# Munging The Sea Level Data

Guest Post by Willis Eschenbach

Guest Post by Willis Eschenbach

mung
/mənj/ [pronounced “munge”]
verb
INFORMAL•COMPUTING
gerund or present participle: munging
to manipulate (data)
EXAMPLE: “you could do what anti-spammers have done for years and mung the URLs”

For more than a decade now, I’ve been wondering about a couple of questions.

First, why does the satellite-based sea-level data show that the sea level is rising so much faster than the rise measured at tidal stations on the coastlines around the world? Records from tidal stations show a rise on the order of a couple of mm per year, a rate which is little changed over the century or so for which we have adequate records. But the satellite record (Figure 1) shows a rise of 3.3 mm/year. Why the large difference?

Second, why does the satellite-based sea-level show such significant acceleration? As mentioned above, the sea-level records from tidal stations, which are much longer, show little or no acceleration. But the satellite record claims that the rate of sea-level rise is increasing by about a tenth of a millimeter per year. That amount of acceleration would double the rate of sea-level rise in about thirty years. Again, why the large difference?

To start with, here’s what the satellite data says, according to the University of Colorado Sea Level Research Group.

Figure 1. University of Colorado sea level data, showing the trend and acceleration. Note that the graph shows no overlap between the individual satellite records. SOURCE: UCSLR Group

I got to thinking about these questions again this week, so I went to NOAA and got their Excel spreadsheet showing the full dataset for each of the four satellites.

I plotted up the NOAA data. But unlike the Colorado data above, I included the full overlap between the individual satellites. I then looked at the rate of sea-level rise shown by each of the four satellites separately. Figure 2 shows that result.

Figure 2. NOAA sea level data, showing the trend of each of the full individual satellite records and the overall trend. SOURCE: NOAA Excel Spreadsheet

Well, go figure … seems like after all these many years of wondering, I finally have an answer to my two questions. The trend is so large and the acceleration is so great for a simple reason. The two most recent satellites show trends that are significantly larger than the earlier two satellites. The first two satellites agree quite well, but they both show a much smaller trend than the latter two satellites. And neither half of the satellite record shows any acceleration.

Now I freely admit, I’m just a guy with no scientific training at all. I took Physics 101, Chemistry 101, and one year of Calculus in college, and that’s it. But the beauty of science is that it’s all about the facts and the evidence, and not about the qualifications of the person presenting the facts. What you see above are the facts as given by NOAA, and I’ve linked to the source of those facts above. And it seems to me that those facts are more than sufficient to entirely discredit the satellite sea-level record.

So if you have an explanation for those facts, fine. But don’t bother busting me because you think I’m not qualified to present and discuss the facts. “The Captain”, my great-grandfather who was a Mississippi riverboat captain, used to say “If you have to hang your diploma on your living room wall, there was something wrong with your education”.

And to return to the issues at hand, why does the NOAA data show an overall trend which is 0.3 mm per year smaller than the Colorado data? It’s because the Colorado data contains what is called the “GIA”, and the NOAA data doesn’t.

“GIA” stands for the “glacial isostatic adjustment”. There’s a description of it on the Colorado site here. It relates to the changes in the earth’s overall shape due to the melting of the huge and immensely heavy ice sheets from the last glaciation.

According to that page, the GIA is “independently estimated from models at -0.3 mm/yr (Peltier, 2001, 2002, 2009; Peltier & Luthcke, 2009)”.

Now, I’ve been programming computers since 1963, coming up on 60 years now. As a result, I’m very aware that a computer model is nothing more than a physical realization of the beliefs, understandings, and in particular the misunderstandings of the person programming the computer. Since Mr. Peltier has been intimately involved in programming all four of the model realizations listed on the Colorado site, their claim that the four models are “independent” is laughable. What we have is one man’s claim that he can calculate the GIA.

What effect does the GIA have? The Colorado site says (emphasis mine):

Prior to release 2011_rel1, we did not account for GIA in estimates of the global mean sea level rate, but this correction is now scientifically well-understood and is applied to GMSL estimates by nearly all research groups around the world. Including the GIA correction has the effect of increasing previous estimates of the global mean sea level rate by 0.3 mm/yr.

So there’s the missing 0.3 mm/year, applied by “nearly all research groups” … I guess NOAA didn’t get the memo.

For me, the claim that a computer model can calculate the changing of the total volumes of all of the world’s various ocean basins to the nearest 0.3 mm per year of sea level … well, let me just say that it strains credulity far beyond the breaking point and leave it at that. Here’s why.

For the land portion of the calculation, these “independent” computer models must be using GPS altitudes. By using split-phase GPS and repeated measurements, these can be as accurate as ± 10 mm or better, an amazing scientific feat … but that’s a far cry from a tenth of a millimetre, and that’s just on land. Not only that, but we don’t have GPS measurements all over the land. They’re concentrated in the US and Europe, with only spotty coverage elsewhere.

At sea, the “independent” models must be using satellite measurements, so we get into the question of the accuracy and precision of the satellite sea level measurements themselves. These measurements are done by bouncing radar waves off of the ocean surface and measuring how long it takes them to return to the satellite. Here, repeated measurements are not possible because the sea level at any point changes constantly, and the satellite is rarely in the same position twice. A recent NASA press release describing the Sentinel-6 satellite, the successor to the Jason-3 satellite, quotes the oceanographer Josh Willis:

Sentinel-6 will orbit about 800 miles up and use radar to measure the surface of the ocean. An instrument on the satellite sends a radar wave down to Earth. The radar bounces off the surface of the ocean and returns to the satellite. By measuring how long it took for the radar to go down and back — and accounting for moisture in the atmosphere which slows the radar down — scientists can measure how far away the surface of the ocean is from the satellite.

In other words, the satellite can tell scientists on Earth how high the oceans are, and how that height is changing over time.

“It’s really kind of an incredible feat of technology,” [Josh] Willis says. “We can accurately measure the water level with an accuracy of 1 inch from 800 miles up.”

An accuracy of 1 inch, that’s 25.4 mm … and they’re claiming they know the annual change in the volume of the oceans from the GIA to the nearest tenth of an mm of sea level height? I know that accuracy is different from precision, and that measurement of changes in length over time (precision) can be an order of magnitude better than the measurements of the length itself (accuracy), but still … sorry, but I’m not buying the GIA claim.

Let me see if I can give you a sense of the difficulty of the satellite measurements of sea level. The satellites orbit at an altitude of about 830 miles, which is about 1.34 billion millimetres. So to measure the change in sea level to the nearest tenth of an mm, we’d need to be able to measure the distance from the satellite to the sea surface to a precision greater than one part in thirteen billion … and that is a hard challenge even in a controlled laboratory setting.

Here are some of the difficulties in that measurement. First is the measurement of the altitude of the satellite itself. Unless we know that to the nearest mm or so for every second of every day, we’re going to get inaccurate answers. Next is the varying composition, temperature, cloudiness, and humidity of the atmosphere. All of these change the time it takes the radar signal to return to the satellite. Then there are the ocean waves, which obviously change the height of the ocean by thousands of mm. Finally, there is “instrument drift”, the inevitable changes that occur to electronic measuring instruments over time.

Net result? Well, the net result can be seen in Figure 2 above, where according to the University of Colorado one satellite says the sea level is rising at 2.5 mm/year, and a mere 8 days after the end of that satellite’s data (the interval between one satellite and the next in the Colorado sea-level record) the successor satellite says it is rising at 4.1 mm/year … no bueno. They’re claiming that in 8 days, the rate of rise jumped by 1.6 mm per year. Note also that there is absolutely no acceleration in either half of the satellite record, just a step-change between satellites. Clearly, they’re not able to measure annual sea-level changes to the nearest millimetre, much less to the nearest tenth of a millimetre.

However, the people working on the project are all true believers. In the same article, Josh Willis is quoted as saying “We know that the oceans are rising because of human-caused interference with the climate.”

Dang humans, always interfering with the climate … for example, our new “Climate Czar”, John Kerry, has 5 houses, a number of cars, a yacht, and a private jet, and he tells us to take the bus to avoid the dreaded “human-caused interference with the climate” … but I digress.

The problem is that starting out with a fixed “scientific” belief like that leads to the people working on the satellite sea-level datasets splicing together what are obviously incompatible satellite results, spreading peanut butter over the cracks so they can’t be seen, and announcing to the world that the satellites show a dangerous acceleration in sea-level rise, so we should all be very worried …

Me, I’ve said for some time that we shouldn’t put any weight on the satellite results. However, I have based this solely on the very large differences in both trend and acceleration between the satellite and the tidal station records, and the known difficulties in satellite measurements discussed above. I investigated this question in several posts, including “Inside The Acceleration Factory” and “Accelerating The Acceleration

But now, at long last, I have the facts to back up my claim. There’s no evidence of any acceleration in the rate of sea-level rise in either the tide gauge or the shabbily-spliced satellite records. It’s been going up at on the order of eight inches (200 mm) per century for quite some time, and there’s no sign of any change in that rate of rise.

So you’re free to do what Obama and Bill Gates have done—buy seaside property. They proclaim loudly that the sea level is rising dangerously, but like the majority of climate alarmists, their actions belie their words.

My very best wishes to everyone in these most strange of times,

w.

PS—My usual request. To avoid misunderstandings, please quote the exact words you are discussing. I can defend my words, but I cannot defend your understanding of my words.

PPS—My post linked to above, “Accelerating The Acceleration“, earned me a laughable “fact-check” on Facebook by some well-meaning folks who were apparently short on cranial horsepower … now that Facebook has announced it’s taking over as the global arbiter of scientific truth, we’ll see what happens to this post.

Article Rating
Inline Feedbacks
Bill Powers
February 21, 2021 2:15 pm

Willis at these most strange of times Oceania is no longer at war with Eurasia. We are now at war with Eastasia. It is important to note that we have always been at war with Eastasia. You can consult Facebook to confirm.

Joel O'Bryan
February 21, 2021 3:55 pm

May a rat chew-off your nose in Room 101.

• Sincerely, The Ministry of Love.
Dave Fair
February 22, 2021 10:12 am

1984 version of cancel culture.

Waza
February 21, 2021 2:28 pm

Willis
Thanks for the article.
The first thing that struck me was the willingness to do a quadratic fit for 30 years of data.
Maybe alarmists can provide a similar quadratic fit for, the last century of CO2, Temperature, and SLR. On the one graph would be nice.

Chris Nisbet
February 21, 2021 2:28 pm

Why the discrepancy?
I just assumed it was due to scientific fraud.

Joel O'Bryan
February 21, 2021 3:49 pm

The on-going climate scam is like observing a magician’s magic trick as a child in awe versus understanding as an adult how the sleight of hand maneuvers are being made to pull off the deception.

And then having to do that unraveling for the dozens of different deceptions that are being employed, deceptions from what is going on with SLR claims to instrumental temperatures-derived anomalies to dozens of supercomputer models. All an elaborate fraud on the public. But when Trillions of dollars are at stake, one can see the motivations.

Last edited 1 year ago by Joel O’Bryan
Geoff Sherrington
February 21, 2021 9:37 pm

There was a neat analogy like your magic trick in the climate blogs the other day.
“The bull in the ring sees only the red cape, not the man behind it with the sword.”
Cannot find it again to attribute. The original had some Spanish in it.
There is a lot of this going on, two different mind sets, one viewing measurements and deductions, the other viewing beliefs and imagination. Geoff S

philincalifornia
February 21, 2021 7:36 pm

That would be a good assumption.

H. D. Hoese
February 21, 2021 2:33 pm

“But the satellite record claims that the rate of sea-level rise is increasing by about a tenth of an mm per year.” In listening to NOAA weather radio for coastal Texas blizzard data, I got a lot of information about local sea levels to two decimal places, possibly my cold brain.

All sea level modelers must first measure sea level from a real ocean vessel, let them have a calm sea even, also view of coastal tide gauge, I’ll wage that their error will be much greater than they model by at least an order of magnitude. Where did they learn about significant figures?

Alan M
February 22, 2021 3:59 am

As is usual I went for my morning beach swim here in Perth (Western Australia), beautiful morning including a dolphin in the waves less than 4m away. Light to moderate offshore breeze as is common this time of year with a small maybe 0.8-1.0m swell. The “ruffle” from the offshore was probably of the order of 5cm and with that small swell can you tell me a satellite that many km above can estimate the sea-level to mm, what about in the afternoon when we get our famous sea breeze where the “ruffle” can be greater than the swell.
As Darryl would have said ” Tell him he’s dreamin’ “

Fred Hubler
February 21, 2021 2:37 pm

From what I’ve been reading there was a steric sea level rise do to the recent very strong El Nino, but since sea level has barely budged in over a year, both NASA and the Univ. of Colorado have stopped updating their graphs. The Univ. of Colorado once said they would update their graph every 10 days, but haven’t updated it for at least 2 months. NASA hasn’t updated theirs since Sep 2020. Ten days is the time it takes to cover the whole Earth once.

https://climate.nasa.gov/vital-signs/sea-level/

Rud Istvan
February 21, 2021 4:03 pm

FH, overall ocean steric rise should not be much affected by Nino/Nina. Those are just where the warm water pool is, western or eastern equatorial Pacific. Will affect local tide gauges, but not the overall Pacific let alone all oceans.

Dave Fair
February 22, 2021 10:17 am

The great and mighty Government Con kills those that bring bad news.

Abolition Man
February 21, 2021 2:42 pm

Willis,
I am very concerned that your continued posting will have the modern day Torquemadas howling for your blood! Please exercise extreme caution and be wary of anyone offering to BBQ some steaks for you; check their spelling!
What a joy it is to have a site with such informative authors and interesting, albeit sometimes snarky, commenters; it makes dealing with the griffters and trolls bearable!

Last edited 1 year ago by abolition man
Rud Istvan
February 21, 2021 2:43 pm

Nice post, WE. For those interested, I have posted concerning satalt—fit for purpose? examining specificallyJason 3, and separately on its new replacement Sentinal-6, and separately on the lack of observational acceleration. These issues and more were also covered years earlier in essay PseudoPrecision in ebook Blowing Smoke.

There is a second way to prove the satalt SLR ismunged. Closure is the idea that SLR must close with thermosteric rise (calculable from ARGO) and ice mass loss, calculable from satellites ICESat and GRACE. (Other contributions are negligible, see PseudoPrecision for some laughable attempts to claim they are not. Closure is about 2.2-2.3mm/year, nowhere near satalt. Covered the specifics here in guest post Sea Level Rise, Acceleration, and Closure a while back.

Fred Hubler
February 21, 2021 2:46 pm

I believe the GIA adjustments around 2011 and the Univ. of Colorado openly stated that it was done to measure change in sea volume, not sea level.

stinkerp
February 21, 2021 2:50 pm

Don’t forget that the precision of the instruments on those satellites is on the order of centimeters. For Jason-3 it’s about 4 cm. That’s 40 mm. And yet they claim to measure sea level trends to a tenth of a millimeter, three orders of magnitude more precision than the instruments can measure.

There are three kinds of lies: lies, damned lies, and statistics. You can see from the Jason data why climate alarmists love statistics.

Last edited 1 year ago by stinkerp
MarkW
February 21, 2021 6:27 pm

2 orders of magnitude, not 3. It’s still an absurd claim.

Brian Jackson
February 21, 2021 8:21 pm

Both you and stinkerp don’t understand the difference between precision and accuracy.

MarkW
February 21, 2021 8:47 pm

You keep using those words, but you so obviously don’t know what they mean.

stinkerp
February 21, 2021 11:17 pm

An instrument that is calibrated to measure to a tenth of a millimeter is more precise than one that can measure only to a centimeter (100 tenths of a millimeter). Taking hundreds of thousands of measurements from both instruments and averaging them will show that the one with greater precision greatly improves accuracy. When the error range is a hundred times larger than the precision you claim, you’re doing voodoo, not science.

Average sea level rise of 3.3 mm/yr ±0.4mm is laughable coming from instruments that can’t measure any more precisely than 40mm. It should be 3.3 mm/yr
±40mm but then everyone would be in on the joke.

fred250
February 22, 2021 2:29 am

That 4cm EITHER WAY,

When you are pretending to measure in mm, that is like having a target on the side of a barn.. and mostly missing the barn.

Neither accurate, nor precise.

More like RANDOM CHANCE.

Last edited 1 year ago by fred250
Paul Penrose
February 22, 2021 2:23 pm

I have to agree with Brian Jackson on this one, you got them reversed. Accuracy is a statement of how far off the real value the measurement is and precision is how much the measurement varies when repeatedly measuring the same value. Accuracy is generally limited by physical characteristics of the measurement device, for example an analog to digital converter has only so many bits with which to represent the measurement (usually a voltage). Precision is basically the noise/error in the system and generally is the combination of the tolerance of all the parts in the measurement system.

Think of it this way, when throwing darts, how close you get to the center is your accuracy. The spread of all your throws is your precision.

TC in the OC
February 23, 2021 1:08 pm

All good but you have to remember that the overall error of the instrument calibration needs to be factored into the multitude of measurements as it just doesn’t go away.

“Accuracy is the closeness of agreement between a measured value and a true or accepted value. Measurement error is the amount of inaccuracy.

Precision is a measure of how well a result can be determined (without reference to a theoretical or true value). It is the degree of consistency and agreement among independent measurements of the same quantity; also the reliability or reproducibility of the result.

The uncertainty estimate associated with a measurement should account for both the accuracy and precision of the measurement.” (From the College Physics Labs – Mechanics UNC Chapel Hill)

My crews use GPS for measuring both horizontally and vertically. To get a really good reading they will let it cook for a period of time over the point being measured. We also will use a GPS network and/or a base station to improve accuracy. Doing all that makes the horizontal location nice and tight but the vertical is still +/- 0.1 feet. Really don’t see how their satellite is more accurate that a group of GPS satellites, networks and base stations.

We do construction staking and never use GPS for vertical staking only hard elevations run with a precision level from a known benchmark. The graders do a good job grading the dirt with GPS but then our tolerance for acceptance and approval of the grading is +/- 0.1′ which is from +0.149 to -0.149 or just about 0.3′.

I really wanted to get into all of the error theory but it takes a long time to understand (and it still makes my head hurt). Let me just say that this article points out differences in measurements between satellites and the trend is always going up. We see the same in Lord Monckton’s recent article about HadCRUT5. What these articles show me is that there are systematic errors in their measurements. By definition (same site)

Random errors are statistical fluctuations (in either direction) in the measured data due to the precision limitations of the measurement device. Random errors can be evaluated through statistical analysis and can be reduced by averaging over a large number of observations (see standard error).

Systematic errors are reproducible inaccuracies that are consistently in the same direction. These errors are difficult to detect and cannot be analyzed statistically. If a systematic error is identified when calibrating against a standard, applying a correction or correction factor to compensate for the effect can reduce the bias. Unlike random errors, systematic errors cannot be detected or reduced by increasing the number of observations.

Now this is all fun but you also need to deal with calibration of the instrument, atmospheric conditions (they do mess with GPS a lot) and also parallax and instrument drift and lag time etc.

And that doesn’t even cover estimating uncertainty, standard deviation and propagation of uncertainty and so much more.

I am so glad I don’t get tested on this yearly to keep my license.

The following is a link to the Chapel Hill article…enjoy. Oh and this is a quick summation of the actual course book.

https://www.webassign.net/question_assets/unccolphysmechl1/measurements/manual.html

garboard
February 21, 2021 9:46 pm

isn’t Josh Willis the same guy behind the Argo buoy temperature adjustments ?

February 21, 2021 10:06 pm

Yes – the very one.

Curious George
February 22, 2021 8:02 am

Have heart. While an individual measurement has an accuracy of 4 cm, you take 10 of them for an accuracy of 4 mm, and ten thousand for an accuracy of 4 micron. The sky is the limit 🙂

Paul Penrose
February 22, 2021 2:26 pm

I hope that was sarcasm because as we all (should) know, there is no mathematically valid way increase accuracy. Accuracy can only be improved by making better instruments.

commieBob
February 21, 2021 2:50 pm

The reason they think they can get remarkably accurate results is the power of averaging a huge number of samples. If your ‘noise’ is randomly distributed, that technique is amazing. It lets me pull a real clean signal out from under 20 db of noise.

On the other hand, if you’ve got red noise, averaging doesn’t help things at all. Given that red noise is common in nature, ‘their’ claims of accuracy are, at very least, highly suspect.

Geoff Sherrington
February 21, 2021 4:44 pm

cB,
Many statistical methods require data to be IID, random variables that are independent and identically distributed. It is not rare to see this fundamental condition precedent violated in climate research. Geoff S

Paul Penrose
February 22, 2021 2:29 pm

Yeah, it’s too bad you can’t improve accuracy by averaging (or any other mathematical technique). Precision can be improved by simple averaging, but as Geoff points out, the data must be IID or you have to adjust for the serial auto-correlation, and that’s not easy.

OK S.
February 21, 2021 2:51 pm

So you’re free to do what Obama and Bill Gates have done—buy seaside property. They proclaim loudly that the sea level is rising dangerously, but like the majority of climate alarmists, their actions belie their words.

Do you know of any way I (or you, also) can get onto the international gravy train to pay for these extravagances? 🙂

Edited to add the smiley face, just in case.

Last edited 1 year ago by OK S.
Dave Fair
February 22, 2021 10:32 am

There are four ways to get on the gravy train: 1. Be born well; 2. Marry well; 3. Provide something of value to people; or 4. Be part of a huge scam. You can pick the numbers yourself for both Obama and Gates. All of the prominent CliSci profiteers fall in one of the four.

Stevek
February 21, 2021 2:52 pm

Seems to me it would be easier to estimate the net changes in glacier volume over last few decades and use that to predict sea level change. Though water can come from elsewhere but surely most would be due to glacier volume change.

Mike Lowe
February 21, 2021 3:11 pm

Aren’t there several other explanations for SLR? I suspect that the volume of melting glaciers would be minute in comparison.

Stevek
February 21, 2021 3:36 pm

The 2 biggest factors are the ice melting and expansion of water that warms. This is why the warmers really want to see sea level rise increasing.

Editor
February 21, 2021 6:58 pm

Lake Eyre in Australia has been blamed for lowering sea levels one year.

Roger Knights
February 21, 2021 11:40 pm

Say, why don’t we run pipes from the sea (or Queensland’s rivers) to Leke Eyre, or from the Mediterranean to the Libyan depression, and fill them both, lowering the sea level considerably?

RelPerm
February 22, 2021 9:29 pm

Also refill Dead Sea, Salton Sea, Great Salt Lake… There are lots of places to store large volumes of water.

Larry Hamlin
February 21, 2021 2:53 pm

Excellent article. I am puzzled by the University of Colorado satellite sea level rise average rate showing a +/- amount of +/- 0.4 mm/yr. while NOAA shows a +/- amount varying much less with an average of +/- 0.02 mm/yr. Why the huge difference?

Rud Istvan
February 21, 2021 6:04 pm

NOAA lies with more precision?

Brian Jackson
February 21, 2021 8:23 pm

ROTFLMAO…….you also are confused regarding the difference between accuracy and precision!!!!!!!

MarkW
February 21, 2021 8:49 pm

One bad example, and the poor fool actually believes himself to be an expert who must never be questioned.

Lrp
February 22, 2021 1:11 am

Do you mean NOAA lies with more accuracy?

fred250
February 22, 2021 2:33 am

Satellite SLR measurements HAVE NEITHER.

Try not to confuse maths and statistics, for your ideology, BF !

Paul Penrose
February 22, 2021 2:31 pm

Brian,
I think Rud knows the difference between accuracy and precision, but is being snarky in implying that NOAA does not.

Rud Istvan
February 22, 2021 2:40 pm

Nope. I just forgot the sarc tag. See my guest on Jason 3, which actually illustrated using shooting targets the difference. To sum that, precision is tight grouping, accuracy is on the bullseye. BOTH is how you win a rifle competition.

Paul Penrose
February 23, 2021 9:49 am

Rud,
I think the most important distinction is that accuracy is applies to single measurements and is generally constrained by the physical properties of the measurement system. Whereas precision is a description of the error in the system after making multiple measurements. Generally this error is from environmental noise and is independent of the accuracy, although they may be related. Because of this, mathematical techniques may be employed to characterize and nullify some of the error. While this can result in an overall better measurement, it can’t improve it beyond the physical accuracy of the instrument.

Take for instance your rifle competition example. No matter how good a shot you are and how much environmental noise (wind, etc.) you control for, you are still limited by the quality of your rifle and ammunition. You simply can’t improve the accuracy of rifle with a badly machined or worn barrel. There is no way to tell beforehand, or compensate for how far off each shot will be. You can characterize the accuracy by taking a lot of shots and then measure the maximum spread, but that’s it.

In a measurement system, the accuracy describes the limits of your knowledge about what is being measured. For example, an analog to digital converter has only so many bits. If you divide the range of voltages it can convert and divide by the number of bits in the result, you get the accuracy in volts per bit. No matter how good your electronics are and how stable the voltage source is, you can never measure that voltage at a better resolution than one bit. You can extrapolate, but that’s really just a guess. You can’t know because information was lost during the conversion. All physical systems are limited in some way, so while the mechanisms change, they all have a physical accuracy which can’t be changed without improving the physical instrument.

Brian Jackson
February 21, 2021 2:54 pm

[deleted per request. w.]

Last edited 1 year ago by Willis Eschenbach
Brian Jackson
February 21, 2021 3:18 pm

Moderator, please delete the above comment, it’s formatted correctly below.

Gordon A. Dressler
February 21, 2021 3:27 pm

“The point is that you have just measured the average height of
adult males to be somewhere between 5.745 and 5.755.”

Absolutely not . . . you have only measured the average height to be somewhere between 5 and 6 feet. “Averaging” the measurements to supposedly obtain “higher precision” assumes that you have, a priori, an accurate idea of the distribution function of the population over the range of measurement (i.e., is it:
Uniform Distribution
Binomial Distribution
Normal Distribution
Poisson Distribution
Exponential Distribution
T Distribution.
Chi-square Distribution.
F Distribution.
etc.?)

Applying statistics to data does NOT create new data.

Last edited 1 year ago by Gordon A. Dressler
Brian Jackson
February 21, 2021 5:24 pm

You confuse “precision” with “accuracy.”

MarkW
February 21, 2021 8:50 pm

Have you created a macro to just keep repeating the same inaccurate claim over and over again?

February 22, 2021 6:53 am

Are you related to Griff?

Gordon A. Dressler
February 22, 2021 7:45 am

Brian Jackson posted “You confuse ‘precision’ with ‘accuracy’.”

Absolutely not! My use of the term precision was correct. You can do any sort of “averaging” (correctly on not) and report the average value to as many decimal points as you want . . . in your case it was to three decimal places (i.e., 5.745, 5.755). You could just as easily have asserted (again, incorrectly) the average to be between 5.74518931642 and 5.75471582775.

In any case, in terms of accuracy the only thing that one could correctly assert in your example of the stick with (supposedly) accurately marked 1 foot intervals—and data recorded to only the nearest foot—is an accuracy of ±0.5 foot, no matter what data distribution is assumed over the 5 foot to 6 foot interval and no matter what “averaging” technique is used.

Dave Fair
February 22, 2021 10:37 am

“Lies, damned lies and statistics.”

Michael E McHenry
February 21, 2021 2:57 pm

It’s just like NOAA claiming it can measure global surface temperatures to 0.01/yr or even 0.001/mo.

Ian W
February 22, 2021 11:03 am

Or even worse NOAA claiming to provide the global 2M temperature anomaly back to 1850 to  0.01C when the Southern hemisphere had at best 100 observation sites mainly in Australia/New Zealand and South Africa. It was only a hundred years later, after the 2nd World War, that things improved and even then not so much. It is difficult to understand how someone can guess/homogenize/grid a temperature out of nowhere and claim an accuracy of a hundredth of a degree for the guess. Even with an observation from a Stevenson Screen taken by the observer at 1am with the sleet blowing down around their neck …. will not be accurate to any better than a degree.

Hotscot
February 22, 2021 11:41 am

My enduring refrain, the example of the Tea Boy being sent out into the rain/sleet/snow/baking sun to take measurements a ‘scientist’ should be doing. Even a height difference between the scientist and the Tea Boy has an influence.

Stevenson screens themselves were designed for local weather forecasting. They were never intended to be part of a global network. There are also several designs, at least one for the US and one for the UK, and we now have a mini version in Australia which, by some accounts, saw considerable jump in temperatures which was not allowed for. I think either Jo Nova or Jennifer Mahorasy have objected to this.

Canvas buckets were still being chucked over the sides of ships to no definable depth, to take sea temperature measurements in the 1960’s. Another task for the Tea Boy, or Cabin Boy in this case.

Willis’ excellent article justifies my unsubstantiated contention that even with multiple measurements taken by satellite over a small area of Oceans, the precision down to even 40mm would be stretching credibility. For example, what is the prevailing weather at any given moment doing; is the sea surface flat calm, choppy, violent, or just with a substantial swell?

We can average out anything we want but much like land surface temperatures, there may not be a square inch of all the Oceans that conform to the ‘average’ other than in momentary passing.

I also believe sea levels are different at each end of the Panama canal.

Brian Jackson
February 21, 2021 3:05 pm

Willis, this will help you understand how the satelittes can measure such small lengths.
.
Get a 10 foot long pole, and make marks on the pole at 1 foot intervals. Now use this pole to measure the height of 20,000 adult males, recording each measurment to the nearest foot. Sum the 20,000 measurments and you will get a number around 115,000 feet. You may get 114,900, or 115,100, but if you repeat the measurment you will get about 115,000.  114,900/20,000 = 5.745 and 115,100/20,000 = 5.755.  The point is that you have just measured the average height of adult males to be somewhere between 5.745 and 5.755.  That is pretty close to the real value of 5.75.  So with a device that measures to the nearest foot, you can measure the average height of adult male to +/- 0.01 feet. You can get more decimal places if you use 100,000 readings instead of 20,000.
.
Learn about the standard error of an statistical estimator.

Last edited 1 year ago by Willis Eschenbach
Brian Jackson
February 21, 2021 3:17 pm

TYPO: should be: “real value of 5.75.”

Gordon A. Dressler
February 21, 2021 3:35 pm

“The point is that you have just measured the average height of
adult males to be somewhere between 5.745 and 5.755.”
﻿
Absolutely not . . . you have only measured the average height to be somewhere between 5 and 6 feet. “Averaging” the measurements to supposedly obtain “higher precision” assumes that you have, a priori, an accurate idea of the distribution function of the population over the range of measurement (i.e., is it:
Uniform Distribution
Binomial Distribution
Normal Distribution
Poisson Distribution
Exponential Distribution
T Distribution.
Chi-square Distribution.
F Distribution.
etc.?)

Applying statistics to data does NOT create new data.

Last edited 1 year ago by Gordon A. Dressler
garboard
February 21, 2021 9:57 pm

is that like the statistician who drowned in water that averaged one inch deep?

Gordon A. Dressler
February 22, 2021 1:39 pm

On this subject, a very wise man said it best:

“It is the mark of an educated mind to rest satisfied with the degree of precision which the nature of the subject admits and not to seek exactness where only an approximation is possible.” —Aristotle (384 B.C – 322 B.C.)

Brian Jackson, take heed.

Gordon A. Dressler
February 24, 2021 2:09 pm

Let’s just put a lid on Brian Jackson’s assertion that one can obtain increased precision of measurement and increased accuracy by his Gedankenexperiment of determining average adult male height by using a large number of measurements with a pole just (accurately) marked with just a 5 foot index and a 6 foot index and (presumably accurately) visually rounding off each measurement and recording it to the nearest foot.”

Here’s how to mathematically show the fallacy of this Gedankenexperiment: let’s assume we add additional accurately positioned marks at 4.50 feet, 5.50 feet, and 6.5 feet and record each measurement both as Brian proposed (rounding and recording to the nearest foot mark) and also rounding and recording separately to the nearest half-foot mark.

In Brian’s case, there are only two bins to record data: the “5 foot bin” and the “6 foot bin”, but with the new modification there are now actually four bins (the “4.5-5.0 foot bin”, the “5.0-5.5 foot bin”, the “5.5-6.0 foot bin” and the “6.0-6.5 foot bin”) to record the measurement data for each individual.

Let’s examine two hypothetical, but entirely possible, cases for data records:

Case 1: the four bins are filled with these numbers in order of increasing height:
4,000, 8,000, 6,000, and 2,000, yielding a weighted average of 5.4 feet for the 20,000 samples, with this average being exactly the same using Brian’s two bin data collection and the modified four bin data collection.

Case 2: the four bins are filled with these new numbers in order of increasing height:
2,000, 10,000, 5,000, and 3,000, again yielding a weighted average of 5.4 feet for the 20,000 samples per Brian’s two bin technique (as expected since the 5 ft/6 ft bin numbers are identical to Case 1 at 12,000/8,000). However, average from the four bin data collection is now different . . . it is 5.5 feet or 0.1 foot greater, rounded to the nearest 0.1 foot.

Why did this happen? It is because the population distribution, when measured at sub-foot resolution, shifted slightly upward in Case 2 compared to Case 1.

So, this simple example of having a “calculated average” change from 5.4 feet to 5.5 foot (again, just considering four sample bins compared to the proposed two bin method) shows the supposed increase in precision obtained just from a large amount of samples is generally invalid (i.e., BOGUS). Likewise, a claimed accuracy of 0.1 foot or better for the two-bin averaging case cannot be mathematically justified.

QED.

Geoff Sherrington
February 21, 2021 4:53 pm

Willis,
Couple of years ago I saved this academic bafflegab to describe mathematics the way BJ seems to imagine.
“The “begetted” eighmess as the system-limit number of the nuclear uniqueness of self-regenerative symmetrical growth may well account for the fundamental octave of unique interpermutative integer effects identified as plus one, plus two, plus three, plus four, as the interpermuted effects of the integers one, two, three, and four, respectively; and as minus four, minus three, minus two, minus one, characterizing the integers five, six, seven, and eight, respectively.”
Geoff S

Brian Jackson
February 21, 2021 8:26 pm

You and Willis have similar styles of prose. 😀 😀 😀 😉

fred250
February 22, 2021 2:56 am

And you are prone to mathematical mis-comprehension.

MarkW
February 22, 2021 7:28 am

If being accurate qualifies as a “style of prose”, then yes they do.
Assuming you could dump the ego, perhaps you could learn from them.

Brian Jackson
February 21, 2021 5:06 pm

“I suspect that I may know more about the subject than you do. ”
.
I disagree.
.
I gave you a physical, concrete, reproducible example of using a measuring device with 1 foot marks to measure an average with 0.01 foot accuracy.
.
Not only is it true in theory, it’s used everyday in practice, especially in particle physics. If you disagree, please tell me what ruler is used to measure the diameter of a hydrogen nucleus?
.
” it assumes that we are repeating measurements using the same measuring stick.”
..
The same satellite passes over the same location multiple times, and generates a data point. They take a lot of data points at the same location to arrive at an average measurement. Not only is this done with one satellite, it is done with all of them. Ask Mr. Spencer about this for his calculation of UAH.
.
” In the real world, this is often not true.” & “it ignores the very real phenomenon of instrumental drift.” …. The satellite’s radar altimeters can be calibrated against known physical altitudes ON DRY LAND.
.
“human rounding errors.” All of the data is collected, stored and processed on computers. Seriously Willis…..please try harder.
.
You can claim “systemic” errors, but you will have to prove they exist over the multiple satellites measuring sea level.
.
“repeated measurements can give us an order of magnitude improvement in our precision,.” ……..wrong word, you confuse “precision” with “accuracy.” Standard error of a statistical estimator gives you higher accuracy to the actual population metric.

Brian Jackson
February 21, 2021 5:20 pm
Brian Jackson
February 21, 2021 7:35 pm

1) Please tell me where on earth you are going to find 20,000 adult males that are ALL  5’6″ and 6′ tall. Really, your claim that nobody in a 20,000 adult male sample will be 5′ 4″ tall? Get real and watch the Wizard of Oz movie. It is obvious to all, you have no concept of the term “random sampling.”

.
2) “I also pointed out that if one of the marks on the stick is incorrect, you won’t get your vaunted accuracy either. Seriously Willis, as a professed builder of his own home, you can’t tell me that a tape measure can’t make 1-foot markings?……if so, please, make sure any/all of the guests you invite into the home you built know this.;
.
2a) We know the diameter of a proton (discussed elsewhere…….pretty sure the marks can be make accurately)
.
3) Thank you, thank you, thank you, thank you for acknowledging standard error in statistical estimators: “Researchers have now measured this in experiments which are ten times more accurate than all previous ones.” LMFAO, they repeated the experiment with increased NOBS to obtain the accuracy. You simply don’t understand that quantum chromodynamics is a STATISTICAL theory. (remember you said: “. I may know more about the subject than you do” which you have just proven you don’t and that you use Google/Wikipedia for references based on your copy/paste).
.
4) “Not true in the slightest” && “back to that exact same spot are vanishingly small.” ….. please talk to Roy Spencer.
.
5) “they are calibrated at only a few points” you posted a link: ” http://file///Users/willis/Documents/%20Internet%20Downloads/remotesensing-10-01679.pdf ” which shows all of us how computer illiterate you are. You are not sharing the hard drive of your computer on the internet, so nobody knows WTF that link points to. Please stop.

6) Radar altimeters work BOTH over lakes and dry land. Do you know how radar works?
.
7) ” I was talking about the general issue of why repeated measurements may not help”…….yes, and I was trying to educate you as to how a radar altimeter with a resolution of cm can measure a population average to less than a mm.
.
8) ” THAT IS EXACTLY WHAT I JUST DID IN THIS POST”…. no, you failed. “none of the nine issues I raised have any validity regarding sea-level satellites”…….you don’t know the difference between precision and accuracy.
..
9) “If there is systemic error”……big if without evidence. Please remember in logic that a false premise allows any conclusion. You have provided no evidence of systemic error, so should I wait for you to post the evidence, or should we assume ther is no systemic error?
.
10) your claim “I suspect that I may know more about the subject than you do. ” has obviously been shown to be false.
1) Please tell me where on earth you are going to find 20,000 adult males that are ALL  5’6″ and 6′ tall. Really, your claim that nobody in a 20,000 adult male sample will be 5′ 4″ tall? Get real and watch the Wizard of Oz movie. It is obvious to all, you have no concept of the term “random sampling.”

.
2) “I also pointed out that if one of the marks on the stick is incorrect, you won’t get your vaunted accuracy either. Seriously Willis, as a professed builder of his own home, you can’t tell me that a tape measure can’t make 1-foot markings?……if so, please, make sure any/all of the guests you invite into the home you built know this.;
.
3) Thank you, thank you, thank you, thank you for acknowledging standard error in statistical estimators: “Researchers have now measured this in experiments which are ten times more accurate than all previous ones.” LMFAO, they repeated the experiment with increased NOBS to obtain the accuracy. You simply don’t understand that quantum chromodynamics is a STATISTICAL theory. (remember you said: “. I may know more about the subject than you do” which you have just proven you don’t and that you use Google/Wikipedia for references based on your copy/paste).
.
4) “Not true in the slightest” && “back to that exact same spot are vanishingly small.” ….. please talk to Roy Spencer.
.
5) “they are calibrated at only a few points” you posted a link: ” http://file///Users/willis/Documents/%20Internet%20Downloads/remotesensing-10-01679.pdf ” which shows all of us how computer illiterate you are. You are not sharing the hard drive of your computer on the internet, so nobody knows WTF that link points to. Please stop.

6) Radar altimeters work BOTH over lakes and dry land. Do you know how radar works?
.
7) ” I was talking about the general issue of why repeated measurements may not help”…….yes, and I was trying to educate you as to how a radar altimeter with a resolution of cm can measure a population average to less than a mm.
.
8) ” THAT IS EXACTLY WHAT I JUST DID IN THIS POST”…. no, you failed. “none of the nine issues I raised have any validity regarding sea-level satellites”…….you don’t know the difference between precision and accuracy.
..
9) “If there is systemic error”……big if without evidence. Please remember in logic that a false premise allows any conclusion. You have provided no evidence of systemic error, so should I wait for you to post the evidence, or should we assume ther is no systemic error?
.
10) your claim “I suspect that I may know more about the subject than you do. ” has obviously been shown to be false.
1) Please tell me where on earth you are going to find 20,000 adult males that are ALL  5’6″ and 6′ tall. Really, your claim that nobody in a 20,000 adult male sample will be 5′ 4″ tall? Get real and watch the Wizard of Oz movie. It is obvious to all, you have no concept of the term “random sampling.”

.
2) “I also pointed out that if one of the marks on the stick is incorrect, you won’t get your vaunted accuracy either. Seriously Willis, as a professed builder of his own home, you can’t tell me that a tape measure can’t make 1-foot markings?……if so, please, make sure any/all of the guests you invite into the home you built know this.;
.
2a) We know the diameter of a proton (discussed elsewhere…….pretty sure the marks can be make accurately)
.
3) Thank you, thank you, thank you, thank you for acknowledging standard error in statistical estimators: “Researchers have now measured this in experiments which are ten times more accurate than all previous ones.” LMFAO, they repeated the experiment with increased NOBS to obtain the accuracy. You simply don’t understand that quantum chromodynamics is a STATISTICAL theory. (remember you said: “. I may know more about the subject than you do” which you have just proven you don’t and that you use Google/Wikipedia for references based on your copy/paste).
.
4) “Not true in the slightest” && “back to that exact same spot are vanishingly small.” ….. please talk to Roy Spencer.
.
5) “they are calibrated at only a few points” you posted a link: ” http://file///Users/willis/Documents/%20Internet%20Downloads/remotesensing-10-01679.pdf ” which shows all of us how computer illiterate you are. You are not sharing the hard drive of your computer on the internet, so nobody knows WTF that link points to. Please stop.

6) Radar altimeters work BOTH over lakes and dry land. Do you know how radar works?
.
7) ” I was talking about the general issue of why repeated measurements may not help”…….yes, and I was trying to educate you as to how a radar altimeter with a resolution of cm can measure a population average to less than a mm.
.
8) ” THAT IS EXACTLY WHAT I JUST DID IN THIS POST”…. no, you failed. “none of the nine issues I raised have any validity regarding sea-level satellites”…….you don’t know the difference between precision and accuracy.
..
9) “If there is systemic error”……big if without evidence. Please remember in logic that a false premise allows any conclusion. You have provided no evidence of systemic error, so should I wait for you to post the evidence, or should we assume ther is no systemic error?
.
10) your claim “I suspect that I may know more about the subject than you do. ” has obviously been shown to be false.
1) Please tell me where on earth you are going to find 20,000 adult males that are ALL  5’6″ and 6′ tall. Really, your claim that nobody in a 20,000 adult male sample will be 5′ 4″ tall? Get real and watch the Wizard of Oz movie. It is obvious to all, you have no concept of the term “random sampling.”

.
2) “I also pointed out that if one of the marks on the stick is incorrect, you won’t get your vaunted accuracy either. Seriously Willis, as a professed builder of his own home, you can’t tell me that a tape measure can’t make 1-foot markings?……if so, please, make sure any/all of the guests you invite into the home you built know this.;
.
3) Thank you, thank you, thank you, thank you for acknowledging standard error in statistical estimators: “Researchers have now measured this in experiments which are ten times more accurate than all previous ones.” LMFAO, they repeated the experiment with increased NOBS to obtain the accuracy. You simply don’t understand that quantum chromodynamics is a STATISTICAL theory. (remember you said: “. I may know more about the subject than you do” which you have just proven you don’t and that you use Google/Wikipedia for references based on your copy/paste).
.
4) “Not true in the slightest” && “back to that exact same spot are vanishingly small.” ….. please talk to Roy Spencer.
.
5) “they are calibrated at only a few points” you posted a link: ” http://file///Users/willis/Documents/%20Internet%20Downloads/remotesensing-10-01679.pdf ” which shows all of us how computer illiterate you are. You are not sharing the hard drive of your computer on the internet, so nobody knows WTF that link points to. Please stop.

6) Radar altimeters work BOTH over lakes and dry land. Do you know how radar works?
.
7) ” I was talking about the general issue of why repeated measurements may not help”…….yes, and I was trying to educate you as to how a radar altimeter with a resolution of cm can measure a population average to less than a mm.
.
8) ” THAT IS EXACTLY WHAT I JUST DID IN THIS POST”…. no, you failed. “none of the nine issues I raised have any validity regarding sea-level satellites”…….you don’t know the difference between precision and accuracy.
..
9) “If there is systemic error”……big if without evidence. Please remember in logic that a false premise allows any conclusion. You have provided no evidence of systemic error, so should I wait for you to post the evidence, or should we assume ther is no systemic error?
.
10) your claim “I suspect that I may know more about the subject than you do. ” has obviously been shown to be false.
.
11) Thank you for acknowledging that I know more than you with regards to precision and accuracy.

Last edited 1 year ago by Brian Jackson
Brian Jackson
February 21, 2021 8:27 pm

WordPress sucks.

KAT
February 21, 2021 10:13 pm

Brian. Please explain how the tidal configuration of the sun and the moon complicates the average height of adult males!

MarkW
February 22, 2021 7:29 am

It’s operator error.

eck
February 21, 2021 10:58 pm

Willis, to quote Bugs B., “Moroons” abound!

BobM
February 21, 2021 11:31 pm

So, you would agree that the step changes in the satellite trend(s) shown by Willis accurately demonstrate that the last two satellites have not been measuring sea level correctly.

fred250
February 22, 2021 2:58 am

“that the last two satellites have not been measuring sea level correctly.”

Very good point. 🙂

Dave Fair
February 22, 2021 11:41 am

Gentlemen, please! Don’t distract the CliSci practitioners with observations. They know that sea levels are accelerating (their models say so) and any discontinuity between satellites can be smoothed to their satisfaction.

Jim Whelan
February 23, 2021 11:29 am

As I caught one warmunist saying, “the theory is so obviously true that measurements which disagree can be safely disregarded.”

Carlo, Monte
February 22, 2021 6:20 am

It is obvious to all, you have no concept of the term “random sampling.”

Sampling isn’t the issue, but rather population. The big stick scenario assumes the heights have a normal distribution. With different distributions, the average could end up as 5.00000 or 6.00000 feet. Precision is not resolution, which is the real issue here.

Clyde Spencer
February 22, 2021 10:21 am

BJ
You remarked, “Radar altimeters work BOTH over lakes and dry land. Do you know how radar works?”

I’m convinced that you do not know how radar altimeters work. They provide a calculated estimate of the distance between the satellite and the surface of interest, based on time of travel of the pulse. The problem is, both the altitude of the platform above the geoid varies with horizontal position, and the ocean surface is continuously varying. The altitude of the satellite over land is not constant, and similarly, the altitude of the satellite over the oceans is not constant!

If one of them were fixed, then you could obtain a reasonably accurate estimate of the other with the systems that are in orbit. However, the oceans have variations that might be as large as 100 feet with storm waves constructively interfering, along with smaller wind ripples of an inch, and tides varying from 1 foot to 50 feet. The satellite is moving up and down in response to changes in the strength of gravity. I’m not going to take the time to try to chase down what that is. I’m going to guess that we are talking about a few feet — much more than 0.1 mm!

You are not being successful in convincing readers that you understand this problem better than Willis does!

Jim Whelan
February 23, 2021 11:26 am

Please! Arguing over the number of angels on the head of a pin.
What does it matter if you make the measurement at the exact same place or not. This is a constantly varying, chaotic surface you are measuring. The surface being measured at one time isn’t the same as the one even a fraction of a second later. Besides you are measuring with an imprecise instrument. the radar wave strikes over a significant area of the chaotic surface and the return signal is spread out, some reflected from troughs, other from peaks.

And you are worried about averaging to a precision that is so much greater than the variability in the surface as to be meaningless.

Last edited 1 year ago by Jim Whelan
MarkW
February 21, 2021 7:14 pm

Willis gave you a detailed explanation as to why your example doesn’t demonstrate what you wish it demonstrated.
Your only come back is to state that since you gave an example, your point has been proven.

Your claim to knowing more than Willis has been well and thoroughly refuted.

Brian Jackson
February 21, 2021 7:55 pm

Willis didn’t know the difference between accuracy and precision and acknowledged it.

rah
February 22, 2021 12:12 am

Sure he does. It has been explained on this blog many times over the years.

fred250
February 22, 2021 2:59 am

Satellite have NEITHER to the level they are pretending.

Dave Fair
February 22, 2021 11:43 am

Brian, why won’t you acknowledge it when you misspeak?

Clyde Spencer
February 22, 2021 9:59 am

BJ
You remarked, “The satellite’s radar altimeters can be calibrated against known physical altitudes ON DRY LAND.”

That is true. However, over the ocean there is no way to know the actual altitude of the satellite because it will change velocity and altitude in response to the Earth’s local gravity field. To do that, they rely on a Gravity Model, with coarse spatial resolution, and almost certainly not to a precision of 0.1 mm derived altitude when both the sea surface and the satellite are changing heights!

It appears that you are unfamiliar with how the altimeters actually work. They aren’t like a speedometer in a car that relies on mechanical calibrations that work on different kinds of pavement. The satellite altitude has to be calculated for every data sample, and that can only be modeled over water and glaciers.

Ron Long
February 21, 2021 5:07 pm

Willis, the ability of satellites to measure seal level changes is even worse. There are several websites that state that the altitude of a satellite can be known to 1 in 1 million, not the 1 in 13 billion required for their stated precision. A satellite changes orbit altitude slightly as it passes over different density areas of the earths crust, which density differences are doe to either composition or thermal variations. Good for you for your fig. 2, the king has no clothes exposed once again.

Brian Jackson
February 21, 2021 5:35 pm

Perseverance hit a sphere 6800 kilometres in diameter from a distance of 54.6 million kilometers……and the target was moving.

Given that the timing of the orbit of any satellite is extremely precise with the accuracy of atomic clocks, I guess NASA does pretty well.

Brian Jackson
February 21, 2021 7:56 pm

Satellites in orbit have thrusters.

MarkW
February 21, 2021 8:54 pm

Satellites in orbit have limited fuel on board. Regardless, they would have to know that they were 2 inches high before they could order the satellite to correct for it.

Gordon A. Dressler
February 22, 2021 8:10 am

Brian Jackson posted: “Satellites in orbit have thrusters.”

And so has every spacecraft ever sent on an interplanetary mission . . . for the purpose of “mid-course corrections” that Willis mentioned above.

But I will also add such thrusters are also needed to orient/reorient such spacecraft for (a) adjusting fly-by trajectories on multi-planet-visitation missions, (b) for breaking into orbits around the target planet or one of its moons, and (c) for proper attitude (heat shield-wise) for breaking entry into the target planet’s atmosphere and/or for direct descent to landing.

Condition (c) above was used in the recent instance of the Perseverance mission landing on the surface of Mars.

Clyde Spencer
February 22, 2021 10:33 am

Satellites that have thrusters use them to maintain a nominal orbit that is decaying from air resistance. They would quickly run out of fuel if they attempted to maintain an exact, unvarying altitude above the geoid. Besides that logistical issue, they would have to have an independent verification of the altitude, which is a practical impossibility over oceans, and even over places like Greenland and Antarctica where the ice is moving, melting, and accumulating snow.

What the radar altimeters are fairly good at is measuring the above-water thickness of pack ice in polar regions, where there is an abrupt change in the time of the return signal.

You are so anxious to be right that you are convincing everyone that you know very little about the topic at hand.

Gordon A. Dressler
February 22, 2021 1:51 pm

Clyde, your main point about satellite thruster use is valid, but I cannot agree with your statement “Besides that logistical issue, they would have to have an independent verification of the altitude, which is a practical impossibility over oceans . . .”

With the implementation of full-Earth coverage of precision GPS location-altitude determination data, all Earth-orbiting satellites have a means of independent verification of altitude for any orbital nadir point.

Clyde Spencer
February 22, 2021 2:05 pm

Gordon,

Point taken that it is possible to use GPS, although high precision requires differential GPS, which isn’t possible with a moving satellite. Typically, a single GPS has a resolution of about 10 meters altitude. That isn’t much help when trying to get sub-millimeter elevation changes.

However, I’m unaware that the radar altimeters are using GPS. When I did background reading previously, it was stated that they were using gravity models to determine their altitude.

Charles
February 22, 2021 7:00 pm

GPS satellite orbit ephemeris is accurate to about 5 cm, updated by range tracking amongst 8-12 remote tracking sites (Diego Garcia has one). This level of accuracy is required for the p(Y) code capable military GPS receiver applications (offensive and general navigation).

Air Force Satellite Control Network – Wikipedia

What isn’t so straightforward is adjusting the raw radar range data with “models” to compensate for:

• Atmospheric air pressure depressions and mounds that affect water height that the beam is illuminating
• Wind fields that do the same
• Currents that do the same
• Atmospheric delays that the two way signal encounters
• Pointing (slant range) errors from geometry (is the beam firing strait down or slightly off Z axis) which requires gimbal control and very accurate satellite body attitude
• Geoid inaccuracies due to local gravitational variances
• Ocean surface diffraction
• Sun and earth tide deformations on top of all of the above

To get sub-cm level accuracy is a bit of a stretch, actually.

Clyde Spencer
February 22, 2021 9:08 pm

Charles
I’m not sure that you are saying what you think you are saying. As I understand your statement, we know where the GPS satellites are to within a couple of inches. However, that apparently is only possible with numerous range tracking stations correcting the ephemeris. The radar altimeters don’t have that level of support and correction. The reference I looked at said that the multiple satellite GPS system is only good for 10-20 meters elevation for the GPS receiver. The question was about using GPS to determine the altitude of the radar altimeters.

https://en.wikipedia.org/wiki/Ephemeris

MarkW
February 22, 2021 2:10 pm

To what degree of accuracy? The accuracy for my car is multiple meters.

Jim Whelan
February 23, 2021 11:56 am

Perhaps you don’t realize that Mars is much larger than a few millimeters as was the designated landing zone.

MarkW
February 21, 2021 7:17 pm

Perserverance had multiple mid course corrections all the way from Earth to Mars.
You are 0 for 2 in coming up with valid examples.

Brian Jackson
February 21, 2021 7:57 pm

LOL….see post above.

MarkW
February 21, 2021 8:55 pm

I’m glad to see that you find your ignorance to be so amusing. Regardless see post above where I once again correct your lack of knowledge regarding space craft.

Rory Forbes
February 21, 2021 5:12 pm

Now, what were you saying about learning about the standard error of a statistical estimator?

That was the best laugh I’ve had in quite some time … many thanks …

And as a bonus your instruction added to my understanding of statistics.

Brian Jackson
February 21, 2021 5:40 pm

Willis confuses “accuracy” and “precision”……still laughing?

Rory Forbes
February 21, 2021 5:58 pm

What I find hysterical is how completely Willis dissected your meager example and you’re still trying to figure it out. Come on man! Stop digging. You’ve already been buried.

MarkW
February 21, 2021 7:18 pm

No he didn’t.

Brian Jackson
February 21, 2021 7:58 pm

Willis acknowledges he “misspoke” which is a clever way to recover when he’s found to be in error.

MarkW
February 21, 2021 8:57 pm

He corrects you on 10 things, admits that one of his corrections might not have been fully accurate, and from that you conclude that you are completely exonerated.
Just how long have you been doing climate science?

Carlo, Monte
February 22, 2021 6:24 am

Jim Whelan
February 23, 2021 11:59 am

The longer one has been “in” climate science, the more cognitive dissonance warps their thinking on the subject.

Last edited 1 year ago by Jim Whelan
MarkW
February 22, 2021 7:31 am

To bad you aren’t even that clever.

February 22, 2021 9:51 am

First rule of holes.

Clyde Spencer
February 22, 2021 10:36 am

I suspect that it will be a “cold day in Hell” before you would admit to misspeaking, let alone actually being “in error.”

MarkW
February 21, 2021 6:58 pm

ninth: It assumes that the height of the people being measured, is the only thing that is changing. What if each person being measured is wearing a different brand of shoe. In the SLR case, things like the actual height of the satellite and the moisture content of the air are different with every measurement made. These values can never be known with perfect accuracy, nor can they assume that any errors in these values will be perfectly random.

Also, has the profile of the wave heights in differing sea conditions has been studied to the point where it can be compensated for in these measurements. Do they even know with any degree of accuracy what the sea conditions directly below the satellite are at any given point in time?

Last edited 1 year ago by MarkW
Brian Jackson
February 21, 2021 8:06 pm

Wrong MarkW, height is known to be a normally distributed human characteristic with or without a shoe.
.
Do they even know with any degree of accuracy what the sea conditions directly below the satellite are at any given point in time?” … irrelevant because they average out over time. For example, tidal behavior is sinusoidal which averages to zero over long time periods.

MarkW
February 21, 2021 9:01 pm

Do you ever get tired of making a fool of yourself. The height variations aren’t random. Beyond that, since these satellites don’t measure sea heights over 100% of the earth’s surface, even if the height differences WERE to be random, the lack of complete sampling would still invalidate your methodology.

If you think gravitational anomalies are related to tides, you know even less than you have indicated.

Clyde Spencer
February 22, 2021 10:49 am

MarkW
I’m beginning to wonder if BJ is one of our better known trolls using an alias. The style reminds me of “weekly rising.”

February 21, 2021 9:12 pm

height is known to be a normally distributed human characteristic with or without a shoe

Whee! In the 1950s and 1960s, women were on average 1″ to as much as 2.5″ ‘taller’ than they are now – if you compare measures with shoes on. In modern times, when most women have stopped the insanity of balancing themselves on approximately 20% of their foot area, their “shod height” has diminished considerably. But, if you actually plotted the current distribution of “shod height,” you would find a very interesting spike (pun intended) at the upper range, consisting of those women who still follow high fashion. (Not that anyone will ever do such a distribution – doctors, military recruitment stations, etc. all measure height with shoes off.)

Second point about that distribution. The shape of the human height distribution has been determined by many millions of measurements that are accurate to 0.1″ (+/- 0.3″). So, you can make deductions about the most likely average of any sufficiently large sample of properly randomized measurements of human heights. Only because far more accurate measurements than “nearest foot” have already been made. (Which begs the question – why ever are you doing this study in the first place???) Now, if you please, provide us, the ignorant masses, with the set of measurements, accurate to within +/- 0.05 mm, of the global sea level, so that we can see that the distribution is what you claim it to be?

February 21, 2021 11:28 pm

Should have noted that myself. I knew the distribution isn’t anything at all symmetric.

fred250
February 22, 2021 3:05 am

Now re-draw the graph in 1 ft intervals, like the measuring stick. 😉

Each “bin” will be basically linear, or maybe undefinable?

A person measuring 6ft, could be anywhere between 5.5ft and 6.5 ft and you have no idea of the distribution within that “bin”

Gordon A. Dressler
February 22, 2021 1:00 pm

Exactly! The marked-every-foot stick and the data-recorded to the nearest foot process results in a measurement set “average” value accuracy of ± 0.5 foot. That will be the case no matter how many measurements one takes and records (be it 10 or even 1 million+) and no matter how many decimal places (i.e., the precision) that are used in reporting the “average” value.

Joao Martins
February 22, 2021 8:18 am

“YOU are wrong!

The height of a person is not constant. This fact is known to Biology since decades long ago: height at wake up and raising from bed DIFFERS from height of the same person late in the evening. By several millimeters (more than 10).”

This is not the same as covered by your sixth point (in comment of February 21, 2021 3:55 pm). It does not refer to the variability of the population, it refers to the variability of each individual. Humans, animals, are not steel rods with fixed dimensions (in the same physical conditions). When measuring the sea level height at a given (geographic position) point in different days (or, more generally, moments) you are not taking measures of different “objects”: you are taking different measures from the same (changing) object. Like when you go to a town at 8 am and start to measure its inhabitants and go on measuring until 7 pm: the “true” (whatever it means) average height of the population could only be obtained by repeating the measurements by a lot of different but simultaneous observers that randomly take the same person at different times during the day…

Clyde Spencer
February 22, 2021 10:53 am

Willis
You nailed him! I wish I could give you multiple thumbs up!

Dave Fair
February 22, 2021 11:51 am

Willis is cheating, again! He brings facts to a cat fight.

February 23, 2021 8:22 am

Sigh… The older you get, the more things you “know” – and the more things you find out that you “know” are wrong. For some reason, I always thought that “left” skew meant the mode was to the left of the mean. Now corrected. (I did have the correct notion for “negative” and “positive” skew, though. Different math teachers? Too far back to assign blame now.)

MarkW
February 22, 2021 7:33 am

height is known to be a normally distributed human characteristic with or without a shoe

First off, height is not known to be normally distributed. In fact as Willis points out, it is known to be not normally distributed.

Secondly, if you can’t accurately subtract the thickness of the shoes, then you aren’t accurately measuring height, with or without normal distribution.

Joao Martins
February 22, 2021 8:03 am

YOU are wrong!

The height of a person is not constant. This fact is known to Biology since decades long ago: height at wake up and raising from bed DIFFERS from height of the same person late in the evening. By several millimeters (more than 10).

Clyde Spencer
February 22, 2021 10:47 am

BJ
“Rogue waves” are not sinusoidal. Actually, tides are known to have long term (~20 yr) trends. Over the length of time that radar altimeters have been in orbit, those long-term tidal trends will not yet have averaged out. Also, sea surfaces respond to wind and barometric pressure, which are somewhat random, but again the issue is whether we have been sampling long enough for them to average out. The Law of Large Numbers says that probabilistic events (such as coin tossing) will eventually approach the theoretical probability. However, that doesn’t preclude long runs of heads or tails. To ‘average’ them out will again probably take longer that the time altimeters have been in orbit.

Are you having difficulty breathing? I’m getting the feeling you are in over your head.

MarkW
February 22, 2021 2:13 pm

BJ reminds me a lot of bethan. Reads one article, and becomes an instant expert. At least in his own mind.

Jim Whelan
February 23, 2021 12:07 pm

The sea surface is completely chaotic not “somewhat” random

Stephen Fitzpatrick
February 22, 2021 12:30 pm

Brian Jackson,
You are making a fool of yourself. It is clear that your practical knowledge of science is limited. Stop digging.

Leonard
February 21, 2021 9:26 pm

Willis
Please consider earth tides and use you excellent skills to let us know how to interpret earth tides and there relevance to the current discussions.
Thank you in advance. And I am asking for a favor knowing you are busy and may not have time.
Leonard

MarkW
February 22, 2021 7:50 am

Willis,
Another source of possible error when trying to measure the height of humans.
How do you handle hair?
Do you measure to the top of hair (TOH)?
Do you squish the hair down? But then those with more hair would still get a couple of extra micrometers of height.
Do you just cut all the hair off of everyone being measured and then measure Top of Skull (TOS)?

Clyde Spencer
February 22, 2021 10:56 am

And thusly, Big Hair styles women favored in years past would not compress down as much as shorter hair styles, suggesting that women have lost height over time.

Jim Whelan
February 23, 2021 12:08 pm

And measuring the sea surface is more like doing the height measurement while the people are jumping on trampolines.

Latitude
February 22, 2021 12:53 pm

….and both of you are missing the obvious/obvious

they keep changing sticks

Willis your 2009 just was a satellite change….when Envisat when out…Jason uses the exact same “Doris”…..

when Jason 3 was brought on board….it’s rate did not match J/2…..instead of tuning 3 to 2….they adjusted 2 up to match 3

now where have we seen that before?

the increase does not actually start until 3

Latitude
February 22, 2021 5:54 pm

but I’ve posted those links so many times to you…asked you to look….and I don’t think you ever look at them..or didn’t catch my posts

Here’s 2 fairly relevant….they explain what they do in their reports
…or put better….explain the jiggling and fenagling

the Envisat working annual report 2009

Merged envisat and jason satellite altimeters using crossovers adjustment to determine sea level variability

https://iopscience.iop.org/article/10.1088/1755-1315/200/1/012035/pdf

…there’s literally 100’s of reports…and funny enough, they admit to everything they are accused of

the only real jump up in sea level is J3….the end of J2 is adjustments

Last edited 1 year ago by Latitude
Latitude
February 23, 2021 5:55 am

absolutely….

The end of J2 is not actual measurements….the end of J2 was adjusted up..after the fact…. to meet with the new J3

rather than tune J3 to match all the previous satellites…they ‘assumed’ all the previous were wrong…and the new and improved J3 was right

…because the previous satellites didn’t show enough sea level rise to suit them and their computer models

there is no real acceleration in sea level rise…because it’s all adjustments

…and that’s why none of it matches any of the tide gauges

Last edited 1 year ago by Latitude
TC in the OC
February 23, 2021 3:17 pm

Willis and Latitude,

In following this thread I agree with Willis that they are not measuring the same spot from the same spot. At the end of my post I will link some fact sheets from NASA for the Topex satellite and the Jason 3 Satellite.

We can talk all day about precision and accuracy but that doesn’t cover the fundamental problems with these measurements. In looking at the Topex fact sheet I see that they have a network much like my surveyors GPS network and that the satellite is just a big receiver. The fundamental difference between how my crews uses GPS for horizontal and vertical locations is 2 part. 1 is my crew uses a receiver and are not bouncing the waves off of an irregular surface and 2 is that the receiver is in a fixed location and static from a minute to upwards of 30 minutes and is not a moving surface being hit with a one-time shot.

I mentioned in a comment above that we have a GPS network and also use a base station to improve the measurements. That being said we do not use GPS elevations for staking…why you may ask…well it is never the same from day to day. My crew can go out to a jobsite and shoot a few control points to make sure everything is still good. What they find is a slight difference in horizontal location from day to day of +/- 0.01 feet at the worst. Vertical location is different every day usually +/- 0.1′ and it is not surprising to see it off by 0.2 feet or more and always in different directions (for example one day it will be 0.08′ high the next 0.04′ low and so on).

I would like to know how they can get a more accurate measurement with their system locating a moving target with one satellite on a one time pass every 10 days than I can get with a fixed location of a longer duration with 4 to 8 satellites broadcasting data to the fixed network we use of 12 constantly correcting stations triangulating with my base station and receiver.

TC

https://sealevel.jpl.nasa.gov/system/documents/files/1674_tp-fact-sheet.pdf

https://www.nesdis.noaa.gov/jason-3/pdf/JASON-3-facts-Apr%2015%20v2.pdf

https://www.nesdis.noaa.gov/jason-3/pdf/Jason-3%20GPSP.pdf

https://www.nesdis.noaa.gov/jason-3/pdf/Jason-3%20DORIS.pdf

Please note in NASA’s words not mine…Topex 10 day repeat of ground tracks +/- 1 KM and the unprecedented accuracy: sea level measurements to better than 5 CM. For Jason 3 The GPSP supports precise orbit determination by the DORIS system. It also helps to improve gravity field models and provides data for satellite positioning accurate to about 50 meters and 50 nanoseconds. Never the same location.

Mr.
February 21, 2021 3:57 pm

C’mon man.
It’s Sunday afternoon past beer o’clock where I am, and you’re asking us to digest your math parables TWICE.

(To Willis’ article itself – I see the tide come in and go out in our estuary every day, and the water levels are always more than 3mm different from the same point they were on the previous high or low tides.
Depends where the wind is blowing from, and how strong.
Poofteenth and bee’s d1ck measurements from space just don’t pass the sniff test.
But they do register on the old

Joel O'Bryan
February 21, 2021 6:29 pm

Absolutely not. If I make any one additional measurement, I cannot be certain of the accuracy of that one measurement of no more than the graduations on the pole.
You never specified those graduations and what system is used to point to the point on the pole. You just made up the sums of 114,900 and 115,100 for convenience.

I award you a D- in Measurement Science 101.

Last edited 1 year ago by Joel O’Bryan
Joel O'Bryan
February 21, 2021 6:40 pm

Consider:
a 500 mL Erlenmeyer flask, 50 mL intervals.
a 250 mL Erlenmeyer flask, 25 mL intervals.

Q: Do they provide the same precision if I want to measure 250 mL?

Gordon A. Dressler
February 22, 2021 1:25 pm

Joel, again careful distinction is required between the words “accuracy” and “precision”.

If the two flask are calibrated so as to the same percent accuracy at the 250 ml mark, they should theoretically provide the same overall accuracy assuming the measuring person uses the same care and technique in conducting the measurements for both flask types (e.g., measuring the liquid meniscus the same way).

The reported precision of the measurement is subject to the whim of the person performing and recording the measurement . . . one observer may be content to record the measurement as 250 ml, while another may want to record it as 250.0 ml, while yet a third observer might feel competent recording it as 250.00 ml if he/she feels the calibration and his/her eyesight measurement technique warrant such. These are different levels of precision and they might, in fact, all be correct (albeit in the last case, manually by eyesight determining a value to one part in 25,000 would not be expected).

MarkW
February 22, 2021 2:16 pm

Does the volume of the flask change as the temperature changes?

Gordon A. Dressler
February 23, 2021 11:04 am

Of course . . . both the volume of the flask and the volume of the liquid within that is being measured change with temperature, probably NOT with the same slope (one would have to compare coefficients of thermal expansion of both over the given temperature range being considered).

However, Joel did not imply, in his question, concerns about differences in temperature or CTE, so this was left unaddressed so as to not over-complicate the fundament issue he raised.

Joel O'Bryan
February 21, 2021 6:43 pm

Do each of these precision marked cylinders provide the same level of precsion for say 50 mL? or compared to the Erlenmeyer flask (above)?

Your comment about a 10 foot pole and a given level of precision measurements is nonsense.

Joel O'Bryan
February 21, 2021 7:07 pm

As a final note on precision versus accuracy.

Precision comes from how “precise” the graduations are on those flasks, and the care I take to read the fluid level to a line.

Accuracy comes from how closely those graduations were calibrated to a standard when they were etched on the glass.

fred250
February 22, 2021 2:34 am

Satellite SLR measuremsnt aren’t accurate enough to know if they have any precision.

M Courtney
February 22, 2021 5:49 am

Would have thought it was the other way round.

<b>Question</b>How do they cope with changes in humidity of the air the radio waves pass through?
If they just average them away then they are assuming humidity is independent of whatever is happening with sea level. That would be assuming that humidity is not related to heat in the oceans.

Which would be a curious approach.

M Courtney
February 22, 2021 3:42 pm

Thank you.
At least they realised that the idea is fatally flawed. And tried to compensate.
Maybe they did.

February 21, 2021 10:15 pm

Brian Jackson is correct.
Of course averaging of measurements increases precision as per the example of the height measurements in feet only.
It is a total error to argue that averaged measurements can’t exceed the precision of a single measurement. They can and do routinely in many forms of measurement.

This brain-fail about precision and averaging is making WUWT look, well, not clever. Stop it! Brian is right. It’s embarrassingly basic.

M Courtney
February 22, 2021 7:31 am

Basically, you need to state your assumptions before averaging out random errors.
When you have that list you may see why Brian Jackson is not right.

Stats is not just number crunching. It is also logic.

MarkW
February 22, 2021 7:38 am

Brian is not right, as many posters have demonstrated.
There are a number of assumptions that must be met before multiple measurements can be assumed to increase accuracy.
When it comes to using satellites to measure sea height, none of these assumptions have been met.

Pat Frank
February 22, 2021 10:26 am

Brian is assuming all the scores on his ruler are infinitely accurate. They’re not. This means round-up and round-down errors will accrue. But you won’t know when, or how many of each.

As a result, systematic error will creep into your set of measurements. But you won’t know how much. The only way out is to calibrate the ruler and report the calibration error statistic with the measurement report.

Brian is correct only in infinitely-accurate-instrument-land. In the real world, his method fails.

That knowledge is basic to experimentalists Hatter.

Clyde Spencer
February 22, 2021 11:14 am

Eggburn
The basic assumptions for improving the precision of a given measurement is that all errors are random. That implies using the same instrument for measuring an invariant property that doesn’t change over the time necessary to take the multiple measurements. If the property changes, then the standard deviation is no longer just of the measurement random errors, but includes a component of the trend, if any. If different instruments are used, they will likely have different calibrations characteristic of the type of instrument and what it is measuring. Differently calibrated instruments will introduce systematic errors, which will also reduce the precision.

If being read and recorded by a human, then it is best to use the same observer to eliminate issues of parallax and bias of interpolation accuracy as to whether or when to round up or down, and how well they can estimate the next unmarked significant figure.

In the case of a radar altimeter, the pulse return-time is the only thing known with high precision and accuracy. All the rest of the necessary model parameters, such as temperature along the air path, air density/pressure, absolute humidity, platform-altitude above the geoid, and wave-form and amplitude/period of the surface are known poorly. Done properly, a propagation of uncertainty analysis of all the relevant parameters should be done for every pulse path.

You don’t seem to understand the problems any better than BJ does!

Gordon A. Dressler
February 23, 2021 5:38 pm

Clyde, if you go back to Brian Jackson’s reply to Willis, giving the sophomoric example of using a marked-every-foot stick to measure an “average adult male’s height” (Feb 21, 3:05 pm), you’ll see this exact verbiage:
“Now use this pole to measure the height of 20,000 adult males, recording each measurment to the nearest foot.”

Overlooking the typo, his Gedankenexperiment did not say that interpolation at the time of measuring and recording data was allowed . . . it was binary, either record the height for each male as 5 feet or record it as 6 feet.

Understanding this, there is no way that n measurements with this defined process will improve the precision of the resulting “average” height to one decimal place or more . . . all one can state definitely is that the average is somewhere between 5 and 6 feet, no matter how many measurements are performed.

Jim Whelan
February 23, 2021 12:12 pm

“averaging of measurements increases precision” But does it increase accuracy? Especially when the measurements are of a constantly changing, chaotic surface.

fred250
February 22, 2021 2:55 am

WAY too many assumptions

Lets look at 10 people, 3 measure as 5 ft, 7 measure as 6 ft

The average is 5.75.

BUT….

…. a 6 ft person could be anywhere between 5.5 and 6.5ft, and a 5ft person anywhere between 4.5ft and 5.5ft

You cannot assume any particular distribution without much more accurate data.

So If all people measured were just at the top of their respective range, the average height would actually be close to 6.25m

Similarly, if all people measured were at the low end of their respective range, the average height would actually be close to 5.25m

So from the information obtained by your actual measurements, the average height is somewhere between 5.25m and 6.25m

Gordon A. Dressler
February 23, 2021 5:46 pm

Well, except for the fact that the high end their respective range would average to 6.49999… feet (rounded off and recorded as “6 ft”) and the low end of their respective range would average to 4.5000000…+ (rounded off and recorded as “5 ft”).

Last edited 1 year ago by Gordon A. Dressler
rbabcock
February 22, 2021 5:46 am

The problem with your example is your 10′ long pole isn’t always 10′. Sometimes it 10′ 3″ and sometimes it’s 9’8″ and so on. The satellite is never exactly “10′” above the Earth. It drifts up and down and you never know exactly how “long” it is.

Last edited 1 year ago by rbabcock
Dale S
February 22, 2021 7:08 am

Get a 10 foot long pole, make marks on the pole at *3* foot intervals. Now use this pole to measure *every last adult male on earth*, recording the measurement to the nearest 3 feet. You will find that the average height of adult males is extremely close to 6 foot, which isn’t even close to +/- 0.01 feet.

Whether the average of a crude measuring method happens to match the average of a more precise measuring method is entirely a function of how the population distribution falls, and taking a larger sample does exactly nothing to affect a distortion from an uncooperative distribution.

Let’s consider an easy distribution. There is a 100-sided die and you wish to determine if the distribution is uniform — let us suppose the die is fair and you are recording each result exactly. As the number of samples increase, the average will inevitably zoom in on 50.5.

Now suppose you decided measurement precision doesn’t matter to accuracy as long as you take enough measurements, and figure recording the nearest multiple of 12 (rounding up) is enough. Now the expected distribution looks like this:
0 – 5%
12, 24, 36, 48, 60, 72, 84 – 12% each
96 – 11%

As the number of samples increase, the average will inevitably zoom in on 50.88 — which is wrong! You might be impressed that the it got within 0.38 of the true answer, but that’s a function of the distribution chosen and has *nothing* to do with the number of samples taken.

Taphonomic
February 22, 2021 8:08 am

Brian, you are totally ignoring the concept of significant figures.

Pat Frank
February 22, 2021 10:36 am

Taphonomic, Brian’s method works when the integer round-ups and round-downs are perfectly accurate and correct, and there are a large number of measurements.

But let any instrumental limits of resolution enter, and the method fails. Systematic error is then always present and never goes away, no matter how many measurements are made.

Taphonomic
February 22, 2021 11:06 am

If you measure something to one significant figure you can not get more than one significant figure out of mathematical calculations.

The answer to his thought experiment will be approximately 6 or approximately 5 depending on rounding.

Calculators and computers can take calculations out to a meaningless string of digits. That doesn’t mean they reflect valid significant figures.

Taphonomic
February 22, 2021 3:58 pm

1.8 inches? Significantly more significant figures. Seems like you’re getting Bayesian.

If you’re going to change graduations might as well just start off with graduations of .01 ft.

Pat Frank
February 22, 2021 5:16 pm

Hi Willis, I had to do this exercise when Richard Garwin challenged my limits of resolution argument at the 2015 WFS meeting in Erice, Sicily.

He used exactly Brian’s argument. We had an email conversation, and I was eventually able to show him that the presence of systematic errors (resolution limits) offset the final result, making the integer mean wrong.

But the method does work, so long as the measurement accuracy is perfect and the instrumental precision is infinite (in some world in another universe). Following your description, I generated 10,000 values of 5.75±0.241667 feet, where the SD was a random distribution (0.241667 feet = 2.9″).

The mean of the double precision 10,000 points was 5.7498. I then rounded the values to a single decimal, which is where the SD made its contribution and where the variation in each value begins. The integer mean after 10,000 values was 5.7490.

With 200,000 values the exact mean and the rounded (to one decimal) were identically 5.7498.

The image shows the 10,000 and 200,000 tests side-by side. The plots are crude (especially by your standards, Willis), but I think they’re adequate to convey the point.

More data points get the convergence closer to exact. But it’s necessary to have absolutely perfect rounding. Any instrumental resolution limit causes rounding errors, inserts systematic error, and Brian’s method does not work.

Pat Frank
February 22, 2021 5:17 pm

By the way, Willis, I meant to add my very best to you, right back. Your posts are terrific.

Pat Frank
February 22, 2021 9:13 pm

Ah, you’re right, Willis. I didn’t read Brian’s method closely enough.

Rounding to the nearest foot won’t work because one foot is greater than the limit of variability among the values.

Rounding to the nearest foot will produce far more 6’s than 5’s if the average height is 5.75 feet.

Good catch! 🙂

Best back, Willis. 🙂

Pat Frank
February 22, 2021 10:11 am

You’re assuming that the 1-foot division marks on your ruler are absolutely accurate, Brian. Typically, that would not be the case.

Suppose they were good to only ±0.5 inches; 4.2% accuracy. Your 5.75 would then nominally be 5.75±0.04 feet.

But it’s worse than that. With a ±0.5 inch limit of accuracy, some measurements will be rounded off in the wrong direction. But you won’t know which ones. And there’s no way to know the round-up errors equal the round-down errors.

That means some unknown systematic error will creep into your measurements. If you don’t account for that in some way, your measurement report will include false precision.

The only way to make a correct report is to calibrate your ruler using some very well-measured standard. Then report the calibration error statistic as part of your measurement report.

Rory Forbes
February 21, 2021 3:09 pm

I’m a strong believer in the ‘hands on’ approach to almost everything. Doing work at a distance always seems to add in errors. Tide gauges are best. They’re easily tested. Hell, we can even look at the Nile gauges to see what the Egyptians had to contend with, 5 millennia ago.

I have been either in, on, near or interacting with the sea nearly every day of my life … on three continents, over 77 years. I’ve had the good fortune of revisiting many places several times at long intervals. What I have noticed (or more aptly failed to notice) is any significant difference in sea level. Strange, that!. Don’t people look? Are they willing to just believe what they’re told? I’ve been launching my various boats at the same launch ramp for over 60 years … I’m sure I would have noticed something had changed.

Dave Fair
February 22, 2021 12:07 pm

Rory, you have succinctly described the difference between living life and mental masturbation. The keyboard cowboys (mental masturbators) incessantly and manically pound on argumentative minutia far beyond any relevance to the current topic. They will never quit and it is futile to try to reason with them. They are ideologically driven.

Rory Forbes
February 22, 2021 12:15 pm

You’re right of course, but I’ll be damned if I let them have the floor to spread their nonsense unimpeded. Besides they’re often such easy targets … and smacking them around can be quite diverting (and cathartic, don’t you find?)

Gordon A. Dressler
February 21, 2021 3:11 pm

Excellent, Willis . . . just excellent!

I agree that any assertion to measure global SLR to an accuracy of +/- 0.4 mm/year is ridiculous, whether it be based on satellite radar measurements or on tidal gauge data.

Also, I would change the article’s title verb from “mucking” to “f**king over” the sea level data . . . but that’s just me.

dk_
February 21, 2021 3:23 pm

Good summary, but don’t forget tides. When researching the cartographic use of navigational chart datum points, I found that sea level at any given point was once (pre-satellite) only considered established after nearly 80 years of more-or-less continuous observation and relative to the altitude of a stable land feature. Turned out that this was associated with early observation of astronomic cycles of the sun, moon, and nearer planets. Unless a raw measurement is compared to the exact same point in the lunar tidal cycle, it is guaranteed to be inaccurate by quite a lot.

Another point is orbital jitter. Turns out that within any orbit, the precise altitude of a satellite can vary quite a bit — a variance several orders of magnitude greater than the accuracy claimed for most of the sea level measurements I’ve seen. Not only are satellites affected by gravitational tides from other heavenly bodies, but the earth, not being a smooth orb, has a varying mass density at every point in the surface. An orbit, even “about 800 miles up” is actually not a regular circle, but a wavy line with variations over distance that is often greater than the surface wave height.

I’ve not seen an explanation yet as to how many islands seem to be increasing in land area, per satellite data, while the ocean rises. It does seem sumhow that dirt has a better radar return than air or water, It might be better to measure land surface area changes than ocean surface. I wonder why land surface area measurement isn’t an easy (and obvious) calibration point for sea surface measurement instruments.

Rud Istvan
February 21, 2021 4:17 pm

There is a long tide cycle caused by the moon, peak to trough about 18 years. I forget its name; is in the SLR, Acceleration, and Closure post. So SLR change needs to be computed from tide gauges with minimum record lengths of 60-65 years according to Nils Axel Moerner, the global expert. Then to be accurate you either need a large sample (presuming vertical land motion washes out) and there isn’t one, or you need difGPS corrected tide gauges with long records. There are about 70 globally. And they show about 2.2mm/year with no acceleration. Details are in the SLR, Acceleration, and Closure post.

Editor
February 21, 2021 7:04 pm

Saros cycle.

Waza
February 21, 2021 7:28 pm

Rud,
It is my understanding that due to wind and river flows, tide predictions are only accurate to a few hundred mms.
Example Key West is currently out by 200mm.

https://tidesandcurrents.noaa.gov/stationhome.html?id=8724580&units=metric

This makes comparing historic peak sea levels at Individual locations impossible.

Q. But does this impact the averaging or rate of SLR over multiple tidal cycles?

TonyL
February 21, 2021 7:34 pm

“There is a long tide cycle caused by the moon, peak to trough about 18 years. I forget its name;”

Nodal precession, period = 18.6 years.
This effect produces a sine wave superimposed on the data.
For the Boston tide gauge, the start of the cycle is at ~1933 and bottoms out at ~1970 and is in a steep upswing up at ~2007, headed to a new peak ~2024. So from 1970 through to today it looks like an acceleration even though it is just the part of the sine wave on the upswing. {the tide gauge “The Battery”, in NYC is nearly identical}
NASA GISS (yes, them) published a graph of SLR *starting* in 1970!
They then claimed the upswing is true acceleration due to AGW.
Either NASA does not know that the moon governs the tides, or they were not truthful.
Either way, it does not look good for them.

TonyL
February 22, 2021 12:37 am

The Moon – Nodal Precession == 18.612958 years.

Look at my chart!
That wave is 74-75 years- Look At It. !!!!!
That is exactly 4 x 18.61 precession cycles.
Not 54 years.
Actually, when I first did this, I did not have a least squares sine wave fitting algorithm. So I used the Least Squares 3rd order and 4th order fits as a *first approximation*. The agreement between the two fits was remarkably good so I went with it. With the equations for the two polynomial fits in hand, along with the linear fit, it was a simple matter to calculate the intersection points between the curves and the linear fit.
The length of the cycle calculated out to 74.44 years +/- 10 days.
I need to say that again: 4 x 18.61 +/- 10 days. With *monthly* data. Not bad, not bad at all.
Here is the Boston tide gauge with all the fits.
The two straight lines are the linear and 2nd order fits (the acceleration).
The two curves lines are the 3rd and 4th order poly fits.

Now I agree that the ultimate waveform probably is not a real sine wave, but for such a noisy data set, Close enough. So I slapped on a sine wave and called it done.
The Saros cycle simply does not fit the Boston data.

TonyL
February 22, 2021 10:46 am

How curious.
Saros cycle gives 54 years, your CEEMD gives 65 years, and my curve fit gives 74 years.
Do we hear 85?? We need a bid for 85. Going once, going twice…..
Nice work with the CEEMD, I had not thought to look at it that way. Boston tide gauge is way more interesting than it looked at first glance. And to think all this started out with some bogus claims of acceleration from UC and NOAA.

As an aside, I did this back in 2017 and did an update in 2019. Guess what?? The data changed. Subtract one data set from the other and a very strange repeating pattern emerges. The pattern is 12 data points long so repeats every year. When another SLR post comes along, maybe I can put up what I found and we can all take a look at it.

Cheers.

Dave Fair
February 22, 2021 12:12 pm

Rud, thanks for bringing facts to a cat fight.

RickWill
February 21, 2021 3:56 pm

There is a NOAA measurement set that gives rise in temperature of oceans in top 0-2000m is 0.11C over the past 65 years. So gone from average of say 4C to 4.11C.

The gain in heat content is equivalent to 22 days sunlight so the slow rate of rise is a mixing issue not an available energy issue.

On average, the surface is about 16C. So if the mixing rate stays constant we could expect thermal equilibrium in about 7,000 years. I have not checked what the constant enthalpy temperature is; just assuming constant temperature top to bottom. Somewhat longer than 7,000 years if the bottom 1700m gets warmed as well.

The question that comes to my mind is – why are deep oceans so cold? They are getting geothermal heat at depth so they should be slightly warmer than the surface not cooler.

Has anyone worked out what the adiabatic temperature profile is for the oceans. That would then give a better idea of how much warming is going to take place before the deep oceans reach thermal equilibrium with the surface. If it all reaches an average of 16C then expect another 5m of sea rise.

Clearly finding missing heat in the ocean is nonsense as it is mixing rate limited and the surface temperature is thermostatically controlled.

We could accelerate the heating process by using the heat differential of about 26C between deep ocean and the tropical warm pools to make a heat engine to generate reliable electricity. About 3000m of well insulated pipe would be the main investment.

Rud Istvan
February 21, 2021 6:18 pm

The problems with the NOAA ocean temp data are legion. Trade route biased, ship lading biased, inaccurate dart probes ( depth guesses)… covered in essay Missing Heat in ebook Blowing Smoke. We only have reasonably reliable data since ARGO. And by the ARGO design, we don’t have a real good handle on actual OHC change and resulting thermosteric rise until about 10 years into deployment, so only about 5 years ago. Hence the closure calculations proving satalt is just off.

RickWill
February 22, 2021 12:27 am

Yes – I found some really interesting work on seawater thermodynamic modelling that makes atmospheric modelling appear simple:
http://www.scor-int.org/Publications/Thermodynamic_TEOS-10_Manual_GOSHIP_19Aug09.pdf

But the simple answer is best in this case as seen in the chart attached. Another unique property of water that gives the temperature gradient at depth.

So we know why the ocean is so cold. We still do not know what the equilibrium state looks like. In fact it would be an equilibrium in steady state flows rather than any static condition.

Just think how long it would take to warm the Earth up with all that cold water sitting below a very thin warm mixed layer.

MarkW
February 22, 2021 7:42 am

Any bottom water that was warmed by geothermal, wouldn’t stay on the bottom.

RickWill
February 22, 2021 12:31 am

The other thing I learnt today was that there is an OTEC plant already generating electricity in Hawaii:
https://www.power-technology.com/projects/makais-ocean-thermal-energy-conversion-otec-power-plant-hawaii/
Will be interesting to get more insight into this one.

Rud Istvan
February 22, 2021 3:03 pm

Sorry for the late reply. Willis explained. I will do a bit more. When sea ice freezes, it exudes brine. (Sea ice is ‘fresh’.) The resulting salt brines make water ‘heavier’, so it then sinks to the bottom of the Arctic/Antarctic depending on season, at the temperature near where it was created. This is called the thermohaline circulation. And the reason we know this is not massively disrupted by geothermal ocean floor energy is that it persists. For reasons having to do with seafloor configuration and continent positions, the present ‘recirculariztion’ is about 800-1000 years.

Ossqss
February 21, 2021 4:06 pm

Seems like the same magic used to measure global temp to hundredths of a degree or perhaps Karl et al is just now dumping all the ship engine intake water they held to usurp the ARGO data?

Last edited 1 year ago by Ossqss
February 21, 2021 4:18 pm

I wish Aviso would follow their lead. They still add a 0.3 mm/yr isostatic adjustment (fudge factor) to all their reported satellite-measured sea-level trends. (It used to be optional, but they removed that option.)

Of course, that sum isn’t really a rate of sea-level rise (SLR). It’s what they think the rate of sea-level rise WOULD be if the ocean floor weren’t sinking. Calling it SLR is misleading.

“it… means their “mean sea level” is now floating, phantom like, above the waves.”Greg Goodman

Last edited 1 year ago by Dave Burton
Clyde Spencer
February 22, 2021 11:36 am

The Rule of Thumb for adding/subtracting numbers with different significant figures is to truncate the right most significant figures to agree with the addend with the least precision.

That implies that to add a GIA with a high precision requires knowing the shape and volume of all the ocean basins to the same level of precision! I don’t believe we know the precision to that level. Our estimates probably aren’t even accurate.

Further more, in the attempt to make that GIA, the land is uplifting where there had previously been thick glaciers, decreasing the volume of the ocean basin(s) with uplift along the margins. However, at the same time, the melted glacier water is weighing down on the ocean bottoms, depressing the bottoms in proportion to the depth, thus increasing the total volume. I’m incredulous that the counteracting forces can be modeled in ocean basins with shapes and volumes known to low precision. These “Rocky Mountain High” modelers are simply having wet dreams thinking they can model to +/-0.1 mm SLR!

February 26, 2021 2:26 pm

The 0.3 mm/year number is from Prof. Richard Peltier. To the best of my knowledge Prof. Peltier and his colleagues have not published uncertainty estimates for their 0.3 mm/yr ocean basin GIA estimate, nor for any of their other GIA / PGR figures. Tamisiea, 2011 gives a broad range: 0.15 to 0.5 mm/year.

Alastair Brickell
February 23, 2021 1:15 pm

Yes Dave a good explanation. They report what would have happened if the floor didn’t move. So non-science or fake news.

I expect when S.L. things slow down a bit they will then just add in a new adjustment “correction” to take account of what would happen if the oceans weren’t evaporating from the surface. That would really produce some scary results. Just think of the funding they could get then!

DMA
February 21, 2021 4:18 pm

A good electronic distance meter for land surveying has a resolution of 3mm plus an about 2 parts per million. They use infrared or visible lasers emitting a pulsed beam and compute the distance to the target by comparing the emitted frequency to the returned. They require a rather focused beam. I don’t know how focused these Jason radars are or how they decide what part of the beam is used to measure but it seems to me that there must be a lot of averaging going on for each data point. But, as Willis says, they are not measuring the same point several times to improve accuracy because the surface being measured is always changing due to winds, tides currents, air pressure and likely several other variables. To add to this complex problem the geoid ( surface of equal gravitational potential) is not smooth but more like a potato. I doubt that the geoid is mapped to the accuracy they claim for sea level so I have always been suspect of the satellite sea level numbers. It did occur to me that it should detect anomalous changes and could possibly be used to track submarines.

Rud Istvan
February 21, 2021 6:21 pm

They arent focussed. if I recall correctly from my Jason 3 post here a few years ago, the ‘aperture’ is about 10km^2. That is a lot of waves.

MarkW
February 21, 2021 7:32 pm

Areas with higher gravity would cause the oceans to pile up, while at the same time cause the satellite to drop lower in it’s orbit.

MarkW
February 22, 2021 7:44 am

Comparing frequencies is used to measure the targets speed. To measure distance you record how long it takes the signal to return.

Clyde Spencer
February 22, 2021 11:38 am

Yes, Doppler Shift!

Bob Ernest
February 21, 2021 4:36 pm

Excellent

Tony Taylor
February 21, 2021 4:39 pm

Is there anywhere that the average punter can look at the coast and see that the sea is clearly rising?

Rory Forbes
February 21, 2021 5:15 pm

No. Only 2 mm annually (approx. 8″ over a century) is imperceptible.

Tony Taylor
February 21, 2021 6:21 pm

Thanks. That should buy me some time to move to higher ground.

Editor
February 21, 2021 7:15 pm

Would a place where you can see the sea level not rise be acceptable?

See http://john-daly.com/deadisle/ describes and analyzes the history behind a sea level benchmark in Tasmania that was set in 1841.

fred250
February 21, 2021 10:35 pm

Interesting discussion

That last bit about a jump in 1950 in Sydney….. seems to have been corrected.

rah
February 21, 2021 4:48 pm

“Now I freely admit, I’m just a guy with no scientific training at all. I took Physics 101, Chemistry 101, and one year of Calculus in college, and that’s it. But the beauty of science is that it’s all about the facts and the evidence, and not about the qualifications of the person presenting the facts.”

That makes you better qualified IMO than many who claim to be “Scientists”.

A scientist is by definition a person who uses the scientific method to discover facts or truth.

One of the earliest fathers of the scientific method was Ibn al-Haytham. He was the first to develop hypothesis by experimentation/observation. He wrote:
“The seeker after truth is not one who studies the writings of the ancients and, following his natural disposition, puts his trust in them,” the first scientist wrote, “but rather the one who suspects his faith in them and questions what he gathers from them, the one who submits to argument and demonstration and not the sayings of human beings whose nature is fraught with all kinds of imperfection and deficiency.”

IOW a scientist must be a skeptic that must develop/confirm his/her hypothesis by experimentation and observation and not base it on the supposed research of others alone.. If a person is not a skeptic then they aren’t a scientist no matter how many degrees or titles they may have.

The so called field of “climate science” is filled with people that rely on nothing more than computer models. Models that have without fail have been terribly inaccurate in their projections so far and yet they still treat them as valid scientific evidence. They are NOT scientists. They’re leaches working flawed academic and government systems for self enrichment.

The most dangerous phrase written by so called journalists today supposedly reporting on climate change is. “Scientists say”.

Geoff Sherrington
February 21, 2021 5:16 pm

There is also the problem that one cannot find a number for the mm of sea level change per degree C change in atmospheric temperature. Just one of a dozen missing numbers that are fundamental to global warming hypotheses.
https://wattsupwiththat.com/2020/09/11/the-dirty-dozen-tests-of-global-warming-science/
Geoff S

Philip
February 21, 2021 5:49 pm

Just out of curiosity how do they account or adjust for swell, waves, and tide ect. much of which depends on sun, wind, near by and direction of storms or weather fronts or condition. Just wandering.

Rud Istvan
February 21, 2021 6:26 pm

P, see my guest post here a few years ago on Jason 3, if I recall correctly titled “Sea level rise from satellite altimetry—fit for purpose?” It was requested by Charles, and It went deeply into all the details you seek. See also my post on successor satellite Sentinel-6, which looked at the ‘improvement’ and concluded still not good enough. WE has this nailed.

BobM
February 21, 2021 7:18 pm

From Wikipedia referencing NASA information on Jason-3: The scientific goal is to produce global sea surface height measurements every 10 days to an accuracy of less than 4 cm. In order to calibrate the radar altimeter, a microwave radiometer measures signal delay caused by atmospheric vapors, ultimately correcting the altimeter’s accuracy to 3.3 cm.

So they measure the same spot once every 10 days, accurate to 3.3 cm. And they call that more accurate than tide gauges?

An orbit takes 112 minutes. Does anyone know how many data points they get per orbit?

Dudley Horscroft
February 21, 2021 8:15 pm

There are 1436 minutes in a sidereal day. Divide by 112, and it appear that there are only 12.82 and a long string of small numbers of orbits in a day. The question is, how many orbits will it take to be a whole number of days? That is the next time when the satellite will be over the same point on the earth’s surface. Exceptions: if (a) the satellite is in exactly the equatorial plane, or (b) the satellite is at the level where it is permanently above some point on the equator. Obviously (b) is restricted to measuring the height above that one spot, which would give a fair estimate of sea level rise for that one spot. But over the years it would give a good estimate of the variability caused by wind, humidity, temperature, and any other atmospheric variables over that spot. Useful to calibrate the atmospheric variability for the other satellites!

BobM
February 21, 2021 10:38 pm

Jason-3 took 9.9 days to get back over the exact same spot, so basically 1 measurement of 1 spot in the ocean every 10 days. Difficult to believe that there isn’t going to be huge variation, for starters.

The radar has a pulse rate that gives X number of “snapshots” per second which can be merged into a visual, but are individual data points. For Envisat, that pulse repetition frequency (PRF) was 1984 pulses per second. From those radar returns they calculated the sea level along the orbit track. Jason-3’s orbit was 112 minutes, so the speed of travel along that track was 40,000 km (circumference of Earth) / 112 minutes = 357 km/min. = 5.95 km/sec. = 5,950 m/sec. If Jason had the same PRF as Envisat, 1984 pulses per second, it would obtain measurements of sea level that were (5,950 m/s / 1984 pulses/s) = about every 3 meters apart, and 20,000 measurements as in the “example” above would take 10 seconds of orbit time and stretch 59.5 km. in a straight line. Doubtful to me that you can average any of it to get accuracy better than a couple of centimeters. That said, I don’t know if Jason has a faster pulse rate than Envisat. Still looking.

P-A Johnsson
February 22, 2021 3:34 am

You also have a pointing error. The radar lobe is rather narow. A 0.1 degree pointing error gives 0.15 m too long measurement.

According to “Attitude Determination
and Control System” https://ntrs.nasa.gov, 0.1 degree is a high end requirement.

Mike Dubrasich
February 21, 2021 8:06 pm

Some clarifications, which I hope help:

The precision of a measurement process is the degree of agreement among measurements. Precision is more or less synonymous with Standard Deviation. In theory the SD should shrink with increasing number of measurements. The opposite of precision is imprecision.

The accuracy of a measurement process is the degree of agreement of a set of measurements with the True Value of the quantity being measured. Accuracy cannot be determined if the True Value is unknown. Calibration (with known True Values) is the method used to improve the accuracy of a measurement process. The opposite of accuracy is bias.

Repeated measurements of an unknown value may have high precision but still be biased. Measurement errors (aka systemic errors) can lead to bias if the mean of many separate measurements differs significantly from the True Value.

The satellite data appear to have bias. Both sets may be precise (or not — the SD’s are not given nor are the number of repeated measurements at any one point), but they differ in means. Either a phenomena occurred in which sea level rise suddenly changed in 2010, or one or both sets are biased. The True Value remains unknown.

fred250
February 21, 2021 10:01 pm

Willis, how come when I shoot, I only get 2 dots at most ? 😉

cerescokid
February 22, 2021 9:19 am

That graphic is a keeper. Thanks for reminding me of it.

February 26, 2021 5:05 pm

I believe the original source for that graphic is here:

http://www.antarcticglaciers.org/glacial-geology/dating-glacial-sediments-2/precision-and-accuracy-glacial-geology/

I modified it slightly, to make the “low accuracy, low precision” example less accurate, and the “high accuracy, low precision” example a bit less precise, and also to add their requested attribution; click here to enlarge it:

February 27, 2021 5:39 am

The problem is that if you remove the painted target, and are left with only the arrow holes (measurements), then you can’t tell the difference between the top row and the bottom row. When multiple measurements are highly consistent, that gives confidence in their precision, but if there’s systematic error in those measurements then their accuracy may still be poor.

OTOH, if your measurements are all over the place, that proves that they are of low precision, but they still might be of good accuracy. If you somehow know, a priori, that they are of high accuracy, and if you can take a large number of measurements, then you can reliably “find the bull’s eye” simply by averaging those measurements, even if few of the individual measurements were on target.

The problem is the temptation to simply assume that the measurements are of high accuracy, out of optimism, or because that makes it easier to justify your grants. That’s how you can end up reporting erroneous results.

Many, many things in climatology look like that. Measurement of sea-level by satellite altimetry is an obvious example, but there are many others.

For example, as reported in AR5, the CMIP5 climate models assume a wide range of values for even such fundamental parameters as radiative forcing from a doubling of CO2 (ERF), and climate sensitivity (TCR & ECS).

That proves that those parameters are known with poor precision. Yet alarmist climate modelers assume that by averaging the models’ outputs, they can nevertheless “find the bull’s eye.” That means they’re assuming high aggregate accuracy of models’ long-term (and thus effectively untestable) predictions, despite the known poor precision of the models’ internal parameters. There’s really no evidence to support such an assumption.

Last edited 1 year ago by Dave Burton
Philip
February 21, 2021 7:08 pm

Thanks very much.

Regan Howard
February 21, 2021 6:22 pm

Willis – This brings back ancient memories. Years ago I worked on ICESat-1 with Scott Luthke. The GIA accuracy numbers aren’t looney and I do remember going through them 20 years ago, but I can’t remember the process now for the life of me. The bigger issue and the source of the arguments is the assumptions going into the GIA estimate. I’ll have to go consult the ancient records.

Rhs
February 22, 2021 6:56 am

Clyde Spencer
February 22, 2021 11:52 am

Please do consult your ancient records. I’d be interested in how such high precision is rationalized.

Mr. Lee
February 21, 2021 6:49 pm

Satellite could say sea level is rising by 1 meter a day, but if the tidal gauges don’t change, then who cares. The gauges give information relevant to the concerns. The satellites do not. IMHO, when the satellite data is cited, it is usually in the form of a FOC attack.

Fear instilled by graphs procured by
Obscure methods, cited by people with impressive
Credentials