Vindication For Suyts?!?! New Tidal Gauge Sea Level Paper Out!! Reports 1mm/yr Sea Level Rise!!!

There’s a new sea level paper out and it’s very interesting!

TIDE GAUGE LOCATION AND THE MEASUREMENT OF GLOBAL SEA LEVEL RISE

It’s also fairly lengthy for a climate paper, and it uses some vernacular and terminology which I don’t often see in these papers.  They also throw in some statistical stuff which I’m unfamiliar with.  So, it may be a hard read for some.  Or, maybe it’s just me.  But, before we go, I’m want to show just a couple of things which may help the reader.  Well, I’ll put them below the meat of the post.  For a novice scroll to the bottom and look at the simple graphs to visualize some of the things the authors are talking about.   Naw, forget that.  They were too simple and too condescending in nature, and took up too much space.

I’m not easily impressed, but, I’m fairly impressed with this.  Mostly because in many ways, the authors echo what I saw.  For readers who weren’t with us in the early times of this blog, one of our first efforts as a blog was to look at our sea levels, both with satellites and looking at the tidal gauges.  I spent countless hours on the gauges.  In the end, I said, you can’t do it.  That is to say you can’t determine sea levels by the gauges without opening yourself up to just criticism.  There’s just too many pratfalls.  The authors, acknowledge this, but, they took it much further, and, may have helped resolve some of the problems.  Here’s the abstract.

Abstract
The location of tide gauges is not random. If their locations are positively (negatively) correlated with SLR, estimates of global SLR will be biased upwards (downwards). We show that the location of tide gauges in 2000 is independent of SLR as measured by satellite altimetry. Therefore PSMSL tide gauges constitute a quasi-random sample and inferences of SLR based on them are unbiased, and there is no need for data reconstructions. By contrast, tide gauges dating back to the 19th century were located where sea levels happened to be rising. Data reconstructions based on these tide gauges are therefore likely to over-estimate sea level rise.
We therefore study individual tide gauge data on sea levels from the Permanent Service for Mean Sea Level (PSMSL) during 1807 – 2010 without recourse to data reconstruction. Although mean sea levels are rising by 1mm/year, sea level rise is local rather than global, and is concentrated in the Baltic and Adriatic seas, South East Asia and the Atlantic coast of the United States. In these locations, covering 35 percent of tide gauges, sea levels rose on average by 3.8mm/year. Sea levels were stable in locations covered by 61 percent of tide gauges, and sea levels fell in locations covered by 4 percent of tide gauges. In these locations sea levels fell on average by almost 6mm/year.

And, this is a problem with the lunatics and their idiotic reconstructions.  While the abstract doesn’t address this, they do in the body.  Not only are the older gauges biased because they were placed where sea levels were rising, their spatial distribution is simply unacceptable for any serious reconstruction.

Any one babbling stupidity about how they know what our MSL was or how much the Sea Level Rise (SLR) was from 1900 is simply a deluded fool or a liar.  Likely both.  The authors are more charitable, but, they also understand something else which was very obvious to me when I was chasing this down.

Researchers are impaled on the horns of a methodological dilemma: to reconstruct or not to reconstruct. There are dangers in both. On the one hand, the conservative methodology might induce sample selection bias. On the other hand, the use of reconstructed data might induce reconstruction bias. This study shows that there is no sample selection bias in the conservative methodology because tide gauge locations are independent of SLR. However, reconstruction bias may be present because, as shown in this study, the global diffusion of tide gauges is correlated with SLR.

Well, that and a reconstruction going back to the early 1900s would be making data out of nothing at all.  They continue …..

Some investigators have used time series data on changes in sea levels rather than absolute levels because it is difficult to compare absolute sea levels in different locations. We show that this common practice might not be innocent; it may generate spurious trends in estimates of SLR. If sea levels are non-stationary, changes in sea levels will tend to be stationary. However, if sea levels are stationary, changes in sea levels might induce spurious time trends.

One of my favorite statements from the authors is a nice and simple one which needs repeated over and over again.

Rejecting the null hypothesis of stationarity is not conceptually equivalent to accepting the null hypothesis of non-stationarity, and vice-versa.

They carry that thought to practice, OMG!!!!  Real science!!!!  Read this!!!

We use both of these tests. However, we prefer the KPSS test for two reasons. First, if the claim is that sea levels are rising, the null hypothesis should be that they are stable. Second, as we explain in section 4, KPSS test statistics are more likely to reject the null hypothesis that sea levels are stable, than the DF test statistics are likely not to reject the null hypothesis that sea levels are rising. Therefore, the KPSS tests are more conservative and are more likely to show that sea levels are rising.

Just having someone in the climate arena acknowledge the “null hypothesis” is shocking enough, but, this outfit is using statistical tests going both ways to reject either way the question is framed!!!!  They get bonus points!  More bonus points for nailing spurious trends!

2.3 Spurious Time Trends
Suppose SLR is zero, and the trend is estimated using data on changes in sea levels as in Church and White (2008). The estimate of the average change in sea level during T time periods is defined as the change in sea level over the entire period divided by the number of observations, i.e.  = (YT – Y1)/T. Therefore, if the last observation in the data (YT) just happens to exceed the first observation (Y1)  is positive. Since SLR is zero a spurious time trend is estimated. The reason is that all the interceding observations, which are stable, have been ignored by looking at changes rather than levels.

Section 2.4 is a great thought experiment.  I’ve never seen a paper do that.  Good on them.  They’re actually giving a class!

They then go into what was very frustrating for me, when I was doing this.

When the data were retrieved (2011) PSMSL (www.pol.ac.uk/psml) comprised (in December 2009) 564,552 monthly observations of 1,197 tide gauge stations observed between 1807 and 2009. ……  The reporting periods for these tide gauges vary and range from 6 months to 203 years, with a mean of 39 years. It is clear from Figure 1 that not only were new tide gauges commissioned, but some tide gauges were also decommissioned. The number of tide gauges decreased sharply after 1995 because of reporting delays which are substantial. We exclude tide gauges with continuous records of less than 10 years since a decade is insufficiently long to check for non-stationarity. Out of the 1197 tide gauges in PSMSL we excluded 197 with fewer than 10 years of consecutive data. This leaves 1000 tide gauges which we use in our analysis.

Well, there’s “continuous” and then there’s continuous.  Our approaches were very similar but, our criteria was a bit different.  But, every one has to draw a line somewhere.  I wanted 30 years but ended up running with 20.  And, I was specifically only looking for recent data.  What I often found was a station with great data continuity and then about 2000 or so, it would just stop.  For my purposes I didn’t include any which didn’t at least get to 2007.

Only 140 (Figure 2) of these 1000 tide gauges had complete data records. Many records are incomplete. Figure 3 presents the distribution of the number of monthly missing values in the dataset. Forty tide gauges are missing one data point. In some cases missing data are clustered, i.e. the data gaps are consecutive, while in other cases they are sporadic. We divided missing values into three categories: i) sporadic missing values, ii) 2 – 12 consecutive missing values (up to 12 months), iii) over 12 consecutive missing values. In the first case we imputed the missing value by the average of the data before it and after it. In the second case imputation is based on interpolation (see Data Appendix). In the third case the data are split and treated as separate segments. These imputations and interpolations should be distinguished from reconstructions because they refer to existing tide gauges rather than to tide gauges that do not exist.

A big, “duh”, but, these guys knew this, but showed it ……

….The main conclusion from the censored regression model (tobit) is that the number of tide gauges varies directly, as expected, with GDP per capita, population and length of coastline. It also varies directly with SLR (as measured by satellite altimetry) but this effect is not even remotely statistically significant.

One of the things I do not like about this paper is that they rely on satellite telemetry for some of their tests for positioning, but, then we see this …..

It is notable that tide gauges with positive trends are co-located with tide gauges that are trend-free. The same applies to the minority of tide gauges that happen to have negative trends. This surprising pattern may also be found in Map 2 which plots the satellite altimetry data where dark blue (large negative SLR) grid points are located in the vicinity of dark red (large positive SLR) grid points.

Now, ask your self if that’s realistic?  Sure, for the gauges that’s entirely possible, because of erosion and silt buildup and all sorts of physical changes to the landscape.  But this?

Yeh, it prolly happened just like that.  There’s a huge ridge off the coast of Japan that was forming for nearly 20 years.

The substantive contribution of the paper is concerned with recent SLR in different parts of the world. Consensus estimates of recent GMSL rise are about 2mm/year. Our estimate is 1mm/year. We suggest that the difference between the two estimates is induced by the widespread use of data reconstructions which inform the consensus estimates. There are two types of reconstruction. The first refers to reconstructed data for tide gauges in PSMSL prior to their year of installation. The second refers to locations where there are no tide gauges at all. Since the tide gauges currently in PSMSL are a quasi-random sample, our estimate of current GMSL rise is unbiased. If this is true, reconstruction bias is approximately 1mm/year.

And, then, here is the approach I should have taken …….

Sea level rise is regional rather than global and is concentrated in the southern Baltic, the Ring of Fire, and the Atlantic coast of the US. By contrast the north-west Pacific coast and north-east coast of India are characterized by sea level fall. In the minority of locations where sea levels are rising the mean increase is about 4 mm/year and in some locations it is as large as 9 mm/year. The fact that sea level rise is not global should not detract from its importance in those parts of the world where it is a serious problem.

The paper is too lengthy for me to cover all that they did, but, is a good read, and by all appearances very supportable.  The issues they ran into and partially resolved, were the very same issues I identified.  You have stations right next to each other one showing a fall and the other showing a rise.  And, you still have another problem.  The map in the paper showing the stations at the year 2000 may be misleading.  Sure enough, the stations are there, off the coasts of Africa and S America.  But, as I recall, there was about 3 with data one could use for current study.  Same can be said for the south pacific.  Which I thought odd, until I remember the huge sums of money at stake for proclaiming that something wasn’t a dire emergency.

I think this paper should generate much discussion.  It utterly destroys the lunatic meme of sea level rise, and pretty much confirms early work done here.  I kinda gives me a warm fuzzy, but, at the same time, I’m now wishing that I hadn’t simply given up.  That said, some of the statistical kung fu they did was beyond my scope of knowledge and ability, so ….

Still, this should make waves!!!  (heh!)

h/t tom0mason