Understanding the impact of climate change
Climate scientists try to understand the impact of climate change by running climate models. There are many different models (and datasets) available from various different institutions which have been developed in different languages. Integrating these models so that they work together would increase confidence in their predictions, and this is the goal of the Enhancing CIAS (ECIAS) project. We are working with ECIAS to enable new ways of coupling the models, and to investigate how they are stored, queried and visualised.
At the University of East Anglia, Rachel Warren’s team developed the Community Integrated Assessment System (CIAS). CIAS is a web portal that brings together the different climate models and datasets. At the core of CIAS is the Bespoke Framework Generator (BFG), which links together pieces of software that model different aspects of the climate. This means that the output of a model that simulates greenhouse gas emissions can be fed into another model that simulates hydrological modelling, surface temperature or various other features of the climate. By linking these models together, climate scientists can study the complex relationships that determine our climate.
The Software Sustainability Institute will help Rachel Warren's team to further develop ECIAS.
The current BFG runs the models in sequence. The enhanced system will permit the introduction of feedback loops. This will allow a whole new class of climate simulations to be run, the results of which can be used to test the models’ robustness and to explore new climate scenarios.
The CIAS portal does not have a queue. This means that researchers have to wait for the portal to finish processing one simulation before starting the next simulation. We will help add the Torque/PBS batch scheduling system, which allow the portal to queue simulations so that many researchers can submit their run models and come back later to collect their results. This capability could be used to farm out simulations to cluster-computing facilities, which will increase computational power so that many independent scenarios can be run simultaneously.
Climate simulations are extremely complex so they can be expensive in terms of CPU and elapsed time. Under these circumstances, it is extremely important that researchers do not repeat simulations that have been completed in the past. We will add a search and query interface to the enhanced portal, which will allow researchers to search previous results to see if someone else has already performed a similar simulation.
The simulations deliver an output in the commonly used netCDF format (www.unidata.ucar.edu/software/netcdf ). Currently, it is up to the user to interpret the output data. The enhanced portal will have a common visualisation tool so that researchers can see their results directly.
Installation, configuration, system design, and backup procedure documentation has been subjected to third party review and has been expanded where necessary. This ensures that the portal can be replicated either on new infrastructure within UEA or on other sites, protecting the service from catastrophic machine failure and allowing other future partners to offer the same service in other parts of the world.
Last updated: Monday 7 November 2011.