Researchers at the University of South Wales have been helping NASA to increase the accuracy of their weather predictions.
In the first few months of 2020, the UK saw unprecedented weather patterns. In February, it experienced the highest amount of rainfall in any month since records began in the 1800s, and widespread flooding devastated local communities.
Combined with the growing concerns around climate change, the importance of being able to accurately predict the weather, in order to prepare and potentially save communities and to plan for the impact on the environment and the economy, has never been clearer.
Maths researchers at the University of South Wales have been working with NASA to refine their atmospheric model, which is needed to make weather and climate predictions. As a United States Government agency, their research is used to influence policy decisions, and is one of several agencies in the US that undertake climate modelling.
Although it is beneficial to use the same numerical methods in both, it is not always possible in practice. For example, the use of non-linear limiters for weather and climate prediction models is considered paramount for obtaining accurate forecasts. Although these limiters work very well in the Dynamical Core, they are not well suited to being linearised for the Linear Model.
To help improve the prediction models currently being used, Dr James Kent, a mathematician at the University of South Wales, worked with NASA to take a different approach.
During initial testing, researchers applied the new model to tracer transport, such as predicting the movement of clouds as well as quantities such as specific humidity. When testing this against the full Dynamical Core, they found that the predictions had a greater correlation, i.e. they were more accurate.
The updated model has also been used to look at the sensitivity to different parameters within weather predictions, including changes to the wind, temperature and water vapour, and what effect this could have. The same model can also be used to look at the concentration of gasses in the atmosphere, helping to monitor climate changes and giving scientists more accurate predictions for the future.
NASA’s atmospheric model, based on a series of mathematical equations, is separated into many components. One is the Linear Model, for data assimilation. This is used to input atmospheric data, such as temperature, wind speed or humidity, into the models. Another component is the Dynamical Core. This solves the governing equations to predict the state of the atmosphere in the future.
Although it is beneficial to use the same numerical methods in both, it is not always possible in practice. For example, the use of non-linear limiters for weather and climate prediction models is considered paramount for obtaining accurate forecasts. Although these limiters work very well in the Dynamical Core, they are not well suited to being linearised for the Linear Model. To help improve the prediction models currently being used, Dr James Kent worked with NASA to take a different approach.
“Instead of taking the numerical methods from the Dynamical Core and linearising them, we looked at what they needed the Linear Model to do in order to achieve as accurate a prediction as possible,” said Dr James Kent.
“We then worked backwards to identify what methods would help to produce this. This change of viewpoint led to a breakthrough and a change in the way that NASA’s model functioned.”
During initial testing, researchers applied the new model to tracer transport, such as predicting the movement of clouds as well as quantities such as specific humidity. When testing this against the full Dynamical Core, they found that the predictions had a greater correlation, i.e. they were more accurate.
The updated model has also been used to look at the sensitivity to different parameters within weather predictions, including changes to the wind, temperature and water vapour, and what effect this could have. The same model can also be used to look at the concentration of gasses in the atmosphere, helping to monitor climate changes and giving scientists more accurate predictions for the future.
