AI-Accelerated Climate Modeling | Tapio Schneider at Caltech | AI FOR GOOD DISCOVERY
0While climate change is certain, precisely how climate will change is less clear. Breakthroughs in the accuracy of climate projections and in the quantification of their uncertainties are now within reach, thanks to advances in the computational and data sciences and in the availability of Earth observations from space and from the ground. Schneider will survey the design of a new Earth system model that the Climate Modeling Alliance (CliMA) is developing. The model exploits tools from machine learning and data assimilation jointly with process-informed models, with the goal of achieving a new level of accuracy in modeling important small-scale processes such as clouds and precipitation.
⏱ Shownotes:
00:00 Opening remarks by ITU
02:29 AI-ACCELERATED CLIMATE MODELING
03:49 Global damages from climate-related disasters
04:45 Mitigating climate change is essential
05:53 Challenges
06:39 2ºC warming threshold
08:24 Primary source of uncertainty
10:23 Clouds in climate predictions
10:39 Small-scale cloud-controlling processes
11:42 Difficulties in modeling low clouds
13:01 Newton’s and Thermodynamics laws
13:13 Computer performance
13:59 Global low-cloud resolving models
16:08 Data and AI to the rescue
16:52 Requirements for data-informed climate models
18:14 Reductionist science with data science to accelerate climate modeling
20:25 Three crucial ingredients
23:05 Abstract model
24:15 How does it actually work in modeling clouds
24:43 How to model clouds?
27:34 Closure functions in the coarse-grained equations
28:46 Low cloud cover bias in a typical current climate model
29:54 Reduced-order model with 9 parameters
33:09 Scaling this success up to an entire earth system
34:09 CLiMA
35:02 Learning from statistics accumulated in time
35:35 Learning from climate statistics
37:23 Our setting for learning about parameters
38:44 Calibration and Bayesian approaches in a three step process
42:23 Proof-of-concept in idealized general circulation model (GCM)
43:10 Calibrate with ensemble Kalman inversion
44:04 Emulate parameters-to-statistics map during calibration step with Gaussian processes
45:28 Uncertainty quantification Bayesian learning at 1/1000th the cost of standard methods.
46:24 Climate predictions from the posterior of parameters for UQ of predictions
48:36 An all-new Earth system model
49:19 CliMA: fine-grained climate projection on demand
51:04 Conclusions
55:28 Q&A session
55:39 How confident are you with learning parameters as the climate shifts?
58:52 Are there ways for startups and the private sector to collaborate on this topic?
1:00:35 Merging physics and ML is more meaningful for short-term/long-term prediction?
1:01:25 Could someone learn more about the model structure itself?
1:03:02 Closing remarks from Philip Stier
1:03:24 Closing remarks by ITU
🔴 Watch the latest #AIforGood videos:
Explore more #AIforGood content:
1️ ⃣ • Top Hits
2️ ⃣ • AI for Good Webinars
3️ ⃣ • AI for Good Keynotes
📅 Discover what's next on our programhttps://aiforgood.itu.int/programme/mme/
Social Media:
Websihttps://aiforgood.itu.int/int/
Twitthttps://twitter.com/ITU_AIForGoodod
LinkedIn Pahttps://www.linkedin.com/company/2651190707
LinkedIn Grohttps://www.linkedin.com/groups/856774848
Instagrhttps://www.instagram.com/aiforgoodod
Facebohttps://www.facebook.com/AIforGoodod
What is AI for Good?
The AI for Good series is the leading action-oriented, global & inclusive United Nations platform on AI. The Summit is organized all year, always online, in Geneva by the ITU with XPRIZE Foundation in partnership with over 35 sister United Nations agencies, Switzerland and ACM. The goal is to identify practical applications of AI and scale those solutions for global impact.
Disclaimer:
The views and opinions expressed are those of the panelists and do not reflect the official policy of the ITU.
#AIforGoodDiscovery #AIforClimateScience