Post sponsored by

Source: United Kingdom – Executive Government & Departments

On 4 September 2019, Jonathan Tawn from the Department of Mathematics and Statistics at Lancaster University, Andrew Parnell from the Hamilton Institute at Maynooth University, and Thordis Thorarinsdottir from the Norwegian Computing Center participated in the Royal Statistical Society conference 2019 session on Climate Change. The contributed session was organised by the Environmental Statistics section of the Royal Statistical Society and made up part of the Environment and Spatial Statistics stream of sessions at the conference.

Speaking first, Jonathan Tawn noticed that when considering environmental data, the focus is often on the mean. His talk sought to answer what happens in the extreme and consider aggregates of regional versions of climate models against their global counterparts. He examined heatwaves and extreme sea waves to gain an understanding of how such events could become more common under climate change. Combining climate models in an ensemble provided extremes increasing as much as three times faster than the mean, in terms of changes in temperature.

Next, Andrew Parnell spoke of how to quantify climate variable values from the past using proxy variables we can measure in the current age. He considered pollen dug out of lakes and noted that in order to obtain climate data from an ice core dating back 100,000 years, you need calibration data from the present day to include the link between the current proxy and climate values. The proxy is considered the outcome while climate is modelled as explanatory variables, which was noted as not being the common thing to do when wanting to predict climate. He noted that proxies should not be modelled individually as they are not independent.

The final speaker was Thordis Thorarinsdottir who showcased the work behind last year’s Significance cover story on sea level threats to the Norwegian city of Bergen (Volume 15, Issue 2). She focused on the adaptation of measures to mitigate climate change and noted that producing evidence is not enough if money budgeted towards implementing measures is not available. She considered uncertainty around future sea level rise around Bergen and considered three placements of barriers that could be built around the city to stop flooding. She modelled the changes in damage costs as well as the changes in climate and examined the effects on the choice of timing and spending when different sources of uncertainty around the estimates were considered.

It was considered important to examine historical data available beyond that found in the time series, as these could extend the window data provides into the past. In discussion it was noted that while records are interesting events, it is better to model the entire process and examine where records lie within the model. It was also noted that there may not always be enough data to reconstruct an extreme climate, so variable selection is key. Finally, the fact that discount rates can have a huge effect on decision-making was mentioned.

MIL OSI United Kingdom