Algorithm reconstructs wind fields quickly, accurately, and with less observational data.
Via Eurekalert, maybe this will work, maybe not. Worth a try.
WASHINGTON, Nov. 19, 2024 – Hurricanes, or tropical cyclones, can be devastating natural disasters, leveling entire cities and claiming hundreds or thousands of lives. A key aspect of their destructive potential is their unpredictability. Hurricanes are complex weather phenomena, and how strong one will be or where it will make landfall is difficult to estimate.
In a paper published this week in Physics of Fluids, by AIP Publishing, a pair of researchers from the City University of Hong Kong employed machine learning to more accurately model the boundary layer wind field of tropical cyclones.
In atmospheric science, the boundary layer of the atmosphere is the region closest to the Earth’s surface.
“We human beings are living in this boundary layer, so understanding and accurately modeling it is essential for storm forecasting and hazard preparedness,” said author Qiusheng Li.
However, because air in the boundary layer interacts with land, the ocean, and everything else at surface level, modeling it is especially challenging. Conventional approaches to storm forecasting involve large numerical simulations run on supercomputers incorporating mountains of observational data, and they still often result in inaccurate or incomplete predictions.
In contrast, the author’s machine learning algorithm is equipped with atmospheric physics equations that can produce more accurate results faster and with less data.
“Unlike traditional numerical models, our model employs an advanced physics-informed machine learning framework,” said author Feng Hu. “Only a small amount of real data is required by our model to capture the complex behavior of the wind field of tropical cyclones. The model’s flexibility and ability to integrate sparse observational data result in more accurate and realistic reconstructions.”
Being able to reconstruct a tropical cyclone’s wind field provides valuable data that experts can use to determine how severe the storm will be.
Feng Hu and Qiusheng Li
“The wind field of a tropical cyclone contains the information of the storm’s intensity, structure, and potential impact on coastal regions,” said Li.
With a more detailed picture of what that wind field looks like, disaster authorities can better prepare for storms before they make landfall.
“With more frequent and intense hurricanes due to climate change, our model could significantly improve the accuracy of wind field predictions,” said Hu. “This advancement can help refine weather forecasts and risk assessments, providing timely warnings and enhancing the resilience of coastal communities and infrastructure. “
The authors are planning to continue to develop their model and employ it to study different types of storms.
“We are planning to incorporate more observational data sources and improve the model’s capability to handle the time evolution of winds,” said Hu. “Expanding the application to more storm events across the world and integrating the model into real-time forecasting systems is also planned to enhance its utility for weather prediction and risk management.”
###
The article “Reconstruction of tropical cyclone boundary layer wind field using physics-informed machine learning” is authored by Feng Hu and Qiusheng Li. It will appear in Physics of Fluids on Nov. 19, 2024 (DOI: 10.1063/5.0234728). After that date, it can be accessed at https://doi.org/10.1063/5.0234728.
I am always wary of the modelling communidee..
Reconstruction of tropical cyclone boundary layer wind field using physics-informed machine learning – “With less observational data.”
It sounds almost like a euphemism for [AI] extrapolation. Extrapolation used to be a dirty word, well, in chemistry at least. If this methodology really is that good, that accurate while using sparse data, filling in the gaps so to speak, I wouldn’t expect them to need any more inputs, but…
“”“We are planning to incorporate more observational data sources and improve the model’s capability”””
Reconstruction of tropical cyclone boundary layer wind field using physics-informed machine learning – “Now with added extra observational data.”
Trump this:
Tees Valley’s Tory Mayor Ben Houchen has fired off a letter to Donald Trump, applauding his stunning “political comeback” and extending an open invitation to visit the North East. Meanwhile, Starmer’s not yet mustered up a formal invite for the President-elect to visit the UK…
Houchen took on board Farage’s advice: roll out the red carpet for Trump. Houchen wrote:
Read the full letter below:
https://order-order.com/2024/11/25/tory-mayor-beats-starmer-to-the-punch-in-inviting-trump-to-the-uk/
It would be hilarious if Trump accepted the offer. Though we know there probably won’t be enough hours in a day, let alone time for frivolous meetings.
Did the ICC put out an arrest warrant on Trump? Maybe this is a ruse to throw Trump into jail.
Nah. But you do have a vivid imagination.
“Vivid imagination” as in no one’s tried to throw Trump into jail?
Not in the UK, no. That’s on your lot. So far, 2,046,321 have signed an effective petition of recall for the governing party And it is very much like the petrol pump going round. In fact since I’ve typed this line its gone up to 2,047,749
And now it’s 2,050,862
https://petition.parliament.uk/petitions/700143
…2,053,446 and counting
Actually, it’s on you. The UK is an ICC treaty signatory.
Apparently the UK is responsible for African slavery, despite a century of wars and ocean patrols to put an end to it.
You’d better get in the queue, mate.
I’m not sure which queue you want me in. The one to the water board or the one to the guillotine?
The queue for reparations.
You do have a vivid imagination. Don’t let it run riot – too much.
“The queue for reparations.”
Huh?
Occam’s razor needs sharpening.
Catholic Church authorized Portugal to start the slave trade.
Maybe he could meet up with Ellen for tea and crumpets (poisoned).
Yes, the dross is landing….
““With more frequent and intense hurricanes due to climate change, our model could significantly improve the accuracy of wind field predictions,” said Hu.”
If the model can improve the wind field predictions for hurricanes, why make a reference to the questionable “more frequent and intense” claim?
Also, this sounds like a good start to a “Hu’s on first” style meme. 🙂
“Hu said there are more frequent and intense hurricanes due to climate change.”
“I don’t know. Who said that?”
“Yes he did.”
“WHO did?”
“Absolutely. It’s in the paper.”
“Who wrote the paper?”
“Hu did.”
“I’m asking YOU!! Who wrote the paper?”
“Correct.”
Well, you might be able to get funding f yr project.
Apparently “physics-informed” (real-world-informed) only applies to the neural network model in the computer, not the one in their brains.
____________________________________________________________________
Sounds like a TV ad for tooth paste.
In the real world:
IPCC TAR Chapter 14 Page 771 pdf3
The climate system is a coupled non-linear chaotic system,
and therefore the long-term prediction of future climate states
is not possible
Did I just use the IPCC as an example of the real world?
They do get some things right and then they use it if it fits
the narrative or ignore it if it doesn’t.
I do like the “with less data” statement. When did they have “more data?”
“10 out of 9 dentists prefer…”
I see a model
I see the weasel word “could”
I do not see any positive results from using the model versus an older methodology
The false study claim: “With more frequent and intense hurricanes due to climate change … “ does not inspire confidence.
This is just model cheerleading by the study authors. Cheerleading is claptrap
This article is premature and does not pass my internal BS test (I have a BS degree, ha ha)
“does not inspire confidence”
_________________________
Pretty much tells you all you need to know.
This could be an April Fool’s article with the statement … “can produce more accurate results faster and with less data.” Less data = less intelligence = less accuracy.
Tastes great! Less filling!
The global average temperature statistic tells us a lot. It tells us the average temperature has barely changed since 1880.
Try it in Kelvin!!
Try a comparison to 1820 based on published reports in that time.
Hi Anthony,
OT, but I thought you may want to update this Q and A in the WUWT About > FAQ page:
Q. How much traffic does WUWT get?
A. In a typical month, about 3 to 4 million page views, about 25-30% if which are unique visitors. As of this writing WUWT is closing in on 150 million views. The current views counter is near the top of the right sidebar.
I just checked and you are up over 540 million views. Well done!
Models used to understand and learn are great.
Models used to predict or project are not.
I disagree. The true power of a model is in its predictive ability.
A model, like set of Maxwell equations, can be used to predict behavior of electromagnetic waves, predicting many things we use, including radio communications. Mendeleev’s Periodic table left blanks cells for yet to be discovered elements. Einstein’ s model of Relativity predicted black holes, gravitational lensing and waves. Based on a model of Quantum Mechanics Einstein predicted possibility of building a laser (quantum light amplifier). Model of Quantum Mechanics predicted quantum entanglement, once controversial, but now used in quantum computing. Drug design relies heavily on modeling.
Pretty much all Physics and Chemistry are models with predictive powers used by engineers to design everything we use.
Grossly inaccurate.
Maxwell’s equations are not a model. They are the equations the theory of electromagnetic radiation derive from. Those equations has been subjected to countless tests and proven reliable and accurate. It has reached the point where Maxwell’s equations could be declared a scientific law.
Einstein did not formulate a model of Relativity. He formulated a theory with concise equations. Much of the theory has been proven and confirmed. No model predicted black holes or any thing else. Those were derived from the equations, not models, and progress is being made verifying through testing of theories for the whole theory of Relativity.
Quantum mechanics is also a theory. Deriving quantum entanglement was not done by models.
I do not know what the drug industry uses. My son got his PhD with a thesis in refining calcium transport simulations to improve computational efficiencies.
Having work in both modelling and simulations for over half a century, I am aware of the causal conflating of the terms.
Finite element analysis is modelling. It has fundamental assumptions such as purity of the material, lack of defect, and primary is grid resolution. Grid resolution assumptions have been tested and a generally accept minimum has been determine via testing. The goal was to optimize computational time versus accuracy and it was determined beyond the established recommended minimum, any gain in analysis accuracy was not worth the cost.
Using a computer to perform analysis is often referred to modelling. I get that. But there really is a nuanced difference between models and simulations. Models have built in assumptions. When I build models, I test the models for sensitivity to the assumptions. When I simulate a circuit, there are no assumptions. All elements are exact and tolerances tested. The equations are exact. The output defines a range of expected performance to allow determination if the circuit does what is intended.
Even with all the computational analysis and simulations, everything is tested. If a model is predictive as you claim, why bother testing?
Pretty much physics and chemistry are established, proven equations, not models.
So, is V = IR a model?
Yes, V=IR is just a model. It predicts that increasing current or resistance increases voltage linearly. The equations are exact, but the prediction can be tested only within experimental uncertainty.
This model is by no means absolute and has limits, just like Newtonian mechanics. Does it always hold? No, it is not valid in superconductors. Does it work inside a black hole? We do not know. So yes, when you simulate a circuit, there are assumptions.
This goes back to what Science i.e. Scientific Method is.
There is no distinction between “theory” and “model”. A theory at an early stage is usually called a hypothesis. Upon sufficient amount of testing it can be elevated to status of a theory. Nevertheless, the gist of any theory is a non-contradictory model tying observations together. Ideally, it is a quantitative model where observations can be related by mathematical expressions. Those expressions provide predictions that could be tested under controlled conditions within experimental uncertainty.
In case of more than one non-contradictory models, the ones requiring more assumptions are ruled out in favor of the ones with fewer assumptions (Occam’s razor).
You really do not understand what a model is.
Just like weather forecasting models, its value should be relatively quickly obvious.
Can we really model hurricanes with only the boundary layer?