From Eurekalert
Public Release: 8-Aug-2017
Untraditional approach expected to save lives, businesses, and communities along East Coast shows rate accelerating at a pace in contrast to previously accepted data
American Statistical Association
ALEXANDRIA, Va. (August 8, 2017) – While the scientific community has long warned about rising sea levels and their destructive impact on life, property and economies of some of the United States’ most populous cities, researchers have developed a new, statistical method that more precisely calculates the rate of sea level rise, showing it’s not only increasing, but accelerating. The research, methodology and current findings was presented by Andrew Parnell of University College Dublin at the Joint Statistical Meetings (JSM) last week in Baltimore.
The new approach contrasts with previous ways scientists analyzed and came to conclusions about sea level rise because it is “the only proper one that aims to fully account for uncertainty using statistical methods,” noted Parnell, principal investigator of the study conducted collaboratively with researchers at Tufts University, Rutgers University and Nanyang Technological University.
By examining two data sets, one that consisted of measurements from sediment along the East Coast from 2,000 years ago and another that included tide gauges around the world dating back to the 1800s, Parnell and his team discovered the data they gathered from years ago contained uncertainties. For instance, with more tide gauges deployed today than hundreds of years ago, recent records yielded more certainty than older ones. The team honed their statistical models to further take into account such uncertainties and possibly created a statistical first. “This likely is the first time a group of statisticians have had really close examination of sea level data,” said Parnell.
Parnell’s team has been able to show that sea level rise on the East Coast has been much less than 1 millimeter (mm) per year for the entire period 0 AD to 1800 AD, and, since then, it’s skyrocketed. In fact, they’ve discovered the rate of sea level rise on the East Coast is the highest it’s been for at least 2,000 years, and the rate of global sea level rise is above 1.7 mm per year, estimated by the International Panel on Climate Change. “Some people argue that sea levels are not rising. We are showing them that sea levels are not only rising, but accelerating,” continued Parnell.
From their analysis, researchers made additional observations, including the following:
- An increase in the rate of sea level change around the time period known as the “Medieval Climate Anomaly”
- A small decrease around the time of the “Little Ice Age”
- A rapid increase after the start of the Industrial Revolution
The new model has recently been put to the test in New York City, where the rate of sea level rise is more than 3 mm per year in an area that currently houses more than $25 billion of infrastructure at less than 1 meter above sea level. Researchers anticipate the model will be rolled out in other cities along the East Coast and hope governments will be receptive and prepared to take the issue of sea level rise seriously.
###
About JSM 2017
JSM 2017 is the largest gathering of statisticians and data scientists in the world, taking place July 29-August 3, 2017, in Baltimore. Occurring annually since 1974, JSM is a joint effort of the American Statistical Association, International Biometric Society (ENAR and WNAR), Institute of Mathematical Statistics, Statistical Society of Canada, International Chinese Statistical Association, International Indian Statistical Association, Korean International Statistical Society, International Society for Bayesian Analysis, Royal Statistical Society and International Statistical Institute. JSM activities include oral presentations, panel sessions, poster presentations, professional development courses, an exhibit hall, a career service, society and section business meetings, committee meetings, social activities and networking opportunities.
About the American Statistical Association
The ASA is the world’s largest community of statisticians and the oldest continuously operating professional science society in the United States. Its members serve in industry, government and academia in more than 90 countries, advancing research and promoting sound statistical practice to inform public policy and improve human welfare. For additional information, please visit the ASA website at http://www.amstat.org.
For more information:
Jill Talley
Public Relations Manager
(703) 684-1221, ext. 1865
More incongruous data splicing:
“By examining two data sets, one that consisted of measurements from sediment along the East Coast from 2,000 years ago and another that included tide gauges around the world dating back to the 1800s …”
The science bar is how low for NY political finger pointing?
I am very skeptical when talking about changes that are so small. The level of accuracy would seem to be a greater factor than the theoretical rate of change. It just seems like an excuse to cry wolf and keep the fear of climate change in the news.
“Some people argue that sea levels are not rising”
Some people argue that climate change is going to kill us all within the next few years. Probably about the same number in both cases.
As authoritative and believable as political “approval” poles.
I love skeptics. They demand review from real statisticians until they read the results.
Then they claim stats are lies.
Tell that to nic lewis or steve mcintyre.
You don’t need stats or tests. Just ask Rud.
He disproved everything by waving his arms.
Data? Don’t need it..It’s all just stats and lies.
Words…
Numbers…
Now… Find an example of something “skyrocketing” at 1.9 mm/yr.
Hey, I guess you would have to read the paper rather tha eyeballing graph.
Makes me sure they don’t know what “accelerating” means.Find accelerating on the graph above.
That’s one heck of a good graph, David. Love the ruler. 🙂
The 1860-1900 ave. is 1.9 mm/yr. The 20th century ave. is 1.7 mm/yr so they claim. It slowed down! LOL
If it ain’t in the graph, then no amount of “words” can make it magically appear.
It is difficult for me to take nerds seriously when they use “skyrocketed” to refer to something rising ten times slower than my fingernails grow.
It always bugs me when they talk about a meteoric rise of something or someone.
SM, I am among various other things a PhD level econometrician (statistics applied to economics) and systems modeler– stuff like recasting the partial differential predator/prey calculus equations as propabalistic Markov chains then proving their model equivalence.
The reason I use only simple irrefutable facts (you know, like bogus regional expectations BEST qc on station 166900) is that most people here are not at that technical level. But they are intelligent thinkers. I do wave my arms at easy to verify, solid facts. Heck, even cite them, and link to the primary sources. Works great for the licensed lawyer I also am. My jury is the WUWT readership (or equivalent Climate Etc denizens). They get to decide whether I have made a case for ‘conviction’ on whatever the topic. Not you alone, as demonstrated many times. For factual examples of this generalization, see my recent guest posts here ‘SLR and Closure’ and ‘Why Models run Hot’. Your to do is to explicitly identify any of my alleged arm waves in those two posts.
Doubt WUWT will hear back from you concerning this modest rebuttal challenge.
Your posts here wuwt ristvan are greatly appreciated by many.
Ditto
Although intelligibility coincides too rarely with relevance and truth in Mr. Mosher’s comments for me very often to divert my little remaining bandwidth to considering them, I am impressed by those who like Mr. Istvan can take the time to do so.
I am also impressed that Mr. Istvan is capable of “recasting the partial differential predator/prey calculus equations as [probabilistic] Markov chains then proving their model equivalence” yet restricts himself to “simple irrefutable facts” for the benefit of readers like me who “are not at that technical level.”
But I hope it will not be taken amiss if I make a modest suggestion regarding his contributions: that he give a little more thought before expounding. The errors that result from failing to do so compromise what confidence at least this layman accords those presumably worthwhile contributions.
I am prompted to make this suggestion by his last comment’s reference to “Why Models Run Hot.” It reminded me of another of Mr. Istvan’s posts, in which he referred to the mathematical derivation of Monckton et al.’s “irreducibly simple” equation as “impeccable.” Although I’m just a retired lawyer, I did not remain completely unaffected by those isolated scraps of mathematics that came my way, and it happens that Mr. Istvan’s post touched on one of the few things that actually stuck. As a consequence, I know that Monckton et al.’s derivation was the exact opposite of impeccable.
In effect the Monckton et al. paper said the response of a time-invariant system with state can creditably be approximated as that of a stateless time-variant system. In the relevant discipline, this is akin to saying that the product of two numbers is creditably approximated by their sum. So it strikes me as startling that someone capable of “recasting the partial differential predator/prey calculus equations as [probabilistic] Markov chains then proving their model equivalence” could pronounce such a remarkable proposition’s derivation “impeccable.”
It is perhaps unjust to Mr. Istvan that this misstep has placed an asterisk next to all of his contributions at least in my mind and possibly in others’. Still, such results follow from human nature, and maybe a little more care would reduce their occurrence.
(Also, providing actual links when he refers to his other writings couldn’t hurt.)
Er no. Sceptics are sceptical. See how that works?
And bad statistics are what we sceptics call bad statistics. See how that works?
But let’s accept your point nevertheless – if we sceptics disagree only with what we don’t like, then you non-sceptics must be the same, yes? Or is your basic argument that you are virtuous and we are not? Those are your two choices: which one are you going for?
I would go with the virtue one.
Killing people is virtuous?
Who mentioned killing people?
Pheonix44 offered a false dichotomy, but if forced to choose I would go with the virtue rather than that everybody disgrees with what they don’t like. Some of us look at all the evidence.
To Little Stevie, nobody is permitted to review and criticize work that agrees with him.
I wonder what the result and what kinda of certainty will be produced by this statistical analyzing if the same data they use is scrambled randomly first……! or if only the 2000 year record mash-scrambled randomly first…!
I suspect that the “mash potato” would not be much different…..
cheers
“The team honed their statistical models to further take into account such uncertainties and possibly created a statistical first. ”
The phrase that pays, “Statistical models”, and the giveaway “Possibly created a statistical first”, all in one sentence.
Looks like lowering the past and raising the present, again.
To take account of uncertainties simply means increasing the possible range of the data. And if you do that, then historical rates of sea level rise could be lower than we thought. Which means sea level could now be rising faster than we thought compared with those rates.
There we go, I’ve written the paper.
I haven’t proven a single thing though, because it is literally impossible to do so with this methodology.
So,,,, would NYC qualify as an Urban Subsidence Island (USI)?
What does all that man made structure do the ground below it? Kinda like man made glacial forcings?
Remember, you heard it here first 😉
Next: the acceleration is accelerating.
BB,
The term for the time derivative of acceleration is the “jerk”. Geoff
Of course, it is strange that the New York City regional tide gauges are all linear, and have stayed that way for over a century. No acceleration visible whatsoever! NONE!!!
In Sweden we have some pretty long sea-level series. This is because the isostatic rise of the land is very noticeable, and has always been known to people along the coast and have attracted the interest of scientists all the way back to Linnaeus. And there are no tides to complicate things in the Baltic.
In southernmost Sweden the rise is so slow that the sea is gaining, but in most of the country the sea is receding. And after 200 years we know pretty well where the zero line is.
Here is sea-level data from Kungsholmsfort, an old coastal fortress situated almost exactly on the zero line:
http://www.psmsl.org/data/obtaining/rlr.monthly.plots/70_high.png
The (annual) relative sea-level in 2016 was 1 mm lower than in 1887. So at least we know for sure that the total change over 130 years is much less than the interannual variation.
Now in recent years GPS have been installed alongside the old (non)tidal gauges, and they show that the absolute rise at Kungsholmsfort is about 2 mm/year. So, yes, sea-level is rising about 2 mm/year. If it was 3,5 mm/year the zero line would be several hundred kilometers further north.
And no, there is no trace of an acceleration in the long series. Well, that is not quite true, in the very longest series there are some evidence for an acceleration around 1850.
That is a beautiful confirmation of Nils-Axel Moerner’s analysis of the diff GPS corrected long record tide gauges. Many thanks. If I write this up again (another ebook coming?) you will be given full credit and a link to this excellent comment. Many thanks for this Swedish fact contribution. Is now permalinked on my iPad and Mac.
Wonderful! I just complained above that no one had done this. I’m happy to be wrong!
What is meant by “absolute” sea level rise? You say there’s isostatic land rise. So is 2mm the apparent rise according to tidal gauges?
Willis E. has posted a couple of times recently showing the acceleration is SLR is not significant.
So I thought I would grab the data set for Boston and try my hand at it. (It goes back to 1921)
Wow, did I ever make a mess.
I loaded the data set into R-Studio, and did a linear fit, then a cubic fit.
All terms are *highly significant*
Apparently, the p values in the fit summary are not the ones I thought they were.
Anyway, here is the fit summary for the linear model:
Call:lm(formula = SLdata ~ xdata)
Residuals:
Min 1Q Median 3Q Max
-0.144354 -0.031264 -0.001789 0.029746 0.212170
Coefficients:
Estimate Std. Error t value Pr(>|t|)
(Intercept) -1.850e-01 2.926e-03 -63.21 <2e-16 ***
xdata 2.353e-04 4.379e-06 53.73 <2e-16 ***
—
Signif. codes: 0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1
Residual standard error: 0.04982 on 1156 degrees of freedom
Multiple R-squared: 0.7141, Adjusted R-squared: 0.7138
F-statistic: 2887 on 1 and 1156 DF, p-value: < 2.2e-16
And for the cubic model:
Call:
lm(formula = SLdata ~ poly(xdata, 3, raw = T))
Residuals:
Min 1Q Median 3Q Max
-0.151159 -0.028853 -0.001482 0.025511 0.203899
Coefficients:
Estimate Std. Error t value Pr(>|t|)
(Intercept) -2.235e-01 5.609e-03 -39.843 <2e-16 ***
poly(xdata, 3, raw = T)1 6.255e-04 4.200e-05 14.893 <2e-16 ***
poly(xdata, 3, raw = T)2 -8.304e-07 8.438e-08 -9.841 <2e-16 ***
poly(xdata, 3, raw = T)3 4.734e-10 4.794e-11 9.875 <2e-16 ***
—
Signif. codes: 0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1
Residual standard error: 0.04787 on 1154 degrees of freedom
Multiple R-squared: 0.7364, Adjusted R-squared: 0.7358
F-statistic: 1075 on 3 and 1154 DF, p-value: < 2.2e-16
As you can see, all the terms for the polynomial fit are utterly improbable. I have no idea where I have gone astray. The thing that is most vexing to me is I *used* to know how to do this. I could set up the ANOVA matrix, calculate the "Sums of Squares of Everything in Sight", and carry on.
Anybody who is handy with R stats, cares to sort this out for me, I would very much appreciate it.
I suggest you look at the data first. If you can’t see a trend there isn’t any. If you think you can see a trend, check it statistically. It quite likely isn’t real. The human eye is incredibly good att spotting patterns, so good that it often sees patterns that aren’t there.
Also hydrological data frequently follow Hurst-Kolmogorov distributions. A Hurst-Kolmogorov distribution (a. k. a. Fractional Gaussian distribution) does not look random, but it is.
TonyL,
Perhaps Mosher will be good enough to show you the error of your ways. He can probably do it in three words or less, as that is his usual mode of communication.
*Thanks*
Actually, I was hoping for Willis E. because I know he uses R programming.
Maybe risvan, because he has looked extensively at sea level, as well as all kinds of other things.
Or anybody well versed in stats or R in the WUWT community.
So far, Zippo.
As the correspondents on rhelp often write, without a sample of the data in the format you employed, there are problems understanding your problem. What is “xdata” for instance? And, given the values of the simple linear fit, why did you bother with a poly fit?
The Ship at Dunwich was fantastic and still I hope (and know) above water!
Statistics are metadata, not new data.
Statistical methods can never add precision or new information to a data set. Only bias.
If I lie often enough for long enough, eventually you rubes will believe it! It’s for your own good!
NASA ‘is’ finally realising it is the ocean currents taking heat pole-wards that are responsible for global warming and not the magic molecule. NASA’s Advanced Supercomputing (NAS) facility has been recruited into the task force to decipher mystery of the ocean currents
https://www.nas.nasa.gov/publications/articles/feature_ocean_vis.html
The paper might be here https://projecteuclid.org/euclid.aoas/1437397101. It’s paywalled. Here’s the abstract.
The word “Crackpot” comes to mind.
Don K,
How can they claim “unprecedented” when they don’t have differential GPS correction for the last 2000 years, nor accurate dates for the sediment proxies for tide gauges?
Don K, thank you for the link.
When did statistics become more accurate than actual observation. Silly me, I thought science involved ACTUAL observation and measurement. Statistics are applied mathematics.
Ground water levels are dropping by meters and sea levels rise by millimeters. Seems to make sense. How might one estimate soil erosion/depletion over the last 200 years?
http://www.waterworld.com/articles/wwi/print/volume-25/issue-5/groundwater-development-flow-modeling/groundwater-depletion-linked-to-rising.html
Has anyone analyzed the surface levels of the earth? It seems to me that shapes of the earth surface (and hence, the overall surface) and relative location of the continents are constantly changing by very small amounts. So if the shape of the bowl is changing and heights of the continents are changing (mountains formed, sub-sea mountains, post-glacial expansion, etc.), it is likely that the global waterline will also change, regardless of the whether or not the volume of water in the bowl is increasing or decreasing. If the mountains are rising at rates of mm/y (or faster for sub-sea), why is surprising that sea levels are changing. Instead of focusing on ice-melt, perhaps we should look at other causes for sea level rise changes.
Seems like we have been over this time after time with numerous “peer reviewed papers”. I’m sticking with 4 to 7 inches per century globally, with no noticeable acceleration in SLR…101.6 mm to 177,4 mm per century.
It’s amazing how many here form their perception of sea level rise without any clear concept of what time-scales are disregarded (what frequencies are filtered away) in the assessment Parnell’s perspective is entirely on the highly smoothed millennial scale of proxy data (see Figure 6 in:
https://www.researchgate.net/profile/Benjamin_Horton/publication/281136939_Relative_sea-level_change_in_Connecticut_USA_during_the_last_2200_yrs/links/55dc3d6408aed6a199ac855d/Relative-sea-level-change-in-Connecticut-USA-during-the-last-2200-yrs.pdf )
All the purported counter-demonstrations here based on trend-fitting unsmoothed tide-gauge data are largely irrelevant to the point that he makes.
Make that Figure 8.
If you cannot measure it you do not know it , but you can ‘guess it’ and throwing statistics at it does not change that . There is no ‘magic’ that can over come past issues with measurements when you simply no idea what those issue even were . You merely add guess work into data which you consider has an problem but you have no idea of the range , direction or size of the problem .
Bunkum …. New York State Sea Level Rise Task Force Report to the Legislature — Dec 31, 2010 [ http://www.dec.ny.gov/docs/administration_pdf/slrtffinalrep.pdf ] gives (when translated to mm) a figure of 2.5 to 2.77 mm average Relative Sea Level Rise since 1960. (The last 50 years or so.)
The Battery at NY City is known from Continuously Operating Reference System (NOAA CORS) to be subsiding at a rate between 1 and 2 mm per year. The latest subsidence numbers for the Battery, NYC come from this paper Using global positioning system-derived crustal velocities to estimate rates of absolute sea level change from North American tide gauge records by Richard Snay et al. at NOAA NGS . A vertical movement of minus 1.35 mm/yr (SD 1.74). The Battery is sinking towards the center of the Earth. It speeds up and slows down, but that is the long term average lately. Some months it is up 10 mm, some months, down 10 mm….the error bars are even larger.
I must point out that statistics are not measurement. — not now, not ever.
Here are the MEASUREMENTS: (This is RELATIVE sea level rise, remember)
“The mean sea level trend is 2.84 millimeters/year with a 95% confidence interval of +/- 0.09 mm/yr based on monthly mean sea level data from 1856 to 2016 which is equivalent to a change of 0.93 feet in 100 years. ”
Not the statistically manufactured “0ver 3 mm/yr”.
Subtracting the subsidence of 1.35 from 2.84 leaves us 1.49 mm/yr of actual rising sea water — less than the Global Average.
Of course, NY City only cares about relative sea level rise — its the only kind they wlll ever experience — but they are experiencing nothing out of the ordinary.
Thank you for this info, Kip!
Actually, 1.49 mm/yr is almost exactly equal to my best estimate of the current rate of GMSL rise.