The next frontier of regional climate modelling is not in producing more data, but in producing more information.
Climate information is becoming increasingly important for effective planning, adaptation and mitigation to future costs and disruptions arising from climate variability and change. However, this increasing demand for climate information is driving an explosive increase in the volume of climate data. This creates obstacles for users who need to access this data. Accessing such a large volume of climate data will require highly specialised skills and tools which can be unaffordable for many users. Therefore, an emerging challenge in climate change research is ensuring that climate information remains accessible to all users and stakeholders. FRONTIER aims to address this challenge through the key issues of data production and data analysis.
FRONTIER focuses on providing a fundamental breakthrough on how climate model data is generated and analysed, so that we can provide more reliable climate information using less data and computer resources, and thereby accelerating time to science discovery by orders of magnitude. In FRONTIER, we aim to quantitatively assess the simulation of high impact weather events in regional climate models, reduce the number of performance metrics for more efficient analysis, and constrain the size of climate ensembles through a novel approach called Design of Experiment (DoE)-based ensemble. We believe that the next frontier of regional climate modeling is not in producing more data, but in producing more information through a targeted reduction of the data volume and by increasing its representativeness. This will make a substantial contribution towards Sustainable Development Goal (SDG) 13 "Climate action" by directly influencing the production of representative and skilful climate information.
The methodological approach of FRONTIER is based on:
(1) Developing novel process-based model analysis metrics in Lagrangian space to identify optimal model resolutions to capture societally relevant processes
(2) Developing a new reduced set of performance metrics using Big Data methods to simplify and improve efficiency in data analysis
(3) Exploring a new approach to multi-model ensembles, which we call Design of Experiment (DoE)-based ensemble, and contrast it with the current 'ensemble of opportunity'
In FRONTIER, we believe that the next frontier of regional climate modeling is not in producing more data, but in producing less (more representative) information and improving efficiency in data analysis. Hence, we propose three underpinning frontier questions: a) can the RCM added value be better detected in a Lagrangian framework? b) can the number of performance metrics be reduced? c) can the ensemble of opportunity be replaced by something better?