Peter Gibson
@SciGibson
Followers
799
Following
8K
Media
34
Statuses
615
NIWA Climate Scientist. Opinions are my own.
Wellington
Joined February 2018
RT @jasonpeterevans: Small Island states need high resolution climate projections in order to understand and plan for climate change @scigi….
0
9
0
Excited to lead this newly funded @MarsdenFund proposal, with help from @neeleshunleashd, working out the how's and whys of atmospheric river changes in a warming world. Despite the devastating impacts of atmospheric rivers, there is still lots we don't know.
They helped make for Auckland’s wettest summer in history - but what will moisture-laden “atmospheric rivers” look like at the end of this century?.
0
1
11
The last 50 years of global warming. Wait till the end to see how ridiculous and widespread September 2023 warmth was. Data NASA GISTEMP. Shown are temperature anomalies relative to 1951-1980 base period. #wxtwitter #climatechange
1
4
7
RT @SciPritchard: Important paper subjecting an AI global weather model to atmospheric dynamics tests. Beautiful work by Greg Hakim at UW a….
arxiv.org
Global deep-learning weather prediction models have recently been shown to produce forecasts that rival those from physics-based models run at operational centers. It is unclear whether these...
0
17
0
One of the benefits of my job running high-resolution climate models: data visualisation of what they can spin up. When we push to high enough resolution we begin to see emergence of cat-5 tropical cyclones like this one😳. Note this doesn't correspond to a real world event
1
4
13
The last 100 years of May temperature anomalies. This is what a developing El Nino at 420ppm looks like. El Nino at 500pm will be something else. Data:
0
1
6
RT @fchollet: I'm told ChatGPT has been upgraded to be able to solve math problems and that is it the future of math tutoring. But my hit r….
0
122
0
RT @niwa_nz: 🚨NEW JOURNAL PAPER ALERT!🚨 . Promising results for NIWA researchers who have been using deep learning to produce high-resoluti….
0
4
0
RT @niwa_nz: Funded by MBIE’s Endeavour fund, NIWA researchers are developing a physics-informed and AI-driven method to vastly reduce the….
0
6
0
Extremely excited and humbled that our Smart Idea proposal (co-lead with @neeleshunleashd) was funded. We will be working on creating a regional climate model emulator from scratch with AI that can better simulate climate extremes 🌩️💦🌨️.
Congratulations to the successful 2022 Endeavour recipients. Read more about successful projects here:
4
1
14
RT @BenNollWeather: New Zealand’s August 2022 atmospheric river (AR) was the strongest August AR in the New Zealand region on record, shown….
0
6
0
There are a lot of things to be worried about with climate change - this by far trumps them all for me.
I missed this when it came out, but this looks quite important to me. Since the classic Sherwood and @climatedynamics paper, 35°C wet-bulb temp has been viewed as the limit of survivability. This new work finds that the limit of human survivability is lower, closer to 31°C.
2
1
3
#climatetwitter Does anyone have a reference estimating how much CMIP6 costs, in terms of model development and runtime etc. I've heard estimates > $1B but can't track down a source.
0
0
3
RT @Weather_West: New work on #AtmosphericRiver families led by Meredith Fish (w/co-authors inc. @Climate_Done, @Weather_West, @AnnaWilsonW….
0
20
0
0
0
2
As more climate models are performing these large initial condition experiments, and making these available, this provides a great opportunity for training ML models on for making seasonal forecasts. 4/4.
1
0
3
It turns out this approach can give you competitive seasonal prediction results, relevant for water resources and drought forecasts. This is a good way to leverage substantial existing investments in climate modelling for seasonal forecasting at no extra cost. 3/4.
1
0
2
Training ML models on observations isn’t really feasible, because you only have 1 data point per season, so a very short record for training. So training on climate models instead has a big advantage 2/4.
1
0
4