Poster: Connecting daily weather to monthly SST

Connecting daily weather to monthly SST

(Originally Model Analogues: a statistical forecasting technique to explore the climate system)

About

Sea surface temperature patterns (SST) such as El Niño and La Niña have been linked to rainfall on seasonal or monthly timescales. I am unaware of studies directly linking SST patterns to particular synoptic events. However, Australian seasonal rainfall is determined by a few very rainy days (or their absence), which implies the existence of a link between global SST patterns and particular extreme weather events. This poster shows preliminary results from a method to explore such a link, and is being shared with the ACCESS community to gain ideas and feedback, and seek out any individuals interested in collaborating on this project.

We start with 27th Feb 2022 as a case study of a particularly rainy day, contributing to extensive flooding on the northern NSW coast. We compare the regional mean sea level pressure pattern of this event to daily ACCESS-ESM1-5 large ensemble output. The 30 days with most similar MSLP patterns are taken to be “analogues” of the 27th Feb 2022 rain event, and we then average the SST pattern for each of these to produce the average SST to co-occur with this particular event.

In future, we would like to explore SST precursors, rather than just contemporary SST, to understand what might predict these synoptic events. We’re also interested in choosing events to study where understanding SST precursors would be of broader value to the community

Poster

240830_weather_analogues_NRI.pdf (10.4 MB)

Note: this topic is part of the 2024 ACCESS Community Workshop Poster session

2 Likes

Interesting work Jemma! And nice poster - very clear and engaging.
Are you planning to look at this in (sub)seasonal forecasts? E.g Are extreme events that match the analogues better forecast?

Thanks Kim,
Can you clarify what you mean? Ie, looking for analogues of S2S forecasts in ACCESS-ESM1-5 output, or of something else in S2S forecast data? Better forecast than what?

Any seasonal or subseasonal forecast would heaps better than an analogue forecast, because the initial conditions are more accurate, but I don’t think that’s what you’re asking

Hi Jemma,
Yeah that wasn’t a well thought out question :sweat_smile:
I guess S2S forecasting seems like the obvious application of your results, so are you intending to assess if the cases of e.g. extreme rainfall over Eastern Australia that match the analogues are better forecast than cases with a different MSLP/SST set up?
And therefore, should we have more confidence in a forecast of extreme rainfall that has a specific MSLP/SST than one that has a different synoptic set up.

I guess you could apply the technique to investigate what makes a forecast more reliable, and see if particular SST patterns or make a S2S forecast more predictable. There’s always the issue that your analogues won’t quite match and the variation will produce noise. I wasn’t planning on directly going down the forecast evaluation path, more blue-sky looking for precursors, but would be happy to chat if you wanted to