Click images for more details



Recent comments
Recent posts
Currently discussing

A few sites I've stumbled across recently....

Powered by Squarespace
« Diary dates, feedback edition | Main | The Old Lady of Eco Street »

I have a computer model

I have a computer model, which I use for predicting the weather. The algorithm is fairly straightforward and goes something like this:

It will rain tomorrow.

(Round where I live that's likely to be a pretty effective prediction.)

Anyway, if I run my computer model repeatedly, I find that 100% of the runs give the same result - "It will rain tomorrow". I conclude, therefore, that we can say with 100% confidence that it will rain tomorrow.

I thought of my little computer model when reading the 2014 progress report of the Adaptation Subcommittee of the Committee on Climate Change, which describes the UK's progress in managing climate risks. In it, we learn that:

It is very likely (90% confidence) that greenhouse gas emissions over the 20th century substantially increased the risk of flooding in England and Wales in autumn 2000.

The citation is to Pall et al, a paper by Myles Allen's group, the key finding of which appears as a footnote in the ASC report:

The precise magnitude of the anthropogenic contribution remains uncertain, but in nine out of ten simulations 20th century anthropogenic greenhouse gas emissions increased the risk of floods occurring in England and Wales in autumn 2000 by more than 20%, and in two out of three cases by more than 90%.

So you see, while you may have sniggered at my little computer model, my approach is legitimised by none other than Lords Deben and Krebs and Professor Myles Allen.

PrintView Printer Friendly Version

Reader Comments (46)

The abstract doesn't even make sense to me. Surely 9 out of 10 simulations had more flooding rather than 'risk of flooding'. So does this imply the models are good at simulating rainfall then??

Currently sitting out the passing of typhoon Vongfong. It is very wet outside.

Oct 13, 2014 at 9:03 AM | Unregistered CommenterRob Burton

Don't forget that 'Tomorrow will be the same as today' is accurate 66% of the time. No computers, no models, just a fact!

Oct 13, 2014 at 9:14 AM | Unregistered CommenterCharlie Flindt

My system for predicting the weather is more low tec but works quite well and does not require much computing power. It involves tossing a coin, heads it is going to rain, tales it will not. However an additional subroutine comes into effect from early October to late April, this says heads it rains, tales it rains. Adjust this subroutine for your particular location. Although sometimes I just look out of the window and then decide. In the past I hear some people used to check out the weather forecast on something called the "BBC", but they now get their weather from a political crowd called the Met Office.

I suppose if someone had enough coins of very high value (each penny worth several million or more ) they could predict the climate/weather several decades in the future. I hear the Met Office do this but as well as the coin subroutines they also have some additional software that gets input from a marxist data file.

Oct 13, 2014 at 9:15 AM | Unregistered CommenterBlack Dog

Should have said "the BBC, who get their weather from a political crowd called the Met Office."

Silly me.

Oct 13, 2014 at 9:19 AM | Unregistered CommenterBlack Dog

It's known rubbish and increasingly difficult to associate to honesty among scientists. Regional models have no known skill. At least "it'll rain tomorrow" is demonstrably correct some of the time.

Climate science corruption is almost complete.

Oct 13, 2014 at 9:24 AM | Registered Commenteromnologos

Last night...BBC World Service (The Fifth Floor) on the subject of Mathematics. One guy popped up with the topic of dangerous climate change (DCC). No definition of course.

What we got was that no matter what the calculation might be using mathematics (not specified) we need to execute precaution. The precaution theory he espoused was that of doing something in relation to countering DCC. The precaution activity was not clearly indicated, but as it went along it seemed that physical evidence would not be taken into account.

I expected by the absence of precautionary detail that whatever is driven by mathematics would likely be what the fools are doing now. And that is not taking the right precautionary actions.

I thought it was a program on Mathematics but as usual BBC programs degenerate into Climate stuff almost by default.

Oct 13, 2014 at 9:30 AM | Unregistered CommenterEx-expat Colin

Dod you get a grant (or a K) for the development of your computer model?

Oct 13, 2014 at 9:37 AM | Unregistered CommenterCapell

"Anyway, if I run my computer model repeatedly, I find that 100% of the runs give the same result - "It will rain tomorrow". I conclude, therefore, that we can say with 100% confidence that it will rain tomorrow."

If it doesn't rain you may explain it away by stating "natural variability".

Oct 13, 2014 at 9:39 AM | Unregistered CommenterJeff Norman

If it doesn't rain you obviously need a more powerful computer. How many £million do you need?

Oct 13, 2014 at 9:47 AM | Registered CommenterPhillip Bratby

Due to natural variability, on some days you get 110% of normal rain. Therefore on average it rains every day, proving your model is correct.

Oct 13, 2014 at 9:51 AM | Unregistered CommenterAndyL

Actually, your model is far more powerful than the Pall et al one, since it makes a prediction about something happening in the future, whereas Pall et al make a prediction about the probability of something that's already happened having been made more likely to have happened by the fact that something else that's already happened has happened.
One' wonders why ten percent of simulations gave the wrong answer. Although I suppose there's a ten percent chance that the wrong answer was the right one.

Oct 13, 2014 at 9:51 AM | Registered Commentergeoffchambers

Given Lords Deben, Lord Krebs and Professor Myles Allen, did you expect real science?

It is clear that Myles Allen sees a funding stream out of the extreme weather meme. Watch out for more of the AGW increased flooding scare.

Oct 13, 2014 at 9:52 AM | Unregistered CommenterCharmingQuark

Yet when these models are given an aerosol cooling parameter near it's upper uncertainty band then 100% of runs predict a new ice age.

Meanwhile, following the driest September on record we can look forward to another drought soon and the subsequent opportunistic clamour to connect it to carbon dioxide - just as they did, en masse, immediately prior to the recent floods. Happily for Myles Allen, the media seemingly have very short memories so he can be as wrong as he wants as often as he wants and still be referred to as an expert in climate prediction.

Oct 13, 2014 at 10:01 AM | Unregistered CommenterJamesG

Nice one. Of course, the difference is they 'know' their assumptions are correct.

Oct 13, 2014 at 10:03 AM | Unregistered Commenteroakwood

My model is superior. It outputs: "It will rain tomorrow catastrophically"

Oct 13, 2014 at 10:05 AM | Unregistered CommenterBrute


Who is to say those 10% gave the wrong answer? Perhaps it's the 90% that gave the wrong answer?



Oct 13, 2014 at 10:05 AM | Unregistered Commentermailman

If someone claims it is _not_ raining they are obviously in denial.

It _is_ raining, hidden in the deep oceans.

Oct 13, 2014 at 10:26 AM | Unregistered CommenterNial

As a professional pedant and amateur but expert semanticist I can assure you that Myles Allen is an expert at making climate predictions.
I would go so far as to say that there are few people in the world (certainly in the UK) more expert at making climate predictions than he.
The standard that he brings to the making of climate predictions is very high, almost awesome.

Please note: Nothing in the making of climate predictions requires that these predictions be in any way realistic and certainly not accurate. Making accurate climate predictions is a different skill altogether.

Oct 13, 2014 at 10:30 AM | Registered CommenterMike Jackson

No global warming link will be declared about the driest September on record simply because everybody enjoyed it. More evidence of abysmal science quality from the MetOffice.

Oct 13, 2014 at 10:42 AM | Registered Commenteromnologos

I wonder if Myles Allen's model for risk of flooding included the effects of flood-plain management/lack of and/or river dredging/not? The 'risk of flooding' is not always directly proportional to the amount of rainfall. There are an awful lot non-natural processes that combine to make the risk more or less likely.

The man was obviously in need of more funds and does what his Masters tell him to do. It so reminds me of the line from Thomas Moore, on seeing that Richie Rich has succumbed to the bribe of Welsh Secretary: '...[sigh] but for Wales, Richard?' One could say of Allen: 'But for Deben, Myles?'

Oct 13, 2014 at 10:52 AM | Registered CommenterHarry Passfield

What a splendid model. No moving parts. Low maintenance. 100% correct all the time at a global level. I think this would be an excellent starter to introduce schoolchildren to the use of models in science, as well as in the intersection between politics and climatology (mental image: two circles almost entirely overlapping).

It seems perverse to want to gild a lily, but there are mischievous people out there who might want to take advantage of the model's transparency. To start with, leave the core algorithm unchanged – that is where the model output will come from in due course, but bury it well, even encrypt it. Add automated data entry routines. Any old data would do, but for verisimilitude, get some meteorological data streams. Remember the advice recounted by Josiah Stamp about data collected for government use: 'collect them, add them, raise them to the nth power, take the cube root and prepare wonderful diagrams '. No end of fun possible here. Hire computer games programmers to have a ball, and make some glorious graphics of cyclones sweeping across the Atlantic etc etc. The kind of thing that so impressed Paul Nurse in the States, and also, it would seem, our own dear Richard Drake at the Slingo talk. Encourage the coders to use as much physics as they like to get the effects you want to see. Out of all this will emerge, like 42 did from Deep Thought , 'It will rain tomorrow'.

Improving regional performance will be easy enough. Just get the data about the number of rain-days in a year, broken down by season as applicable, and set your random number sub-routines to match those frequencies. You aim is to impress the folks who live in the Sahara as much as the ones who live in Sumatra.

Coming down to the local level, the level where of course we all live, you might then bury a slightly more sophisticated algorithm into the dark heart of this now enormous, all but impenetrable, piece of bloatware: 'Tomorrow will be much the same as today.' This is third generation climate system modelling. By now the grants will be flooding in, provided you have taken the precaution of softening up the political classes. The UN organisation has proven to be a handy route for doing this provided you speak English saturated with pompous and portentous phraseology with links to genuine issues of real concern. That can take a few years , but, my goodness, the rewards!

Oct 13, 2014 at 11:03 AM | Registered CommenterJohn Shade

Many years ago I lived in Crowthorne in Berkshire, a few miles SW of Bracknell - where the Met Office then had its HQ. The weather forecasts on the wireless were uncannily linked to whatever the weather was doing in Crowthorne, which could of course be seen by looking out of the windows of Met Towers towards the SW.

Oct 13, 2014 at 11:06 AM | Unregistered CommenterVictoria Sponge

I've just read this excellent monograph PRECIPITATION, DELUGE, AND FLOOD: observational evidence and computer modelling'.

Forget about tomorrow's weather! The big money is in the climate game. That monograph shows just how rubbish the competition is.

Get some ideas out quick. Maybe 'It will rain an awful lot more 50 years from now'. Get those games programmers on to it. Right up their street. Floods and mayhem. Ker-ching!

Oct 13, 2014 at 11:23 AM | Registered CommenterJohn Shade

And thus, you join the 99% consensus that the models are right, +/- 1%

Oct 13, 2014 at 11:42 AM | Unregistered Commentercedarhill

Ignoring all the model twaddle, Box 1.2 on page 25 of the report contains a surprising amount of honesty.

It is littered with phrases like "not quantified", "difficult to quantify", "less certain", "less well understood" and even

"Changes in future climate extremes are very uncertain."

In respect of "Adapting to flood risk" the question "Can we quantify current vulnerabilities?" is answered with "Yes, maturing models of flood risk available." This is certainly true, in that land-drainage and sea-defence agencies have modelled areas prone to flooding for many years before global warming became the cause of everything under the sun.

In the absence of a 21st century schizoid climate model, they probably also used to look at Ordinance Survey maps and gain additional knowledge from local residents and land-owners.

Oct 13, 2014 at 12:04 PM | Unregistered Commentermichael hart

Victoria Sponge@ Oct 13, 2014 at 11:06 AM

Once upon a time I used to drive regularly from Brum to Portsmouth using the A303. The A303 aligns almost directly to the SW of England and includes the Somerset Levels. Too often while on that road I was hit with massive rainfall. The view to me was almost pre-historic, heavily laden black clouds off the Atlantic. There is a massive rain splat there right now, over the whole of SW England -

So does rain plenty and at times very heavily. If those flood plains/drains don't work its all too obvious whats to happen. And I have experienced a slight car float or two on those Levels.

Oct 13, 2014 at 12:26 PM | Unregistered CommenterEx-expat Colin

Whatever happened to Barometers?
Do you remember those, the Victorians used them to quite good effect when looking to see what the weather "might" do over the next few days.

Take a look at a local BBC 5 day weather forecast, they even modify them on the same day and always modify them over the 5 days,

Oct 13, 2014 at 12:29 PM | Unregistered CommenterA C Osborn

I have an even simpler device for forecasting.
Its a brass plate in my bar (show off..!), which reads: 'Free beer tomorrow'...
Its never been wrong yet...

Oct 13, 2014 at 12:42 PM | Unregistered Commentersherlock1

colin @ 12.26: Brum to Pompey via the Somerset levels? Are you a taxi - driver by any chance?

Oct 13, 2014 at 1:04 PM | Unregistered Commenterdiogenese2

My elementary school science teacher in Canada had been a jet fighter pilot. The science curriculum had a section on weather that he saw as trivial and useless. From his air force days, he still had access to the film courses that were used to teach pilots about the weather and that is what he taught us. He taught us the types of clouds and how they related to weather phenomena like warm and cold fronts. From that we knew and could predict the general shape of the weather from a few hours to a few days out. My mother still talks of the time that I told her, on a bright sunny morning, not to hang the clothes on the line because it was going to rain.

So one good way of predicting the weather is to look outside and see what kind of clouds are there are.

This also shows that even ordinary children can be presented with and benefit from challenging topics

Oct 13, 2014 at 1:11 PM | Unregistered CommenterTAG

@diogenese2 Oct 13, 2014 at 1:04 PM

Don't be daft...used the A303 from M3 to turning for Tidworth (& visa-versa). That was enough.

Car surfing the Somerset Levels was in a VW back in 60's and likely when the drains were working?

Oct 13, 2014 at 1:20 PM | Unregistered CommenterEx-expat Colin

A C Osborn

"Take a look at a local BBC 5 day weather forecast, they even modify them on the same day and always modify them over the 5 days,"

I have seen it change within a few hours. Also I have seen the forecast that it should be clear & sunny but when I look through my window I see thick cloud and heavy rain.

Oct 13, 2014 at 1:20 PM | Unregistered CommenterCharmingQuark

AC Osborn

"Whatever happened to Barometers?"

This happened.

Although strangely, Mercury is still allowed in dental fillings and fluorescent lamps!

Aneroid and electronic barometers are OK, so we may still know the pressure, but it makes me want to go out and buy an antique one just to annoy a Eurocrat...

Oct 13, 2014 at 2:54 PM | Registered Commenterjamesp


"He taught us the types of clouds and how they related to weather phenomena.."

But, but - if you did that, children might think that weather extremes occurred naturally and were nothing to do with AGW!

Oct 13, 2014 at 2:58 PM | Registered Commenterjamesp

why not 2 models?
the other telling it is not going to rain

this way, wait for it, the models outcomes will be cough be CONSISTENT with the observations.
then pitch in any conclusion "its worse than we thought" we need more funds to tell you
it is MUCH worse than we thought.

Oct 13, 2014 at 4:02 PM | Unregistered CommenterMichaela Ffrau

11:06 AM Victoria Sponge

You will be aware, of course, that Crowthorne, and thus Bracknell, is very close to a certain institution for the criminally insane - perhaps that is where they get their ideas from ?

Oct 13, 2014 at 4:04 PM | Unregistered CommenterJohn Constable

Belief that what comes out of unvalidated computer models is 'evidence' is one of the strange characteristics of so-called climate science.

Oct 13, 2014 at 4:36 PM | Unregistered CommenterMartin A

I think I'll stick with Big Ears the pheasant as my forecaster. Mind you, he's got a bit of a strop on at the moment, must be due to the recent heavy rain. This afternoon, he pecked Noddy (the number 2 pheasant) on the bum and pushed him off the roof.

Oct 13, 2014 at 5:47 PM | Registered CommenterSalopian

Oct 13, 2014 at 4:04 PM | Unregistered CommenterJohn Constable

Yes - I remember it well. If I recall correctly they tested the "Escape" siren at 10am of a Monday. Which, as no-one took any notice, was exactly the time to escape...

Oct 13, 2014 at 5:52 PM | Unregistered CommenterVictoria Sponge


"Although strangely, Mercury is still allowed in dental fillings and fluorescent lamps!"

Back in the late 1800's there were a number of gold mines in the Mawddach valley in Snowdonia. One company tried a mechanical amalgamation machine to recover the gold. It was a total failure and 5 tons of mercury was lost into the river and was never recovered. It is still there to this day, and yet the EA recently banned gold-panning in the river on the grounds that it was disturbing freshwater mussels - doh!

Oct 13, 2014 at 6:14 PM | Registered CommenterSalopian

Myles Allen's group first needs to show there has been a statistically significant change in the frequency and intensity of rainfall in Great Briain and that their model can correctly hindcast observed changes in the frequency and intensity of rainfall. Then they still need to ask if the weather at the time of the recent flood ACTUALLY differed in the expected manner from that typicall observed earlier in the century. (For example, the models predicted more intense precipitation from a warmer ocean to the east, while the recent flooding was caused by moist air arriving from the southeast.

Oct 13, 2014 at 8:07 PM | Unregistered CommenterFrank

Allen and Deben and Krebs...Oh, my!

Oct 13, 2014 at 9:14 PM | Unregistered Commenterjorgekafkazar

This reminded me of something I said elsewhere,some time ago, it goes something like...........

"I have a computer model that suggests the planet will be attacked by wave after wave of aliens and it's hopeless because every time we destroy one incoming wave the next is stronger and faster, the consequences are inevitable - we will lose and we will all be killed - the model proves it. Oh wait - I've just realised I was playing space invaders. Panic over."

Oct 14, 2014 at 12:07 AM | Unregistered Commenterjaffa

My model is even more superior.

Its output: "It will rain tomorrow, catastrophically, and it's all our fault"

Oct 14, 2014 at 7:55 AM | Unregistered CommenterAndrew Duffin

Robert Brown on models (inter alia)

I am more than 97% confident that Lord Deben will never read this

Oct 14, 2014 at 4:21 PM | Unregistered CommenterH2O: the miracle molecule


"I am more than 97% confident that Lord Deben will never read this"

Having been involved in providing Deben with written briefing during his time as a Minister at MAFF, I am more the 97% confident that reading was never one of his best subjects.

Oct 14, 2014 at 5:00 PM | Registered CommenterSalopian

PostPost a New Comment

Enter your information below to add a new comment.

My response is on my own website »
Author Email (optional):
Author URL (optional):
Some HTML allowed: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <code> <em> <i> <strike> <strong>