Thursday, July 06, 2017

More About RSS's Large Changes to Their Temperatures.

RSS has posted a FAQ about their version 4.0 changes for the lower troposphere temperature anomaly, here.

Roy Spencer says, yabutt, their new numbers (+0.18°C/decade, up from +0.12°C/decade) are still nowhere close to the model calculation of 0.27°C/dec. I don't know where that number comes from, so I asked him. Will let you know.

Frankly, I'm starting to wonder if either of these satellite datasets can be useful. Roy writes:
In general, it is difficult for us to follow the chain of diurnal corrections in the new RSS paper. Using a climate model to make the diurnal drift adjustments, but then adjusting those adjustments with empirical satellite data feels somewhat convoluted to us.
This doesn't sound good. If the one set of supposed experts can't follow what the other set of supposed experts are doing, then who are we to possibly judge?

Maybe it's time to just forget about the satellite measurements of the atmosphere and focus on surface, where measurements are much easier. That's where we all live, anyway.

Anyway, here are the data for RSS v4.0's 12-month moving average:

Pretty obvious warming.

Also, whereas RSS LT v3.3 shows 2016 to be the warmest year by, like UAH, 0.02 C, the new version v4.0 shows it to be the warmest year by 0.16 C. Huge and indisputable.

These differences will probably remain for some time -- I doubt UAH will do another entirely new version before Christy and/or Spencer retires, after which the UAH dataset will probably, unfortunately, fade into insignificance. It's now clearly the outlier when compared to RSS and the several surface datasets.


JoeT said...

David, I would be interested in hearing an answer as well. Considering that Roy's most important point is that "both RSS and UAH lower tropospheric trends are considerably below the average trends from the climate model." one would think he would be thrilled to provide the data.

If you look at Gavin Schmidt's post here, you can find the CMPIP5 data as well as John Christy's estimate of the average trend as 0.214 C/decade. I would be curious why that isn't the trend Roy is referring to.

David in Cal said...

I find this adjustment suspicious, for the following reasons:

1. The adjustment has a substantial impact on the rate of warming.
2. The people who do RSS are no doubt very competent. If there was some significant flaw in their methodology, why didn't they notice it years ago?
3. The adjustment process is complex, thus difficult to evaluate.
4. The adjustment process appears to be ad hoc, I think.
4. AFAIK the adjustment process hasn't been validated, by comparing adjusted temperatures to temperatures at the same point in time that are known to be accurate. In fact, I suspect that this sort of validation wouldn't be possible.
5. There have been a number of adjustments to various temperature readings, land, sea, and air. It seems that all the adjustments result in a higher rate of warming. If there was no bias, one would expect half the adjustments to reduce the rate of warming. In fact, the Urban Heat Island effect may be the biggest data problem, so more than half the adjustments should reduce the rate of warming.

Layzej said...


According to Carl Mears himself: “I would have to say that the surface data seems like its more accurate." and "You get a much bigger spread of results from the satellite data than you do from the surface data".

Layzej said...

Here's a comparison of trend in various versions of UAH and RSS. looks like some adjustments increased the trend while others decreased, though RSS updates almost always decreased the trend while UAH updates almost always increased the trend. The range is huge - about 0.13C/decade between the lowest and highest if you look around 2002.

They list "Some things to note":

UAH and RSS agree better than they ever have, but only through about 2000. After that, they diverge fairly rapidly.
Sometime in 2009, the previous versions of the RSS and UAH datasets agreed exactly (for global TLT trends).

Ned said...

The comment above by David in Cal is an embarrassment to this blog. Its six points contain so many errors and misleading statements, all of which strangely incline in the same direction, that it can only be considered to be propaganda.

Although I suppose the fact that DiC can't even count to six without making a mistake should be taken into consideration.

Yes the RSS data have been validated using radiosondes. No, the improvements are not ad-hoc. No, not all changes in temperature data sets increase the trend; in fact the net effect of all adjustments (land + ocean) to the surface temperature records is to reduce, not increase the trend.

I am guessing that DiC expressed exactly zero "skepticism" about the adjustment of UAH from v5.6 to v6.0, which produced a large reduction in the warming trend.

Because with David in Cal, it's not about science, he's purely here to disseminate propaganda.

Ned said...

Here are a couple of links that would have answered David in Cal's "questions", if he were actually looking for answers.

Remote Sensing Systems: FAQ about the V4.0 TLT update

Mears and Wentz 2017 (preprint)

Basically, he posted a bunch of really dumb comments here wihout bothering to look at either the FAQ or the paper.

Ned said...

Note that the FAQ I linked to is also mentioned in the very first line of David Appell's post. It's astonishing that David in Cal would ignore the FAQ and jump right in with easily answered questions like:

"AFAIK the adjustment process hasn't been validated, by comparing adjusted temperatures to temperatures at the same point in time that are know to be accurate. In fact, I suspect that this sort of validation wouldn't be possible."

This point is addressed nicely in the FAQ, under the heading What about comparison with radiosondes (weather balloons)? The short version: RSS compared their new and old versions of TLT (and the UAH version as well) against four different radiosonde data sets. Note that the radiosonde data were not used in processing or calibrating the satellite measurements, so they do represent a fully independent source. The new RSS version 4 is a better match to the radisonde data than either the older RSS TLT or the UAH version.

David in Cal said...

Ned - you say the net effect of all adjustments reduce the trend. OTOH a recent paper says that the adjustments to land temperatures increased the trend a lot.

Figure IV-2 on page 12 in particular purports to show the adjustments both upward and downward by year.

You guessed wrongly about my attitude toward adjustments to UAH. My attitude is that any adjustment is uncertain. It's a kind of educated guess. So, all these adjustments tell us that there is very large uncertainty as to what the true historic temperatures were. This uncertainty adds to the uncertainty of the model structures, which helps explain why we don't have a single reliable predictive global warming model.


Layzej said...

Calling a Wordpress blog "a recent paper" is a bit of a stretch. No?

Ned said...

Come on. That is not "a recent paper" it's unpublished blog propaganda from Anthony Watts's buddy Joe D'Aleo and others of his ilk.

But I like the way you try to quietly move the goalposts. I wrote:

"in fact the net effect of all adjustments (land + ocean) to the surface temperature records is to reduce, not increase the trend."

You try to rebut that by pointing to land-only temperatures, ignoring the 70% of the surface that is oceans. The net effect of adjustments on land is an increase, but the net effect of adjustments on oceans is a decrease, and because the oceans cover a much larger area, the overall impact of all adjustments on global mean surface temperatures is to decrease the trend.

This is nicely illustrated in figure 2(b) from Karl et al. 2015:

The figure is also available here:

And, again, this is only one of the multiple errors in your original comment. The changes to the RSS TLT algorithm are not "ad hoc". The new algorithm is validated using independent in-situ observations. It is not true that the effect of all adjustments is to increase warming; as I told you, the net effect over the entire globe is to decrease, not increase warming. UHI is not the "biggest data problem" (UHI is nonexistent at sea and even just on land there are more important factors in adjustments, e.g. TObs).

As for your second comment, I don't recall ever seeing you criticize the change from UAH v5.6 to UAHv6 in terms like this, despite the fact that it had a large cooling effect on the trend. If you can show me a comment where you did so, I'd appreciate it. Thanks.

Ned said...

It is not true that all (or even a majority) of adjustments to the temperature record are upwards. But it does appear to be true that all of David in Cal's "mistakes" tilt in the same direction. Funny how that works.

David in Cal said...

Yes, Ned, my "mistakes" all point in the same direction, namely that there's great uncertainty in climate science.

Your ad hominem hardly refutes the published paper, written by experts in the field.

Regarding sea surface temperatures, the method was changed, perhaps for the better, but that change did indeed increase the warming trend:

"A buoy-only sea surface temperature record supports NOAA’s adjustments.

Significant recent media and political attention has been focused on the new NOAA temperature record, which shows considerably more warming than their prior record during the period from 1998 to present. The main factor behind these changes is the correction in ocean temperatures to account for the transition from ship engine room intake measurement to buoy-based measurements and a calibration of differences across ships using nighttime marine air temperatures (NMAT). Here we seek to evaluate the changes to the NOAA ocean temperature record by constructing a new buoy-only sea surface temperature record. We find that a record using only buoys (and requiring no adjustments) is effectively identical in trend to the new NOAA record and significantly higher than the old one.

Ned said...

That's not a published paper and it's embarrassing that you keep thinking it is.

You also are either confused about or willfully misrepresenting Karl et al. 2015. The revised sea surface temperature data set does have a tiny effect on trends post-1998, compared to the previous version, but it's dwarfed by the much larger cooling effect of all adjustments to SST data earlier in the century. As can be seen clearly in Karl et al. Figure 2b, linked above, going from the raw SST observations to your "adjusted" version greatly reduces the long term warming trend. If you were to insist on only using unadjusted data for the surface temperature record, the 20th century warming trend would be much much larger.

Ned said...

Layzej wrote: "Calling a Wordpress blog "a recent paper" is a bit of a stretch. No?"

It's an unpublished and unreviewed "report" hosted on the Heartland Institute's website. As such, it's a natural source for David in Cal's purposes.

Layzej said...

"A semi-regular reminder that adjustments in surface temp analyses for non-climate-related biases LOWER 20th C trends. You're welcome."
- Gavin Schmidt‏ tweet including a nice graph illustrating.

David in Cal said...

Thanks Layzej. I don't know how to reconcile the apparently contradictory comments by experts. Compiling adjustments should be an objective, determined process. My guess is that Schmidt is right and Wallace, et. al are also right, but they're measuring different things. If they're wrong, they should be easy to refute. I don't know of anyone who has refuted either paper.

I wonder whether Gavin Schmidt included the switch in method for sea surface temperature.

Schmidt says his graph reflects "adjustments in surface temp analyses for non-climate-related biases." Does that mean that there were also some other adjustments that were climate-related which Schmidt didn't include? OTOH his graph says, "The net effect of all data adjustments."

Layzej said...

"non-climate related biases" are what you are trying to remove with the corrections. No -
there are not climate-related adjustments excluded from Schmidt's graph.

Joe d'aleo is not an expert except at how to lie with statistics. He's found some subset of some record that shows net warming as a result of corrections. Can't find it globally in CRU? Try land only. Still no? Try NOAA. Still no? Try GHCN. Ah ha!

Some corrections will result in more warming. Some less. The question is only whether the adjustment is necessary, or whether some necessary correction is missing. If they have a valid concern with any one of the corrections they should publish. Showing that some subset of some record goes one way or another is not interesting or surprising. The net effect is much LESS warming as a result of the corrections.

Ned said...

Gavin's tweet is referring to the net effect of all adjustments to land and ocean observations. The graph you pointed to by Joe D'Aleo etc. is referring only to land temperatures. There is no conflict, and both of these facts have been known and widely discussed for years.

There are plenty of sincere, realistic explanations of the adjustments for surface observations, for example here:

Among the largest factors requiring adjustments are changes in time of observation, changes in instrumentation, and changes in station location. The different surface records (NOAA, Berkeley Earth, GISTEMP) handle these somewhat differently, so the fact that they come up with very similar answers provides a measure of robustness.

Bringing this back to the topic of this tread, the surface temperature data clearly have a higher signal to noise ratio than the satellite troposphere temperature do. Nonetheless, it's nice to see RSS continuing to work on their algorithms.

Layzej said...

It looks like they've discovered some subsets of the data where the non-climate-related biases reduce the apparent warming in the record and ignore the rest. Most importantly, they ignore the fact that overall the corrections have drastically reduced the apparent warming.

The papers that originally identified and provided corrections to each bias would undoubtedly have highlighted the impact of the bias. There is nothing new here, though it is a good opportunity for innuendo.

Ned said...

Layzej, Nick Stokes put together a nifty Google Maps application that lets you map GHCN stations based on the effect of adjustment on their trends:

Have you seen it? It's really cool.

The controls are at right. Click the radio button for "Trend_Adj > 0" and then the radio button for "Pink". Then switch the Trend_Adj to " < 0" and click "Cyan".

Result is a map showing stations in pink where the adjustments warm the trend, or in cyan where they cool the trend.

It's a nice counter to the claim that "all the adjustments result in a higher rate of warming."

Layzej said...

Good resource. Thx.

Layzej said...

Snopes has weighed in on the D'Aleo blog:

Some good points made.