OBSALL: DestinE climate simulations in observation space

Destination Earth has given rise to a new generation of global climate models capable of resolving kilometre-scale processes. These storm-resolving models are deployed operationally to produce multidecadal climate simulations within Climate Adaptation Digital Twin. OBSALL conceptualise and implemente an Earth observation-based system for monitoring the quality of the climate simulations. Technically, it is a one-pass algorithm in which observation models operate online on the state vector of the simulation model and generate a full-resolution trace in the observation space. This enables real-time monitoring of the simulation and posterior evaluation right after its completion, thus enhancing the overall resilience of the Climate DT workflow.

OBSALL has the projection and monitoring pipeline implemented for three observation modalities: surface variables with SYNOP weather stations, vertical profiles with TEMP radiosoundings, and satellite products with AMSU-A. OBSALL consumes climate model states in the form of stream of Generic State Vectors, projects the model states in to the observation space spanned by activated observation modalities, stores the projected states in Observation DataBase files and intermittently produces a suite of monitoring plots comparing the climate simulation to climatology of observations.  All modalities share the same general workflow structure facilitating implementation of new modalities or features. Finally, OBSALL has been designed to allow deployment on a range of environments from personal computer to HPC infrastructures.

This functionality will be part of the EU’s Climate Adaptation Information Service and it is intended to benefit Users of the Service, although it also holds strong research appeal. From the User perspective, presenting the simulation in observation space is advantageous. Earth observations, especially the in-situ components, are intuitive and easy to interpret. Therefore, observation-space projections provide a concrete handle on adaptation information and enhance the User relevance of the DestinE Climate DT. These projections tend to lower the threshold for users to engage with the Information Service. Therefore, we see future development of user-oriented tools using the projection data as input as a promising strategy to attract new users. Extending the projection to other user-relevant observation types, such as long-term wind mast measurements, would be beneficial.

Here, we showcase the current capabilities of the climate model projection and monitoring software OBSALL from the view-points of runtime simulation monitoring through different observation types, and observation-based posterior validation, which offers a versatile way to validate process-level features in storm-resolving climate models.