Lindzen on Climate Feedback

Update- Some changes over the last few hours
Update2- Anthony Watts has contacted Lindzen for clarification
Update 3 on bottom

Science is an evolving process. Data, models, and methods all evolve in time, correcting errors along the way, and building more questions or robust conclusions in the process. It is necessary for people with expertise in a particular area to keep up with the data and any changes that may be made to it, as well as the underlying problems that may exist within the data. In some unfortunate cases, people lose objectivity and will use only a particular dataset (or version of that dataset) that re-inforced some point they are trying to make. In presenting information, not telling people outside of their field what the possible caveats are in a dataset, or explaining revisions that were made to data they show is not generally taken well in academic setting. So it is with a recent example.

Richard Lindzen has made his presence felt at Anthony Watts blog in “Lindzen on Negative Climate Feedback”. Accuweather also has a recent blog post on it. In the comments, it is being hailed as the new found gospel truth of negative climate feedbacks and low sensitivity. Unfortunately, the problem with most skeptical arguments is not what we are told, but rather what we are not told. So what aren’t we told?

Lindzen’s method is easy to understand. You can read it at Watts’ blog. The image of emphasis is,

Lindzen’s emphasis is on the outgoing LW flux at the top-of-the-atmosphere, and the fact it is large compared to models. This is inferred to mean less of a “blanketing” effect from greenhouse gases, and therefore feedbacks which are less positive than models suggest (or negative in this case). At “RealClimate”, gavin pointed out a fixed graph that appeared in Science after a comment by Kevin Trenberth. The changed figure is,

However, there have been major revisions to this data since 2002. Specifically, there was a significant correction for changing satellite altitude in the computer code that was not turned on which led to erroneous results in the earlier work (Bruce Wielicki, personal correspondence). In many other cases, this kind of thing would be something WUWT would be on top of, so it is unfortunate that it is not mentioned. There is an inverse square dependence on the amount of energy received at the Nonscanner WFOV instrument and the distance from the planet’s center; improperly accounting for altitude change led to spurious results for the TOA longwave and shortwave fluxes. This has been documented in Wong et al 2006, Journal of Climate, a paper not even mentioned by Lindzen. ERBE S10N_WFOV ERBS Edition3 Data Quality Summary is available to cover further issues. They state

Algorithm Changes between New Edition3 and Previous Edition2 Release
The main difference between new Edition3 and previous Edition2 release is in the treatment of TOA radiative fluxes resulting from changes in the ERBE nonscanner processing algorithm to account for decay in satellite altitude over the data period.

During an instrument performance study, Lee et al. (2003) discovered that the ERBE nonscanner inversion algorithm did not correctly account for the decay in the ERBS altitude over its mission lifetime; this can have a small but significant effect on the reported decadal changes of nonscanner TOA fluxes. The ERBE nonscanner inversion algorithm is used to convert nonscanner measurements at satellite altitude (approximately 611 km at the start of the mission) to TOA measurements at a reference altitude of 30 km. While these altitude changes over the 15-year period are small (on the order of 25 km) and do not affect the overall quality of the large regional fluxes, they do, however, have significant effect on the smaller changes associated with the observed large scale decadal changes in Earth radiation budget (Wong et al., 2005).

This satellite altitude related problem is unique to the ERBS nonscanner instrument and does not affect the quality of the ERBS scanner data product. The nonscanner is a hemispheric instrument which views the entire Earth disk along with the small portion of the deep space surrounding the Earth itself. As the satellite altitude dropped over its mission, the small portion of the deep space partially viewed by the nonscanner began to be filled in by the Earth view itself, resulting in more energy being recorded by the nonscanner instrument. Since the original and the Edition2 release data did not account for these subtle altitude changes, there is a small effect of artificially increasing the reported longwave, shortwave, and net fluxes over the mission lifetime. Specifically, the overall effect of this altitude change is a small increase (~0.6%) in both longwave and shortwave radiation over the 15-year period.

To minimize errors in the ERBS nonscanner data product, an altitude correction algorithm to the Edition2 data was developed and applied to the entire Edition2 data set. The result is the new Edition3 data set.

Actual ERBS Edition 3 data is available at Nasa langley. I will leave readers to explore further details. Assuming he was familiar with these updates, Lindzen should have at least told his readers why he felt the older version was better for his analysis. The experts working on it apparently do not think so. An examination reveals that these corrections eliminate most of the signal that Richard Lindzen was using (see figure 5).

Section 5 of the 2006 paper also does comparison with ocean heat storage data, where the two agree within the uncertainty of the ocean data sampling. This is pretty neat given the independent nature of the ERB data and ocean heat content measurements, an example of robustness that distinguishes results appearing in peer-reviewed documents vs. those in blog protocol.

In short, Lindzen’s analysis is based on outdated data that has been revised since 2002, and these revisions are not exactly recent, so he should have been aware of them. Using the more recent data would not allow him to make his argument as presented as WUWT. It would be nice to see an update at WUWT reflecting these changes.

I’ve said quite a bit about feedbacks lately and it’s a little old now, but many WUWT commenters still seem confused about how postive feedbacks relate to an unstable system. Lindzen has recently been using the “gas pedal” analogy (not only in Watts’ post, but at the skeptic conference) in which positive feedbacks are supposed to be analogous to someone changing the gas and brake pads in your car. If you want to slow down, you actually speed up. Apparently it follows that climate does not act this way. Actually feedbacks don’t really act this way either. If we let the moving car roll on a flat, frictionless surface (with no influence from the tires or air resistance) in the absence of any net force change, it will roll forever by Newton’s laws. Think of this as some equilibrium condition, with the climate analog being radiative balance. Pushing your gas or brake is more like the “radiative forcing” on the car which essentially puts it off of its current course. Positive feedbacks simply let the planet equilibriate at a higher temperature than the sensitivity from CO2 alone, but the same principle that balance is acheived still applies. Feedbacks go up like a converging power series and therefore never get strong enough to override the fourth power dependence of thermal radiation and trigger a “runaway.” In short, positive feedbacks can be stable and don’t require any runaway scenarios.

Update 3– Lindzen has responded to Anthony Watts at his blog post. I wish that more was to address, but to me he didn’t really say anything meaningful, but that is for readers to make judgments on. Essentially Lindzen has set up the usual attacks that adjustments are always made to favor “alarmism” (which is incorrect, if he bothers to read the standard literature from HadCRUT, GISS, etc on their methods; perhaps if his claim was more specific, he knows it would be that much easier to invalidate). A reduction in the LW flux at the TOA can be interpreted in other ways as well, some might argue for less overall warming in the 20th century for instance.

Lindzen once again claims that the changes still imply negative feedbacks, which is a rather dubious claim, given the discussion and comparisons with models in Wong et al. I also do not believe the full range of sensitivity can be evaluated from these results, but even so, the justification for strong negative feedbacks has vanished.