Badenoch Strikes a Blow

On Thursday, Mark linked to an opinion piece by the Guardian’s Zoe Williams about Kemi Badenoch’s “Net Zero by 2050 is impossible” speech. Mark didn’t like Williams’ piece, and when I read it, I didn’t like it, either. I’ve been thinking about it off and on since, so you could at least call it thought-provoking – although not in a good way. In what follows I will note and pass comment on some of Williams’s more outré remarks.

The title of the piece, filed under the category of “climate crisis,” is:

Badenoch’s attack on net zero is ridiculous. But so were the right’s Brexit claims, and look where they left us

Williams is going to draw a line between Brexit and opposition to Net Zero.

The subtitle:

The run-up to 2016 shows ‘common sense’ isn’t enough. Even ignorant, reactionary arguments must be properly countered

The diatribe that follows consists of 9 paragraphs.

Paragraph 1: Williams snarks about Badenoch’s lack of expertise in climate science. This invalidates Badenoch’s assessment about the viability of Net Zero. But it only takes a moment’s thought to realise that the same objection can be served up to every politician who advocates for or against Net Zero. You may retort that the same could be said about any topic, and I might agree. How many of today’s Cabinet are domain experts?

While I’m on this point, I will make the observation that climate scientists should be judge-like, disinterested. Too often they stray from reporting the facts as they believe them, to demanding preferred policies (like Net Zero). Science is an attempt to understand the Universe, and the Universe does not have morals, only Laws. There is nothing demanding Net Zero, even if there does happen to be a “Climate Crisis” (there doesn’t). It is for politicians to decide what to do with the dispassionate information that the disinterested scientists have provided to them.

Paragraph 2: Williams notes that the attack on Net Zero has been foretold. Well done, Nostradamus! No-one could have seen that coming! Strange that the sceptics’ line has long been: “As soon as the pips start squeaking, the opposition to Net Zero will begin to grow.” Maybe in Williams’s ivory tower, the pips aren’t squeaking yet. I suppose she does not worry about paying her next energy bill.

Paragraph 3: Williams cites some guy from Led by Donkeys, and his description of the evolution of climate denial, which originates elsewhere. (Maybe from Oreskes, please enlighten me if you know.) There are unpleasant overtones in the description, which goes: first they denied the science. Then they minimised the seriousness of the consequences. Then they said we couldn’t afford to do anything about it. Those are the first three ditches of climate denialism, and there is a fourth, and last, which we’ll come to in due course.

In Paragraph 4, Williams uses this extraordinary expression (as noted by potentilla here):

“Because even while outlets such as GB News have been preaching climate impossibilism for some time, it has until now been broadly disallowable in mainstream political discourse.“

Well of course the Guardian doesn’t like GB News. But can an argument that Net Zero is impossible really be disallowable? What if Net Zero really is impossible? We had better hope that it isn’t, for as a civilisation, we need to move onto other energy sources sooner or later, since hydrocarbons will eventually run out. (Note: these new energy sources will not be weather dependent, not if we want our descendants to live, not merely exist.) But can an opinion about any policy which involves the imposition of frankly draconian measures on the populace be disallowable?

Paragraph 5: Williams thinks that opposition to ULEZ, LTNs and 15-minute cities has something to do with Net Zero. Maybe people don’t want ULEZ because they can’t afford a ULEZ-compliant car? Maybe they don’t want to be locked in their neighbourhood by bus gates? Such policies are “pretty anodyne.” Yes, for you they are, but not for some of us.

Paragraph 6: Opposition to ULEZ etc “takes on the heft of an imagined constituency, people who are fed up with environmentalists.” Do not conflate advocates of Net Zero with environmentalists. True environmentalists oppose the destruction wrought in the name of Net Zero, the forests of wind turbines, the hungry maw of Mr. Drax. But Williams is unaware that there are any negatives at all about the pursuit of Net Zero. That is the only conclusion I can draw from her dismissal of opposition to it.

Paragraph 7: Sane people agree with Williams. The public are still in favour of Net Zero: a facile point, since Net Zero is more than a principle. This is where pollsters tilt the answer in the direction they want. You don’t get Net Zero for nothing. You have to exchange something for it. And it looks as if the payment will be so high that only a fool would agree to the deal. Or someone so insulated from the real world that acceding would cost them naught personally.

Then, disgracefully, we have this: “Getting into the weeds of Badenoch’s own character, a debate is playing out that is also deeply familiar – is she saying this because she’s enchanted by dark money, or is it because she’s an “irresponsible, ignorant, reactionary fool”, as one journalist put it.” So no-one sane can believe these things; anyone who claims they do, is either doing it for money, or because they are an “irresponsible, ignorant, reactionary fool.” Badenoch may be pleased to note that she has struck a nerve. There seems to be no answer other than character assassination.

Paragraph 8: Williams draws an entirely specious parallel between Brexit, i.e. “what happened last time,” and the current slow wave building against the absurdity of Net Zero. No conspiracies are needed to explain either occurrence: there are valid reasons to support both. What rational country would surrender itself to the laws of an undemocratic supranational state, outside of defeat in war? And what rational country would agree to immiserate its population for a policy that will have no measurable effect on climate change?

Then Williams has the gall, or maybe blissful unawareness, to start wondering about where the money was (in Brexit) coming from and by implication where it is coming from now to gin up opposition to Net Zero. Try looking at the funding for the very many groups advocating for this policy, funding that is, in effect, aimed at crushing what is left of a once Great country into a thin paste. (Mark’s recent Avarice in Funderland might be a good place to start.)

In Paragraph 9, mercifully the last, the Donkeys guy talks about the “last trench.” This is the stage at which we sceptics admit that the science was always right, the consequences were always apocalyptic, and that we really could afford the cost of Net Zero, but that, thanks to all the spanners we managed to toss into the works, “it’s too late.”

By that time, one presumes, rats the size of cats will be taking shifts to roam the streets with packs of rabid dogs, our roofs will have been blown off by a tremendous gale, and Williams and the Donkeys guy will be fighting tooth and nail for a lick of a wind-blown piece of paper that three months ago was wrapped around a Big Mac. Then, from stage left, er, right, out will pop a climate sceptic, and he will say, “I admit it. I was lying all the time, but it’s too late now.”

And the righteous will turn on him with knives in their eyes.

Then, thanks to the collapse of civilisation, the emissions of CO2 will drop to trivial levels, and it will turn out that it wasn’t too late after all, because in due course, everything will return to normal. Except for the rats the size of cats. Our sleek brown friends are staying in my fantasy dystopian future.

Oh, a final note. As far as I heard, Badenoch did not say anything controversial in what might be termed “the real world.” In fact, on my reading, she did not repudiate Net Zero, only its timing, and the lack of a Plan for it. That her speech brought out such agitation in certain quarters only shows how weak their position is. They can’t argue the facts, so have to rely on a code of consensus from which no-one respectable must deviate. But respectable people must try to grab the wheel, for the present course is destruction.

PS. If a pollster asked me if I was in favour of Net Zero, I would not know how to answer. I would have no reason to oppose it in principle. I have reasons to oppose it in practice. A fantasy Net Zero in which there are no negative environmental consequences, no negative consequences for the UK’s wealth, and no negative consequences for the UK’s security, I could not oppose. But only a fool would believe that version of Net Zero exists. In the real world, the people of the UK are better off killing Net Zero rather than themselves.

via Climate Scepticism

https://ift.tt/xLYTl4g

March 23, 2025 at 12:53PM

The Uranium – Hydrogen – Carbon Energy System of the Future

David Archibald

Following is a lecture given on 13th March at the Energy Exchange Australia convocation in Perth. It was a 20 minute slot.

This graph shows the last 26 years of the National Electricity Market which is the power grid on the east coast of Australia. The vertical blue bars are generating capacity. The orange line is demand. Demand used to closely track capacity up to 15 years ago. Now capacity is 80% higher than demand. The increased supply should have resulted in lower prices due to oversupply. Prices should have gone down, surely?

Power prices didn’t go down though, they doubled instead. The blue line is Australia’s Consumer Price Index from 1980. The red line is the Power Prices Index. They used to track closely up to 2008 where they parted company. We are now paying twice as much for power as we should be. And this is for stuff that is going straight to landfill. A power price twice as high as it should be is the price of stupidity.

The result of that is that Australia’s economy is shrinking. We have had three years now of declining GDP per capita. We are getting a lower standard of living.

This was entirely predictable, because it was predicted by a bloke called Brian Fisher in 2019. He predicted a GDP contraction of at least 10% for Labor’s then Net Zero target of 45%. We are feeling that contraction now.

Australia’s four remaining aluminium smelters account for 10% of the power demand on the east coast National Electricity Market. Almost all their production is exported. Power is their biggest operating cost. Neville Wran, when Premier of New South Wales, said that ‘aluminium is solidified electricity’.

So how do the aluminium smelters keep going after doubling the price of power? They are kept going by enormous subsidies paid by taxpayers.

With the result that our trading partners view those subsidies as an assault on their economies. Thus the US has applied tariffs on Australian aluminium equivalent to the subsidies. The same has happened to steel made in Wollongong. This hasn’t been explained to the Australian public.

The jerry-rigged power market fantasy of Net Zero has been smacked by reality. Net Zero is a house of cards.

Why are renewables and Net Zero so cursed? One reason might be that renewable energy is an ideology invented by the Nazis in Germany. Renewables was a major plank of Nazi ideology.

What you see on the left is the cover of a book published in Germany in 1933, the year that Hitler came to power.

The title says ‘Technif und Wirtschaft im Dritten Reich’ which translates as ‘Technology and the Economy in the Third Reich’.

On page 47 you read “The renewable energy is flowing and free.” You read exactly the same thing in today’s renewables propaganda. The opposite is true of course. Renewables may cost us everything.

And just like today’s renewables freaks, the Nazis were also big hydrogen enthusiasts. On page 60 you read the Nazi plan to use wind power to make hydrogen.

And why, why are we destroying our economy and the country? Ostensibly it is all about reducing our carbon dioxide output. The graph shows the carbon dioxide output of China, the United States and the EU since 1850.

What you see is China going flat out at putting carbon dioxide into the atmosphere, now outproducing the US and the EU together.

China knows that the carbon dioxide scare is a sham or they couldn’t care less, or both.

To add insult to injury, we export coal to China but don’t burn it ourselves.

In World War Two, former Prime Minister Robert Menzies was known as Pig Iron Bob because he allowed pig iron exports to Japan despite Japan’s invasion of Manchuria. We got some of that pig iron back in the form of bombs dropped on Darwin.

Next time round we will be getting some of the coal we export to China back in the form of the explosive filler of the bombs China will drop on Darwin.

China uses cheap coal power to make solar panels to sell to us. Solar panels last 20 years before going to land fill. They don’t have enough value to be worth recycling and the cadmium loading of up to 10 grams per square metre means that they have to go to an engineered repository.

Windmills are the same.

Has anyone thought this through, really? Solar panels are made in China using cheap power from coal at five cents per kilowatt hour.

Under the most ideal conditions on the planet, in the deserts of Australia, the same panels produce power at a cost of 20 cents per kilowatt hour. Then the panels go to landfill.

If you wanted to make solar panels using power from that first generation of solar panels, what would be the cost of power from that second generation of panels?  It would be at least 80 cents per kilowatt hour and so on to infinity. And all the panels end up in landfill.

Solar panels are neither renewable or sustainable. They are simply an artefact of cheap Chinese coal power.

And that coal won’t remain cheap for much longer. This is a diagram from a Chinese paper of a couple of years ago. It shows how much of China’s initial coal reserves have been mined to date, by depth.

Down to 600 metres, they have mined 80% of what they started with.

From 600 metres to 1,000, 60% is now gone.

Typically in resource extraction, once you have exhausted half of what you started with, the cost of mining starts rising.

China has now burnt through half of its initial coal reserves and the cost of doing everything in China will now rise. It follows that to rely upon China as a source of solar panels to replace the ones going to landfill would be unwise.

A bloke called Santayana said way back in 1905 that “Those who cannot remember the past are condemned to repeat it.”

So what is the best historic analogy that we can draw guidance from in terms of what Australia is doing to itself with Net Zero?

That would be the cattle-killing frenzy of the Xhosa tribe of South Africa in 1858.

A teenage girl called Nongquause had gone down to the river to fetch water. When she got back, she told the tribal elders that spirits had told her that if the Xhosa killed all their cattle, the spirits would replace them with bigger and better cattle.

So the Xhosa tribe killed all their cattle and three quarters of them promptly died. These are people who had been farming cattle for thousands of years but it did not stop them from doing something very stupid with their food supply.

We are doing the same with doing the same with our energy supply and it is within the realms of possibility that the result could be the same with 75 percent of Australians dead.

That is because Australia won’t be able to defend itself when the economy keeps shrinking due to an idiotic cost burden.

Two hundred years after the Enlightenment, we have chucked over science and gone back to a form of paganism.

And for what? We can quantify what would be achieved in destroying our economy.

This is a graph I popularised 20 years ago. The greenhouse gasses, water and carbon dioxide, keep the planet 30 degrees warmer than it would otherwise be.

So the average temperature of the Earth is 15 degrees instead of minus 15 degrees. Carbon dioxide contributes 10 percent of that which is three degrees.

The graph shows the heating effect of carbon dioxide in 20 ppm increments. Lo and behold, the first 20 ppm is worth half the heating effect to date. The heating effect of carbon dioxide is logarithmic, not arithmetic, so it drops away rapidly after that.

From the current level of 421 ppm in the atmosphere, each extra 100 ppm is only good for 0.1 degrees of heating.

When we have dug up all the rocks we can burn, and burnt them, that will add another 200 ppm to atmospheric carbon dioxide. That will be only 0.2 degrees of warming. If you are scared by that prospect, you are beyond saving and I can’t help you.

So, we have established that renewables and Net Zero are ultra-stupid and we should be doing something else. The situation is a bit more complicated than simply going back to fossil fuels. The fossil fuels are running out and we need to leave fossil fuels faster than they leave us.

Take the case of oil which lubricates every activity in our economy. What this graph shows is US tight oil production since 2007. It started out with the Bakken Formation in North Dakota and now most production is from the Permian Basin of Texas and two counties in New Mexico.

Peak oil was supposed to arrive in 2005 but the supply increase from tight oil kept the oil price cheap for another 20 years.

Let’s look at Texas in particular. This graph shows the production profile of the top eleven oil-producing counties of Texas along with the rest of Texas. The biggest producing counties have tipped over into decline while total Texan production continued to rise.

In particular, Howard County tipped over into decline from May 2023. And the rate of its production decline is a lot faster than the rate of production increase up to May 2023. There was no production plateau for Howard County and that implies no production plateau for the whole of the Permian Basin.

This graph shows why Howard County tipped over into decline so abruptly. The X access shows the monthly gas/oil ratio from January 2018 to November 2024. This is the amount of gas that has come out of solution in the oil, in thousands of cubic feet of gas per barrel of oil produced. The Y axis measures total oil production for Howard County.

The gas/oil ratio was rising slowly until oil production peaked and then broke the trend shown by the red line. After that the gas/oil ratio rose rapidly as oil production fell. All this is explained by the reservoir pressure falling below the bubble point after which gas bubbles form in the reservoir. The energy that pushes oil out of the formation starts dropping faster and oil production falls.

And this is another graph of Howard County showing that gas production, the red line, held up for a while as oil production, the blue line, plunged. The two biggest oil producing counties in Texas, Midland and Martin counties, have now also broken through their bubble points.

The implication of all this that US tight oil production, which kept oil cheap for the last 20 years, has tipped over into decline and that there will be no happy plateau in production. The decline will be as fast as the rise at about one million barrels per day per year. As someone who lived through the oil shock of 1973, I can tell you that this has profound implications.

Now look at a bigger picture in terms of the energy available to humanity. This graph shows US oil and production, and Chinese coal production, from 1900. All in oil equivalent barrels so that apples can be compared with apples.

The US tight oil and gas phenomenon, big as it is, is dwarfed by the increase in energy supply due to Chinese coal production. In energy equivalent terms, Chinese coal production equates to oil production of 55 million barrels of oil per day. This is almost twice the energy from US oil and gas production. The rest of the world gets some enjoyment from Chinese coal production in the form of cheap goods. That will be no longer and standards of living will effectively fall.

Going back to coal won’t help much even for those countries, like Australia, that have a lot of coal. We know this because of a lesson from the 2008 commodities boom.

In 2008 there was a cargo of LNG that was imported into Thailand at above the then oil price. And the lesson learnt was that during tight oil supply, things that can substitute for oil will go to the oil price less the cost of conversion.

For a big chunk of the world, the cost of natural gas, imported as LNG, has gone to the oil price in energy content terms.

Coal is destined to also go to the oil price one day because coal can be converted to diesel and petrol using the Bergius process. So if you are relying on coal to keep the lights on and the wheels of industry turning, you will end up with a cost of doing that much the same as if you were burning diesel.

Now we need diesel to keep the economy lubricated and so making it from coal is a good thing to do. Bergius discovered how to do that via hydrogenation in 1913 and this figure is from his 1931 Nobel Prize acceptance speech. Simply, adding 5 kilos of hydrogen to 100 kilos of coal produces 100 litres of liquid fuel and another 20 kilos of hydrocarbon gasses.

A big chunk of the capital and operating cost is the steam reforming of part of the product stream to make hydrogen.

Thanks to nuclear power, we can skip that bit and make hydrogen from the electrolysis of water instead. This will make our coal reserves last 20 percent longer. We should do what we can to conserve our coal so that we can convert it to what which is most precious – liquid hydrocarbon fuels.

In the 1960s, Australian taxpayers used to subsidise oil exploration because everyone knew that having your own oil supply was essential to national survival. When the Bass Strait oilfields were discovered in the late 1960s, they weren’t competitive against cheap crude from the Middle East so Australian motorists happily paid a levy on petrol to get them developed. That paid off in spades during the oil crisis of 1973 when the rest of the world had shortages and Australia didn’t.

The last time we were near self-sufficient was 25 years ago as shown in this graph. The green is Australian oil production in millions of barrels per day and the orange is the balance of demand that is imported. Most of those refined product imports come from the same region where our next war will be fought, which is a very stupid situation to be in.

Note that once we start making enough for ourselves, we can start exporting diesel and petrol to our friends. Instead of being just another mendicant begging for supply, we can be a saviour to our neighbouring countries in the Pacific, which will bring a lot of diplomatic leverage with it.

In fact, if Australia’s current level of coal exports were converted to diesel and petrol, that would amount to seven million barrels per day – over half the rate that Saudi Arabia is producing oil at.

This graph illustrates how it would be achieved for the lowest capital cost and operating cost. In hydrogenating a reactive coal such as the Latrobe Valley lignite, you are adding another 5 percent hydrogen to something that is already 8 percent hydrogen to make diesel which is 13 percent hydrogen. Diesel is 13 percent hydrogen by weight but contributes 39% of diesel’s energy content.

The electrolysers which will convert nuclear power to hydrogen can be turned down to 25% of capacity without affecting conversion efficiency. That allows us to run our nuclear reactors at a steady state despite the big diurnal fluctuation in demand from the power grid. As demand from the grid drops off with the setting of the Sun, the electrolysers can step up and store hydrogen in gasometers.

This graph isn’t diagrammatic. It is scaled to Australia’s daily power demand and current fuel demand. We need 50 gigawatts of nuclear.

We have established that what we need is our nuclear future as soon as possible. There is no alternative, as a famous lady once said. It is a case of either going nuclear or it’s back to horse drawn carts and a seventeenth century standard of living at best.

But what sort of nuclear? It should not be the current dominant nuclear technology of U235-burning light water reactors. These have dominated since the original one in the first nuclear submarine, the USS Nautilus of 1956. The technology of nuclear reactors hasn’t changed much in the last 70 years.

This graph shows why that is such a bad thing. Producing one gigawatt of power continuously over a year requires the fissioning of one tonne of something. To achieve that in light water reactors, you start with 250 tonnes of uranium as mined out of the ground.

You then concentrate the U235 component up from 0.7 percent to 3.5 percent in 35 tonnes of what you started with. This is done at some expense and relies upon the 1.1 percent mass difference between U235 and U238. Nevertheless, 29 percent of the U235 you started with gets thrown out in the other 215 tonnes of uranium which is then called depleted uranium.

The 35 tonnes of enriched uranium are made up into fuel rods clad in zirconium. After three years in a reactor, they are pulled and put into long term storage. They contain one tonne of fission products, 300 kilos of unburnt U235 and another 300 kilos of plutonium that has bred from uranium atoms capturing a neutron. By the time the rods are pulled, half the energy being produced is coming from the plutonium created in the rods.

The whole process only uses 0.4 percent of the energy contained in the 250 tonnes of as-mined uranium. This is one two hundred and fiftieth.

Nobody could be proud of such a wasteful technology. Thankfully, there is a better way.

Before we get to that, there are two problems with nuclear energy that nobody on either side seem to want to talk about.

The big one is that once a reactor has settled down to a continuous power output, seven percent of that power is coming from nuclei that have absorbed a neutron but have yet to split and release energy. They will split, perhaps months or years later. That means you can shut a reactor down by with its control rods but it will still be producing heat.

So a 1,000 MW thermal reactor, which would produce 300 MW of electrical power, will be producing 70 MW of heat from delayed fission reactions. That drops away rapidly but remains significant. If the coolant pipes are broken or the coolant pumps have lost power, or the generators to power those pumps have run out of diesel, then the core heats up and the remaining steam reacts with the zirconium of the fuels rods to produce hydrogen. The hydrogen accumulates in the containment building and explodes.

That is what happened to the three operating reactors at Fukushima after the earthquake and tsunami on 11th March, 2011. Reactor No 1 exploded the following day with No2 and No 3 a couple of days later. Making your reactors bigger to achieve economies of scale makes the problem of shedding heat harder to overcome.

Also, most of the spent fuel rods produced each year around the world are not reprocessed. In the United States they are piling up at 2,000 tonnes per annum with the total now at 98,000 tonnes.

There is one reactor type that solves both those problems as well as the problem that the current dominant technology only uses 0.4 percent of our uranium endowment.

This is the plutonium breeder reactor which has been successfully operated in Russia for decades, and also successfully put into operation in France.

There is no waste. Everything is recycled until it is burnt up. They can be online two years after the concrete foundations have been laid. There are several designs available. We should install them all and see what works best.

The economics of plutonium breeder reactors are competitive with current coal-fired costs.

Adopting plutonium breeder reactors will unlock an even bigger energy resource. As well as breeding U238 to plutonium, thorium can be bred to U233.

Breeding U238 to plutonium has a theoretical margin of 30 percent. That is, under ideal conditions, you will produce 30% more nuclear fuel than you started with. Breeding U233 from thorium has only an eight percent margin which might disappear with neutron losses to the containment vessel and so on.

Every spare neutron from breeding plutonium from uranium should be applied to breeding from thorium. This means that our nuclear endowment would last 1,250 times longer than if just used the current U235-burning technology.

There is a better future, and there is an even better future. That is the promise of thorium.

Nuclear technology hasn’t changed for 70 years. We are just using variations of the first nuclear reactor that went to sea. The space business was the same with cost of getting stuff into orbit unchanged for decades. Then Space X came along and dropped costs by 90 percent. The same potential for much lower costs also occurs in nuclear power.

How fast can we do the nuclear reactor rollout we need to do as soon as possible? Let’s take the example of France after they decided to go all-in on nuclear in the 1970s.

This graph shows the contribution from hydro power in blue, yellow is the nuclear component, brown is fossil fuels and green represents the hopeless renewables.  The blue bar is 245 terra watt hours which is Australia’s current power consumption.

France was able to install 245 terrawatt hours of nuclear capacity over the twelve years from 1980 to 1992. To suggest that Australia couldn’t also install 245 terrawatt hours of nuclear capacity over twelve years would mean that the French are better than us, even the French of 40 years ago before all the improvements in manufacturing technology since. That is just not possible so we can also do it in twelve years.

The four pillars of civilisation are diesel, plastics, steel and concrete. All those things need carbon if made with current technology. So, when the fossil fuels run out, where will the carbon come from? That will require plantation forests of eucalypts. This graph shows the square metres of eucalypt plantation required per capita for each pillar of civilisation. All up it is about 300 square metres or less than a small suburban block, and eminently doable.

In summary, we will be going to Net Zero whether we like it or not. Because one day we will run out of coal and oil and all the other good stuff from the Earth.

But we have a perfectly wonderful future to look forward to with nuclear power by the breeder reactor route.

Instead of burning coal in power plants, it should be saved for making synthetic diesel. Our motto should be ‘conserve to convert”.

Hydrogen enables the conversion of electric power from nuclear reactors to chemical potential. Hydrogen will be a big part of our future.

Carbon is the carrier molecule that enables hydrogen to be used at room temperature and pressure. The sooner we start down that route, the safer we will be.

David Archibald 13th March, 2025


Discover more from Watts Up With That?

Subscribe to get the latest posts sent to your email.

via Watts Up With That?

https://ift.tt/fbJPyLg

March 23, 2025 at 12:08PM

Can You Trust an AI/ML Model to Forecast?

The latest fashion in model building is adding AI/ML (Artificial Intelligence/Machine Learning) technology to numerical models for weather forecasting.  No doubt soon there will be climate models also claiming improved capability by doing this.  A meteorological example is called Aardvark Weather and a summary is provided at Tallbloke’s Talkshop Scientists say fully AI-driven weather prediction system delivers accurate forecasts faster with less computing power.

Like all inventions there are weaknesses along with the claimed benefits.  Here’s a short list of the things that can go wrong with these new gadgets. The concerns below are listed along with some others in a paper Understanding the Weaknesses of Machine Learning: Challenges and Limitations by Oyo Jude. Excerpts in italics with my bolds.

Introduction

Machine learning (ML) has become a cornerstone of modern technological advancements, driving innovations in areas such as healthcare, finance, and autonomous systems. Despite its transformative potential, ML is not without its flaws. Understanding these weaknesses is crucial for developing more robust and reliable systems. This article delves into the various challenges and limitations faced by ML technologies, providing insights into areas where improvements are needed

Data Quality and Bias

Data Dependency

Machine learning models are highly dependent on the quality and quantity of data used for training. The performance of an ML model is only as good as the data it is trained on. Common issues related to data quality include:

Incomplete Data: Missing or incomplete data can lead to inaccurate models and predictions. Incomplete datasets may not represent the full spectrum of possible inputs, leading to biased or skewed outcomes.
Noisy Data: Noise in data refers to irrelevant or random information that can obscure the underlying patterns the model is supposed to learn. Noisy data can reduce the accuracy of ML models and complicate the learning process.

Data Bias

Bias in data can significantly impact the fairness and accuracy of ML systems. Key forms of data bias include:

Selection Bias: Occurs when the data collected is not representative of the target population. For example, if a model is trained on data from a specific demographic group, it may not perform well for individuals outside that group.
Label Bias: Arises when the labels or categories used in supervised learning are subjective or inconsistent. Label bias can skew the model’s understanding and lead to erroneous predictions.

Model Interpretability and Transparency

Complexity of Models

Many advanced ML models, such as deep neural networks, are often described as “black boxes” due to their complexity. The lack of transparency in these models presents several challenges:

Understanding Model Decisions: It can be difficult to understand how a model arrived at a specific decision or prediction, making it challenging to diagnose errors or biases in the system.
Trust and Accountability: The inability to interpret model decisions can undermine trust in ML systems, particularly in high-stakes applications such as healthcare or criminal justice. Ensuring accountability and fairness becomes challenging when the decision-making process is opaque.
Explainability:  Efforts to improve model interpretability focus on developing techniques and tools to make complex models more understandable. Techniques such as feature importance analysis, surrogate models, and visualization tools aim to provide insights into model behavior and decisions. However, achieving a balance between model performance and interpretability remains an ongoing challenge.

Generalization and Overfitting

Overfitting

Overfitting occurs when a model learns not only the underlying patterns in the training data but also the noise, resulting in poor performance on new, unseen data. This issue can be particularly problematic with complex models and limited data. Strategies to mitigate overfitting include:

Cross-Validation: Using techniques like k-fold cross-validation helps assess model performance on different subsets of the data, reducing the risk of overfitting.
Regularization: Regularization methods, such as L1 and L2 regularization, add penalties to the model’s complexity to prevent it from fitting noise in the training data.

Generalization

Generalization refers to a model’s ability to perform well on unseen data that was not part of the training set. Achieving good generalization is crucial for the practical application of ML models. Challenges related to generalization include:

Domain Shift: When the distribution of the data changes over time or across different domains, a model trained on one dataset may not generalize well to new data. Addressing domain shift requires continuous monitoring and updating of models.
Data Scarcity: In scenarios where limited data is available, models may struggle to generalize effectively. Techniques such as data augmentation and transfer learning can help address data scarcity issues.

Comment:

Many of these issues have been raised against climate models, undermining claims their outputs are valid projections of future climate states.  For example, the issue of detailed and reliable data persists.  It appears that even the AI/ML weather forecasting inventions are dependent on ERA5, which has a record of only ~40 years to use for training purposes.  I’m suspending belief in these things for now–new improved black boxes sound too much like the Sorcerer’s Apprentice.

Disney’s portrayal of the Sorcerer’s Apprentice in over his head.

via Science Matters

https://ift.tt/BLj8SW1

March 23, 2025 at 10:48AM

Morpeth: Cockle Park DCNN 2084 – A Venerable old station no longer properly cared for.

52.212586 -1.6854742 Met Office CIMO Assessed Class 4 Installed 1897

Cockle Park is 262 hectare farm owned and managed by Newcastle University for agricultural research. Despite its original “Second Order” status, this is an important weather station site hosting its own 128 year old Phenological Garden and “conducting the longest running grazing and hay cutting experiment in the world that was started in 1896”. The data from here is valuable and should be carefully protected – but then this is the Met Office so expect the worst.

The Met Office proclaims Cockle Park as one of “their” longest running time series and it is heavily featured on their website https://weather.metoffice.gov.uk/learn-about/how-forecasts-are-made/observations/recording-observations-for-over-100-years

Technically it was not a creation of the Met Office (most old sites were not) and they were simply accepting the data from the research centre in much the same way as at Rothamsted . The researchers involved had no motivations other than to record, as accurately as they were able, the exact weather conditions (note “weather”) and those readings were later analysed against noted crop growth and development. The term “Anthropogenic Climate Change” would not have been featured in their vocabulary.

It is important to note that when installed all UK sites recorded in Fahrenheit. Some stations recorded to the nearest whole degree F (i.e. 0.55°C) though Cockle Park recorded to the nearest 0.5°F (0.278°C) and a very few fully manned “First Order” sites did for a period try to register to the nearest 0.1°F. This latter accuracy would have required a very high level of observation skill to avoid parallax reading errors and is no longer considered within the ability of the instrumentation in use back then. When compiling historic comparisons these step change conversion differences must be borne in mind given that modern day readings are to 0.1°C. {and some silly Met Office representatives claim to the 5th decimal place!} A typical 19th Century hand written return from Cockle Park was as below.

Note how appropriate the location was for their specific purposes from this aerial image indicating a 100 metre radius.

There really is nothing to mark down this site apart from the difficult to assess issue of changing ground cover. There are certainly no UHI effects to consider, the area is flat, no shading nor any extraneous heat sources. I would rate this as definitely a Class 1 site – so why is it only assessed as Class 4 by the Met Office? The answer is incredibly simple…..this is NOT the current site of the weather station.

The original site above operating from 1897 was a manually recording unit with an impeccable readings record typical of research sites such as at Cawood. From 2014 the Met Office deemed the site should be automated and seemingly to facilitate that (for ease of electricity and data comms access) relocated the site as below. The former screen is still visible as the small square in the field to the north west.

An object lesson in how to degrade a site from almost certainly pristine Class 1 to an unreliable Class 4. From a trained, dedicated research observer to an automatic “2.5 Class 4 (additional estimated uncertainty added by siting up to 2 °C)” in one swift jump. 117 years of excellence are now “dumbed down” to a subsequent 11 years (and counting) of inaccurate data that was only ever going to result in higher readings caused by the compromising buildings to three compass points.

A cynic, like me, might suggest ulterior motivation in such a site degradation, however, another institution also seemed to be underwhelmed by this relocation. The UK Centre for Ecology and Hydrology has both a Met Office weather station and its own separate and highly sophisticated automatic unit at their Wallingford site. Subsequent to the Cockle Park Met Office relocation on 19/2/2014, the UKCEH opted to open their own separate site at Cockle Park on the 21/11/2014.

This is where this independent automatic reporting unit is located.

The UKCEH managed to get a fully functioning site that (ground cover issues aside) operates perfectly well and is significantly over 100 metres from any problems. Why could the Met Office not manage the same feat? I do not know the reason behind the UKCEH decision to install a new unit but the timing does seem to be remarkably coincidental.

Considering how the Met office opts to represent Cockle Park’s readings is equally illuminating. They operate a Historic Station Data web page which includes some sites with data only going back to the latter 1970s but completely overlooked Cockle Park with one of the longest of all records. However, what really makes the vaunting of the site’s long history so strange is the Location specific Long term Averages data page for Morpeth.

Cockle park is not shown at all. This page gives climate averages covering rolling 30 year averages from 1960 to 2020. Cockle Park could not possibly be any more “Location specific” for Morpeth and the Met Office have certainly confirmed it is very “Long Term” indeed…..so why not just show its figures, it really cannot be that difficult can it? Well, obviously it is given that none of the alternative “data” offered is actually genuinely “Location specific ” nor adequately “Long Term” The numbers offered are, by definition, fabrications.

Albemarle is 13 miles away, Class 5 junk, and did not come into existence until 18/2/2002. Any “average” for the period from 1960 to 2020 is pure invention – it is physically impossible to derive any 30 year rolling average from a specific location that has not existed for any 30 year period. Clearly the numbers presented are made up from unspecified locations that the Met Office have been proven to be unable to verify. elsewhere.

Tynemouth is 15 miles distant, or actually not, because it does not exist. This coastal site, that bears no climatological similarity to Cockle Park at all, closed in 2001. Every number since is a work of pure fiction.

Boulmer (aka RAF Boulmer) 18 miles distant is absurdly claimed to be Class 4 but is really more Class 5 junk (will be reviewed shortly) and also did not come into being until 1975 meaning there are 15 years of “creative accounting” in compiling the first two the claimed 30 years average concoctions…..and where did those numbers come from? I doubt the Met Office can (or will not) supply that information.

Redesdale Camp 25 miles distant, a mere 400 feet higher elevation yet again Class 5 Junk and this one only came into existence in 1979 so almost 2 decades of nonsensical comparative calculation.

Durham 28 miles distant, yet more Class 5 UHI ridden low quality but at least it was around for the entire time.

In summary Cockle Park used to be, until very recently, a good quality reliable site that was well maintained with an impeccable readings record. The Met Office chose to relocate it to a poor quality site on automation for no apparently necessary reason – it could have been automated where it was. Other institutions have felt it necessary to subsequently install better quality weather stations to maintain the reliable site history. The Met Office themselves decline to present the available raw data as a simple rolling period average preferring to portray “peer reviewed” computer modelled nonsense from sites not even in existence for the whole period. In an exact replica of Cawood the Met Office openly declines to provide simple averages that would either confirm or refute their “peer reviewed” concoctions – I wonder why?

via Tallbloke’s Talkshop

https://ift.tt/WtV3Q0p

March 23, 2025 at 10:43AM