NASA GISS shows 2009 as tied for 2nd warmest year on record

The usual prefatory stuff: We're talking about global temperatures, the globe is three-quarters water, the US land area is maybe 2% of the globe's surface, and so on. So presumed error in the US land trend, if such error exists, will have only a small impact on the global temperature trend. Just get that out of the way.

First, you again fail to mention that NOAA staff literally tested the notion that the low-quality stations distort the trend. They literally used Watts list of high-quality stations. They got the same temperature trend with the full dataset and with the stations that Watts approved of. This is a non-issue.

The brief NOAA report is a good read if you actually want some understanding of the issues. They cite peer-reviewed study of this exact issue; they talk about the new climate monitoring network that is going into place; and then they empirically test Watts exact list of stations to demonstrate that they get the same trend with Watt's 70 or so high-quality stations or with the full network.

http://www.ncdc.noaa.gov/oa/about/response-v2.pdf

Second, there is no mention of the actual underlying analytics. There is a huge amount of redundancy in the climate network as a whole, because there are very few true degrees of freedom in the temperature readings. That is, there is very high correlation of temperature trends for stations near one another, and a reasonably high correlation of temperature trends even for stations at fairly large distances from one another. So if you have 1200 thermometers, you can't get 1200 arbitrary readings. For example, if it's a colder-than-average winter in Sterling, VA, there's a very high probability that it's a colder-than-average winter in Richmond, VA, and a decent probability that its a colder-than-average winter in Philadelphia, PA.

(Another way to understand that is just to look at any gridded graph of the current world temperature anomaly. It never looks like a checkerboard (alternating cool and hot squares), it always looks like it has been painted with a broad brush. If it's warmer-than-average in Paris, it's highly likely to be warmer-than-average in Brussels, etc.)

In fact, for the globe as a whole, depending on how you measure it, there are only about 200 true degrees of freedom in temperature readings. That is, you could adequately monitor global average temperature (to within 0.05 degree C accuracy or thereabouts) with just 200 well-placed thermometers. The fact that you have many times that many temperature observations means you have redundancy, and can use that redundancy to clean up the existing data.

That spatial correlation is largely what drives the NOAA result cited above. The 70 high-quality stations, by themselves, provide nearly as good an estimate of US average temperature trend as the full network does. (Using the full network reduces the variance of the estimate).

And that is, in part, how NOAA (and independently NASA and others) go about cleaning up the data. In addition to literally knowing the station metadata (e.g., did the station location change), they use the high degree of spatial correlation to find the thermometers that are out-of-whack and exclude them, or, possibly, apply empirical adjustments to bring them into concordance with the rest of the network. For example, and most basically, they check for outliers -- if you have N nearby stations, N-1 agree, the Nth one is grossly different, they toss the outlier. So if somebody fires up the barbeque pit, and screws up the reading, it'll get tossed. Another example is NASA's adjustment for urban heat island effects -- they find the rural stations nearest to an urban station and adjust the urban trend to match the rural trend. If there is no nearby rural station, they toss the observation. Does it matter if they toss some observations in this fashion. No, probably not. Why not? Because the temperature trends are highly spatially correlated -- so if you have to gap-fill Richmond, VA's trend with Sterling, VAs trend plus those of other nearby stations, you probably have made only the tiniest of errors, averaged across the entire database.

Well, why bother, I don't think this is going to change anybody's view anyway. So, let me finish up.

Second, you cite the analysis of some unknown with a BS in applied physics. As if I'm supposed to be believe that some amateur newbie knows more about how to do this properly than the people who spend their careers doing it. I don't think so.

Third, the historical network is used for temperature measurements because that's what we've got. Yes, now that climate is a significant issue, the US is putting stations into place specifically to monitor climate. But all you've got for historical data is what was on the ground at the time.

What are you suggesting -- that we go back in time to make those stations right? Or just pretend that we have zero information about historical temperatures?

Or, just maybe, the sensible thing to do is to extract as much useful information as possible out of the available data, in as sensible a fashion as possible. And, ideally, to have different groups use different methods, to see whether or not those methods matter materially. That is, maybe what we're doing right now is the most sensible course of action.

Finally, you trust the Christy satellite data. I'll just note a few things.

A) Producing temperatures from the satellite observations is a difficult thing to do, as I've posted here before. At least four groups do it and get different estimates.

B ) Christy/Spencer are on verision 5.2 of their temperature time series.

C) They twice got the trends very wrong, and in a very public fashion. First, saying there was no trend, second saying there was very little trend. The second time, the people at RSS had to point out their math error to them. Maybe they are as right as they can be now.

D) If you look at the various entities that produce temperature trends from the satellite data, you can see a lot of disagreement as to methods. RSS, for example, flatly states that you cannot use the satellite data above about 85 degrees N and S latitude, or for high elevations (I think it's anything over 3000 meters, but I'd have to check that). Christy/Spencer, by contrast, just blow right through that and calculate temperatures for the entire surface. That's odd, but consistent with what appears to be a fairly cavalier attitude toward calculation errors.

E) If you look at the Christy/Spencer UAH data in detail, it's really hard to make much sense of it. For example, the GISS data show much more warming over land than over water. As I understand the theory, that's what the data should show. And they show that consistently for the Northern Hemisphere and the Southern Hemisphere. In fact, if you adjust the S.H. data to account for Antarctica (land with zero warming -- that is, you try to look at the S.H. land that is comparable to the N.H. land), you find that GISS gives a remarkably consistent estimate of land warming, ocean warming, and the difference between the two. With the Christy/Spencer data, by contrast, in the Northern Hemisphere, land warms just slightly faster than water, while in the Southern Hemisphere, water warms slightly faster than land.

Once again, this is not to say that any one of these series contains no information. Just to say that the satellite data certainly are not inherently preferable to the ground-based readings.

 

I think this is as fine an example of the quality of your postings as any.

So, first, show me where James Hansen himself literally wrote that Manhattan would be flooded by 2008.

I don't think you can. Because I don't believe Hansen ever wrote that. I'm not looking for conditional statements (e.g., if we had a repeat of historical episodes, we might see rises of 3 feet per decade.) None of that. I want to see something written by Hansen where he says Manhattan, flooded, 2008.

OK, if there's no direct evidence from Hansen's own writings, then who says he said it? Where does that come from. What's the qualify of the evidence?

Well, look at your favorite site:

A little known 20 year old climate change prediction by Dr. James Hansen – that failed badly Watts Up With That?

Here's the source of information. In 2001, A former newspaper reporter (called Rob Reiss by Watts, but actually identified in the Salon interview as Bob Reiss) was flacking his forthcoming book on global warming (called "Stormy Weather" by Watts, but in fact identified in the Salon piece as "The Coming Storm: Extreme Weather and Our Terrifying Future,"). In an interview on Salon, he said that he recalled a conversation with Hansen 12 or 13 years ago (the uncertainty is Reiss', not mine), in which Hansen appeared to misunderstand the question, and then made a prediction of flooding.

OK, put aside the fact that Watts did not bother to get either the author's name or the book title correct. Let me just focus on substance.

Let me see if I have this straight about what it is I am supposed to believe here. First, I'm supposed to believe that Hansen makes this sensationalist prediction, and this newspaper reporter just sits on that for --- 12 to 13 years? Never mentions it in print anywhere. Then, as he goes out to promote his book, then he recalls it, in great detail. Second, I'm supposed to believe that Hansen made that prediction in total contradiction to his scholarly work, where he was on record projecting average global temperature increases on the order of 0.15 to 0.20 degrees centigrade per decade?

In short, the story I'm supposed to believe is that Hansen ignored his own published work to make this startling prediction of flooding to a newspaper report. This newspaper reporter mentioned this to nobody for more than a decade. Now, he can't quite recall exactly when the prediction was made (12 or 13 years ago), but he remembers it in exquisite detail -- in 2001, as he goes out to flack his forthcoming book.

Oh, yeah, and the quote on Watts site says 20 to 30 years. And Reiss says he asked Hansen again in 2001 and Hansen says he'd say the same thing. So Reiss claims Hansen was happy to predict massive flooding (and water shortages, and rising crime rates) within an apparent seven years. Oddly, again, this newspaper reporter made absolutely nothing out of those apparent dire and immediate predictions -- until he went out to sell his book. And the only thing Watts posts is an interview with Riess on Salon, so apparently Riess didn't think it worth putting Hansen's prediction into his book (or Watts didn't bother to read his book, also possible).

Yeah, OK, whatever.

Hansen has in fact made the point that the IPCC sea level rise projections are based on there being no large dynamical changes in the ice sheets. (Translation: they stay in one piece and melt in place.) But the paleological record clearly shows surges in sea level that could only have occurred, in the past, if there had been massive, short term collapse of the land-based ice sheets.

So Hansen is one of a few scientists who point out that there is an asymetric risk function here -- while the mean projected rise is low, over any small time interval there is a non-negligible risk of rapid sea level rise, and, based on the paleological record, over longer intervals (century scale) it's a pretty good bet that you'll see periods of rapid sea level rise. If the paleological record replays itself, it's a pretty good bet that some generation in the near future will get to experience rapid sea level rise. We shouldn't look at the IPCC mean and, based on that, feel that there is no risk there. And for doing that, for reminding us of that possibility, I think Hansen does us a favor.

But, a firm prediction, Manhattan, 2008, nah, I don't believe it. Show me some hard evidence. Not some hearsay about a 12 or 13 year old discussion that was kept quiet by a newspaper reporter, until he remembered it in vivid detail as he went out to advertise his forthcoming book.

 

Well, I'm sorry I have to spell this out so bluntly, but what I was saying was, put up or shut up.

You've had a loud and obnoxious mouth here regarding James Hansen, calling him, at various times, "moron", "psychopath", and so on. And you keep asserting that he's stupid because, among other things, he "predicted" that Manhattan would be flooded by 2008.

I went back to your source on that information (you posted the pictures from Watt's site), and laid out the correct story, as posted just above. All the issues from the the small (Watts didn't get either the name of the author or the name of the book correct), to the middling (the actual Salon quote was 20 to 30 years, not 20 years), to the large (we're supposed to believe that a newspaper reporter kept this prediction secret for 12 or 13 years, only brought it up in an interview flacking his book, but did not actually put this in his book, and that Hansen made this prediction in total contradiction to his own published work on the likely speed of climate change).

So, put up or shut up. If you want to assert that Hansen is foolish because he predicted a flooded Manhattan by 2008, please show me where James Hansen himself actually wrote that he expected Manhattan to be flooded by 2008.