Several comments from my original post suggested using the more recent (2010) Church & White (C&W10) sea level data set instead of the 2006 data set. Others raised questions about the potential impact of water accumulation in reservoirs (which would artificially reduce the measured sea level rise) and ground water depletion (which would artificially inflate the measured sea level rise). Vermeer and Rahmstorf (2009.. V&R) included a contribution due to water accumulation in reservoirs, but not ground water depletion. B. F. Chao, et al (2008) estimated the impact of reservoir accumulation on sea level. Since V&R was published, Wata et al (2010) published an estimate of ground water depletion from 1960 to present, along with an estimated impact on sea level.
Impact of C&W10 Data Set
The graph below shows the 2006 and 2010 data sets, but with the 2006 data set adjusted with a constant off-set to match the basis of the 2010 data set.
The principle difference is in the 1915 to 1945 period, where the 2010 data set has less of a downward deflection in the trend. The newer data set goes until 2009, but drops the 1870 to 1879 data that was present in the original (2006) data set. To simplify my calculations (since everything was already set up to calculate starting in 1870), I (rather rashly?) added the original 1870 to 1879 data to the more recent data set, with an appropriate off-set. I then used the total set (1870 to 2009) to redo the calculations from the original post. The figure below shows the model fit, along with the mass-only portion of the model.
The fit to the updated C&W data yield somewhat less acceleration (compare the quadratic fits to those from the original Figure 5). The fit was identical in R^2 to the original model – 0.987. The figure below shows the resulting projections of sea level rise based on the newer data set.
All projections are reduced somewhat, but with the largest reductions for the higher assumed warming rates (especially 0.25 and 0.3 C per decade). For no assumed warming, the rise is approximately 180 mm (about 7 inches), versus about 250 mm (about 10 inches) in the projection based on the earlier data set. The projection for 0.3C per decade warming rate is about 550 mm (about 19 inches), instead of about 690 mm (about 27 inches) based on the earlier data set. The lower rates are the result of a smaller contribution of melting to the total rise.
Adjusting for Reservoir Accumulation and Groundwater Depletion
I (rather crudely) converted the graphical data form sea level contribution from Chao et al into a four linear segment approximation to their original curve. The Wata et al estimate of groundwater contribution covered only 1960 to 2000, so I extrapolated the (almost linear) annual contribution trend backward to 1925 (where it was ~0) and forward to 2009, and then integrated to yield an (roughly!) estimated contribution to sea level over time. The graph below shows the two separate contributions, along with their sum.
The sum of the two adjustments was added to the C&W10 data set to generate a “corrected” C&W10 data set.
The adjustment makes a relatively modest difference, with the biggest change in the 1980’s. I then repeated the calculations for the model using the adjusted C&W10 data, yielding the results in the following two graphics.
The model R^2 remained 0.987. Adding the combined adjustments for reservoir accumulation and groundwater depletion increases the projected sea level rise only slightly compared to the unadjusted C&W10 data set. In any case, substitution of the more recent C&W data yields a significant reduction in projected sea level increases through 2100. The maximum projected increase is only slightly greater than the IPCC AR4 range of values. None of these results supports the extreme sea level increases (1+ meter) suggested by V&R, among others. Extreme rates of sea level rise would appear unlikely based on all the above results. Of course, these projections are only as reliable as the model formulation and the C&W10 data set.
Steric Only Contribution
Steve Mosher asked me to generate the steric contribution (that is, total sea level increase less the mass increase calculated by the model). The graph below shows this data.
References for adjustments to Church and White:
Global depletion of groundwater resources, Wada et al, GEOPHYSICAL RESEARCH LETTERS, VOL. 37, L20402, doi:10.1029/2010GL044571, (2010)
Impact of Artificial Reservoir Water Impoundment on Global Sea Level; B. F. Chao, et al., Science 320, 212 (2008)
From the end of 2008 to the end of 2009 C&W 2011 tide data shows a 25mm rise.
Are they ill?
2007.9583 41.8 8.2
2008.0417 42.8 8.2
2008.1250 41.6 8.2
2008.2083 46.2 8.2
2008.2917 49.0 8.2
2008.3750 54.4 8.2
2008.4583 54.7 8.2
2008.5417 56.5 8.5
2008.6250 53.7 8.5
2008.7083 51.5 8.5
2008.7917 47.1 8.5
2008.8750 46.4 8.5
2008.9583 43.8 8.8
2009.0417 44.7 8.8
2009.1250 43.0 8.8
2009.2083 46.4 8.8
2009.2917 47.7 8.8
2009.3750 53.4 8.8
2009.4583 57.0 8.8
2009.5417 58.4 8.8
2009.6250 59.1 8.8
2009.7083 60.9 8.8
2009.7917 62.3 8.8
2009.8750 64.6 8.9
2009.9583 68.3 8.9
When I download the annual data (admittedly not the monthly) the largest difference in Tide Gauge data was 6.5mm.
The average was pretty close to zero.
PS Their satellite data shows a 6mm rise.
2007.9583 46.4
2008.0417 50.8
2008.1250 47.5
2008.2083 51.6
2008.2917 47.4
2008.3750 53.7
2008.4583 50.4
2008.5417 53.9
2008.6250 49.0
2008.7083 51.4
2008.7917 51.5
2008.8750 54.8
2008.9583 52.3
2009.0417 55.8
2009.1250 51.9
2009.2083 56.2
2009.2917 53.9
2009.3750 54.2
2009.4583 52.4
2009.5417 54.6
2009.6250 53.7
2009.7083 57.2
2009.7917 57.8
2009.8750 56.7
2009.9583 58.2
Colorado showed a 15mm rise from end of 2008 to end of 2009 and then a 6mm drop the following year.
2008.9769 37.892
2009.9813 52.715
2010.9858 46.482
Can any of this data be trusted?
Bruce,
When you are trying to figure a long term trend in a noisy system, very short term trends like you show don’t by themselves contain much useful information; too much short term noise mixed with the longer term (secular) trend. You need to look at longer periods (where short term noise tends to average out) to have useful information about long term trends. Just to be clear, “noise” includes any uncertainty in the measurement process, but more importantly, also causal factors which are not part of the model calculation. In the case of sea level, a multitude of potential short term factors could be involved which are not included in the model.
Bruce: Please plug your numbers into excel and make a chart. Given what you already have, you can do it in 15 minutes. Put the date and data in two columns in excel, select the data column, push F11, tailor to taste. You’ll need to select “source data” and then select your date column for the X axis.
Hmm, I thought it was being called Church and White 2011. But I guess they submitted in 2010 and their data goes through 2009. It’s a little odd, because they reference Woodworth et al 2011 and they talk about 2011 satellite data but only use it through 2009. But that’s off subject.
In their paper Church and White say this:
“A significant non-climatic influence on sea level is the storage of water in dams and the depletion of ground water from aquifers, some of which makes it into the ocean. Chao et al. (2008) estimated that about 30 mm of sea-level equivalent is now stored in man-made dams and the surrounding soils; most of this storage occurred since the 1950s. Globally, the rate of dam entrapment has slowed significantly in the last decade or two. The depletion
of ground water (Konikow et al. personal communication; Church et al. in preparation) offsets perhaps a third of this terrestrial storage over the last five decades and the rate of depletion has accelerated over the last two decades. We remove this direct (non-climate) anthropogenic change in terrestrial water storage
(both dam storage and aquifer depletion) from our observations to focus on the sea-level change related to climatic influences.”
So, my impression was that C&W had already taken care of the dam storage, aquifer depletion issue. And their figure 7 shows the difference that it made.
Stevef, it isn’t the trend. It just looks like a fudge factor was tacked on to get the 1992 to 2009 trend as close to the satellite trend as possible.
Tilo Reber,
I did not read the paper, I only downloaded the data. It seems they came to a very different conclusion about the magnitude of ground water depletion than did Wada et al. In any case, it looks like Figures 5b and 7b above then show the model run with Church and White’s ‘corrected’ sea level trend, rather than an uncorrected trend. Since they disagree with Wada et al, it would be good to see the Church and White data without any corrections applied. I don’t know if that is available.
Sorry, that should have been “Figures 5a and 7a’, not 5b and 7b.
So sea level rise really is a non-event if you don’t consider the desintegration of the WAIS like Rahmstorf does…
Bruce: “It just looks like a fudge factor was tacked on to get the 1992 to 2009 trend as close to the satellite trend as possible.”
Bruce, Church and White say this:
“Calibrations of the TOPEX/Poseidon data against tide gauges have been performed by Gary Mitchum and colleagues (see, e.g., Nerem and Mitchum 2001). Here and in earlier publications, we have used the calibrations up to the end of 2001 (close to the end of the TOPEX/Poseidon mission).”
So it sound like some blending effect was done, at least up to 2001. That’s probably why the transition from tide gauge to satellite doesn’t look as radical as if you just stopped one and went on with the other. A plot of Church and White satellite era data versus a plot of CU satellite data for the same time period does look a little different. The Church and White data increases it’s rate more slowly for the satellite era – but it ends up with about the same rate. This means that it dishes up rather than going straight up like the CU data. And it means that the C&W satellite data alone has more of an accelerating look to it than the CU satellite data alone.
That said, other sources show the tide gauge data alone accelerating during the satellite period. But it looks like accelerations of that magnitude and duration have happened before without being permanent or changing the trend. The thing that I find interesting is how dead flat the trend stayed for 72 years between 1920 and 1992. It seems like it would be hard to explain that if one wants to correlate global temperature to global mean sea level. I think that it indicates that some level of temperature rise is required just to maintain a given level of sea level rise. I’m also stumped by why the sea level trend seems to change by inflection points.
Thanks Steve. Using the newer data and accounting for the “corrections” from GWD and reservoirs makes your result more “robust” 😉 I now think your estimate looks pretty soid, although one could quibble with how uncertainty of the model fits translates into the uncertainy in projections. In short, nice work!
Hans Erren (Comment #79911
Actually, the V and R paper from 2008 is not based on disintegration of the WAIS (which I think most people recognize is a bit far fetched any time in the near future). The V and R paper is based on ‘normal’ melting processes, just as I used in generating my model. What constitutes a “non-event” probably depends on the POV of the person. A rise of 30cm may be a non-even in Weert, but something of an event near Amsterdam.
How high are the dikes in the Netherlands? 😉
I don’t think anyone can convince me a 25mm change is possible in one year from the data I downloaded.
It would be nice if the Church & White data download included the actual tide gauges used and the amount each rose.
Instead we just get a “result” that doesn’t pass the smell test.
Topex was kind of “drifty” …
http://www.aviso.oceanobs.com/es/kiosco/newsletter/newsletter05/altimeter-drift-from-tide-gauges/index.html
Bruce,
I think you have made that clear in several earlier comments; no need to say it again.
I suggest that you try to keep in mind the reported ‘variation’ includes a combination of factors, including uncertainty in the measurements themselves and some relatively short term factors which are not accounted for, including things as simple as weather patterns. Focusing on the short term data in a noisy system is usually misleading, and often causes people to draw incorrect inferences. You need to take a longer view when you are thinking about long term variation.
SteveF, do you have any idea how they could get 25mm?
SteveF, It’s an interesting analysis. The switch from old to new versions of Church and White seems to have made an interesting difference to the results, which suggests some other tests that you could try.
1. The HadSST3 data are available now (http://www.metoffice.gov.uk/hadobs/hadsst3/) and my eyeball estimating reckons there will be some differences between your estimate and theirs. They also have a range of bias corrections you could try.
1.25 C&W also have uncertainty estimates.
1.5 That doesn’t account for other measurement uncertainties in the earlier data. I’m not sure how that could be tackled easily tho.
2. Removing ENSO effects from global temperatures is not without difficulty. There are other methods for estimating de-ENSOd global time series around that you could try. (e.g. http://www.atmos.colostate.edu/~davet/ThompsonWallaceJonesKennedy/)
3. Much of the ice is on land and at high latitudes. SST measurements are made at sea and at lower latitudes. In the past few years SSTs have been flattish while land temperatures have continued to increase, due mainly to continued high rates of warming over the Arctic.
4. Like Steve Mosher, I’d be interested to see what the SST series backed out from C&W would look like. Can that be done? Does that give reasonable grounds for assuming that SST data sets are biased mid-century?
Nebuchadnezzar (Comment #79928),
I separated the thermal expansion estimate from the model already (see post).
All the other issues may be of some technical interest, but I don’t think I want to spend the time to investigate them. My objective was not to produce a perfect projection of future sea level rise. Rather, it was to show that a reasonably implemented model using the “melt front” assumption for glacial melting (as Vermeer and Rahmstorf used), but where thermal expansion is properly separated from melt contribution, yields estimated sea level increases which are much lower than the (rather alarming!) increases projected by Vermeer and Rahmstorf. Which is to say, the Vermeer and Rahmstorf projections are almost certainly very wrong (much too large); nobody should be evaluating risks posed by sea level increase, and formulating public policy in response to those risks, based on the Vermeer and Rahmstorf projections.
“How high are the dikes in the Netherlands? ”
I have been seconded at two waterboards (Waterschappen), responsible for the loading of the detailed dike elevation data that was measured by helicopter lidar. I can inform you that every 5 years there is a dike MOT when the dikes are audited against the latest current models based on observed statistics.
On design dutch dikes have a safety margin included of 50 cm. Given the fact that storminess is not expected to increase with warming of the north pole, and that the dikes are reviewed every five years, the dutch are ample prepared to the rising seas of the coming century.
Also: a mean doesn’t say much. Have a look at the enormous observed variation from the mean.
http://sealevel.colorado.edu/content/interactive-sea-level-time-series-wizard
As we say in Holland: “The soup ain’t that hot”
Nebuchadnezzar: “In the past few years SSTs have been flattish while land temperatures have continued to increase, due mainly to continued high rates of warming over the Arctic.”
The Arctic warming that Hansen has in the GISS record is not quite real, as explained here:
http://reallyrealclimate.blogspot.com/2010/01/giss-temperature-record-divergence.html
Right now, HadCrut3 is dead flat since 98. UAH and RSS are diverging for unknown reasons, with UAH showing a slight increase and RSS showing a slight decrease:
http://reallyrealclimate.blogspot.com/2011/07/rss-and-uah-divergence-charts.html
It’s the second chart.
neb:
What i was hoping to compare was this:
Steve Fs model for steric versus the estimate of steric that is obtainable by differencing Grace and topex.
Might be a small bit of data, but going forward that is how I would judge the model. V&R’s model clearly has that non physical issue
that is rather bothersome.
SteveF (Comment #79916) -“What constitutes a “non-event†probably depends on the POV of the person. A rise of 30cm may be a non-even in Weert, but something of an event near Amsterdam.”
For the local changes in sea level, subsidence and uplift can and indeed have been much larger effects on the actual levels than climatic changes in mean sea level. People have coped with these changes and will continue to have to do so. an extra .3 meters is not going to make that big of a difference to the fact that these changes are happening and will happen regardless of the future evolution of the Earth’s climate (short of another Ice Age, that is). Heck, the Netherlands proves that humanity is capable of adjusting the “effective” sea level all on our own, by raising the land in areas to actually claim them from the ocean. Over the course of a hundred years this can be done again several times over with modern technology. The original case was done long before modern technology.
Andrew_FL (Comment #79948),
Sure, I agree with most of that, but I don’t think the “raising the land in areas to actually claim them from the ocean” part is correct. The Dutch are indeed industrious, but they build dikes to keep the sea, and the rivers that must pass through, out and then pumped out the water to make dry land… even the Dutch don’t move that much earth around! The final planned stages in the process (yes, more shallow lakes/sea was to be drained) may never happen; too much enviro/political opposition these days. http://en.wikipedia.org/wiki/Polder#Polders_and_the_Netherlands
The reclaimed lands actually subside after they are separated from equilibrium with the sea (these are soft marshy lands – lake floor/sea floor/river deltas). The subsidence takes the reclaimed land ever further below sea level…. just like New Orleans.
steven mosher (Comment #79947),
I understood your objective.
Please keep in mind that the short term response of the ocean heat diffusion model is probably the least certain part of the model. I suspect the upper layer mixing rate in the ocean is a little higher and the deeper mixing rate a little slower than in the model. In developing the model it looked to me like I would need to integrate the diffusion calculation using considerably shorter time steps and use somewhat different diffusion constants for different depths to get a more realistic short term response, but I figured for my purposes (long term trend projection) the simple model was probably good enough, even if the short term response looks a little sluggish. Decade and longer comparisons like you want to do are probably OK, but I wouldn’t count on short term comparisons telling you very much.
By the way, going forward we will have Agro data for measured OHC, so I think the comparison of the model with the measured OHC will be a very robust test… I assume that soon Argo data from 700-2000 meters will be regularly included in total heat.
Andrew Fl: “Heck, the Netherlands proves that humanity is capable of adjusting the “effective†sea level all on our own, by raising the land in areas to actually claim them from the ocean.”
When you look at how much of the buildings, factories and infastructure of the world have been built in the last 100 years, I’ve always thought that we could rebuild most of civilization in a different place if we really needed to in the next 100 years.
SteveF (Comment #79951)-Yes, my description of how the land was claimed was inaccurate. At the time I wrote my comment I was going “off the cuff” and hadn’t remember the history quite right. I believe modern Boston, however, expanded over the sea by landfilling.
Tilo Reber (Comment #79953)-We can and probably will massively move people and infrastructure around within a century, regardless of whether we need to or not. Human beings are quite industrious!