Betting

April UAH TLT Bounces up: Current value + 0.295C

69 Comments

Roy has posted: UAH took a large positive bounce in April hitting 0.295C up from 0.110 C in March. (Oddly, Chicago March was very anomalously; April didn’t see quite so warm.) Of course this means that this will be another year where the more virulent warmers point out it really truly is warming suggesting things are worse than we thought, model-watchers will wait to see if the 12 month average manages to pierce the multi-model mean projection from the AR4 and coolers will continue to explain it’s weather noise. For now, let’s look at the data since 1980. I’ve superimposed the line as a counter balance to Roys 3rd order polynomial:

Who won?

Despite the size of the upward bounce, the size of the move did not leave the bettors in the dust: four bet for even higher anomalies. Three of the four betting high were in the money. John Norris took first place by betting the anomaly would hit 0.282C. Paul S and Peter took 2nd and 3rd this month

Winnings in Quatloos for UAH TTL April, 2012 Predictions.
Rank Name Prediction (C) Bet Won
Gross Net
— Observed 0.295 (C)
1 John Norris 0.282 5 81.628 76.628
2 Paul S 0.28 3 39.182 36.182
3 Peter 0.31 2 20.897 18.897
4 Pavel Panenka 0.321 3 25.076 22.076
5 Bob Koss 0.337 5 33.435 28.435
6 Ray 0.24 5 26.748 21.748
7 Cassanders 0.223 5 21.398 16.398
8 Tim W. 0.22 5 17.119 12.119
9 Greg Meurer 0.22 3 8.217 5.217
10 Owen 0.213 5 5.692 0.692
11 Robert Leyland 0.204 4 0 -4
12 Boris 0.201 5 0 -5
13 Jeremy Harvey 0.198 2 0 -2
14 MichaelP 0.195 3 0 -3
15 EdS 0.18 5 0 -5
16 plazaeme 0.18 1 0 -1
17 MikeP 0.18 5 0 -5
18 Skeptikal 0.178 4 0 -4
19 Scott Basinger 0.175 1 0 -1
20 ivp0 0.173 5 0 -5
21 BenjaminG 0.159 5 0 -5
22 Jeff Condon 0.151 5 0 -5
23 Freezedried 0.15 3 0 -3
24 Nyq Only 0.15 5 0 -5
25 Paul Butler 0.144 5 0 -5
26 Anthony V 0.14 5 0 -5
27 Don B 0.138 4 0 -4
28 Anteros 0.137 5 0 -5
29 PaulS 0.135 4 0 -4
30 YFNWG 0.135 5 0 -5
31 nzgsw 0.132 5 0 -5
32 Arfur Bryant 0.132 5 0 -5
33 Jarmo 0.132 5 0 -5
34 AMac 0.131 3 0 -3
35 Earle Williams 0.123 5 0 -5
36 MarcH 0.123 5 0 -5
37 pdm 0.121 5 0 -5
38 sHx 0.121 5 0 -5
39 Rick 0.12 4 0 -4
40 CoRev 0.12 5 0 -5
41 John Knapp 0.117 3 0 -3
42 denny 0.115 3 0 -3
43 RobB 0.113 5 0 -5
44 ErnieP. 0.112 4 0 -4
45 SteveF 0.101 5 0 -5
46 AndrewKennett 0.1 4 0 -4
47 dallas 0.1 5 0 -5
48 Tamara 0.095 5 0 -5
49 Big Bear 0.089 5 0 -5
50 Pieter 0.079 5 0 -5
51 Jefff 0.07 4 0 -4
52 MDR 0.065 3 0 -3
53 Lance 0.061 5 0 -5
54 hswiseman 0.06 5 0 -5
55 Niels A Nielsen 0.055 5 0 -5
56 Paul Ostergaard 0.05 5 0 -5
57 mike worst 0.05 5 0 -5
58 Steve Taylor 0.04 4.25 0 -4.25
59 mct 0.017 5 0 -5
60 Hal 0.01 5 0 -5
61 N (bot?) 0 1 0 -1
62 DocMartyn 0 5 0 -5
63 diogenes -0.01 2 0 -2
64 ob -0.011 1 0 -1
65 KÃ¥re Kristiansen -0.042 5 0 -5
66 Gary Meyers -0.054 3.14 0 -3.142
67 IainT 0.65 3 0 -3

The net winnings for each member of the ensemble will be added to their accounts.

The betting script for May will be up tomorrow. (Note: No bots bet. The bet of 0C was claimed, I just forgot to edit the database. )

69 thoughts on “April UAH TLT Bounces up: Current value + 0.295C”

  2. Whoohoo!
    My first winnings!
    I didn’t expect to be in the range.
    I notice Roy Spencer mentions the RSS anomaly is lower
    than UAH, after adjustment to the same base period.
    Is he having doubts about is own figures?

  3. Is he having doubts about is own figures?

    Not necessarily. He may just be pointing this out because “some people” are always accusing UAH of being biased low due to the beliefs of Christy and Spencer.

  4. Billc (Comment #95378)
    May 9th, 2012 at 9:14 am
    “2012 will have a long way to go to pierce the multi-model mean.”
    ——————————
    True. We are now coming out of a 2-year double-dip and quite strong La Nina (see http://www.esrl.noaa.gov/psd/enso/mei/) that has persistently depressed temperatures, and are currently just entering ENSO neutral conditions. If we do in fact transition this summer to a full-fledged El Nino, I will bet on record-breaking anomalies in the fall and winter.

  5. @Ray

    I think he’s extra pointing it out this time because a recent publication that just came out by Fu (endorsed by Trenberth) accuses UAH of being the coldest dataset and out of line with everyone else; thus needing some unusual adjustments they propose to bring it in line with the climate models. In fact that isn’t true, and while that publication states that RSS is hotter than UAH, it’s the other way around.

    Lot of craziness going on lately. Lucia is also very right, as Roy has tried to point this out in the past due to those “UAH is anomalously cool” accusations that keep flying.

  7. Also, Lucia, does that temperature graph look anomalous to you? So April was a +0.295C anomaly, but when I look at the graph and use a ruler to line up the latest data point with the Y-axis, it looks like it’s falling at 0.395C, right beneath the 4, and way above the midway point between 2 and 4 which it should be just beneath instead.

  8. Ged–
    Because the various agencies pick different baselines, I rebaseline everything to 1980-1999. Roy uses the most recent baseline of all groups and that’s a reason my plots will always shift his reported anomalies up.

  10. Ged,

    The Fu study is more about the TMT (mid-troposphere) data than the TLT record featured here and by Roy, though they do suggest there are implications for TLT trends too – not sure how that would work out given that UAH and RSS have about the same trend over the whole record.

    Lucia’s anomaly is different because she’s using a different baseline.

  11. Ged says “In fact that isn’t true, and while that publication states that RSS is hotter than UAH, it’s the other way around.”

    Paul S says “UAH and RSS have about the same trend over the whole record.”

    Both true, and the corollary is that RSS ran a hotter trend in the early years (not referring to the UAH pre-corrected data, but the current/corrected dataset). To what extent is this picked up on in Fu et al?

  12. Ged, thanks for the explanation.
    I haven’t read that publication, but from my own calculations, after adjustment to the same baseline, I would say that UAH is generally above RSS and the average of other series.
    Of course, due to the later baseline, UAH can “appear” lower, but presumably they took that into account?
    It would be much simpler if everyone used the same baseline.

  13. Spencer and Christy really do need to release their computer software. The bias is in changing temperature of the NOAA-9 satellite so it is difficult to see how this could NOT affect the TLT product and as the last line of the abstract puts it

    ====================
    This warm target factor bias directly affects the UAH lower tropospheric (TLT) product and tropospheric temperature trends derived from a combination of TMT and lower stratospheric (TLS) channels.
    ===================

  17. Lucia,
    “rare disagreement”
    With Eli? I hope you meant to say just the opposite!

  18. Re: lucia (May 9 11:01),

    That would be agreement, yes?

    Not much point in releasing the software until version 6 is implemented, whenever that happens. I would also take anything Fu publishes about satellite temperature with a 50 pound sack of salt.

  19. I hope this isn’t too off-topic (or is this an off-topic thread?) for you guys, but I happened to follow a link to a post on Open Mind, and some of the graphs there confused me. I was wondering, could anyone pop over there and tell me if they make sense?

    Specifically, the ends of the graphed lines seem off to me. If you look at the end of USHCN record, there are four years running low compared to the years before them. Despite this, Tamino’s smoothed lines show a rising end. I know there are issues with endpoints in smoothing, but how does one manage/justify that? Moreover, he says:

    Lest you object that I’ve applied some fancy smoothing method designed to get what I wanted, let’s apply the simplest “smoothing” method of all. Let’s compute 5-year averages rather than 1-year averages. We get this:

    The last five year average he shows has a huge increase over the one before it, but as far as I can tell, we should see the exact opposite. How does that happen?

    My apologies if I’m just missing something obvious.

  20. Hmm.. surely Peter and I are tied for second place, or have I won out through alphabetic means? If so, I’m “Aardvark” next month.

  22. Eli,

    Here is an offer. If you want to write a request to Roy I will Join you in the request. Also, you should have a look at what magicjava was able to do in his attempt to get the software.

    You want to get ALL the software all the way down the whole processing chain. Basically, you want voltages as inputs.

    So, if you write the letter asking for the code, I’ll gladly sign on.

  23. oh, I forgot, we need to ask RSS as well. They are right up the road from me. occupy RSS? how about a tent city of geeks demanding RSS code. Eli, you up for it.

  25. Brandon–
    I think, but I’m not sure that’s he’s showing the ending with years that are multiples of 5.

  26. Clearly, just eyeballing, the 5 year average of 2006-2011 would be lower than 2005-2000. I guess we could make a graph where we plot Tamino’s choices vs. other choices for years of 5 year averages to see if we get any difference what the graphic suggests.

  27. ok…I am wearing a sweater (unusual) in May and the world is, supposedly warmer and all the temperatures of cities I visit in Europe have shown very cool temps throughout May and one would suspect cooler temps in the S hemisphere. Where was the hot-spot?

  28. Right now according to Policlimate, the hotspot dragging up planetary temps is Antarctica.

  29. lucia:

    Is this the data?

    I don’t think so, but I could be wrong. I know he is using the same data used on this page, but when I followed the link to where they say the data is, I couldn’t find an individual series. All I found was station data, which is valuable, but not what he used. But I did get the same results as him when I used that site to plot the data.*

    I think, but I’m not sure that’s he’s showing the ending with years that are multiples of 5.

    When you say “the ending.” you mean the ending of the periods being averaged, right? If so, I think you’re right. If you draw a line straight up from the years labeled on the chart, you’ll cross halfway between points each time. It makes sense since the red, smoothed line continues past the last point. The answer to my confusion is the last two years just aren’t included in that because there isn’t a full period for them. If he centered his periods two years later, the same thing would happen, just at the beginning of the line.

    But I’m still at a loss on the smoothing he did. He extends that line all the way to the end of the data, meaning it should include the last few years. By the looks of it, it doesn’t. The last four years are equal or lower than the ten years before them, so there’s no way a smoothed line including them should steadily rise at the end. And if it does exclude the points at the end, why does it extend all the way to 2012? It probably doesn’t matter, but it strikes me as very weird to see his results go up while the data goes down.

    And suddenly, it occurs to me I could just post at his blog and ask.

  30. lucia:

    Clearly, just eyeballing, the 5 year average of 2006-2011 would be lower than 2005-2000. I guess we could make a graph where we plot Tamino’s choices vs. other choices for years of 5 year averages to see if we get any difference what the graphic suggests.

    I don’t think would really impact the point he was making with the graph. He was saying modern temperatures were notably different than temperatures ~80 years ago, and that would be true regardless.

    However, there would be a notable change in the visual impact. The last point would be lower, and the point before it would be higher. That means the end of the graph would probably level out, or even show (slightly) decreasing temperatures.

  31. Bandon–
    At the bottom of that page you (and Tamino) linked I found “Please see our ftp site for statewide temperature and precipitation data
    which are used in making these charts and tables.”. I clicked “temperature” to get to the page I linked.

    However, there would be a notable change in the visual impact.

    Yep. And depending on what arguments are “out there” the change in visual impact can “matter” when the image is circulated. I don’t know why Tamino makes the sorts of graphical choices he makes. His habits tend to create distrust in people who begin by saying “huh” and then think “Oh. That was a big deceptive” after figuring out what he must have done.

  33. But I’m still at a loss on the smoothing he did. He extends that line all the way to the end of the data, meaning it should include the last few years. By the looks of it, it doesn’t. The last four years are equal or lower than the ten years before them, so there’s no way a smoothed line including them should steadily rise at the end. And if it does exclude the points at the end, why does it extend all the way to 2012? It probably doesn’t matter, but it strikes me as very weird to see his results go up while the data goes down.

    I read the article and he doesn’t seem to say. He just tries to convince you whatever choice he made was not tricky because if he plots 5 year smooth, it still ends by going up. But of course, when you look at the 5 year smooth and the 1 year data, your first thought is: “Huh? Is there some sort of computational error”. Then the next one is: “Oh. He made it end ‘up’ by not showing the most recent 5 year average.” Hmmm….

  35. Nick

    thanks for the chart…but the jumps seem to have been in really remote places. Do the terrestrial records, if there are any in these places, show the same kind of heating effects?

  37. Lucia…did you check the map that Nick posted? The areas covered by the hotspot seem at first sight to be places that are not inhabited by very many people. So my ref to central asia was to a part of central asia where few people actually reside and pay taxes, unless they subscribe to realclimate. but the real questions are how would we detect a big hotspot in the Siberian tundra and what would it matter? Did the thermometers and trees detect it too?

  38. diogenese,


    I think plenty of people live in the dark red blob on the right. I think quite a few people live in the red blob in Africa. I realize the “flyover states” in the midwest are not thickly populated as New York city or LA, but people do live there.

    If you mean that some of the dark red is in the arctic— sure. But people in Provo and Salt Lake are under the impression the cities they live in exist.

  40. #95419
    Actually, diogenes has a point. The map is a spherical harmonics fit, and it can wobble a bit where readings are sparse. But I’ve published these for nine months now, and they generally line up fairly well with the later GISS plots.

  41. lucia:

    At the bottom of that page you (and Tamino) linked I found “Please see our ftp site for statewide temperature and precipitation data
    which are used in making these charts and tables.”. I clicked “temperature” to get to the page I linked.

    You’re absolutely right. When I looked for a link for data, I saw the link above the form, and I didn’t look any farther. That was silly of me!

    I read the article and he doesn’t seem to say. He just tries to convince you whatever choice he made was not tricky because if he plots 5 year smooth, it still ends by going up. But of course, when you look at the 5 year smooth and the 1 year data, your first thought is: “Huh? Is there some sort of computational error”. Then the next one is: “Oh. He made it end ‘up’ by not showing the most recent 5 year average.” Hmmm….

    I asked him what he did, and he told me:

    The smoothing method was a modified lowess smooth (I use a different weighting function than the usual tricube) applied to the monthly data (rather than annual averages). Therefore it includes the most recent 4 months, which is astoundingly warmer than any preceding third-of-a-year (4.1 standard deviations above the long-term mean).

    Intuitively, I assumed two lines said to be representing the same data would represent the same data, but it appears the reason for my confusion is he included more data in one line than the other. Personally, I think that makes this comment of his peculiar:

    The smooth contradicts intuition simply because it’s right and the “intuitive result” is wrong.

  42. lucia:

    Bandon–

    It’s not every day I get called something I’ve never been called before!

  43. Personally, I think that makes this comment of his peculiar:

    The smooth contradicts intuition simply because it’s right and the “intuitive result” is wrong.

    The truth is the smooth contradict intuition because unless revealed by the person making the graph people tend to assume the smooth curve is based on the unsmoothed data.

    That still leaves open the question about the 5 year averages. Am I correct that they don’t involve the more most recent years?

  44. Has any bettor performed well enough over time to suggest he or she has supernatural talent?

  45. lucia:

    The truth is the smooth contradict intuition because unless revealed by the person making the graph people tend to assume the smooth curve is based on the unsmoothed data.

    Technically, I think using annual averages counts as smoothing. If so, the problem is that for one line, he handled the endpoint issue by excluding data, but he included that data for the other.

    That still leaves open the question about the 5 year averages. Am I correct that they don’t involve the more most recent years?

    I believe It doesn’t include the last 1 1/3 years, stopping after 2010.

  46. Brandon,

    Here is what Wikipedia says about the Lowess smoothing (used by Tamino, I assume, on both 1-yr and 5-yr data:
    ” It does this by fitting simple models to localized subsets of the data to build up a function that describes the deterministic part of the variation in the data, point by point. In fact, one of the chief attractions of this method is that the data analyst is not required to specify a global function of any form to fit a model to the data, only to fit segments of the data.”
    It creates a function that describes the secular trend which is not necessarily related to individual points at any part of the curve.

  47. OK, I went back and re-read Tamino – I don’t know what he did with the 5 year data to get the upturn.

  49. Owen:

    It creates a function that describes the secular trend which is not necessarily related to individual points at any part of the curve.

    It sounds like what you’re describing is curve fitting, not smoothing. If so, I think you got confused because you read what the article said is done to “localized subsets of the data.” as what is done to the whole series.

    LOWESS is definitely a smoothing function. In fact, it can actually be just a (weighted) moving average if you use a zero degree polynomial for your fitting.

  50. Ray (and others..)

    With all the talk about alleged ‘differences’ between UAH and RSS is it just a wild (and happy) coincidence that their trends from the beginning of the satellite era are the same? I mean ‘same’ as in identical to three significant figures? [0.134C per decade]

  51. Anteros, The difference in the tropics seems to be a big deal. It looks to me to be related to the issues with ozone changes in the tropics and subtropics which may be having a greater than anticipated (as in modeled) cooling impact. While the global trends are pretty close, the tropics trends are significantly different early in the data records, (there is about 0.05 degrees per decade difference over the whole satellite era).

    There are a couple of interesting emails on the issue with Susan Solomon appearing to believe the models do not very well represent the ozone part of the atmospheric picture.

  52. Anteros,
    In my case, any differences I was referring to were in the actual anomalies, not the trends, and I was referring to recent figures. I haven’t done any detailed calculations, but I think it is possible (hypothetically), for the UAH anomalies to be higher than those for RSS, but for the trends in those anomalies to be identical.

  53. Brandon/Lucia,

    How are the past 4 months ‘astoundingly warmer than any preceding third-of-a-year (4.1 standard deviations above the long-term mean)’ when the anomalies are below 2010 for each of the 4 months? I’m looking at UAH, maybe that’s my problem ;).

  54. Billc

    How are the past 4 months ‘astoundingly warmer than any preceding third-of-a-year (4.1 standard deviations above the long-term mean)’ when the anomalies are below 2010 for each of the 4 months? I’m looking at UAH, maybe that’s my problem 😉 .

    That would be your problem. Tamino was looking at USHCN, which only covers the continental United States. There can be huge differences in the two at times.

    As an update, I’ve pretty much sorted everything out with the graphs in Tamino’s post. I need to talk to lucia about some things first, but I should have it posted later today. It’s worse than I thought.

  55. I was just reading UAH’s response to the Fu paper:

    http://www.drroyspencer.com/2012/05/our-response-to-recent-criticism-of-the-uah-satellite-temperatures/

    This particular section jumped out at me:

    ”
    To compensate for this error, we devised a method to calculate a coefficient that when multiplied by the hot target temperature would remove this variation for each satellite. Note that the coefficients were calculated from the satellite data, they were not estimated in an ad hoc fashion.

    The calculation of this coefficient depends on a number of things, (a) the magnitude of the already-removed satellite drift correction (i.e. diurnal correction), (b) the way the inter-satellite differences are smoothed, and (c) the sequence in which the satellites are merged.
    ”
    Shouldn’t the hot target coefficient be applied independently and then the inter-satellite differences smoothed? My reading is that this is just another smoothing coefficient. Maybe I’m missing something.

  59. Tamino usually uses simple moving averages without explanations. That is why I am not even going to bother reading it. Its been pointed out before that when he doesn’t have enough data points he adds them as if they will be the same just to fill out his chart. He could use a dotted line or cut it off at the last valid point, or something, or put a note in, but if he likes the way it looks he just uses it.

  60. RE: Steven Mosher (Comment #95421)
    May 9th, 2012 at 5:13 pm

    “where is paulK2 to drive another nail in the lukewarmer coffin?”

    🙂 LOL!

  61. lucia:

    I’m sending Brandon an author log in. 🙂

    Now if I can only figure out how to make images in R turn out prettier. Or at least get the axes labeled right.

  62. Yikes. I totally forgot being logged in to write a post would change what I post comments as.

    Sh. You did not see my actual account name.

  63. I got curious about how often the anomalies move in either a positive or negative direction in sequential months. So, assuming I got the code right, I took the first order differences and out of 399 intervals I got 196 positive moves, 189 negative moves and 14 zero moves. I then wondered how often it moves in the same direction twice or more in a row. I ran the rle function against an array of positive, negative and zero moves and found that it moved 96 times in the same direction in successive months and 60 of those where it was of length two, 25 of length 3. Does this have any meaning besides an exercise in R? Don’t know, but what I take from it is that 147 times out of 399 intervals the temp moved in the same direction twice in a row or about 37 percent of the time. So you have about 2 to one odds that the next temp will be below the present value for you bettors.

  64. BarryW, wow! And I thought my method was complex. Sharpen dart. Raise/lower target on board according to best guess (I mean complex scientific calculation.) Shoot. It’s science don’cha no?

  66. Billc:

    duh. I missed an obvious one there.

    Don’t worry. That happens to all of us.

  67. I’m still on track to make a small fortune in Quatloos, out of a large one.

  68. CoREv

    Oh, I don’t bet. I’m retired so I have to conserve my Quatloos. Just wanted to add another level of complexity to the betting. Of course, that analysis gives you a clue as to which way it might go but doesn’t help you with the size of the change.

Comments are closed.