Sea Level Rise: U.S. East Coast (II)

In the last post I defined the four regions of the U.S. east coast for which I’ve created a regional sea level estimate (since 1950). Northernmost is my New England (NE) region, which includes the coasts of Rhode Island and Connecticut, even the tip of Long Island:

And here’s the regional monthly sea level:

Interesting stuff, but at the monthly time scale there’s so much fluctuation it’s hard to see what’s going on. The information is still there of course, but it’s not always clear to the eye. A graph which may be more informative visually (but is certainly no more informative mathematically) is that of yearly averages rather than monthly averages:

Of course sea level is on the rise. But it’s not rising at a steady rate (i.e. following a straight line), there’s more than that going on. A simple way to show this is to fit a quadratic curve, a parabola; it’s strong statistical significance means we know right off the bat that it’s not just a linear trend plus noise.

Some of the better estimates of the actual trend are from a good smoothing method (I like the lowess smooth) and from a piecewise-linear fit, a function consisting of two straight lines which meet at their endpoints. These two strategies give similar estimates:

The piecewise linear fit has its “turning point” in 1990. The high quality of this fit doesn’t mean that the rate of sea level rise suddenly jumped to a higher level and has since held steady at that higher rate — even though that’s the basis for the piecewise linear model. What it really signifies is threefold: first, its statistical significance proves (again) that the trend is more than just a straight line; second, it shows that the “sudden jump to higher rate” model is at least approximately correct, i.e. there’s something to be learned from it; and third, it gives us an estimate of the average rate of sea level rise during the two disjoint periods, before and after the turning point.

Those two rates (of sea level rise) for New England turn out to be 1.5 +/- 0.8 mm/yr during the 1950-1990 period, but 4.1 +/- 1.2 mm/yr for the period since 1990. We can also estimate the rate from the lowess smooth, and compare that to the piecewise-linear model (smooth in red, piecewise linear in blue):

The agreement is pretty good, but keep in mind the wise words of statistician George Box, that all models are wrong, some models are useful. Bottom line: the NE region has shown acceleration of sea level, acceleration which is substantial.

More to come …

This blog is made possible by readers like you; join others by donating at My Wee Dragon.


12 responses to “Sea Level Rise: U.S. East Coast (II)

  1. But Judith Curry says…never mind. Just kidding by bringing up JC and her “paper” (a manuscript, apparently.)

    • We know land ice is retreating in both the Arctic and West Antarctic.
      The latest research also suggests East Antarctic is beginning to react to rising ocean temperature .
      As tamino has repeatedly shown we see the expected result in sea level measurements.
      Yet still the fringe dwellers fool them selves that we are not facing a ongoing retreat from coastal regions world wide .
      Deniers produce ranges that are nonphysical using all the available real world data. Currys range starts at 200mm by the end of the century.Sea level rise will be a lot more than the present global 3.4mm a year.
      Retrograde beds below sea level in the already effected glacial regions make it a certainty that the retreat will not only continue but must accelerated with time.

  2. Looking forward to the North Carolina Outer Banks report.
    I realize that’s a moving target:

  3. David B. Benson

    This is sea level relative to the local land. Some of that is, I opine, isostatic adjustment which ought to fit a sinusoid, close enough. But the recent increase in rate is something else, l should think.

  4. Minor nit: *its* strong statistical significance

  5. I think you need to be careful before reaching a conclusion here, not that you are doing that at this point. If you extended your analysis back another 30 years to 1920 or so, you would see a period of higher sea level rise rate during the 1920-1950 period similar to what is happening from 1990 thru the present. Go to the NOAA site:
    Click on one of the tide gages, then click on linear trend, then the Variation of 50-year RSL tab. For those stations in the Tamino group with a long and complete history, each one shows the higher sea level rise rates in the 1920 – 1950 period than in the 1950 – 1990 period. While there is a long-term warming trend in the data, as shown by a linear trend line, there appear to be multi-decadal oscillations above and below that trend line. What that signifies I do not know, but something is going on in addition to a recent global warming link to sea level rise.

  6. My somewhat trained eyes do not see an acceleration in the data annual mean data. So I would be interested in some more technical details what you did.

    Did you test a linear trend with AR(1) noise against a bi-linear trend with unknown breakpoint and AR(1) noise? And was the fit of the latter so much better that it justified the more complex model?

    Did you do this testing on the monthly or on the annual data? The monthly data is quite noisy and the noise is not very (short-range) autocorrelated, could be measurement uncertainties. The annual data has quite strong autocorrelations, which I would expect to be real and partially due to ocean currents.

  7. Bear in mind that , concomitant with lower sea levels at peak glacation, the glacial load from which this region is rebounding ( and the terminal moraines) once extended as far out as the Hudson Canyon-