Skip to content

Search for Research Data

Search the ANDS Site


Maximising the benefits of high-resolution climate modelling


A story about building bridges

By Ian Macadam

How will climate change affect the things that we care about? Will water be scarcer? Will our crops be affected? Will our ecosystems be damaged? Will our health suffer? These are all questions that we need to answer if we are to adapt to a changing climate. As a result, there has been a surge in demand for assessments of the possible impacts of climate change in recent times.

It is the global climate system that is changing, and researchers have been seeking to understand this global system for some time now. Global Climate Models have been helping them in this quest. These computer representations of the climate system have proved useful for allowing researchers to unpick the behaviour of the real climate. They have also been pressed into service to generate projections of future climate conditions. In this mode, they are used to answer increasingly relevant "what if" questions about the future - "What if greenhouse gas emissions continue to increase?", "What if greenhouse emissions start to decline sometime this century?". The "answer" from a Global Climate Model often takes the form of a climate map of the world, but unfortunately the map lacks detail at a regional scale. Large-scale features of the climate (e.g. the main climate belts and the contrasts between continental and oceanic climates) are visible, but the more local effects of mountain ranges and coastlines are not. However, these effects are important if we want to know how the climate will impact us. It is no use feeding data from a Global Climate Model representing average conditions over ten thousand square kilometres into a crop modelling system designed to simulate crop yields from a paddock lying in the lee of a mountain range.

Regional Climate Models are an additional tool in the climate modelling arsenal that can help us to address this dilemma. Instead of simulating the climate of the whole globe at a coarse resolution, they simulate the climate of a continent or smaller region at a much finer resolution. The grid cells of a Global Climate Model are around a hundred kilometres apart, whereas the grid cells of a Regional Climate Model can be 50km, 10km or even just a few kilometres apart. Regional Climate Models are therefore able to better simulate the local climatic effects of mountains and coastlines and this can enable them to provide more realistic climate data to climate impacts assessments.

This is all well and good so far - climate impacts researchers can use output of Regional Climate Models. Unfortunately this is more easily said than done. There are some practical problems to be overcome.

Firstly, Regional Climate Model datasets are vast. Not only are they high-resolution, but they contain data for many climate variables, and values of some of these are stored for many altitudes in the atmosphere, and not just for the Earth's surface. Furthermore, although these data may not cover the globe, they may cover the continent or a large region. This means that the entire dataset is not very "portable". Climate impacts researchers therefore need a way of interrogating the data and extracting just the variables that they are interested in for just the areas that they are interested in.

Secondly, Regional Climate Model output is generally stored in a binary file format called "netcdf" and data values are often only stored on the grid on which the model calculations are done, which is often not a standard grid (e.g. a standard regular latitude-longitude "geographic" grid). Climate modellers use specialist software to handle these netcdf files. Impacts researchers also use specialist software, but this often cannot handle netcdf files or non-standard grids.

The "Climate Model Downscaling Data for Impacts Research" (CliMDDIR) ANDS Applications project has developed software that resolves this problem. The result is a portal that allows users to extract data from a climate model dataset stored in netcdf and download it in a more widely used format. It was developed by a team at the University of New South Wales' node of the Australian Research Council's Centre of Excellence for Climate System Science (ARCCSS). The team worked closely with impacts researchers in the fields of agriculture, ecosystems, human health and hydrology to make sure that the portal met the needs of potential users.

CliMDDIR outputs ASCII files that can be opened by numerous software packages, including GIS software. A user-friendly interface allows users to select their climate variables of interest and CliMDDIR returns data for these for either a set of user-selected locations or a user-selected region of interest. CliMDDIR therefore helps to bridge the gap between climate modellers and impacts researchers.

The software behind CliMDDIR is freely available to others and comes with documentation. The software has been designed with multiple climate model datasets in mind. It can be configured to serve data from virtually any climate model dataset stored in netcdf. The idea is to give a solution to anyone facing the problem of making climate model data widely available. This problem is likely to become more and more widespread in the future as increasing computer power allows modelling groups to run more and more climate model simulations. These will include many more Global Climate Model simulations but also Regional Climate Model simulations at the high spatial resolutions that are relevant for assessing the impacts of climate change.

An especially important aspect of CliMDDIR is that it allows users to assemble their own collections of data and describe and expose them to other users of the portal, and more widely through the ANDS Research Data Australia website. This makes CliMDDIR unique. Not only does it give users access to climate model output without the need for them to acquire and learn how to use specialist software, it helps them curate their data and share it with others.

As well as the portal itself, the ANDS project has helped to build a community of scientists, software developers and public servants who are now interacting to solve the issues of making "big" climate data accessible to those who would benefit from it. In addition to enhancing the relationships between ANDS, ARCCSS and other interested organisations, such as CSIRO and NSW Government Office of Environment and Heritage, it has led to more involvement of ARCCSS in big data initiatives such as RDS and SPEDDEXES.

The governments of NSW and ACT have embraced regional climate modelling and have embarked on the NSW and ACT Regional Climate Modelling project (NARCliM). Under this project, data from a whole set of Regional Climate Model simulations performed by a team led by Jason Evans at the University of New South Wales have been used to underpin the latest climate projections for the state and territory. It is important that impacts researchers are able to access the data and use it in their assessments of the impacts of climate change. Therefore the CliMDDIR development team have worked with NSW Government Office of Environment and Heritage to hitch CliMDDIR to the NARCliM data. CliMDDIR now forms part of The NSW Climate Data Portal, allowing numerous impacts researchers to access data from a major regional climate modelling project.

An archived version of the CliMDDIR and video demonstration of the web portal is available here.