Ronan Connolly On Sea Levels
By Paul Homewood
Dr Ronan Connolly has a thoughtful essay on his website, which goes into some depth on the subject of sea levels. It’s well worth a read. Although very detailed, it is still very readable.
The section that really took my eye though was the one on satellite measurements, in particular this bit:
So, are the satellite estimates reliable? Well, in order to answer that, we have to learn a little bit about how they were actually constructed.
Unfortunately, satellite altimeters don’t actually measure sea levels directly. Instead, they measure the length of time it takes light signals sent from the satellite to bounce back. In general, the longer the signal takes, the further the satellite is from the sea surface. So, in theory, this measurement could be converted into a measure of the sea surface height, i.e., the mean sea level.
However, the conversion is complicated, and a number of other factors need to be estimated and then taken into account. For instance, the distance of the satellite from the Earth’s surface varies slightly as it travels along its orbit, because the gravitational pull of the Earth is not exactly uniform – see the Wikipedia page on “geoid”, and the maps in Figure 19.
So, in order to convert a particular “satellite-sea surface distance” into a sea level measurement, the “satellite-Earth’s surface distance” also needs to be independently measured, e.g., using the DORIS system.
Another complexity is that light takes slightly longer to travel when travelling through water vapour than dry air. So, the water vapour concentrations associated with a given satellite reading also need to be estimated, and accounted for.
As a result, satellite estimates of sea levels involve the use of complex models, approximations, other measurements and calculations. Unfortunately, this means that if there are problems in any of those stages, it could introduce artificial biases into the estimates, possibly making them unreliable… or even worse, wrong.
Mörner, 2004 (Abstract; Google Scholar access) managed to track down a graph of the raw satellite trends from the TOPEX/Poseidon satellite up to 2000. When he looked at this graph (Figure 20), he didn’t see much of a “sea level rise”. Instead of the +2.8 or +3.1 mm/yr trends commonly reported, it appeared to him that sea levels had been essentially constant from 1993 to 1996. He agreed that from 1997 to 1999, there were considerable sea level changes. But, they comprised falls as well as rises, and were probably related to the unusual 1997-98 El Niño event.
Mörner, 2004 was a controversial paper, and several of the researchers involved with the TOPEX/Poseidon analysis objected to Mörner’s analysis, e.g., Nerem et al., 2007 (Abstract). However, surprisingly, these objections were not over his claim that the raw satellite data showed little trend. They agreed with Mörner that the original satellite data didn’t show much of a sea level rise. Instead, their objection was that he should have used their adjusted data. They felt the raw data was unreliable, and had developed a series of adjustments which they believed made the trends more realistic.
For example, Keihm et al., 2000 (Abstract; Google Scholar access) had decided that the TOPEX satellite was showing an instrumental negative drift of 1.0-1.5 mm/yr between October 1992 and December 1996. So, they adjusted the data by adding a positive trend of 1.0-1.5 mm/yr to that period. Chambers et al., 2003 (Abstract; Google Scholar access) decided that even more negative biases were introduced when the TOPEX satellite switched to its backup instrument in February 1999. So, they introduced more adjustments. This set of adjustments increased the apparent sea level rise from +1.7 mm/yr to +2.8 mm/yr. Neither set of adjustments affected the period January 1997-January 1999, but as Mörner had noted the raw data already showed significant variability for that period due to the 1997-98 El Niño event. Finally, they believe that an adjustment of +0.3 mm/yr is necessary to account for Peltier’s Glacial Isostatic Adjustments (see Section 4).
It turns out that almost all of the +2.8 mm/yr (or +3.1 mm/yr if Peltier’s post-glacial rebound adjustments are applied) sea level rise in the 1993-present satellite estimates are due to adjustments! The raw data (which no longer seems to be in the public domain) doesn’t show much of a trend, after all.
It is plausible that the unadjusted trends are unreliable, as Nerem et al., 2007 claimed. However, that doesn’t automatically mean that Nerem et al.’s adjustments are valid
The paper by Chambers et al, used as an example, is here, and this is the Abstract:
During the calibration of the Jason-1 altimeter, it was discovered that the Jason-TOPEX sea surface height (SSH) residuals contained a trend correlated with significant wave height (SWH), indicating an error in the sea state bias (SSB) model for one or both of the altimeters. After updating the SSB model for Jason the trend remained, which pointed to an error in the TOPEX model. Since two different TOPEX altimeters (TOPEX_A and TOPEX_B) have operated during the mission, we estimated new SSB models using data from each one. The estimated SSB model for TOPEX_B is significantly different than the model provided with the data, which was estimated using only TOPEX_A data. Replacing the SSB model on TOPEX with the new TOPEX_A and TOPEX_B models not only removes the correlation with SWH in the Jason-TOPEX SSH residuals, it also removes most of a bias between TOPEX_A and TOPEX_B that has been observed in tide gauge calibrations. The magnitude of the change between TOPEX_A SSH and TOPEX_B SSH is of the order of 10 mm globally, with TOPEX_B SSH increasing when the new SSB model is applied. The application of this improved model will increase the rate of observed global mean sea level rise from 1.7 mm/year when the original TOPEX data are used to 2.8 mm/year when the updated SSB models are applied.
Most people probably assume that satellite estimates of sea level change are based on accurate, actual measurements. They would likely be shocked to find that, instead, they are based on dodgy data, that is then subjected to massive adjustments, bigger than the observed changes.