Weather system models used by engineers, architects and planners to create and maintain major infrastructure including flood defences are having to be renewed to reflect climate change.
More sophisticated extreme rainfall model
Engineers and computer scientists in the UK’s University of Bath and Turkey’s Erzurum Technical University are creating a new, more sophisticated extreme rainfall model, using machine learning to fully understand the risks posed by future floods.
The model, the first national-scale system of its kind, is being created to help Turkey’s hydrologists, constructors and planners prepare for likely extreme rainfall in the future, helping them to design and protect infrastructure accordingly.
The research team, funded by the British Council, intend for their findings to be integrated into future planning guidelines.
Dr Thomas Kjeldsen, one of the study’s leaders and a Reader in Bath’s Department of Architecture and Civil Engineering, said: “The existing weather and rainfall models used that underpin design and planning guidelines are increasingly underestimating extremes due to climate change.
“We are using new machine learning methods developed here in Bath in combination with extensive historical weather data and top-level climate modelling to better inform the risk assessments and guidelines of the future.”
Professor Fatih Tosunoğlu, based in the Department of Civil Engineering at Erzurum Technical University, is also a leader of the project. He said: “Extreme rainfall events are becoming more frequent and intense in Turkey due to climate change. Flooding in the Black Sea region in 2021 led to the deaths of 97 people, as well as hundreds of injuries and large-scale evacuations.
“Having a cutting-edge model that will help us to reduce the risk of extreme weather to people and key infrastructure is crucial to preparing for the future.”
The research will analyse extreme rainfall events across the country to develop new precipitation Intensity-Duration-Frequency (IDF) model. This will provide more accurate information for designing infrastructure under climate change.
Precipitation IDF curves, which generally assume that extremes are stationary and do not significantly change over time, are commonly used to design and operate critical infrastructure. However, as climate change is altering extremes in weather – a concept termed ‘non-stationarity’ – IDF data must be updated too.
Prof Tosunoğlu added: “Our work will assess non-stationary IDF models, determine if they provide improvements over stationary models, and work out how they can be implemented for effective risk management and design.
“I am delighted to help to lead such exciting this research collaborating Erzurum Technical University and University of Bath. I strongly believe that this research will enhance the academic cooperation between the two universities and countries.”
Diverse terrains
Dr Kjeldsen said that in addition to the increasing severity of flooding in Turkey, the country’s diverse terrains – coastal plains, mountainous regions and high plateaus – mean it is an ideal place to create the first nation-wide rainfall modelling system of this kind.
He added: “The project will lead to better understanding of the conditions that create extreme rainfall in Turkey – tracking where moisture arrives from, and how this affects different regions. This work will allow us to categorise flooding events.
“In the UK for example, we have already been able to link global climate drivers with local-scale rainfall or floods: rainfall can be caused by moisture blown over from the continent, or from the North Atlantic, which create different effects in our regions.
“Understanding these drivers for Turkey, and their likelihoods, will help us create guidelines that reflect the new extremes of weather, which need to be understood to properly design buildings, bridges dams and other infrastructure. In future, we hope to repeat this work in other nations as well.”
The project, Rainfall frequency models for critical infrastructure design in a changing environment, will run for two years and will receive £76,000 in funding from the British Council.