Posts Tagged ‘VR2009’


Uh, Oh! Karl, et. al., is bad news for Stefan Rahmstorf’s sea level rise rate.

September 25, 2015

Conclusion first

When the 20th century GISS temperature is modified according to Tom Karl,, it causes the 21st century sea level predictions of Vermeer’s and Rahmstorf’s semi-empirical model to go down!


I have written extensively about “Global sea level linked to global temperature,” by Vermeer and Rahmstorf (which I will refer to as VR2009).

VR2009 was a widely cited claim of using historical 20th century sea level and temperature data to calculate parameters that could be used to build a model to predict 21st century sea level rise for various 21st century temperature scenarios.  I reproduced the VR2009 model based on their description.  My code was verified by reproducing the VR2009 results using the same inputs that they used.

I spent a lot of time pointing out some of the bizarre results of their model that surely disqualified it form being taken seriously, some of which can be seen here, here, and here.

I also spent a lot of time pointing out that the VR2009 choices of 20th century sea level data sources left much to be desired.  For example, they used the 2006 Church and White sea level data that was already outdated.  If they had used the revised Church and White data, then their resulting sea level rise predictions for the 21st century would have been much lower.

They happily modified Church’s and White’s outdated sea level data by subtracting a reservoir correction (Chao, et. al.), which made their 21st century predictions for sea level rise go up. But they made no attempt to estimate a groundwater depletion correction. It turns out, unsurprisingly, that the groundwater depletion is of the same magnitude as the reservoir correction (Wada, et. al.), and including it would have made their 21st century predictions go down.

Nevertheless, Rahmstorf would later claim that his modeling approach was “robust!”  That is, it would give essentially the same result for the 21st century given different sources of 20th century sea level data.

So, I also implemented the VR2009 technique using several different sources of sea level data, which should have given similar results, according to Rahmstorf’s claim of robustness.  In fact, they gave widely varying results, and every combination of sea level data, reservoir data, and groundwater depletion data that I tried gave lower results than VR2009’s chosen combination.

New Temperature Data!

The widely reported nearly two decade long pause in global warming was causing suicidal ideation among hard-core global warming alarmists.  Something had to be done to stop them from slitting their wrists with shards of glass from their shattered thermometers.

Just in the nick of time – revised temperature data!   Like all proper revisions of temperature data, this revision caused the reported temperature change of the 20th century to go up.

This was a result of a paper by Tom Karl, et. al. (Nature) based on very thin reasoning (see for example) that argued for such revision.  The folks at GISS (who provided VR2009’s temperature data) glommed onto Karl’s logic and subsequently revised their temperature data accordingly.  Other temperature data source like UAH and RSS did not.

Which means we must ask ourselves, what happens to 21st century sea level rise predictions based on the VR2009 model using the now modified GISS data?

VR2009 applied their model to six families of temperature scenarios for the 21st century form the IPCC’s 4th Assessment Report.  Let’s see what happens to each of those scenarios when we update the 20th century GISS temperature data.

The IPCC temperature scenarios that VR2009 used for prediction of 21st century sea level rise.

Case 1.

Sea level inputs are identical to what VR2009 used: Church’s and White’s sea level with the Chao reservoir correction.  The old GISS temperature data is replaced with the new GISS temperature data.  The table below shows that the new GISS data yields 21st century sea level rises that are about 17% less than when the old GISS data is used.

Old GISS vs New GISS

It is a shame that after Tom Karl went to all the trouble to increase the temperature rise of the 20th century it just makes VR2009’s model predict LOWER sea levels for the 21st century.  This must be a great disappointment to Vermeer and Rahmstorf, so you can be pretty sure they will never tell you this result. But I just did.

Case 2

As I pointed out previously, VR2009 chose to use outdated 2006 Church and White sea level data, instead of Church’s 2009 data.  They also neglected a groundwater depletion correction.  When these improvements are included the VR2009 model yields 21st century sea level rises that are only about 55% of VR2009.  When the new GISS temperature data is included in the mix this drops to about 45%.

New GISS CW2009 Chao Wada

Case 3.

Lest Vermeer or Rahmstorf argue that their large sea level rise rates are saved by another update of the Church and White data in 2011, I have include these results also.  The difference between 2009 and 2011 Church and White sea level data was small.  Here is how the 2011 Church and White sea level data version plays out in the VR2009 model. The resulting 21st century sea level rise predictions are only about 43% of the VR2009 predictions.

New GISS CW2011 Chao Wada

The trend continues.

It seems that no matter what combination of inputs that are used in the VR2009 model, the predicted sea level rise for the 21st century is always smaller than with VR2009’s choice of inputs.  I wonder what that implies?


Rahmstorf (2009): Off the mark again (part 13), 21st century projections with gamma = 1

December 16, 2011

Recall the six IPCC families of temperature scenarios, summed up in the following IPCC figure.  VR2009 applied these temperature scenarios to their model to yield corresponding sea level rise rates.  Let’s consider the A1F1 and A1T temperature scenarios.

Figure 1. (top) This is figure 10.26 from the IPCC AR4 Chapter 10, "Global Climate Projections." It shows the temperature projections for each of the six IPCC SRES emission scenarios averaged for the 19 AOGCM models and 3 carbon cycle feed backs and the standard deviations. (bottom) Zoom in on A1F1 and A1t averages.

Here are the resulting VR2009 sea-level rise rates for the A1T and A1F1 scenarios…

Figure 2. Resulting sea level rise rates when the VR2009 model is applied to the A1T and A1F1 temperature scenarios.

Figure 2. Resulting sea level rise rates when the VR2009 model is applied to the A1T and A1F1 temperature scenarios.

Nothing really surprising so far. The sea level rise rates look more or less like the temperatures. 

Now consider some the following hypothetical 21st century scenarios.  Note that they can’t be considered “extreme” when compared the 21st century temperature scenarios already used by VR2009.

Figure 3. The same IPCC temperature scenarios, A1T and A1F1, as in figure 1 and three hypothetical temperature scenarios from Moriarty.

Here are the resulting sea level rise rates…

Figure 4. The sea level rise rates due to the A1T and A1F1 temperature scenarios and three the hypothetical temperature scenarios from Moriarty.

Where are the sea level rise rates for Moriarty’s hypothetical temperature scenarios?  They are perfectly hidden below the sea A1T sea level rise rate.  How can that be?  Because they were designed to be that way to make a point.  See the math here and let γ=1 in equation (VIII) and you will get the idea.  This is not some mistake in my math, but rather a direct consequence of the VR2009 and one more illustration of the bizarre consequences of their model.


Rahmstorf (2009): Off the mark again (Part 6). Satellite data

May 27, 2010

Let’s consider the equation that relates sea level rise rate to temperature from Vermeer’s and Rahmstorf’s 2009 PNAS paper, “Global sea level linked to global temperature” (referred to as VR2009 for the rest of the post) again. 

where H is the sea level and T is the global temperature.  VR2009 has already told us the values for a, b, and To (a = 05.6 mm*a-1K-1, b = -49 mm*K-1, and To = -0.41 K). 

Some of the bizarre consequences of this formula have already been discussed extensively.

Applying VR2009 to satellite sea level data.

 Here I apply their formula to the best sea level data available, from the Topex and Jason-1 satellites.

 The graph below shows the result of using the GISS monthly temperature data, covering 1996 to the present, to calculate the sea level from VR2009’s equation.  It also shows the actual satellite sea level data covering the same time period, and a version of the sea level data that has a reservoir “correction” applied. Click on the image to enlarge.

The thing that should jump out at you is that the calculated sea level shows an increasing rise rate, while the actual satellite derived data for the same time period shows a decreasing sea level rise rate. This may look like a subtle point, but predictions of sea level rise in the 21st century are extremely sensitive to these types of differences. A model that is applied to the best data of the last one and a half decades and cannot even get the sign of the change of the sea level rise rate (d2H/dt2) correct cannot possibly be trusted to draw conclusions about the remainder of the 21st century.

Reversed look: getting temperature from the sea level data

In my previous posts I have used VR2009’s equation on various temperature scenarios to generate sea levels.  We can reverse this process by re-arranging VR2009’s equation to use sea levels to generate temperature….

Fitting the satellite sea level data, H, to a function, inserting the function’s derivative (dH/dt) into equation 2, and solving the differential equation will give an analytical solution for the temperature.  Reasonable initial conditions can be extracted from the GISS temperature data.*

The graph of temperature vs. time, below, shows the result of this process.   It shows the GISS 5-year mean temperature, the GISS monthly land+ocean temperature (which covers 1996 to present), and the 7-month running average of the GISS monthly land+ocean  temperature.   Contrasted to these three versions of the GISS temperature data is the temperature that the VR2009 model (equation 1, above) would require to yield a sea level that looks like the actual satellite measured sea level (from 1996 to present).

So, VR2009 would require a ridiculously rising temperature to generate the measured sea level rise of the last decade and a half.  This bizarre result is one more reason that VR2009 should be rejected


* In my example, the satellite sea level data from 1996 to the present was fit to a quadratic, as shown in figure 1…

The initial condition was chosen as 0.44 degress in the year 2001.5, as indicated by the GISS 5-year mean.