Next Generation Weather & Climate Prediction
Technical developments in computing and in global scale observations create an environment of enormous opportunity for improving capabilities for weather and climate prediction, but there are major scientific and technical barriers to realizing this potential.
This programme aims to addresses two key areas where scientific and technological advances offer opportunities to maintain UK leadership in environmental prediction:
- Goal A - Resolution of small scale weather systems in the atmosphere and ocean
- Goal B - Use of observations to initialise climate predictions
One issue requiring urgent attention is the selection of numerical algorithms currently used to simulate the atmosphere in climate prediction models.
Many of these algorithms are unable to take advantage of the additional processing power of massively parallel computers moving towards petascale and exascale resources.
The development of new algorithms involves research challenges in numerical and computational methods.
It is proposed that a consortium is created, between NERC, STFC and the Met Office to research, design, and develop a new atmospheric dynamical core for a next-generation weather and climate prediction system, to eventually replace the dynamical core of the Unified Model (UM), the principal UK tool for weather and climate prediction.
The corresponding issues for ocean models appear to be less immediately pressing, based on current understanding of needs.
However there is an important need to develop a longer term view based on the emerging scientific requirements for ocean modelling (eg integrated modelling of deep ocean and shelf processes), and a 'roadmap' of needs for massively parallel ocean modelling systems will be developed.
This programme is essential to ensuring that the UK has access to world class tools for climate prediction in the future.
A second strand of this action supports research to develop and evaluate methods for initialization of climate predictions, addressing issues of coupled data assimilation and the design of ensembles.
For predictions of regional climate on seasonal to decadal timescales, much of the signal comes from internal variability of the system, rather than changes to anthropogenic forcing.
Hence climate predictions on these timescales inherently need to be initialised with the observed climate state.
The science of how to do this is at a very early stage of development and a number of fundamental questions must be answered in order to derive the most possible information from the available observations.
The development of properly initialized climate predictions is central to exploiting the opportunities to improve climate models and predictions that arise from the wealth of new observations from Earth-based and space-based platforms.