And What Are Your Error Sizes?

You gotta wonder with a story like this:

The new estimate shows that polar melting contributed about one-fifth of the overall global sea level rise since 1992; other factors include warming that causes the seawater to expand.

The study does not seek to forecast future change.

Supported by US and European space agencies Nasa and Esa, the research brought together data from satellites measuring the surface altitude, the flow of the glaciers and the gravitational effect of the ice mass to produce the first joint assessment of how the ice sheets are changing….The study’s headline conclusion is that the polar ice sheets have overall contributed 11.1mm to sea level rise but with a “give or take” uncertainty of 3.8mm – meaning the contribution could be as little as 7.3mm or as much as 14.9mm

The combined rate of melting from all the ice sheets has increased over the past 20 years with Greenland losing five times as much now as in 1992.

I wonder about that 3.8mm error number, I mean Hurricane Sandy made the tide rise between 8-11 feet HIGHER than normal. That’s a massive low pressure system and a massive surge, but think of how little it would take to account for 4mm of ocean difference. When the Earthquake hit Costa Rica there were a ton of landslides and cliff faces that fell into the ocean. Hell it makes me wonder how stable the Moon’s orbit is, when it comes to planetoids 4mm is a rounding error.

I know a bunch of people who are dieting who weigh themselves every day and will get excited or depressed over a half-pound change in the numbers. But the human body fluctuates so much with food and water weights that even a pound or two can fluctuate in a day. I personally only call victory or defeat in increments of 5lbs. If I gain or lose 5lbs from a previous weight that’s from my total body mass, not my blood volume or gut-contents.

Still in the end, even if they’re right the best they can claim is 14.9mm in sea level change in 20 years time. Here’s a fun exercise. Pretend an Alien Stealth Craft sneaks into the Pacific ocean and dumps its cargo undetected into the sea. Whatever this may be, its enough to raise the sea level TWENTY millimeters! TWO WHOLE CENTIMETERS! ALMOST AN INCH!!!

How long do you think it will take people to figure out that it happened?

uh huh!

This entry was posted in Biology. Bookmark the permalink.

11 Responses to And What Are Your Error Sizes?

  1. Jack says:

    Now take this error and uncertainty analysis and apply it to temperatures.

    How do you measure temperatures and how do you get a “representative” temperature. And how does said temperature track with previous temperatures.

    And this is just getting good observable data.

    Model validation is a whole other kettle of fish.

    And which is why climatology at best operates with a foot in a bucket. Like astronomy, there’s no experimental data. No calling up McMaster and ordering a pallet load of earths to test.

    And before anyone pipes up about “computer modeling does let us have a ‘crate of earths’ ” how do you think those computer models are *validated*?

    But unlike astronomy, which has a whole universe to get observational data, climatology is limited to one earth like planet and some very, very limited temperature data on Mars, Venus, and Mercury.

    Having a global warming mode is not like having a metallurgical model or an aerodynamic model. You can’t just run a test on a bunch of steel rods or put your new wing design in a wind tunnel.

    There’s no, if I vary X in my model how do my predicted results compare to the results if I run the experiment for realsies.
    And that’s kinda important. Scientifically speaking.

    Oh but there I go again, questioning the methodology and rigor of Climate Science, how “unscientific” of me.

    • Weerd Beard says:

      If I called this comment “Fantastic” or “Best Ever” I’d be selling it short. So well said, man!

    • Weerd Beard says:

      If I called this comment “Fantastic” or “Best Ever” I’d be selling it short. So well said, man!

    • Geodkyt says:

      There IS a way to validate the models. You feed them raw data for various scenarios where you KNOW the historical temps through some kind of observational method. You do not allow any “tweaking” of the data to force-fit it. See if the model accurate “predicts” the results that you have already measured.

      Of course, when you do that, pretty much ALL the long-range climate models (for anything remotely habitable by known life) that rely on atmospheric composition fall right the Hell apart. Climate models based on solar output and tectonic plate effects on ocean currents tend to work — of course, neither of those is remotely susceptible to human intervention, and one of them operates in geological timeframes so slow that it is effectively a non-issue.

      • Jack says:

        Indeed they do!

        And that’s also a weak form of validation.

        As your inputs and outputs are limted to different portions of a single runtume.

        Even if the models *could* make predictions that matched existing data, they would still be very limited in that they could only predcit phenomina within a narrow band.

  2. Stuart the Viking says:

    Last I read (admittedly a few years ago) the margin of error on the satellites that measure the sea level was somewhere around a FOOT. Hey, maybe they have gotten better satellites since then, but they were using those satellites at the time to “measure” increases in sea level in the range of inches. Unless they’ve come up with something MUCH more accurate, measuring something like 14.9mm (particularly the tenths of a millimeter?) is so amazing that it is really hard to believe.


    • Jack says:

      That makes me wonder if they had a whole bunch of +/1 1mm measurements (which in and of itself is breathtaking 1mm is SMALL. Water tension alone will play havoc on getting a reading to that level) and they simply averaged ’em and kept the 0.9 because it looked better than rounding to 15

      Basic data lesson” You cannot AVERAGE data and come up with a *higher* accuracy than your source data. An average of {15, 15, 15, 15, 14, 15, 15, 15, 15, 15} does *not* equal 14.9

      If it did, I could another measure of 15 into the mix and get an average with infinite accuracy! (14.909090…)

      In fact, you cannot increase the accuracy of your observational data in any way shape or form. If your gauge only goes to three significant figures then you can’t go around claiming you got a bulk temperature of 435.2 K

      This also shows how vital rigor and transparency is in science. Because it is very, very easy to cut corners and shave the edges. Especially if you have a non-falsifiable model built on opaque data that is “reviewed” by complicit peers and distributed by credulous reporters.

    • Weerd Beard says:

      On a much simpler example I was given a cohort of mice and a digital scale to weigh them. The scale read down to a tenth of a gram…only with my mice on the scale the round numbers were reasonably stable, but the tenths place just gave me number-salad.

      So I marked down the round number and moved on.

      Of course somebody asked me to record to the tenths. I noted with the the equipment I’d simply be guessing. They told me that was fine. (Sigh).

      At least for this experiment the results are broad enough that this little error won’t make-or-break.

  3. Bubblehead Les says:

    Ah, but we need to dig into some of the data. Notice that they said that Polar Ice Melt has contributed to 20% of the Sea Water Rise. So that means there was about 55 mm Total Rise over 20 years. That means that it averaged about 3mm a year. So, to get 4-5 feet of Sea Water to rise to flood the East Coast and put Manhattan under water, using their numbers, it should take about 8 years to rise an inch. Or 96 years to rise a Foot. Or about 400-500 years to sink Wall Street. IF the rate stays steady.

    But I thought Algore said that we’d be under water in about 40 years. Hmm. Seems someone can’t add.

    That is, if one believes the Data in the First place. Yet how many Billions of Dollars has been spent so far to tell us that WE’RE ALL DOOMED!

    Stupid Greens.

  4. Cargosquid says:

    Except for one small problem…

    There hasn’t been any polar melting. The North pole could melt completely and not raise the sea level one inch. The Antarctic has been GAINING ice. Greenland has not lost noticeable mass.

    If the sea level is rising at 3mm per year…then HEAVENS! OH NO!
    Its right on track. Sea level has been rising at about that rate for thousands of years.

  5. Archer says:

    Don’t forget “normalized” data. This is the phenomenon where the climatologist says the average temperature at X location for the last 10 years, in order, is {15, 15, 14, 13, 13, 11, 12, 11, 10, 10} (trending overwhelmingly downward), but when “normalized” shows an increase of 3 degrees over that span.

    At which point my brain goes, “What. The. F*%k.”

    I’m becoming convinced “data normalization” is the pseudo-scientific code term for, “The data didn’t fit our model, so we changed it.” If I hear data has been normalized, the study becomes suspect unless they justify EXACTLY why it needed normalization and explain their methods of doing so.

    Come to think of it, this would also explain how the antis can take Linoge’s “Graphics Matter” numbers and conclude that 1) “More Guns = More Crime”; 2) the overall firearms ownership is declining; and 3) the gun industry is dying.

Leave a Reply

Your email address will not be published. Required fields are marked *