I've seen the Matrix. We need clean reliable energy sources to empower our AI overlords and solve the tricky questions regarding climate change, therefore I humbly propose plugging climate activists into the human battery pod to power this important research. It's for Gia innit.
My question is why didn't they use the ER5A reanalysis data that goes from 1950 to 2023? I am not as suspicious of the data as you but I am naively assuming that they verify the data before it is used in this type of application.
Probably because data from the pre-satellite era is not considered to be accurate enough.
"The original ERA5 release contained data from 1979 onwards. The final ERA5 back extension for 1940-1978 has been produced and is available alongside the original/main release.
An ERA5 back extension 1950-1978 (Preliminary version) was produced. Although in many other respects the quality was relatively good, this preliminary data did suffer from excessively intense tropical cyclones. This dataset is now deprecated."
I don’t see how anyone could trust an AI analysis of climate change. Would it use raw data rather than dishonestly adjusted data? Would it use UHI-corrupted data? Would it take into account all climate variables (which the UN IPCC doesn’t): solar variations, planetary orbital and gravitational variations, solar/ocean-driven ENSO, PDO and AMO cycles and much more including the data which shows that an increase in atmospheric CO2 follows the rise in global temperature rather than coming before it and causing it?
In other words, would it be programmed by the same corrupt deep state professional liars who are pushing the fake climate change so-called emergency?
It uses ERA5 reanalysis data, which is actually output of a simple model, not the raw data itself, although it's a close approximation to the actual raw data and is often used by scientists as such, in order to compare actual 'observations' with the output of the global circulation models, when performing extreme weather attribution analyses. It's all a bit suspect quite frankly. Even the raw data can be corrupted by station siting, as you point out. But, that said, if AI forecasts are proving to be more accurate than the numerical prediction weather models, then they must be getting something right.
"GenCast is trained on 40 years of best-estimate analysis from 1979 to 2018, taken from the publicly available ERA5 (fifth generation ECMWF reanalysis) reanalysis dataset27. Reanalysis provides a reconstruction of past weather by computing analysis for historical dates and times."
Computer games - er sorry, models based on physical processes such as radiative physics, thermodynamics and Navier-Stokes equations are not capable of predicting future climate.
“In sum, a strategy must recognise what is possible. In climate research and modelling, we should recognise that we are dealing with a coupled non-linear chaotic system, and therefore that the long-term prediction of future climate states is not possible.”
So said IPCC Working Group I: The Scientific Basis, Third Assessment Report (TAR), Chapter 14 (final para., 14.2.2.2), p774.
Anyone who claims that a purported computer game - oops, climate simulation of an effectively infinitely large open-ended non-linear feedback-driven (where we don’t know all the feedbacks, and even the ones we do know, we are unsure of the signs of some critical ones) chaotic system – hence subject to inter alia extreme sensitivity to initial conditions, strange attractors and bifurcation – is capable of making meaningful predictions over any significant time period is either a charlatan or a computer salesman.
Ironically, the first person to point this out was Edward Lorenz – a climate scientist.
Lorenz’s early insights marked the beginning of a new field of study that impacted not just the field of mathematics but virtually every branch of science – biological, physical and social.
In meteorology, it led to the conclusion that it may be fundamentally impossible to predict weather beyond two or three weeks with a reasonable degree of accuracy.
Some scientists have since asserted that the 20th century will be remembered for three scientific revolutions–relativity, quantum mechanics and chaos.
The weather model vs. climate model is the initial value vs. the boundary value problem. Numerical weather models start with estimates of the initial state of the system based upon observations and then they forecast the state of the system at some future point. Because of the inherent chaotic instability of the system, they are only really accurate for a maximum of a few days ahead. Climate models are very different. They are not programmed with an initial state but are constrained by boundary conditions, which are the climate forcings - solar, volcanic, greenhouse gases, internal variability, etc. plus feedbacks. In that respect, the way AI is predicting weather, based upon past averages, seems more like a boundary value problem.
If the past proves to be a good guide to the future, then it might suggest (I put it no higher than that) that climate change isn't so extreme as is suggested. It's certainly interesting that AI analysis of past weather produces more accurate weather forecasts than fancy computer models
I've seen the Matrix. We need clean reliable energy sources to empower our AI overlords and solve the tricky questions regarding climate change, therefore I humbly propose plugging climate activists into the human battery pod to power this important research. It's for Gia innit.
This method assumes the conclusion in its premises, namely that global weather patters are short in the scope of time, rather than spanning millennia.
My question is why didn't they use the ER5A reanalysis data that goes from 1950 to 2023? I am not as suspicious of the data as you but I am naively assuming that they verify the data before it is used in this type of application.
Probably because data from the pre-satellite era is not considered to be accurate enough.
"The original ERA5 release contained data from 1979 onwards. The final ERA5 back extension for 1940-1978 has been produced and is available alongside the original/main release.
An ERA5 back extension 1950-1978 (Preliminary version) was produced. Although in many other respects the quality was relatively good, this preliminary data did suffer from excessively intense tropical cyclones. This dataset is now deprecated."
https://confluence.ecmwf.int/display/CKB/ERA5%3A+data+documentation
I don’t see how anyone could trust an AI analysis of climate change. Would it use raw data rather than dishonestly adjusted data? Would it use UHI-corrupted data? Would it take into account all climate variables (which the UN IPCC doesn’t): solar variations, planetary orbital and gravitational variations, solar/ocean-driven ENSO, PDO and AMO cycles and much more including the data which shows that an increase in atmospheric CO2 follows the rise in global temperature rather than coming before it and causing it?
In other words, would it be programmed by the same corrupt deep state professional liars who are pushing the fake climate change so-called emergency?
It uses ERA5 reanalysis data, which is actually output of a simple model, not the raw data itself, although it's a close approximation to the actual raw data and is often used by scientists as such, in order to compare actual 'observations' with the output of the global circulation models, when performing extreme weather attribution analyses. It's all a bit suspect quite frankly. Even the raw data can be corrupted by station siting, as you point out. But, that said, if AI forecasts are proving to be more accurate than the numerical prediction weather models, then they must be getting something right.
"GenCast is trained on 40 years of best-estimate analysis from 1979 to 2018, taken from the publicly available ERA5 (fifth generation ECMWF reanalysis) reanalysis dataset27. Reanalysis provides a reconstruction of past weather by computing analysis for historical dates and times."
Computer games - er sorry, models based on physical processes such as radiative physics, thermodynamics and Navier-Stokes equations are not capable of predicting future climate.
“In sum, a strategy must recognise what is possible. In climate research and modelling, we should recognise that we are dealing with a coupled non-linear chaotic system, and therefore that the long-term prediction of future climate states is not possible.”
So said IPCC Working Group I: The Scientific Basis, Third Assessment Report (TAR), Chapter 14 (final para., 14.2.2.2), p774.
Anyone who claims that a purported computer game - oops, climate simulation of an effectively infinitely large open-ended non-linear feedback-driven (where we don’t know all the feedbacks, and even the ones we do know, we are unsure of the signs of some critical ones) chaotic system – hence subject to inter alia extreme sensitivity to initial conditions, strange attractors and bifurcation – is capable of making meaningful predictions over any significant time period is either a charlatan or a computer salesman.
Ironically, the first person to point this out was Edward Lorenz – a climate scientist.
Lorenz’s early insights marked the beginning of a new field of study that impacted not just the field of mathematics but virtually every branch of science – biological, physical and social.
In meteorology, it led to the conclusion that it may be fundamentally impossible to predict weather beyond two or three weeks with a reasonable degree of accuracy.
Some scientists have since asserted that the 20th century will be remembered for three scientific revolutions–relativity, quantum mechanics and chaos.
http://news.mit.edu/2008/obit-lorenz-0416
You can add as much computing power as you like, the result is purely to produce the wrong answer faster.
But for some climate “scientists” I suppose it pays the mortgage…
The weather model vs. climate model is the initial value vs. the boundary value problem. Numerical weather models start with estimates of the initial state of the system based upon observations and then they forecast the state of the system at some future point. Because of the inherent chaotic instability of the system, they are only really accurate for a maximum of a few days ahead. Climate models are very different. They are not programmed with an initial state but are constrained by boundary conditions, which are the climate forcings - solar, volcanic, greenhouse gases, internal variability, etc. plus feedbacks. In that respect, the way AI is predicting weather, based upon past averages, seems more like a boundary value problem.
Climate and weather are indistinguishable, purely a matter of scale.
If the past proves to be a good guide to the future, then it might suggest (I put it no higher than that) that climate change isn't so extreme as is suggested. It's certainly interesting that AI analysis of past weather produces more accurate weather forecasts than fancy computer models