Science of Global Warming And Possible Man Made Causation

It's difficult to tell from that graph with any certainty. A fair bit of the datapoints in the middle of the plot lie below the line (best-fit linear?). The later points do seem to be "taking off" but may not signal a significant rate increase just as those points steadily below the line during some years are no evidence of slowing.

Actually I would think it's the initial 2 to 3 years that are influencing the trend the most. I would like to see the fit when the data is plotted from 96. But that would only indicate an even faster increase of sea-level.

I was going to ask if there was a graph with data further back but I realized that it would not resolve this question. You simply need more datapoint at this point, not back then.

I don't totally agree. the trend is clear, and the fit seems quite good. The authors' number doesn't seem to be in discussion, unless one is to say that it's to conservative . But the claim I responded to was that

mhaze said:
There never was a faster increase, and all the old numbers were wrong.

The number for sea level is 1.3mm. Period. Going back to 1900 or so.
 
I don't totally agree. the trend is clear, and the fit seems quite good.

All I'm saying is that, to my best guess, you would find most points on that graph to fit within one standard deviation of the best-fit line, even the recent stuff. What appears to your eye to be a recent departure from the predicted trend may not be as significant as it appears.

I don't have the numbers so I can't put error lines on the graph. Is there a version with error lines?
 
I didn't find one, sorry. I only had time to browse through the original paper that was the source of the picture, and it didn't show the error bars either.

But anyway, I think we can agree that the rate of recent increase in sea-level is not 1.3mm/y, and that the value of 2.8 +-0.4mm/y follows from the data, no?
 
But anyway, I think we can agree that the rate of recent increase in sea-level is not 1.3mm/y, and that the value of 2.8 +-0.4mm/y follows from the data, no?

I agree that the slope of the line is 2.8 mm/y.

While eyeballing the slope, I noticed something weird, though. The ordinate is labeled "delta MSL", which I interpret to be the rate of change of the Mean Sea Level. If that is true, a constant change would be represented by perfectly horizontal line on the graph. Also, it would explain the negative values of some of the data but that would mean that the MSL was receding at some point in time. Am I misinterpreting this?
 
The graph comes from here

E.W. Leuliette, R.S. Nerem, G.T. Mitchum, Results of TOPEX/Poseidon and Jason-1 calibration to construct a continuous record of mean sea level, Mar. Geod. 27 (2004) 79–94.

I don't seem to have access here to that paper unfortunately.
 
I agree that the slope of the line is 2.8 mm/y.

While eyeballing the slope, I noticed something weird, though. The ordinate is labeled "delta MSL", which I interpret to be the rate of change of the Mean Sea Level. If that is true, a constant change would be represented by perfectly horizontal line on the graph. Also, it would explain the negative values of some of the data but that would mean that the MSL was receding at some point in time. Am I misinterpreting this?

I noticed that, but it seems like a weird notation for globaly averaged. It's not my field, so I'm not familiar with it.

The yy values are in mm, so it wouldn't make sense to be a rate. Especially since you extract the rate from the slope. I also don't get why the scale starts at -20mm. I'll email Fu to ask about it.

I have to say that such notation should have been explained in detail in the figure legend, or in the text. Bad form, at least from where I stand. The reviewers of my papers would slap me across the head for something like that.
 

Thanks for the link. Went there, scanned the article but could not find any explanation as to why the MSL term is preceded by a delta sign. I suspect that it's merely a designator that the data have been corrected per the various errors described in the article but I'm not sure. I agree with Megalodon that it's most likely a graph of MSL vs time.

This isn't my field either so I can't be of much help. I'll reread the paper when I'm not "at work" (;)) to see if I can spot a good explanation.
 

oh nice. :)

Figure 10 shows the record of global mean sea level change...
I guess they have taken an average of all readings, called that zero, and then plotted deviation from that in mm against time. So the delta is just to signify change from their arbitrary mean 'cos they aren't plotting absolute sea level. And that makes sense because they are only realy interested in the change in sea level, not the absolute level defined against some other level which would be fairly pointless.

ETA: No tide gauge calibration applied which means that it is unaffected by Woppelmanns paper as far as I can tell. In anycase they used GPS to determine motion of the crust, where available, to make the tide gauge calibrations.
 
Last edited:
oh nice. :)

I guess they have taken an average of all readings, called that zero, and then plotted deviation from that in mm against time. So the delta is just to signify change from their arbitrary mean 'cos they aren't plotting absolute sea level. And that makes sense because they are only realy interested in the change in sea level, not the absolute level defined against some other level which would be fairly pointless.

ETA: No tide gauge calibration applied which means that it is unaffected by Woppelmanns paper as far as I can tell. In anycase they used GPS to determine motion of the crust, where available, to make the tide gauge calibrations.

Used GPS when available? I read that paragraph to indicate that they did not because with insufficient GPS coverage at that time, relying on those stations would introduce more error than not using them. (your article by Leuliette). Can't quote now, on the road, but here is my response to the various comments.

FSOL - I said "recent accurate analysis of the satellite data by Woppelmann et. al." (MY WRONG!)Yes of course it was tide gauges.

Woppelmann corrects tide gauges from approximately 1900 to 2006. He does his correction on perhaps 50 different sets of accumulated data (Table 1, pp. 399). There is considerable variation between sets, however the 1.3 is the global summary. Quoting:
(pp. 402) Two important hypothesis underlie our exercise of combining tide gauge and GPS results to derive "absolute" trends in sea level (a) land motions are extremely low frequency in character so that the current GPS vertical velocities can be applied for the last century (b) the vertical velocity observed at the GPS station applies to the tide gauge site.
​
This is the data that is on the IPCC chart. Certainly they say Topex, and note that "values after 1993 are satellite data". What the IPCC did was abandon the tide gauges in 1993 and go to the satellite altimeter data - TOPEX.

That would be the data on the right hand side of the chart that I posted. Note that the "faster increase" occurs exactly at this time, with the change of measurement methods.




TOPEX is calibrated against the tide gauges but not GPS referenced. It should be less accurate than Wuppelmann's work.

Doesn't it make sense that Woppelmann is more accurate? The tide gauges were actually more accurate, but IPCC believed the radar altimeter without adequately cross checking the two. Had the IPCC been more conservative they would have noted there was an unresolved discrepancy starting in 1993 with the substitution of the radar altimeter data. But they didn't do that.

There is considerable misrepresentation of the contents of IPCC Chapter 5 in the Summary for Policymakers. Chapter 5 goes into a lot of these issues, SPM just shows the "alarming" spin on the issue. As I noted.

Postscript

The future of sea level radar altimetry is for it to be calibrated not against tide gauges, but against the GPS corrected tide gauges. Errors in measurement should shrink down even further. The corrections should be capable of being applied to historical data using the assumptions (a) and (b) above noted.

References:
Wöppelmann, G., B. Martin Miguez, M.-N. Bouin, and Z. Altamimi. 2007. Geocentric sea-level trend estimates from GPS analyses at relevant tide gauges world-wide. Global and Planetary Change, 57, 396–406.

E.W. Leuliette, R.S. Nerem, G.T. Mitchum, Results of TOPEX/Poseidon and Jason-1 calibration to construct a continuous record of mean sea level, Mar. Geod. 27 (2004) 79–94.
http://sealevel.colorado.edu/MG_Leuliette2004.pdf
Fu & Le Traon (2006)
IPCC SPM Feb-2007
IPCC Chapter 5
 
Update...

Lee-Lueng Fu said:
Dear [Megalodon],

Thanks for your interest in my article. The sea level shown in Figure 3 is referenced to a time in the middle of 1997. The values are not absolute numbers. The figure basically shows that the mean sea level is rising, relative to a given time. So the negative numbers simply mean that the sea level at the beginning is lower than that in 1997. Hope this helps.

Lee-Lueng Fu
 
Used GPS when available? I read that paragraph to indicate that they did not because with insufficient GPS coverage at that time, relying on those stations would introduce more error than not using them. (your article by Leuliette). Can't quote now, on the road, but here is my response to the various comments.

They use two estimates of land motion. The GPS where available and additionally a second estimate is made using the complete sea level time series at the gauge.

Leuliette et al said:
Mitcham (2000) describes the land motion correction s in detail and only the basic idea is repeated here. These corrections are obtained by using two different estimates of land motion. The first estimate is only available where there is GPS or DORIS recievers near the gauges, and is basically the vertical motion observed at the reciever. In addition, at most gauges a second estimate is made by looking at the complete sea level time series available at the gauge, which is generally substantially longer than the overlap period with altimetry, estimating a vertical rate, and then interpreting this as th esum of land motion and a global sea level rise taken to be 1.8 mm/yr (Douglas 1997) Error estimates are made for both of these land motion estimates and at stations where both are available , the final estimate is a weighted average of the two, with the weights inversly proportional to the error variences for the estimated land motion rates.

They continue by saying that ideally they would have used only GPS for all gauges but that narrowing the set of gauges used in that way would introduce random errors greater than those when they use both sets together.

Leuliette et al said:
Since the Douglas rate is a global average sea level rise rate, two types of errors can occur in the sea level based land motion estimates. First, deviations of the local rate romthe globally averaged rate introduce random errors. More importantly, since the Douglas rate is used at all stations, error in this rate adds a bias error (ie., it is the same at all guages and therefore does not average down). At present this bias error dominates the final altimetric drift rate estimate. An obvious way to eliminate this error would be to use only tide gauges where GPD or DORIS measurements were available. Unfortunately, at present few gauges meet this criterion and narrowing th erange of gauges used results in random errors that are much increased reletive to the present analysis because th enumber of degrees of freedom in the global average is greatly reduced. In the future however, GPD or DORIS rates should be used at every gauge and work along these lines is proceeding
FSOL - I said "recent accurate analysis of the satellite data by Woppelmann et. al." (MY WRONG!)Yes of course it was tide gauges.

Woppelmann corrects tide gauges from approximately 1900 to 2006. He does his correction on perhaps 50 different sets of accumulated data (Table 1, pp. 399). There is considerable variation between sets, however the 1.3 is the global summary. Quoting:
(pp. 402) Two important hypothesis underlie our exercise of combining tide gauge and GPS results to derive "absolute" trends in sea level (a) land motions are extremely low frequency in character so that the current GPS vertical velocities can be applied for the last century (b) the vertical velocity observed at the GPS station applies to the tide gauge site.​
The first assumption they deal with farily well referencing work that backs their hypothesis up. The second assumption they admit requires some looking into. Which is fair enough.

This is the data that is on the IPCC chart. Certainly they say Topex, and note that "values after 1993 are satellite data". What the IPCC did was abandon the tide gauges in 1993 and go to the satellite altimeter data - TOPEX.

That would be the data on the right hand side of the chart that I posted. Note that the "faster increase" occurs exactly at this time, with the change of measurement methods.
The faster increase doesn't occur exactly at that time. What is happening is that the rate of sea level rise increases over time from year to year (that's kind of ugly but I hope you can follow my point). Depending on the period of time you sample over, your results change.

So...

In the table the column on the left states +1.8 +-0.5 mm/yr for 1961 - 2003

As far as I can tell from the table that figure contains both the tide gauge data and the satellite readings. Suffix a states that at least.

The right hand column is averaged values over the time scale 1993 - 2003 and gives a figure of +3.1+-0.7 mm/yr.

That does not mean that there was a step change in 1993.

If you keep reducing the time scale and so increasing the resolution, you'll find that you get different values for sea level rise again, some of them will even show a decrease in sea level height.

They use long time scales so that they can filter out other natural cycles with shorter periods from the results. The IPCC and Fu et al) both state clearly that the figure for 1993 - 2003 might just be the result of one of those other shorter cycles. It is also for that reason that the IPCC at time of writing AR4 didn't consider GPS to be as reliable as traditional corrections to the tide gauges because they didn't have a long enough period of data to be sure and f course there was not GPS data attached to all the tide gauges. Their conservatism is what leaves them with +1.8+-0.5 mm/yr and a descrepency of +0.7+-0.7 mm/yr from the estimated climate contributions.

If anything Woppelmanns paper gives further support to the idea of melting land ice raising the sea level and I suppose the IPCC take on it. It clears up a descrepency that everyone agreed was there. If the work is repeated, which it no doubt will be, and the results are similar then I think that little mystery could be solved.

http://www.internationalskeptics.com/forums/imagehosting/thum_142244731ed7796761.png


TOPEX is calibrated against the tide gauges but not GPS referenced. It should be less accurate than Wuppelmann's work.

Doesn't it make sense that Woppelmann is more accurate? The tide gauges were actually more accurate, but IPCC believed the radar altimeter without adequately cross checking the two. Had the IPCC been more conservative they would have noted there was an unresolved discrepancy starting in 1993 with the substitution of the radar altimeter data. But they didn't do that.
As I have shown they did use GPS where available so your argument is based on inaccurate information. I hope I have also shown that there isn't an unresolved discrepency starting in 1993 and that the IPCCs conservatism is what made them use the +1.8+-0.5 mm/yr figure, which we both agree is probably a bit high.

It turns out that Leuliette et al decided that they don't think it necessary to correct the altimetry with the tide gauge calibrations anymore, but that they are useful for diagnostics.

Leuliette et al said:
With the current set of corrections to T/P data, the tide gauge calibration for T/P is reasonably "flat," and thus we no longer consider it necessary to correct the altimetry using the calibration values. Nethertheless, we consider the calibration valuable in assessing altimeter measurement errors and develop error bars for the sea level measurements. At present the gauges indicate that the drift error is consistant with zero but only to a precision of about plus or minus 0.4 mm/yr. Hence the estimated bias error due to the altimeter drift is 0.0+-0.4 mm/yr. The calibration error could become much better, because the error model is almost entirely dependent on errors in the land motion estimates used for each of the tide gauges Mitcham (2000). Including improved land motion estimates from GPS and DORIS should reduce the uncertainty in the altimetric rates.

On further reading I am not as sure about my quick edit statement from my last post that "no tide gauge calibrations were applied." I haven't dug into the paper enough to see if the "current set of corrections to T/P data" are related to past tide gauge calibrations or not. Though earlier they do say this...

Leuliette et al said:
THe calibration of the altimetric series used for global sea level change calcualtions is essential, but the errors from vertical land motion at the tide gauge locations are apprently larger than any remaining sources of drift in the T/P global sea level change estimates

...which makes me think I may have been correct in my initial assumption.

In any case Figure 10 that the +2.8+-0.4 mm/yr comes from explicitly contains no calibration from the tide gauges and that is where I got my quick edit statement from.

Leuliette et al said:
Figure 10: Global 10-day mean sea level variations from TOPEX/POSEIDON and Jason no tide gauge calibration applied...

So, I am not sure how Woppelmann affects the data here except to reduce the bias error from +-0.4 mm/yr to something less significant. I don't think that even using all GPS/DORIS recievers at tide gauges is going to show that the rate of sea level rise hasn't increased over the last ten years or so.

There is considerable misrepresentation of the contents of IPCC Chapter 5 in the Summary for Policymakers. Chapter 5 goes into a lot of these issues, SPM just shows the "alarming" spin on the issue. As I noted.

Postscript

The future of sea level radar altimetry is for it to be calibrated not against tide gauges, but against the GPS corrected tide gauges. Errors in measurement should shrink down even further. The corrections should be capable of being applied to historical data using the assumptions (a) and (b) above noted.

References:
Wöppelmann, G., B. Martin Miguez, M.-N. Bouin, and Z. Altamimi. 2007. Geocentric sea-level trend estimates from GPS analyses at relevant tide gauges world-wide. Global and Planetary Change, 57, 396–406.

E.W. Leuliette, R.S. Nerem, G.T. Mitchum, Results of TOPEX/Poseidon and Jason-1 calibration to construct a continuous record of mean sea level, Mar. Geod. 27 (2004) 79–94.
http://sealevel.colorado.edu/MG_Leuliette2004.pdf
Fu & Le Traon (2006)
IPCC SPM Feb-2007
IPCC Chapter 5
 
All I'm saying is that, to my best guess, you would find most points on that graph to fit within one standard deviation of the best-fit line, even the recent stuff. What appears to your eye to be a recent departure from the predicted trend may not be as significant as it appears.

I don't have the numbers so I can't put error lines on the graph. Is there a version with error lines?

Are there versions of ANY GW-related graphs that have error lines on them? Especially the graphs from the past, that show how warm it has been historically?
 
The right hand column is averaged values over the time scale 1993 - 2003 and gives a figure of +3.1+-0.7 mm/yr.

That does not mean that there was a step change in 1993.

If you keep reducing the time scale and so increasing the resolution, you'll find that you get different values for sea level rise again, some of them will even show a decrease in sea level height.

So, I am not sure how Woppelmann affects the data here except to reduce the bias error from +-0.4 mm/yr to something less significant. I don't think that even using all GPS/DORIS recievers at tide gauges is going to show that the rate of sea level rise hasn't increased over the last ten years or so.

Look at the right hand side of the table. The number for thermal expansion is 1.6 plus or minus 0.5. This is cumulated in the "sum" and differenced from the "observed". Yet the 1.6 is in excess of that which we have figured from the GPS corrected, more accurate number for the entire sea level rise - 1.3 plus or minus a bit.

Isn't it obvious that the number 1.6 for thermal expansion, is just wrong, and thence the sums of 2.8 and 3.1 are also wrong? If not, why?

You've made other interesting comments and observations that I appreciate, and want to respond to later. Also I agree - there are several paragraphs in the papers that are quite convoluted here and can be construed several ways.
 
Are there versions of ANY GW-related graphs that have error lines on them? Especially the graphs from the past, that show how warm it has been historically?

What would be the meaning, if any, of an "error bar" for a "multi proxy reconstruction?"
 
Look at the right hand side of the table. The number for thermal expansion is 1.6 plus or minus 0.5. This is cumulated in the "sum" and differenced from the "observed". Yet the 1.6 is in excess of that which we have figured from the GPS corrected, more accurate number for the entire sea level rise - 1.3 plus or minus a bit.

Isn't it obvious that the number 1.6 for thermal expansion, is just wrong, and thence the sums of 2.8 and 3.1 are also wrong? If not, why?

I think you misunderstand what the numbers mean. the 1.3 mm/yr is the rate for 1961 - 2003 it is not the rate for 1993 - 2003. That is, according to the table, 3.1 mm/yr. If the rate is increasing, which it seems to be from the data, then the results make total sense.

As way of an example.

Lets for arguments sake take a "hockey stick" shaped graph. It isn't "that" hockey stick, just any old hockey stick shape will do. I am not saying that sea level rise is hockey stick shaped, just that I want to describe the shape of a curve that is easily recognisable by us all.

If I fit a straight line to the whole curve, I get one value for the gradient. This is the equivalent to the figure we have for the 1961 - 2003 rate of sea level rise. This is what the table says is 1.8 mm/yr and Woppelmann has at 1.3 mm/yr

If I fit a straight line to the curve ignoring the flat bit at the beginning of the curve and just use the bit of the curve that sticks up, I get a different value for the gradient. This is the equivalent to the 1993 - 2003 rate of sea level rise. This is what the table says is 3.1 mm/yr

That is all that the table shows. Both figures can be correct and there is no need to think that they contradict each other.

You've made other interesting comments and observations that I appreciate, and want to respond to later. Also I agree - there are several paragraphs in the papers that are quite convoluted here and can be construed several ways.
 
Last edited:
The number for sea level is 1.3mm. Period. Going back to 1900 or so.

That's the average since 1900, presumably. What's relevant to my point is the average over 1993-2003. By most reports that was greater than 1.3, but not by your report. I'm still left wondering what data you're reporting.
 
Last edited:
That's the average since 1900, presumably. What's relevant to my point is the average over 1993-2003. By most reports that was greater than 1.3, but not by your report. I'm still left wondering what data you're reporting.

The data I have reported is the latest and most accurate. I've built a spreadsheet to compare all the data series and am trying to see if there is any way that Fsol could possibly be correct. So far, no luck. It's a bit technical.

The issue has to do with

a) correcting the tide gauge readings with the satellite radar altimeter
b) correcting the tide gauges with GPS stations that are closeby

And these "corrections" are done to the historical data that we've already got. But the recent "increasing rate of sea level increase" from the 1.8 mm number that everyone has always heard about, to the 2.8 / 3.1 mm reported for 1993 onward, is at issue.

That comes only from (a), and it appears to be in error. More later.
 
Are there versions of ANY GW-related graphs that have error lines on them? Especially the graphs from the past, that show how warm it has been historically?

I'm a duffer at this stuff, casebro. I don't remember but mhaze has a point in that error bars for data collected with dissimilar methodology would be a little silly.

In this case, though, I think it would be especially appropriate because, judging from the data points in the original paper, the standard deviation is significant.
 

ISF - Join now!

Every member here is approved by hand. No bots, no spam, just people who care about evidence and honest debate.

Membership is free!

Create your free account

Back
Top Bottom