The Backbone of the Destination Earth Climate Adaptation Digital Twin

Since the first concerns were raised in the 1980s that the climate may undergo catastrophic changes caused by the increasing greenhouse gas emissions[1,2], a number of multilateral efforts have been brought to life for investigating the future climate evolution. The most popular of such joint ventures are the Coupled Model Intercomparison Project (CMIP), and the Intergovernmental Panel on Climate Change (IPCC). The aim has always been to underpin the evidence of climate change, and providing information for policymakers to deal with the resulting consequences. However, the data from those initiatives lacks interactivity, it is generated once, based on a specific simulation protocol and scenario, and is only available at relatively low resolution and frequency due to storage limitations.

The Destination Earth initiative of the European Commission (DestinE) is building upon those efforts while trying to operationalize the generation of global climate projections, iteratively adapting the simulations to new requirements posed by data consumers. The Climate Adaptation Digital Twin (CAT) of DestinE promises interactive access to data generated by global operational Earth System Model (ESM) ensemble-simulations at very high resolution (4-5 Km atmosphere and 5-10 Km ocean) and high frequency (including variables available at sub-hourly frequency). It incorporates what-if scenarios as well as time-slice experiments. 

To deal with the considerable amount of generated data, CAT adopts a streaming approach, where the full model output data is exposed to the data consumers in continuously progressing windows as the climate model runs, erasing older data after having stored a user-defined subset of variables. The data is exposed as a generic state vector (GSV) providing a common interface to the data coming from different models (IFS, ICON, NEMO, FESOM) and realms (atmosphere, land surface, sea ice, and ocean). At the heart of the project lies the interaction with the data consumers driving the GSV design in terms of resolution and frequencies.

In this talk, we explore the opportunities that DestinE provides and the sophisticated software machinery that is required to make it happen. We will give an overall picture on the structure and vision of CAT, and highlight the challenges posed by interactivity, reproducibility, provenance, data accessibility, synchronization, and quality monitoring. In particular, we will showcase diagnostics from decadal high resolution climate simulations, report on the replicability of a coupled model running the ocean in a mixed precision implementation, and detail on the model workflow employed for operationalization of the climate projections. We will further give an outlook on the different kinds of simulations that are planned and explain how the data is provided to the data consumers. 

References

[1] Hansen, J., Fung, I., Lacis, A., Rind, D., Lebedeff, S., Ruedy, R., Russell, G., & Stone, P. (1988). Global climate changes as forecast by Goddard Institute for Space Studies three-dimensional model. Journal of Geophysical Research: Atmospheres, 93(D8), 9341–9364. https://doi.org/10.1029/JD093iD08p09341

[2] Manabe, S., & Stouffer, R. J. (1980). Sensitivity of a global climate model to an increase of CO2 concentration in the atmosphere. Journal of Geophysical Research: Oceans, 85(C10), 5529–5554. https://doi.org/10.1029/JC085iC10p05529