Sunday, 4 August 2013

Plagiarism, Lies and Bullshit - Part 1

Back in April this year, Anthony Watts gave prominence to a "paper" by Nils-Axel Mörner & A. Parker - Australian sea level data highly exaggerated, only 5 inches by 2100. It's clear that neither he, nor any of the commenters on that post actually appear to have read the "paper", or if they did didn't do so with a critical or sceptical eye. Are they all sceptics in the general sense, or not? I'm an AGW sceptic, and I'm also a sceptic in the general sense - I believe nothing I read until I've checked out its conclusions, claims, methodology, and sources in detail. I've done all those checks on the "paper" (I use inverted commas deliberately), and have found all that this post's title implies. The "paper" is titled "Present-to-future sea level changes: The Australian case", and can be found here.

Mörner & Parker claim that the Australian government, aided and abetted by Australia's National Tidal Centre. a branch of the Bureau of Meteorology, have adopted an "exaggerated" figure for recent sea-level rise as "official". None of their cited sources contain the figure, nor statements of recent acceleration as claimed  - it's a "straw man". Indeed, the NTC takes pains to stress the short-term nature of the trends reported for its 16 SEAFRAME stations managed by their Australian Baseline Sea Level Monitoring Programme (ABSLMP), and warns against placing reliance on the data at present. There's a term which describes making unsupported claims, as M&P have done here, especially when the cited sources actually say just the opposite.

Astonishingly, this "paper" contains no data whatsoever to support its "conclusions". Despite having claimed to analyse "86 stations", there is no list of any stations analysed, no tables of results and just one chart for Fremantle, which in fact neatly refutes one of the claimed results (and more besides). Several "analyses" are claimed to have been carried out, but aren't mentioned again, and so no results or summaries are given. Not a single station trend for any relevant period is quoted.

Even more astonishing is not even a list of the 16 "ABSLMP" (Australian Baseline Sea Level Monitoring Programme) stations, which are supposedly central to the "paper" is given. The reader is left in the dark; reproducibility, a central requirement of published papers, would be impossible without a deal of research on the part of anyone attempting the process. I'm not in that position however - I've been studying all aspects of Australian sea-levels in some depth over recent years. Because of that knowledge, and the spreadsheets and charts to hand, I knew instantly that the claimed summary results were fiction. How much of a fiction will be seen below. The author(s) appear to believe that 1990-2010 spans 20 years, a figure they mention several times.

This is their claimed methodology:
We fit linear and 2nd order polynomial lines to the sea level data recorded along the coasts of Australia in order to assess the accelerating trends and to compare with the reconstruction of Church and White[12]. 
If Y is the mean sea level (MSL) and X is the year, then clearly the sea level rise is SLR=dY/dX and the sea level acceleration is SLA=d2Y/dX2. The linear fitting gives the average SLR over the observation period. The 2nd order polynomial fitting gives the average SLA over the observation period.
There's no further mention of  any "2nd order polynomials" at all. "The 2nd order polynomial fitting gives the average SLA over the observation period." - no it doesn't, it gives the equation for a curve.

What about "rebutting" several prominent authors on sea-level rise?

Out of this highly variable spectrum, Douglas [16] selected 25 records and arrived at a mean sea level rise of 1.8 mm/year, Church et al. [11] selected 6 records and arrived at a value of 1.4 mm/year, and Holgate [20] selected 9 records and arrived at 1.45 mm/year.

The word "selected" is meant to infer some kind of "cherry-picking". However, two of the three cases, you won't find the claimed figures for number of stations "selected", and in none  the claimed statistic. Comparing Australian data to global data puts it in context, and no more.

Douglas didn't "select" anything. There were only 23 (not 25) long-term stations available to analyse, and his result was global. Douglas didn't estimate a figure for overall sea-level rise. Peltier (2001) used Douglas's data, corrected for GIA, and came up with 1.84 mm/year.

Church & White (hardly "et al." at all) used 290 stations, and don't mention a set of 6 stations in any context. You'll search in vain for the "1.4" figure, their figure is 1.7 ± 0.3 mm/year.

Amazingly enough, Holgate did actually select 9 stations for analysis; his intention was to test whether a small set of high-quality, continuous (no gaps) records to "composites" of many sets of world-wide gauge data of varying lenght and quality. His estimate was 1.74 ± 0.16 mm/year, and not 1.45.

Here's a list of claimed analyses, and results given for the Australian stations:

70 "non-ABSLMP" stations 1990-2010, linear trends: average 0.1 mm year
2nd. degree polynomials for same: none
70 "non-ABSLMP" stations 1990-2000 linear trends: none
2nd. degree polynomials for same: none
70 "non-ABSLMP" stations 2000-2010: none
2nd. degree polynomials for same: none
16 ABSLMP stations 1990-2010 linear trends: none
2nd. degree polynomials for same: none
All 86 stations 1990-2010, linear trends: average 1.5 mm/year
and "No fitting produces a sea level rise in excess of 1.5 mm/year".

Those last two claims alone demonstrate that the "results" aren't results at all. - they're fabricated. If the average of the 86 is 1.5 mm/year, then some must have been lower, and some higher, but we're told none was higher than 1.5 mm/year. Therefore they must have all been identical at 1.5 mm/year, yet we're told that 70 of the 86 averaged 0.1 mm/year. Also, of the 70 stations that averaged 0.1 mm/year, some must have been higher, and some lower, with many negative. Now that would have been remarkable, in the scientific sense - news to broadcast. M&P haven't even realised the implications of their laughable "statistics".

Check out that last claim against my analysis of the "86 stations" - actually 85. Six stations are so remote from Australia they aren't relevant at all - thousands of km away. I've included them however, to get overall statistics to compare. I've started my list beginning on the north Queensland coast in the Gulf of Carpentaria, progressing clockwise via Tasmania to Milner Bay, Northern Territory, which is also on the Gulf, ending with the remote and decidedly non-Australian stations.

Station 32 is Sydney, which we're told had a "negative trend", though we aren't informed what it was -  details later.

Stations  65-79 are on the W and NW coasts; 66 is Fremantle, 77 is Darwin. 80-85 are "off the map" and very distant. Spaces on the charts indicate that no meaningful analysis for those stations was possible because of very limited data for the period. There were no zero trends, and none was negative.

Full station details are shown in the table at the end of this post.

The average for all stations is 4.05 mm/year, with standard deviation 2.51, minimum 0.17 mm/year, maximum 9.38 mm/year. The average for the 16 ABSLMP stations is 4.35 mm/year (not quoted by M&P, though central to their theme), and that for "non-ABSLMP" stations is 3.90 mm/year. Contrast those figures with the claims of M&B-P, 0.1 mm/year for the "non-ABSLMP" stations, and 1.5 mm/year overall.

There's a claim that the trend fo Sydney was negative over this period, but In fact it's positive over 1990-2010, and more positive over the actual 20 years 1991-2010. Here's the proof, with both trends:

Trends for Sydney, 1990-2010 and 1991-2010                                            Data Source: NTC
Strangely no trend figure is given for one of the most analysed stations worldwide, though perhaps not so strange because it must by now be clear that M&P haven't analysed any Australian stations over "20 years", nor over 21 years either. I intend to keep this post reasonably short and to the point, but what about the "plagiarism" and "bullshit"? Disinformation, and its partner in crime, unadulterated bullshit, is present throughout, like some constant and annoying in-store muzak. Here's a glaring (and hilarious) example:
The sea level changes along the Australian coastline have been measured at many locations starting in the late 1800s. In the early 1990s, the Australian Baseline Sea Level Monitoring Project was designed in order to monitor the sea level changes around Australia and to identify decadal trends with respect to the enhanced greenhouse effect. A sequence of SEAFRAME[36] stations (SEA-Level Fine Resolution Acoustic Measuring Equipment) was installed on 16 South Pacific islands to measure the sea level and to record meteorological parameters (both at stations previously covered by standard tide gauge equipment and stations previously not covered by tide gauges). The vertical stability of the gauges is surveyed by State organizations using GPS.
That's novel, installing tide-gauges on Pacific Islands to measure sea-level around Australia. This gives "telemetry" a whole new meaning. I'll poke around in some more ordure in Part 2 which will follow very shortly.  What about plagiarism? surely not? On page 3 is stated (my bold):
Previously , the National Tide [sic] Centre analyzed all tide gauge data from stations having more than 25 years of recording. This survey ended in year 2003, and was replaced by the ABSLMP data set containing the measurements restricted to the 16 ABSLMP stations. Nowadays, NTC neglects all the data previously measured at these stations as well as at other sites, many of which exceed 25, and sometimes 50, years of recording.
The NTC replaced the National Tidal Facility of Australia (NTFA ) in January 2004. Why is he scolding the NTC for ceasing the issue of reports at all? Why is he bemoaning the loss of reports with linear trends? Trends which he says are "misleading"? His charge against the NTC is that they produced a "misleading statistic" concerning the ABSLMP stations. They did not, and would not have done, as I'll explain later - his citations are fake - they don't contain what he claims, nor anything remotely resembling what he claims.

The reason for his charge of "neglect" will soon be clear. Figure 1 in the "paper", is said to represent the full-record rates for 39 stations with "long-term" (25 years or more) of data in 2009. If data was available to end 2010, you might ask as I did, why quote the number of such stations for 2009 if you had data for "all stations" to 2010? Also why analyse that data to 2009 - a long-term station in 2009 is also a long-term station in 2010. "many of which exceed 25, and sometimes 50, years of recording" - doesn't he know how many in each category? In fact it's most, not many which exceed 25 years in 2010, and it's just two which exceeded 50 years - Fremantle and Sydney.

Here's Figure 1, with the legend "AUSTRALIA" in large friendly letters, in case anyone might be unable to identify what it represents.

There's already a problem - Mörner&Parker (Parker is actually Alberto Boretti, more in Part 2) couldn't have analysed the 39 stations on that map; one of them (Port Adelaide Inner Harbour) ceased operation after 2008, and since the data was virtually identical to Port Adelaide Outer Harbour (hardly surprising!) the NTC pulled the record from the data webpage in 2010, well before M&P say they accessed the data. How do I know that the station is represented on the map? Is my knowledge of Oz stations encyclopaedic? Yes and no, it's extensive (ahem!), but I know exactly which stations that map shows, because as it happens, I have a copy of the map myself, a nice crisp and unfuzzy one, along with a table of station results, in a July 2010 report produced by the NTC. It's a report which M&P (or just Mörner?) took pains to claim (above) wasn't produced after 2003, and is titled "Australian Mean Sea Level Survey 2009"; the graphic is on page 8:

Note the caption - Mörner's map is captioned
Figure 1 : Distribution of tide gauge station [sic] in Australia. Location and average rates of the 39 tide gauge stations in mainland Australia having a period of recording of at least 25 years. The mean rate of all 39 stations is 0.9 mm/year.
That caption implies by omission that the 39 figure is the most recent, though it applies to 2009 only, as the text states. Ignoring the fact that "mainland Australia" on the map includes three islands (Tasmania's hard to miss, and the names "Booby Island" and "Lord Howe Island" are a bit of a give-away). Mörner's graphic is fuzzy, with degraded colours and showing all the signs of having been taken from a screenprint, with aliasing artefacts in the image. I'll compare sentences (in sequence) from the two paragraphs preceding Mörner's map (M&P) with bulleted points from page 1 of the NTC report (NTC) reproduced below. You'll note from the page shown below that the sequence of statements is identical:
M&P:  In 2009, there were 39 sites on the Australian mainland (Figure 1), where relative sea levels had been measured for at least 25 years and with the average length being 42 years .
 NTC:  There are 39 Australian locations where relative sea levels have been measured for at least 25 years. The average length of these records is 42 years.

M&P:  The average trend of all the 39 stations is 0.9 ±1.9 mm/year.
NTC:  The average trend from all 39 stations is 0.9 mm/yr with a standard deviation of 1.9 mm/yr.

Following that statement, the NTC report says
Some of the stations exhibit unrealistic trends due to undocumented datum shifts. A more realistic average trend obtained from 29 stations within 1 standard deviation of the mean is 1.4 mm/yr with a standard deviation of 0.7 mm/yr.
Mörner missed  that bit out. I wonder why?

M&P:  The geographical pattern of relative sea level trends around the Australian coastline is fairly uniform (Figure1).
NTC:  The geographical pattern of relative sea level trends around the Australian coastline is fairly uniform in general.

M&P:  Parts of the Australian coastline are strongly affected by the ENSO events.
The longest sea level records show quasi bi-decadal sea-level oscillations.
NTC:  Annual mean sea levels around the Australian coastline are strongly correlated with the El Niño-Southern Oscillation (ENSO) signal. Annual mean sea levels generally fluctuate in accordance with the Southern Oscillation Index (SOI).
The longest sea level records show decadal sea-level oscillations with periods of around 20 years.

NTC: The Australian Mean Sea Level Survey is updated annually.

There's nothing wrong of course, with using graphics, nor quoting text, from someone else's publication(s), if citations are given. In this case however, Mörner has not only failed to cite the source, he's explicitly said it doesn't exist! There's only one word for that - plagiarism, compounded by a blatant lie to cover his tracks. He's simply pinched the summary stats from the NTC report to 2009, including the standard deviation. None of the other claimed results include a standard deviation. Plagiarism is intellectual theft, and when blatant lies are added, the result is fraud.

Would a "sea level expert" really need to lift almost an entire page from a NTC report? From an organisation he's criticising? The hypocrisy is staggering! Not only that, but the number of long-term stations, and the statistics themselves, which apply to the period to 2009, and aren't exactly relevant here. Data was available to 2010, so why not do the trends and quote results to 2010? Mörner claims to have analysed 86 stations to 2010, so the spreadsheets should have been available. I say "claims", because there's not one scrap of evidence any such analyses were done, in fact just the opposite.

In a "paper" which claims to have analysed "all 86" Australian stations, 39 overall and 86 over the last two decades to 2010, amounting to some 297 trend analyses and 258 polynomial fittings, you'd expect several tables, a few sample charts, even a list of stations at the very least, wouldn't you? They're supposed to be there to inform the reader, summarise the results, and not least to show evidence that the work's been done, and that the summary results are valid. The sum total is one long-term chart for Fremantle, and a rather small bunch of totally unsupported statements. Statements which don't in fact  reconcile. And this heap of crap was given prominence on WUWT, of which a commenter there recently said "Everyone knows nobody gets away with bad science or math here".

Why on earth try it on to this extent? Mörner's relying on his target audience being uncritical and chauvinistic, generally ill-informed about sea-level in general, Australia in particular, and he's also confident that not one of them will check on anything he writes. None of those characteristics apply to yours truly. I've found evidence that Boretti-Parker is not an honourable man either, inclined to bend truth into fiction. He (as Boretti) is listed on the Principia Scientific website as a contributor - it's the lair of the "Sky-dragon slayers". Enough said, I think.

There are other claims in the "paper" - it might have been sub-titled "This paper is bought to you by the number 1.5", including a ludicrous claim to have analysed "all 2059 PSMSL stations". I'll shred that in Part 2. What's really ironic is that along the the Pacific coast of N & S America, tide-gauges do indeed show very low rates of rise, even a fall over recent decades, something Mörner denies. That's a strange contradiction, as he also denies any increase in global sea-levels. Unfortunately, no-one informed all the other tide-gauges worldwide of this "great truth"; for them it's "business as usual".

The M&P "paper" isn't just bad science - it's a total lack of science backed up by no data whatsoever - science fiction. The clues, in the form of "bad math", inconsistencies and disconnects are evident to anyone with a critical and sceptical eye. Those on WUWT, posters and commenters, who uncritically accepted a "paper" so thin you can read small print through it, need to re-examine their critical and sceptical faculties and motives.

Part 2, in which I turn over the ordure with a long stick, and explode more Mörner myths, coming very soon.


NTC data for all stations here, and PSMSL data and charts here - search the page for Darwin; stations are listed clockwise from there. Values are mm/year.

Averages for all stations, clockwise

1 Mornington Island, QLD Insufficient data 8.61
2 Karumba 7.94 7.94
3 Weipa 6.92 7.43
4 Booby Island 6.25 7.04
5 Goods Island 7.67 7.20
6 Thursday Island 2.00 6.16
7 Turtle Head 6.96 6.29
8 Ince Point 8.20 6.56
9 Port Douglas 3.85 6.22
10 Cairns 1.83 5.74
11 Mourilyan Harbour 1.40 5.30
12 Lucinda 3.11 5.10
13 Townsville 2.63 4.90
14 Cape Ferguson * 2.78 4.73
15 Abbot Point Insufficient data 11.7
16 Bowen 2.29 4.56
17 Shute Harbour 2.33 4.41
18 Mackay 2.08 4.27
19 Hay Point 1.24 4.09
20 Rosslyn Bay * 3.36 4.05
21 Port Alma 1.60 3.92
22 Gladstone 1.43 3.79
23 Bundaberg 0.94 3.66
24 Urangan 0.28 3.50
25 Mooloolaba 1.52 3.41
26 Brisbane 3.19 3.41
27 Gold Coast Seaway 3.22 3.40
28 Norfolk Island, NSW 7.04 3.54
29 Yamba 0.93 3.44
30 Lord Howe Island 3.41 3.44
31 Newcastle 1.66 3.38
32 Sydney 1.25 3.31
33 Botany Bay 2.45 3.28
34 Port Kembla * 1.6 3.23
35 Eden 1.23 3.17
36 Stony Point, VIC * 1.71 3.13
37 Lakes Entrance Insufficient data
38 Hovell Pile 5.31 3.19
39 Melbourne 2.38 3.17
40 Point Richards Channel Insufficient data 5.14
41 Geelong 2.29 3.14
42 Point Lonsdale 0.35 3.07
43 Port Welshpool Insufficient data
44 Queenscliff 2.93 3.07
45 West Channel Pile 2.30 3.05
46 Lorne * 2.46 3.03
47 Portland * 2.84 3.03
48 Devonport, TAS Insufficient data
49 Low Head Insufficient data
50 Spring Bay * 3.11 3.03
51 Hobart 0.17 2.96
52 Granville Harbour Insufficient data
53 Burnie * 2.98 2.97
54 Victor Harbor, SA 2.02 2.94
55 Port Stanvac * 4.52 2.98
56 Port Adelaide 5.08 3.02
57 Port Giles 4.40 3.05
58 Wallaroo 2.53 3.04
59 Port Pirie 6.96 3.12
60 Whyalla 0.59 3.07
61 Port Lincoln 3.34 3.07
62 Thevenard * 3.34 3.08
63 Esperance, WA * 4.10 3.10
64 Albany 4.50 3.12
65 Bunbury 3.47 3.13
66 Fremantle 5.27 3.16
67 Hillarys * 7.84 3.24
68 Geraldton 6.12 3.29
69 Carnarvon 5.69 3.33
70 Exmouth 6.42 3.38
71 Onslow 6.93 3.44
72 King Bay 6.83 3.49
73 Cape Lambert 7.43 3.55
74 Port Hedland 8.12 3.62
75 Broome * 9.32 3.71
76 Wyndham 9.38 3.79
77 Darwin, NT * 7.78 3.85
78 Gove Harbour 5.69 3.87
79 Milner Bay * 7.54 3.92
80 Cocos Islands * 7.86 3.98
81 Macquarie Island 6.84 4.02
82 Casey, ANT 6.22 4.05
83 Commonwealth Bay Insufficient data
84 Davis 2.29 4.02
85 Mawson 5.59 4.05
All stations average 4.05
ABSLMP average 4.35
Other stations average 3.90
Standard deviation 2.51
Minimum 0.17
Maximim 9.38

No comments:

Post a Comment