October 3, 2023
Professor Jason Evans from the University of New South Wales Climate Change Research Centre, the Climate Systems Hub and the ARC Centre of Excellence for Climate Extremes spoke at Climate Adaption 2023 about the next generation of rainfall models that will help warn communities of flash flooding and in the long term, help develop stronger rainfall and runoff guidelines for more resilient infrastructure.
Thunderstorms, fronts and east coast lows are all examples of regional rainfall extremes.
Current climate models are fairly reliable when it comes to predicting daily or monthly rainfall projections, but are less reliable for shorter-term rainfall extremes – that is to say, hourly or sub-hourly.
Shorter duration rainfall extremes are important to predict because they lead to flash flooding in urban areas, and can cause gutters to overflow or roofs to leak. This can pose real risks to safety. Good projections of these extremes are needed so we can design our infrastructure and structures to deal with these extremes safely and minimise this risk.
Longer duration rainfall extremes are associated with quite big weather systems, such as east coast lows; our current regional models can replicate those systems to a reasonable extent. Shorter duration events are often produced in spatially smaller storms; storms that are not resolved by current models due to their resolution.
Our next generation of models are going to be at kilometre-scale resolutions which is in the right scale for these smaller systems. That’s where we’re going with the next generation models.
Convection permitting models
Storms are caused by a parcel of air near the ground which is hotter and moister than its surroundings and rises. This is convection, as this parcel of air mass rises in the storm it produces updrafts and generates rainfall. This research is aiming for is a model that can properly simulate this process of convection and updrafts within storms.
Current regional climate projections models are in the order of 20 km resolution, with some going down to 4km resolution, just on the edge of convection permitting.
The next generation of models that predicts convection permitting rainfall is when our models will have much better projection capability for these short duration extremes.
Our observations of these extremes are better than they have ever been, which helps this development. The Bureau of Meteorology has a network of stations measuring rainfall at minute intervals. On top of that, we have got a good weather radar network around coastal areas that has captured the right time and space scales, kilometres resolution and 5-10 minutes in time. Most stations show short duration extremes are increasing but stations miss 90+% of sub-hourly extremes. Using weather radars, we find 10-minute extremes increasing by at least 20% per decade in the Sydney region.
A challenge Australia does face is that with our sparse population, our observational station network is sparser than other developing countries and other radar networks. We have good coverage in densely populated areas but less coverage inland.
Australian Rainfall and Runoff Guidelines
Urban areas are where we often see the most dramatic impact from extreme rainfall events.
The Australian Rainfall and Runoff Guidelines provide the best information and standards for the development of robust structures and infrastructure that can withstand extreme rainfall.
These guidelines are currently based on the historical record of those extremes.
However, we know the climate has changed and continues to change, and that is changing the rainfall extremes.
When the last version of the guidelines was developed 10 years ago, climate change was acknowledged but at the time the scientific evidence wasn’t strong enough to convince engineers that the guidelines needed to change. Any response to climate change was considered voluntary.
There is now significantly more scientific evidence in relation to climate change and extreme events, including observed and projected changes in rainfall extremes, and the guidelines are being updated to include climate change as a factor in the way we design structures.
Recent experience of flooding and damage caused by extreme rainfall events showed that structures that were designed for the historical one in 100-year event may not be adequate for present and future extremes.
In place of the historical one in 100-year event is now a one in 50-year event and may become a one in 20-year event. Structures need to be designed for the climate they will experience over the next 100 years, and designing structures to handle historical extremes does not achieve this.
We need to address this in relation to infrastructure in two ways:
- We have to retrofit existing infrastructure to make it more robust. This is often a more costly approach.
- Any new structures need to take account of the increase in extremes due to climate change. These structures may be more costly to build but will perform better and reduce costs from damage when extreme rainfall events occur.
The good news is this is no longer about whether the guidelines should be changed, but how.
Where to next
The space and time scales that our models can now simulate includes the atmospheric phenomena that contributes to rainfall extremes across a wide range of durations. These have the potential to substantially improve our confidence in projections of rainfall extremes at daily and sub-daily durations.
One of the limitations in applying them, though, is that these models take a long time to run and are computationally expensive. Given our resources, we are limited with what we can do with these models.
One possible answer to that is machine learning which allows us to develop a climate model simulator that is much faster and cheaper to run.
AI or machine learning methods along with convection permitting models (CPM) provide an avenue to extend our knowledge and understanding.
These new avenues will yield new insights into future changes in short duration rainfall extremes in the next five years.
______________
Further reading:
Hooman Ayat, Jason P. Evans, Steven C. Sherwood, Joshua Soderholm (2022) Intensification of subhourly heavy rainfall. Science. 378 (6620), 655-659 DOI: 10.1126/science.abn865
Australian Rainfall and Runoff website here
Professor Jason Evans’s presentation at Climate Adaptation 2023 is available on the conference page on our website here.