simpler” model depends on a study’s precise goals.

We developed the Pathfinder model to remedy a deficiency within the spectrum of existing SCMs. Pathfinder is a compilation of existing formulations describing the climate and carbon cycle systems, chosen for their balance between mathematical simplicity and physical accuracy. The resulting model is simple enough that it can be used with Bayesian inference algorithms for calibration, which enables integration of the latest data from CMIP6 Earth system models and the IPCC AR6, as well as a yearly update using observations of global temperature and atmospheric CO2. The model’s simplicity also enables coupling with integrated assessment models (IAMs) and their optimization algorithms, or simply running the model in a backward temperature-driven fashion. In spite of this simplicity, the model accurately reproduces behaviours and results from complex models – including uncertainty ranges – when ran following standardized diagnostic experiments.

Here, we will briefly describe the Pathfinder model, demonstrate its performance, and illustrate its strengths and potential with two example studies. The first one combines a very large-scale ensemble of climate change scenarios generated procedurally, and the physical uncertainty sampling extracted from the Bayesian calibration, to determine which future CO2 emissions pathways remain compatible with the Paris agreement. The second one couples Pathfinder with a stylized IAM and climate impact emulators, to generate cost-effective pathways that limit permafrost carbon thaw, sea level rise speed, and ocean surface acidification.

How to cite: Gasser, T.: Pathfinder: a simple yet accurate carbon-climate model to explore climate change scenarios, EGU General Assembly 2022, Vienna, Austria, 23–27 May 2022, EGU22-12530, https://doi.org/10.5194/egusphere-egu22-12530, 2022.

13:38–13:44
|
EGU22-8001
|
ECS
|
Virtual presentation
Jingying Zhou Lykke, Mikkel Bennedsen, and Eric Hillebrand
In this paper, we propose a state space representation (EBM-SS model) of the two-component energy balance model (EBMs). The EBM-SS model incorporates three extensions to the two-component EBM. First, we include ocean heat content (OHC) as a measurement of the temperature in the deep ocean layer. Second, we decompose the latent state of radiative forcing into a natural component and an anthropogenic component. The anthropogenic component is modeled as a random walk process with a local linear trend to represent the deterministic and stochastic trends of anthropogenic forcing, while the natural component captures the variations in solar irradiance and transitory episodes in forcing following major volcanic eruptions. Lastly, we use multiple GMST anomaly data sources from separate research groups as measurements for the latent state -- the temperature in the mixed layer in the two-component EBM. 
We estimate the EBM-SS model using observations at the global level during the period 1955 -- 2020 by maximum likelihood. We show in empirical estimation and in simulations that using multiple data sources for the latent process reduces parameter estimation uncertainty. When fitting eight global mean surface temperature anomaly observational series, the physical parameter estimates are comparable to those obtained by using datasets from Coupled Model Intercomparison Project 5 (CMIP 5) in other literature.  We find that using this set of parameter estimates, the GMST projection results under Representative Concentration Pathway (RCP) 4.5, 6.0, and 8.5 scenarios considerably agree with the outputs from the climate emulator Model for the Assessment of Greenhouse Gas Induced Climate Change (MAGICC) 7.5 and CMIP 5 models. We show that utilizing a simple climate model and historical records alone can produce meaningful GMST projections.

How to cite: Lykke, J. Z., Bennedsen, M., and Hillebrand, E.: Global Mean Surface Temperature Projection Constrained by Historical Observations, EGU General Assembly 2022, Vienna, Austria, 23–27 May 2022, EGU22-8001, https://doi.org/10.5194/egusphere-egu22-8001, 2022.

13:44–13:50
|
EGU22-5223
|
ECS
|
On-site presentation
Roman Procyk and Shaun Lovejoy

We present the fractional energy balance equation (FEBE) which is a generalization of the standard EBE. The FEBE can be derived either from Budyko–Sellers models or phenomenologically by applying the scaling symmetry to energy storage processes. It is easily implemented by changing the integer order of the storage (derivative) term in the EBE to a fractional value near 1/2. 

The model used a Bayesian framework based on historical temperatures and natural and anthropogenic forcing series for parameter estimation. Significantly, the error model was not ad hoc, rather predicted by the model itself: the internal variability response to white noise internal forcing, a fraction Relaxation noise (fRn). Due to computational constraints, we employ a block bootstrapping method to calculate the likelihoods of our parameters in the Bayesian scheme. Notably we estimate the regional relaxation time directly from empirical data, generally it is calculated for various discrete surface types using heat capacities or globally from fitting a two-box model to GCM outputs, which to the authors knowledge has not been estimated prior to this study. 

The FEBE historical reconstructions (1880–2020) closely follow observations (notably during the “slowdown”, 1998–2015). We also reproduce the internal variability with the FEBE and statistically validate this against centennial scale temperature observations. We show the FEBE to plausibly reproduce the annual cycle at monthly resolution, in particular to explain the lag between the temperature maximum and the maximum in the radiative forcing. 

Using the calibrated FEBE we made temperature projections to 2100 using both the Representative Carbon Pathways (RCP) and Shared Socioeconomic Pathways (SSP) scenarios, shown alongside the Coupled Model Intercomparison Project Phase (CMIP) 5 and 6 multi-model ensemble (MME) at global and regional scales.  

How to cite: Procyk, R. and Lovejoy, S.: The Regional Fractional Energy Balance Equation: Projections to 2100, EGU General Assembly 2022, Vienna, Austria, 23–27 May 2022, EGU22-5223, https://doi.org/10.5194/egusphere-egu22-5223, 2022.

13:50–13:56
|
EGU22-6522
|
ECS
|
Presentation form not yet defined
Kalyn Dorheim, Ben Bond-Lamberty, Leeya Pressburger, Dawn Woodard, Skylar Gering, and Alexey Shiklomanov

Hector is a carbon/climate model capable of emulating Earth System Model outputs at the global scale and is able to reproduce historical observations well. Like other participating models of the Reduced Complexity Model Intercomparison Project, Hector is a computationally efficient source of climate projections and thus has a wide range of applications such as scenario generation, coupling with integrated assessment models, outreach, education, and policy making. Hector version 3 includes a number of new features: carbon tracking, permafrost, improved land-ocean warming contrast, and a web browser-accessible interface. Here we summarize these developments and discuss how they improve the model’s performance and broaden its potential user base. 

How to cite: Dorheim, K., Bond-Lamberty, B., Pressburger, L., Woodard, D., Gering, S., and Shiklomanov, A.: New features, broader accessibility, and improved performance of the Hector v3 simple carbon/climate model, EGU General Assembly 2022, Vienna, Austria, 23–27 May 2022, EGU22-6522, https://doi.org/10.5194/egusphere-egu22-6522, 2022.

13:56–14:05