Wednesday, 23 September 2015

"Hotspot of accelerated sea-level rise on the Atlantic coast of North America" - finally laid to rest. RIP

There's been some discussion about sea-level rise on the Eastern seaboard of the U.S. in the comments on the recent "Doubling up the sea level scare for Paris using the old ‘one-two punch’ line" post at WUWT. I posted a couple of charts there, and I intend to return to the Sallenger et al article in this post's title. They used windows of  60, 50 and 40 years for their analysis, and showed this chart for New York; incidentally the only time-series chart in the entire article - rather surprising since they analysed dozens of gauge stations around North America.
Supplementary Figure S7.  New York City annual average sea level data with three
regression results used in this study for the most recent 60-year subsample (1950-2009).
The mean elevation of the annual NYC sea level data (1893-2009) was removed and time
t = 0 was set at year 1950 (t = year – 1950) for these regressions.

I'm in the habit of using a variety of techniques to analyse gauge records, especially long ones like New York, as I did in a previous post about the "Hotspot". Those two 30-year regression lines above prompted me to plot a 30-year sliding window for New York, using the data from 1893-2013, slightly longer than Sallenger et al, who used PSMSL annual average data to 2009 - I've used annual averages calculated from monthly data. Their data had three recent years missing because a few monthly data points are missing, and PSMSL exclude such years. I suspect the differences are small though.

Data from PSMSL

The first thing that's obvious is that 30-year periods ending in the 1950s actually had rates higher than the 1980-2009 period shown on their chart above. So the modern acceleration is nothing new; in fact the earlier acceleration was more rapid. It also shows that projecting a 30-year rate forward to 2100 is both unscientific and ignoring history completely. I certainly don't expect future years to mirror the big fall in rates after the mid 1950s, but I do expect something of a drop. Sallenger et al just didn't look for any cyclical pattern in the gauge data. They mentioned it was possible that there were such patterns, but their analysis was effectively designed to mask it by comparing differences in rate between adjacent windows. If that was plotted, the 1950s peak above would be a trough! Why do I expect a gradual drop in rate? Because it's already underway!

Rates (mm/year) for 10 year (121 month) sliding windows.

Note how irregular peaks "kicked in" at the end of the 1960s; the latest peaked after Sallenger et al's 2009 end date, in 2010/11. Sandy Hook is an island S of the entrance to NY harbour. A more up-to-date (to 2014) plot shows the descent from the last peak more clearly:

Rates (mm/year) for 10 year (121 month) sliding windows.

Finally, here's the rate evolution, Rates are computed from 1950; the final point is 2014. When I've time, I'll update all my NY charts to date, and replace any above that need it.

In 2009, the rate was still rising; now it's levelled off. Sallenger et al's projection is already out-of-date. RIP

No comments:

Post a Comment