Member log in

Just looking out the window – can Metservice really do any better?

Over the years I’ve taken several swipes at the New Zealand Metservice. The thrust of these critiques has been that their weather forecasting results are little better than tossing a coin or looking out the window.

To make the point I’ve invited readers to take note on a Monday of the forecast for the following Monday, then watch how that original forecast gradually changed during the week to conform to reality.

On several occasions I’ve received replies from the Metservice’s ‘Weather Ambassador’, the lovely and hugely entertaining Bob McDavitt.

Bob usually points out that the closer you are to a weather event the more likely you are to get the forecast right. That is undoubtedly beyond question.

But you might have thought that with all the electronic wizardry that’s available to today’s forecasters and with our planet circled by weather satellites, they might be able to get it pretty well right a week ahead and absolutely right for tomorrow.

But they can’t and they often don’t.

I’m in a better position to pontificate on this now than I was a decade ago. A decade ago I took almost no exercise. Today Judy and I walk for at least an hour every day. So I’m interested in and familiar with the weather. I am, if you like, a weather-watcher.

This has led me to the conclusion that Metservice getting the forecast exactly right is the exception rather than the rule; getting it more or less right is closer to the norm; and getting it wrong far too frequent to justify the term ‘science’.

The Metservice has unwittingly assisted me in making these judgements by providing an excellent little app, which allows me to check the 10-day forecast for Auckland as I lie in bed each morning going through my emails and news on my iPad.

In particular, the on-line ten-day forecast is graphically presented with golden suns, fluffy clouds and rainy teardrops, either alone or in combination.

As I write this (on Monday), today’s Auckland forecast is shown as a golden sun with a fluffy cloud, which I take to mean ‘dry but some cloud’.

Looking out the window I see that it is indeed cloudy but not raining. I could just look out the window  anyway, but I don’t want to carp. At the moment the forecast for today is correct. (Based on our walking experience, I’m going to predict that the cloud will disappear later in the day, but I’m not putting money on it. The only satellite I have access to is the Sky dish on our roof.)

But today’s 10-day Metservice forecast for Auckland also gives me (and those of you who live here) a rare chance to check on my claim that they often get it wrong. You see, today’s graphic consists solely of nine golden suns, indicating nine fine, cloudless days starting tomorrow.

Here’s my prediction: there won’t be nine consecutive fine, cloudless days starting tomorrow in the City of Sails; There’ll be cloud and/or showers on one or more of those days. If you  live in Auckland and you’ve nothing better to do, check out the Metservice’s prediction and mine for the next nine days and see who was right(er). (You can do the same thing for yourself in your area – their prediction versus yours.)

Now I want to finish on a conciliatory note. It isn’t my view that the Metservice regularly gets it horribly wrong. They don’t. But nor do they get it acceptably right. ‘Sunny’ days turn out to be cloudy days; ‘cloudy’ days turn out to be showery days; ‘rainy’ days turn out to be cloudy days etc.

The trouble with forecasting that, our walking tells me, is more often partially right than totally right, is that it makes planning for the weather – from the layman’s point of view surely the most significant reason for checking the forecast – almost impossible. Which is why I’m so often chilled on ‘sunny’ days and char-grilled on ‘cloudy’ days, not to mention occasionally drenched on ‘dry’ days. 

Take it from me, there’s nothing more certain to turn you into a weather sceptic.

Got to go now – the sun’s coming out.

UPDATE: What actualy happened Monday
Metservice predicted sun and cloud; I predicted cloud would clear later in the day. It did. We were both right.

Media trainer and commentator Dr Brian Edwards blogs at Brian Edwards Media

Comments and questions

Agree with this. Yesterday was a fine example. Cloudy, even slightly foggy in Wellington yesterday afternoon and the Metservice had a bright full sun on the forecast. I'm looking at the website then looking up to see out of the window and they completely contradict what I'm seeing.

I remember this also happened when my parents were here some time ago. For three days the forecast said full sun, clear skies and for those three days I did not see any blue sky at all, it was complete cloud cover.

I know NZ weather is very changeable and they will never get it right all the time, but you're right. You can basically ignore any forecast more than two days out.

My pet cockroach kicks the Metservice's butt any day of the week.

Too much use of the words '"may" and "could" and the phrases "above or below average for this time of the year" sees this service consigned to the take it with a grain of salt brigade.

I agree totally agree with you, Brian. The Metservice predicted that Friday the 18th would be the best weather day in Masterton for the wonderful airshow so I went. It was the worst day. Grrrrrr...

Therefore, why does anyone believe in the 100-year temperature forecasts by weather experts?

100-year temperature forecasts are performed by climatologists - a different science.

Climatology has never been, and will never be, what is known as one of the hard sciences - like chemistry, physics, etc, with deterministic results. Chaos theory, the dynamics of non-linear systems, has proven that climatological forecasting will always be unpredictable. Ref. Edward Norton Lorenz, mathematician and meteorologist.

A funny sort of science that does not use the scientific method.
Falsifying the null hypothesis would be fundamental to showing that any warming has been caused by additional CO2.
But we can't do that without a validated climate model that accurately predicts temperature.
Of course, finding a correlation between the level of atmospheric CO2 and global average temperature would be a start, but we haven't even got that.
Science, you say?

I can remember, a little while ago now, a weather report for the next day during an eight-week drought: "Fine with, a 5% chance of rain." Talk about covering one's butt.

Instead of the TV weather lady starting her report by telling us what the weather was like today (we already know what is was like because we were there), I'd like her start by telling us how accurate her prediction from the previous day was.

Metservice is too busy being a typical government department and being too scared now to really tell the truth - too much PC, care of Chairman Helen and as now carried over by Key.
As any boatie/fishing person knows, the forecast is absolutely worthless, and has been for a long time. Time for Metservice to own up to playing safe rather than accurate.
Other countries can supply a accurate forecast. Heck, even Australia's Metservice has a more accurate forecast of NZ than ours.
There is no excuse. They are just slack. No wonder people get caught out here as no one believes a word they say any more.

Plenty of confirmation bias going on here. You remember when the forecasts are 'wrong' but don't really notice when they get it 'right'.

Unless you have robust data, with clear parameters for what constitutes a correct forecast, the above is simply an anecdote.

Yes, indeed. Unless the author can come up with some data to support his contention then this is so much biased waffle. Can't help feeling that the Metservice is unpopular with some of these characters because it correctly adheres to the man-made origin of climate change.

"Man-made origin of climate change." There is as yet no science which supports that contention.
There is conjecture, but nothing more than that.

Yet these same people and their friends at NIWA seem to think they can forecast "weather", ie climate, well into the future. Good lefties, like Brian, are happy to believe them, and make all manner of policy and tax changes based on this.

Instructive that the Met Service in the UK has backtracked on previous climate forecasts and is now predicting "no change" for the next half decade. Perhaps they just tossed a coin, then decided with even odds they would take the easy path, and avoid all the criticism that has come their way for their way off predictions of the past, driven by their fervent belief the climate was warming.

Perhaps they, too, have looked out the window, seen the snow (that was predicted a decade ago to become a very rare sight) and have decided the climate and weather will do its own thing and forecasting beyond what can be seen out the window is a fool's game.

Here's a forecast: "Tomorrow's weather is going to be the same as today."

I'm willing to bet that forecast has been more accurate than Metservice for any period longer than three months over the last decade.

What do you mean by "Tomorrow's weather is going to be the same as today"? The same amount of rain, sun, cloud, wind (both direction and speed) and at the same temperature at the same time each day? Or are you talking in loose, ambiguous generalities? You would certainly lose your bet for the former and the latter are called seasons.

Well Mr Weatherman-whose-feelings-have-been-hurt:

If its fine today it will be fine tomorrow. If it is raining today it will rain tomorrow.

That's more accurate than the meaningless rubbish (like your response) that you spit out.

What do you mean by fine? Just that it doesn't rain? Does cloudy count, or is that not fine? What if it doesn't actually rain but there is low cloud or mist which is precipitation? What if it rains for an hour, or more, and then 'fines' up for the rest of the day? I could go on but I think I've illustrated how meaningless your original comment was.

Other than that your Ad hominem arguments of your response are flawless.

I couldn't disagree more. All the information that I need to make vital planning decisions is clearly set out with probabilities. I rely totally on it, especially during hay making. But then, I never think that a few spits of rain during an otherwise fine day is a national disaster.
Townies :-)

This is the typical kind of criticism by people who have absolutely no idea what they are talking about, nor any kind of scientifically supportable evidence to back up their claims. It is entirely unfair to criticise when you don't understand the science behind weather forecasting.
I would claim the opposite - that it is remarkable that the forecast is often very close 4-5 days out. What else in life can you predict with any accuracy? Those satellites that you are talking about are not time machines that can see weather in the future.
If either the author or any of the other wise people commenting here have a better method of weather forecasting than is currently used by Metservice then you should offer your services, or set up another agency.
And no, I don't work for Metservice, but I do appreciate the difficulty of what they do.

Never mind forecasts for five or 10 days out as they're presumptuous tosh and as reliable as a Kathmandu waterproof jacket.

My bugbear is the use of vague terminology for next-day forecasts. When is "the evening" in mid-summer? Any time between 5 and 10pm? When is it in mid-winter? Does "in the morning" mean any time from midnight to midday, or between sun-up and midday, or .....? Is "mid-morning" at 6am?

I prefer the Californian (and probably all USA's), use of percentages. 10% chance of rain, 30% chance of cloud, 60% chance of sunshine. So you carry 10% of an umbrella, just in case.

Have a look at a Metservice rural forecast. Or their seasonal forecast.

Really Brian? Have you got nothing better to write about? We are weeks aways from the Labour Party leadership coup and you are fluffing about the weather. Have you been spending time with Lindsay Perigo? ere's an idea for a story. What about writing about the United Nations Development Programme projects and how poorly they are doing under the stewardship of H1 & H2?

Some Metservice employees have been down the comments by the looks of it.

Climate, or long-term weather, is a chaotic non-linear system with endless variables.

The data (satellites) and its processing (computer models) has advanced enormously in recent years but the strike rates of predictions are no better than they were 50 years ago.

Those strike-rates are about 50% out to 4-5 days and then fall precipitously.

Predictions of the 2012 global average temperature made 20 years ago have proved no more accurate than a chimpanzee throwing darts. Forecasts of steady detectable warming have been wrong for over 15 years, and scientists expect this cooling period to continue.

It's not the science that is at fault. It's the politicians who claimed that the prophesies were reliable and that we could make our own weather.

My first-ever science report was on the accuracy of the weather forecast over a one-month period during the late 1990s. Just temperature and I think rain prediction (it wasn't that flash). Only required a copy of each day's newspaper.

Good times.

Do you make this stuff up as you go along? Do enlighten us with an extensive list of those reputable climatologists who think "the cooling period" is going to continue. Each of the last 12 years (2001–12) features as one of the 14 warmest on record and yet still one hears this nonsense.

A cool period after a warm period will begin at the part of the curve with the highest temperatures, won't it?
But simply knowing the global average temperature after the addition of CO2 from fossil fuels tells us absolutely nothing if we don't know what the temperature would have been without those additions of CO2.
And nobody has the faintest idea what that temperature would have been.
We don't have a validated climate model which accurately predicts global average temperature.
So how do we know that additional CO2 is causing anything?
We don't yet. Maybe in 100 years we will be able to do it.

Consider the impact on tourism and the economy. My prediction is that Metservice will forecast cr*ppy weather for the coming Auckland Anniversary long weekend, and more than a few unwitting punters will cancel their travel plans as a result.

You are wrong. The forecast is for an extended period of fine weather. Perfect for haymaking in the North Island until the middle of next week.

The problem is that with global warming the seaweed is too dry and difficult to read. Also, with budget cuts they have to use the same piece all week.

Brian, love your columns but please try to be a little less "Jaffa". The Metservice success rate outside Auckland is a very good one. As a sailor who depends on accurate weather forecasts, I think they do very well.

Who was it who said that weather forecasters go through the same training as economists, because they rarely get it right, either.