Titusville

It’s been awhile since I have shown new USHCN stations, and its not for lack of material. But I got busy with the UCAR conference, publishing a slide show, and other things. But this morning, über volunteer Don Kostuch sent me a note on his latest survey in Titusville, FL near Cape Canaveral and KSC. I’d like to point out that Don has traveled further and surveyed more stations in the USA than anyone. He is a surveying machine. He wrote this in his email to me:

“On your scale of 1 to 5, this is an 8. Peace, Don Kostuch”

Ok in the past we have seen stations on rooftops, at sewage treatment plants, over concrete, next to air conditioners, next to diesel generators, with nearby parking, excessive nighttime humidity, and at non-standard observing heights.

Imagine a USHCN station that embraces all of that. I give you the Titusville, FL USHCN station:

Ever thorough, Don also provided photographs of the Climate Reference Network site, just 7 miles east at KSC, which demonstrates the correct environment for measurement of near surface air temperature:

Now I know there will be the usual critics who will jump in and say “This can be adjusted for!”. Ok here is your chance, show me the equations to untangle Titusville’s temperature record from microsite bias. Personally, it looks FUBAR to me.

59 Comments

  1. CO2Breath
    Posted Sep 19, 2007 at 12:56 PM | Permalink

    heh, heh, heh, he said FUBAR.

    No more comments necessary.

  2. MarkW
    Posted Sep 19, 2007 at 12:58 PM | Permalink

    What’s that roofing material? Could it be tar? It’s dark enough.

  3. Posted Sep 19, 2007 at 12:59 PM | Permalink

    It’s easy. Asphalt has increased at 2% per year. Simply lower all surface temperatures at the rate of 2% per year. Math is fun!

  4. VirgilM
    Posted Sep 19, 2007 at 1:00 PM | Permalink

    How long has this MMTS been at that location? Virgil

  5. Gary
    Posted Sep 19, 2007 at 1:04 PM | Permalink

    Does rocket exhaust raise or lower temperatures?

  6. MarkW
    Posted Sep 19, 2007 at 1:06 PM | Permalink

    Pat,

    Would that be Celsius, Fahrenheit, or Kelvin?

  7. CO2Breath
    Posted Sep 19, 2007 at 1:07 PM | Permalink

    #4 Virgil,

    The i’net is your friend:

    http://mi3.ncdc.noaa.gov/mi3qry/search.cfm?PleaseWait=OK

    1996.

  8. Larry
    Posted Sep 19, 2007 at 1:22 PM | Permalink

    Some of the sewage treatment equipment can be significant heat and humidity sources. I can’t make out what that rectangular structure is that the MMTS is mounted on. I see some sort of exhaust fan, and vent. It would be useful to know what that is. Possibly some sort of odor control system.

  9. Anthony Watts
    Posted Sep 19, 2007 at 1:31 PM | Permalink

    RE4, 7 The measuring location at the Waste Water Treatment Plant has been there since 1982. The CRS was replaced with MMTS in 1996.

    Here is the aerial view:

    http://maps.google.com/maps?q=1105+Buffalo+Rd,+Titusville,+FL+32796,+USA&ie=UTF8&t=h&z=16&om=1

  10. David
    Posted Sep 19, 2007 at 1:37 PM | Permalink

    I know how to adjust for it: set it up right to begin with. Also, it seems like it would be easier and more accurate to adjust for moving it x meters to a better location rather than to try to adjust for this messed up site!

  11. Geoff Olynyk
    Posted Sep 19, 2007 at 1:38 PM | Permalink

    To me, that time-series graph seems to correlate well with the large-scale temperature trends seen worldwide. It doesn’t look that FUBAR to me. The pictures, though, are another story.

  12. steven mosher
    Posted Sep 19, 2007 at 2:15 PM | Permalink

    The charts are very telling.

    I loaded up a few nearby rural/periurban sites with long records.. I tried to stay with 150KM or so.

    So, Titusville, Ft Pierce, Federal Point, Inverness, Saint Leo, Bartow. I probably should have picked
    up Ocala.

    Then I ran differences Titusville – ft Pierce; Titus-Federal pt; Titus-Inverness; Titus -Saint Leo;
    Titus – Bartow.

    If the pair of sites see the the same climate forces over the century of data you would expect the
    difference ( Titus – X) to have a zero trend. If the climate warms, it warms for both.

    If Titusville warms MORE than its neighbors you will see a positive trend for all the differences.

    I think you’ll see this.

    Then you can check neighbors against each other..

  13. MarkW
    Posted Sep 19, 2007 at 2:20 PM | Permalink

    Geoff,

    Since it appears that the worldwide trends are constructed using sites almost, if not as, badly FUBAR’d as Titusville, claiming that this site matches the ROW is not exactly something to brag about.

  14. henry
    Posted Sep 19, 2007 at 2:22 PM | Permalink

    Were there two charts (one chart for each of the stations)? Would be nice to see if there’s a difference between them.

    “Geoff Olynyk says:

    September 19th, 2007 at 1:38 pm
    To me, that time-series graph seems to correlate well with the large-scale temperature trends seen worldwide. It doesn’t look that FUBAR to me. The pictures, though, are another story.”

    Without a trend line, it’s hard to see, but I didn’t think there was a 2C rise in global temps (only .6).

    But then, this station only represents about 1/1200 of 2% of the earth’s surface, so it has no real effect on global averages…

  15. steven mosher
    Posted Sep 19, 2007 at 2:27 PM | Permalink

    RE 11.

    The first step is to compare the site with its neighbors not the globe.

    The site warms relative to all of its neighbors that are clasified as Rural.

  16. Geoff Olynyk
    Posted Sep 19, 2007 at 2:40 PM | Permalink

    Hey, far be it from me to defend a site that looks like that. I can see from the pictures that it’s going to be useless as a long-term reference.

    I’m just sayin’, the time-series shows the general shape that we would expect if the station was correlated to the global average — the rise in the early part of the 1900s, then decline after WWII to the 1980, then a quick increase since then.

    MarkW and henry both have good points though. (1) a correlation to the (possibly useless) GISTEMP global average perhaps doesn’t mean much at this point, and (2) I never looked at the vertical scale.

  17. Sam Urbinto
    Posted Sep 19, 2007 at 3:11 PM | Permalink

    BUT IT CAN BE ADJUSTED FOR!!!!!

  18. Posted Sep 19, 2007 at 3:29 PM | Permalink

    perhaps i am going blind.

    i am staring at that graph, searching for the 20°C jump, but i can NOT see it.
    not even a 5°C jump. none.

    can anyone help me out?

    wasn t the error of a type 5 station to be bigger than 5°C?

    ps: you might want to take a look at Christian enlightened answer:

    http://www.climateaudit.org/?p=2069#comment-138764

  19. Posted Sep 19, 2007 at 3:36 PM | Permalink

    Sam,

    It doesn’t need to be adjusted – evaporation from the sludge ponds dissipates the heat from the AC exhaust and concrete before it reaches any sensors. OK, maybe not perfectly but well within the NASA quality limits…

  20. steven mosher
    Posted Sep 19, 2007 at 3:42 PM | Permalink

    Anthony.

    If you have the ablitity to post charts on your sites I would suggest this.

    Take the rural sites close to Titusville, I mentioned them above. Do a simple average
    they are all with 100-200KM or so.

    Take a 5 year Moving average of this series ( trailing)
    Do the same thing for TitusVille.

    Difference them. Area – Titus.

    I’m going to figure out how to use JohnVs program to do these kind of Comparisons.

  21. Sam Urbinto
    Posted Sep 19, 2007 at 3:57 PM | Permalink

    #21, Rick, I’m kidding dude.

    #20, sod, if I’m always 10 degrees off, it looks the same as always being 10 off. But it gives the reading a lot of room to move and become “warming” or “cooling”. How much of that graph is the error moving and how much is the anomaly changing.

    To answer your question, I don’t know.

  22. Sam Urbinto
    Posted Sep 19, 2007 at 3:57 PM | Permalink

    #21, Rick, I’m kidding dude.

    #20, sod, if it’s always 10 degrees off, it looks the same as always being 10 off. But it gives the reading a lot of room to move and become “warming” or “cooling”. How much of that graph is the error moving and how much is the anomaly changing.

    To answer your question, I don’t know.

  23. Sam Urbinto
    Posted Sep 19, 2007 at 3:57 PM | Permalink

    #21, Rick, I’m kidding dude.

    #20, sod, if it’s always 10 degrees off, it looks the same as always being 10 off as if it’s always 0 or 5 off. But it gives the reading a lot of room to move and become “warming” or “cooling”. How much of that graph is the error moving and how much is the anomaly changing.

    To answer your question, I don’t know.

  24. steven mosher
    Posted Sep 19, 2007 at 4:01 PM | Permalink

    RE 20 … Sod… ( shows sod a shiny thing) over here…

    The 5C figure represents a estimate of BIAS. For a similiarly situated site ( same lat/lon elevation)
    that does not have the microsite issues.

    Homework for you.

    Here is a list of data. from 1901 to present.
    X = ( Average temp around titus) – (Average temp at titusville)

    If the average temp around titus was constant and titus got warmer than its neighbors
    what would be see?

    Area 10, 10, 10 ,10 ,10
    Titus 5, 6, 7, 8 ,9
    Difference 5, 4, 3, 2 ,1

    So Go plot this.

    1.739633333
    1.050966667
    0.9523
    0.6792
    0.5896
    0.6002
    0.806
    0.8736
    0.7728
    0.493
    0.0886
    0.011733333
    0.1209
    0.2278
    0.5737
    0.6981
    0.8553
    0.8713
    0.8709
    0.7744
    0.7264
    0.6516
    0.6488
    0.6924
    0.7076
    0.7202
    0.7647
    0.7939
    0.8315
    0.7775
    0.7073
    0.7104
    0.7004
    0.6296
    0.632
    0.6808
    0.6476
    0.6804
    0.722
    0.68
    0.5992
    0.5524
    0.45
    0.3776
    0.1804
    0.3212
    0.2432
    0.2688
    0.376
    0.612
    0.5516
    0.632
    0.6352
    0.5268
    0.5552
    0.5348
    0.5984
    0.6032
    0.5828
    0.5584
    0.5236
    0.5236
    0.4952
    0.4752
    0.4828
    0.4528
    0.4472
    0.454
    0.4936
    0.4901
    0.4849
    0.4337
    0.4369
    0.3453
    0.3208
    0.3396
    0.3008
    0.1636
    0.278133333
    0.299866667
    0.545066667
    0.572
    0.638
    0.5796
    0.6564
    0.462
    0.4012
    0.4416
    0.3024
    0.2196
    0.1936
    0.1288
    0.032
    0.1338
    -0.0286
    -0.055
    -0.0162
    -0.0446
    -0.0281
    0.1235
    0.0403
    -0.1269
    -0.2041
    -0.3452
    -0.4548

  25. Posted Sep 19, 2007 at 4:15 PM | Permalink

    The 5C figure represents a estimate of BIAS. For a similiarly situated site ( same lat/lon elevation)

    the term that christian used was ” possible instantaneous error in one day”.

    If the average temp around titus was constant and titus got warmer than its neighbors
    what would be see?

    Area 10, 10, 10 ,10 ,10
    Titus 5, 6, 7, 8 ,9
    Difference 5, 4, 3, 2 ,1

    i am very glad that you understand, that the important point is the TREND.

    but i thought the station was moved to that position only in 1982. so we would expect a pretty SIGNIFICANT jump around that year.

  26. jae
    Posted Sep 19, 2007 at 4:31 PM | Permalink

    sod, if the temperature IS actually increasing through time, then the presence of asphalt, etc, would amplify that trend, due to more heat storage, radiation, etc. The “trend” trick doesn’t necessarily solve all the issues.

  27. Posted Sep 19, 2007 at 5:11 PM | Permalink

    Geoff Olynyk September 19th, 2007 at 1:38 pm,

    To me, that time-series graph seems to correlate well with the large-scale temperature trends seen worldwide. It doesn’t look that FUBAR to me. The pictures, though, are another story.

    Let me see. Obviously bad site correlates to ROW.

    What can it mean?

  28. Anthony Watts
    Posted Sep 19, 2007 at 5:19 PM | Permalink

    RE27, it was CRS, not MMTS that year, MMTS at current location did not happen until 1996. CRS was likely mounted in a different location. MMTS by virtue of cable has limits to placement, and tends to get put in compromising locations. Also, CRS needs more space, and when CRS placements were being done. COOP managers didn’t have to worry about cable, so better locations got chosen, even if overall site are wasn’t the best they could often pick the least potentially biased location.

    So no, you would not expect bias jump to be as large when a CRS to CRS move was done, of course there are exceptions, like Tucson.

  29. steven mosher
    Posted Sep 19, 2007 at 6:02 PM | Permalink

    27.

    5C in one day.? Some UHI effects exceed this on a daily basis.

    Here is a test Sod. We will see if Sod should sod off or not.

    TOBs. Time of Observation. you have a thermometer. It records the MAXIMUM in the
    last 24 hours and the MINIMUM in the last 24 hours.

    The climatalogical day ends at midnight.

    At midnight you get out of bed and write down MAX and Min.

    I’m your next door neighbor. I record my thermometer at 7AM

    1. What is the average difference of our measurements?
    2. What factors influence this difference.

    Answer these questions and you pass go. Ignore it, don’t let the door knob violate
    your nether regions on the way out the door.

  30. David
    Posted Sep 19, 2007 at 8:41 PM | Permalink

    Even in the best case scenario, if this site and everything close to it were always like this, then one could possibly make use of the deltas over the long term and be fine. Consistency is the key here for that case. An issue arises when you want to use the actual measurements and compare them with other sites.

  31. Jonde
    Posted Sep 19, 2007 at 8:55 PM | Permalink

    Sod is still mistaking the whole principle of classification of class 1 to 5. What the class 1 to 5 are showing is “The magnitude of probability of error in temperature measurements”. It does not mean that everyday the error is over 5C or more in class 5 sites. It only represents the possibility of error and it’s magnitude. It could be said that class 1 site “very unlikely” cause error in temperature measurements and class 5 “very likely” cause error. Also, magnitude of possible error is “low” in class 1 and “high” in class 5.

    Still, hypothetically it is possible that under certain conditions the error in class 1 site could be bigger then in class 5 site, but this is “highly unlikely” situation.

    I hope sod now understands how the system is meant to work.

  32. Ian McLeod
    Posted Sep 19, 2007 at 9:16 PM | Permalink

    #31, lol

  33. cce
    Posted Sep 19, 2007 at 11:48 PM | Permalink

    I’m no mathematician, and this may be naive, but shouldn’t you be able to identify problems by comparing the instantaneous trend of each station within a given radius? Calculate the 5 year (or some other) trend for each site. Rank the sites from smallest trend to largest trend. Plot the trends on a graph with the rank as the x axis. You will likely get some kind of s curve. Find the site with the flattest slope in the s curve and use that as the “master” station for that area. Move on to the next area.

  34. Jaime
    Posted Sep 20, 2007 at 8:48 AM | Permalink

    Gents, no matter if you discuss this to death with applied statistics and physics thrown in in spades, there’s just one conclusion that a scientifically naive but intelligent person with good olfactory abilities will easily reach.

    The whole thing stinks to high heaven!

  35. MarkW
    Posted Sep 20, 2007 at 10:49 AM | Permalink

    Rick:
    Sludge ponds give off heat. All that biotic activity.

    Mosher:
    To amplify on what you said.
    If the error is caused by an asphault pad, then on those days without much sun, the error will be less than on days that are completely cloudless.
    If the error is air conditioners, then the error is more in the summer than in winter.
    Etc.

    Sod:
    It’s mighty rare to have a site that is grassy one day, then asphault as far as the eye can see the next.
    They degrade slowly, over years.
    If you are looking for sudden jumps, you aren’t going to find them. (At least not very often.)

  36. DocMartyn
    Posted Sep 20, 2007 at 10:52 AM | Permalink

    Anthony, I don’t know how you maintain your network of little helpers, but it might be an idea to contact the ones who have servayed sewage plants. Try to get them to take photos of the plants in Decenber or January when these has been some snow fall. You can have them take pic’s along the approach to the plant and the actual site.
    The sewage plants can actually generate a lot of heat (due to the bacterial activity). The factory my father worked at was very near a sewage plant and on clear frosty mornings it would be covered in clouds. In snowy days you would see a ring of wet ground surrounding the plant. The mix of heat and lots of water increases the local temperature on non-windy days.

  37. Posted Sep 20, 2007 at 10:58 AM | Permalink

    MarkW,

    I need to find the irony tags around here. I thought within NASA quality standards would have been enough. I shall exercise more care in the future.

  38. steven mosher
    Posted Sep 20, 2007 at 11:37 AM | Permalink

    RE 37. YUP!

    First. The Temperature Mean used by GISS and others is this: (TMAX+TMIN)/2

    Second: Site contamination will not impact TMAX, unless the site is shaded. Site contamination
    impacts TMIN. This halves the impact of site issues.

    Third: You won’t see effects every day. effect are modulated by wind
    and clouds and season.

    Fourth: The error is not cumulative.

    I said early on that I expected the diffeence between GISS and CRN12 to be on the order
    of 10-20%. So if GISS showed a trend of .7C since 1975, CRN12 would show .55C to .63C
    trend.

  39. Chris Kaiser
    Posted Sep 20, 2007 at 12:53 PM | Permalink

    Has anyone managed yet to identify a list of “golden sites” that are: class 1 or class 2, rural location (lights =0?) and a long history without station moves?

    I’m sure I’m not alone in wondering what the long term temperature trend is just from such reference sites.

  40. Posted Sep 20, 2007 at 3:01 PM | Permalink

    sod, if the temperature IS actually increasing through time, then the presence of asphalt, etc, would amplify that trend, due to more heat storage, radiation, etc. The “trend” trick doesn’t necessarily solve all the issues.

    a good thing to test. perhaps Christian can provide us with winter data from that meteo france test site.

    i doubt you will find any 5°C jumps though.

    TOBs. Time of Observation. you have a thermometer. It records the MAXIMUM in the
    last 24 hours and the MINIMUM in the last 24 hours.

    The climatalogical day ends at midnight.

    At midnight you get out of bed and write down MAX and Min.

    I’m your next door neighbor. I record my thermometer at 7AM

    1. What is the average difference of our measurements?
    2. What factors influence this difference.

    i am aware of the TOBs corrections done, for example by hansen. what is your point with this post?

    t’s mighty rare to have a site that is grassy one day, then asphault as far as the eye can see the next.
    They degrade slowly, over years.
    If you are looking for sudden jumps, you aren’t going to find them. (At least not very often.)

    on the long time graphs we re looking at, a development over a few years will LOOK like a jump.

    First. The Temperature Mean used by GISS and others is this: (TMAX+TMIN)/2

    Second: Site contamination will not impact TMAX, unless the site is shaded. Site contamination
    impacts TMIN. This halves the impact of site issues.

    Third: You won’t see effects every day. effect are modulated by wind
    and clouds and season.

    Fourth: The error is not cumulative.

    I said early on that I expected the diffeence between GISS and CRN12 to be on the order
    of 10-20%. So if GISS showed a trend of .7C since 1975, CRN12 would show .55C to .63C
    trend.

    this is all very reasonable.

    why don t we hear more about a 0.07 change, and more about the +5°C or even +20°C thing?

  41. Posted Sep 20, 2007 at 3:43 PM | Permalink

    ref 42 Sod,

    Glad you asked! From the AMS articles I have been reading the MMTS shelters can produce errors in that range under low wind or shield conditions. That would be rare, but there seems to be a definite warming error in the MMTS shelters under high solar irradiance provided there is no snow on the ground for some odd reason.

    That makes me wounder why it has a warming correction?

  42. Sam Urbinto
    Posted Sep 20, 2007 at 4:40 PM | Permalink

    I posted a less coordinated version of this earlier as a rather disjointed 2 posts in a different thread in a kind of random order. Thought I’d put it here in a thread about siting and what to expect. Just thoughts, nothing official.

    Sensors need to have a high enough degree of resolution to track temperatures to .01 C

    Sensors need to be regularly calibrated.

    Sensors need to be shielded from wind, snow, rain, dust, and heat, and from possible human or animal interference or general damage.

    Shields need to not heat or cool the air or otherwise disrupt the free mixing of the surface’s thermal characteristics with the air.

    Sensors need to all be at a uniform height (5 feet?) consistent with measuring the Atmospheric Boundry Layer (ie not so low they measure the surface itself only directly under the sensor and not so high they’re not in the surface layer of the ABL.)

    Sensors need to be far enough away (100 feet?) from obstructions and other materials that might interact with or influence the surface/air mix.

    Sensors need to be located in a place indicative of the majority of the area being sampled.

    The area being sampled needs to be large enough to be meaningful, but not so large there are too many varied conditions in the area. (eg 5×5 is too big IMHO)

    The area being sampled needs to be of a known size, so the sensor’s contribution to the whole can be weighted properly. (Why compare 10×10 to .5x.5?)

    The location of the sensor has to be of a known altitude and there needs to be 1 hygrometer at the site to also track humidity.

    There will be a minimum of 3 sensors at least 10 feet apart. (YMMV)

    Sensors will take frequent samples and report them via radio, IR, satellite, wireless Internet or burried cable to a recording station.

    Humidity readings will be transmitted at the same time as temperature.

    Recording stations will collect the temperature and humidity information and in conjunction with altitude, generate an additional figure of effective temperature (or to normalize to some standard humidity and altitude?).

    There will be 1 recording station per 5 sites, which will correlate the data from the sites with each other to ensure nothing abnormal is going on (and/or perform calculations on them?).

    There will be an automated process to collect the data from the recording stations at 1 or more locations, where any further needed processing will take place at as many levels and to the degrees required. (I know, exceedingly vague, but I’m talking about mainly like factoring out water in the grid, combining with sea temps, etc etc etc whatever)

  43. MarkW
    Posted Sep 20, 2007 at 4:46 PM | Permalink

    sod,

    That’s only if all of the development occurrs at once. Typically it’s one building at a time. Each jump is only a few hundredths of a degree.

  44. Kristen Byrnes
    Posted Sep 20, 2007 at 5:35 PM | Permalink

    I posted something similar to this in the Second look thread.

    The reason that you will not find large differences in the CRN 5 stations is because everyone is using USHCN version 2. The first step in creating USHCN v2 is:

    Quality Evaluation and Database Construction

    First, daily maximum and minimum temperatures and total precipitation were extracted from a number of different NCDC data sources and subjected to a series of quality evaluation checks. The three sources of daily observations included DSI-3200, DSI-3206 and DSI-3210. Daily maximum and minimum temperature values that passed the evaluation checks were used to compute monthly average values. However, no monthly temperature average or total precipitation value was calculated for station-months in which more than 9 were missing or flagged as erroneous.”

    That is why John V’s CRN 1,2 and CRN 5 graphs are so close. The big temperature problems like you might expect with this and other CRN 5 stations are already adjusted out.

  45. John V.
    Posted Sep 20, 2007 at 8:57 PM | Permalink

    Kristen:

    I just realized that you’re cross-posting lies about my analysis to multiple threads. That’s disgusting. I now must cross-post myself to set the record straight.

    What gave you the idea that I’m using USHCNv2 data? Every one of my plots is clearly labelled “Rural CRN12 Stations (GHCNv2 Raw) and GISTEMP (Sept 12, 2007)” in bold letters across the top. I do not mention USHCN or TOBS in any of my posts. USHCNv2 is not even available online yet (there website states “Version 2 station data will be made available in July 2007″ but there is no link that I could find).

    Here’s what it looks like to me. My results don’t fit with your world view, so you are convincing yourself there must be a problem with my analysis. Two days ago you were “ROLF”. Now you’re spreading false information about my analysis, and questioning my integrity by implication.

    Here is the data file I used:
    ftp://ftp.ncdc.noaa.gov/pub/data/ghcn/v2/v2.mean.z

    Here is the readme file that explains that it is raw data:
    ftp://ftp.ncdc.noaa.gov/pub/data/ghcn/v2/v2.temperature.readme

    Download the data for yourself and try it out. This is what you’ll find for the USA lower 48:

    1. GISTEMP agrees very well with the rural cRN12 stations;
    2. Nearly all of the trend disagreement between GISTEMP and CRN12 is prior to 1935;

    (See my posts 266, 267, and 271 in http://www.climateaudit.org/?p=2069 for details).

    PS:
    I did consider the possibility that I accidentally used the wrong data file. After double-checking I know that is not the case.

    PPS:
    I would appreciate it if somebody could confirm just how “raw” the GHCNv2 raw data is. Thanks.

  46. Kristen Byrnes
    Posted Sep 20, 2007 at 9:30 PM | Permalink

    Quick cross-post:

    John V,
    You may or may not be aware that the GLOBAL HCN data comes from 30 different sources. Where does the source of data come from for the US 48 states? Let me help you; USHCN. (that’s the data set you are using)

    Do you remember not too long ago when people on this blog were complaining about the data being switched on them from NASA? They went from a SHAP to FILNET. If you look at the description of the USHCN v2 link that I provided you might notice that the USHCN origonal version used SHAP, USHCN V2 uses FILNET. That is because NASA switched to the special USHCN V2 data set provided to them by the same folks at NOAA. It appears that the GHCN set you are using was changed in the same way, without notice.

    Your apologies are accepted in advance. I am here to help you, not the other way around

  47. Steve McIntyre
    Posted Sep 20, 2007 at 9:51 PM | Permalink

    #47. JOhn V and Kristen, please dial back the rhetoric.

    John V, you ask:

    What gave you the idea that I’m using USHCNv2 data? Every one of my plots is clearly labelled “Rural CRN12 Stations (GHCNv2 Raw) and GISTEMP (Sept 12, 2007)” in bold letters across the top. I do not mention USHCN or TOBS in any of my posts.

    In this case, I think that you’re making a distinction without a difference. GHCN v2 Raw is obtained from USHCN for the stations at issue. IT is as you say USHCN v1, not USHCN v2 which is not released yet, but it is USHCN data nonetheless.

    Kristen says:

    If you look at the description of the USHCN v2 link that I provided you might notice that the USHCN origonal version used SHAP, USHCN V2 uses FILNET. That is because NASA switched to the special USHCN V2 data set provided to them by the same folks at NOAA. It appears that the GHCN set you are using was changed in the same way, without notice.

    I’m unaware of any recent switch in the GHCN data. USHCN has several versions of their station data: raw, TOBS, SHAP, MMTC, FILNET and urban. My understanding is that GHCN Raw matches USHCN Raw, but this is from memory and I don’t guarantee this. NASA used SHAP until Sept 7, when it switched to FILNET.

    Of course, none of this matters for non-USHCN stations and as SChmidt points out, the US is only 2% of the world’s surface.

  48. Anthony Watts
    Posted Sep 20, 2007 at 11:08 PM | Permalink

    Just for grins, why not run it with data from the Version1 folder just to eliminate the issue?

    ftp://ftp.ncdc.noaa.gov/pub/data/ghcn/v1

  49. Anthony Watts
    Posted Sep 20, 2007 at 11:21 PM | Permalink

    Another option might be to run the USHCN folder data rather than GHCN set which can be found here:

    ftp://ftp.ncdc.noaa.gov/pub/data/ushcn/

    just to see if anything changes. I’m not trying to make John jump through hoops, but given what we’ve seen lately with GISS data switcheroos, and exhaustive testing of different data sets might be a good idea.

    You probably already know this, but for the benefit of all whom might be reading, a whole sea of data can be found here:

    ftp://ftp.ncdc.noaa.gov/pub/data/

    Just paste that URL into Internet Explorer and you’ll be able to browse them.

  50. D. Patterson
    Posted Sep 20, 2007 at 11:30 PM | Permalink

    Re: #49

    A suggestion…

    is there someone who can maintain some type of a racing form or scorecard that maintains and explains the datasets in play and each of their incarnations? The folks working with them have been frequently confused and the visitors and lurkers are hopelessly lost. It will almost certainly pay dividends down the road for data users and the audience watching the spectacle of the dataset morphing unfold.

  51. John Lang
    Posted Sep 21, 2007 at 6:46 AM | Permalink

    D. Patterson, #49

    You can see how the different versions relate to one another in this chart.

    In terms of how each version’s adjustments add to the overall trend since 1900 in temperature:

    – TOBS adds 0.35F to the RAW data

    – MMTS then adds 0.05F to TOBS

    – SHAP then adds 0.25F to MMTS

    – FILNET then adds 0.1F to SHAP

    – FINAL reduces 0.1F from FILNET

    Needless to say, it is hard to believe that almost all the adjustments add so consistently to the trend. This cahrt stops in 2000 and we know there has been more changes in the trend post-2000.

    Now add some Hansen GISS adjustments to this and what do you get? Global warming.

    John V. use the RAW data so we can see if there really is some difference betwen CRN 1,2, and 5 stations before all the variation is adjusted out by global warming adjustments.

  52. Posted Sep 21, 2007 at 8:03 AM | Permalink

    1) Do we know exactly where the site was, and what that area looked like, before its current position?

    2) Assuming that, before 1920, the site had no asphalt, no AC, and no sewage treatment plant (reasonable assumptions, I believe), couldn’t we eliminate pretty much all of these microsite issues by setting up a temporary thermometer site with similar equipment (even if only for a day in a couple of seasons, or much better for a year) in the open field about 400 feet to the southeast of this site, and comparing the readings (at least for days without wind from the NW)? We wouldn’t know exactly when the microsite skew occurred, but we would know its magnitude for the recent period vs. the 1900-1920 period. And shouldn’t the government be doing this with at least several such sites, to see if their microsite adjustments are reasonable?

  53. John V.
    Posted Sep 21, 2007 at 8:21 AM | Permalink

    #298 Kristen:
    (last cross-post; let’s move this to the Second Look thread)

    I absolutely apologize for the tone of my last post.
    I should know better than to post at the end of a long and frustrating day.
    Now, let’s figure out where to find the raw data so my results can be updated if needed.

  54. Posted Sep 21, 2007 at 8:59 AM | Permalink

    That would be rare, but there seems to be a definite warming error in the MMTS shelters under high solar irradiance provided there is no snow on the ground for some odd reason.

    That makes me wounder why it has a warming correction?

    i have no idea. but the 20°C argument is based on a very bizarre situation (sunlight hitting the heat sensor DIRECTLY.)

    That’s only if all of the development occurrs at once. Typically it’s one building at a time. Each jump is only a few hundredths of a degree.

    how many buildings do you want to build around the sensor????
    again, why not hear more about those “few hundredths” and less about the 5°C thingy?

    i really like the notion of a 10 to 20% difference between GISS and type12. my guess is slightly lower, but i think those are reasonable numbers!

    lots of people with zero understanding are following these discussion, then run away with a completely useless set of numbers on their minds.

    Now, let’s figure out where to find the raw data so my results can be updated if needed.

    John, i have immense respect for the work you do and how you do it!

  55. MarkW
    Posted Sep 21, 2007 at 10:28 AM | Permalink

    I want to build no buildings around the sensor.
    The problem is that for many of them, lots have been built.

    And lots other things as well. Parking lots, concrete pads, air conditioner units, diesel generators, trash burn barrels, chimneys.

    For many of these sites, they have been degrading over years, sometimes decades. 5 degrees spread out over 20 years doesn’t leave very many big jumps tin the record.

  56. steven mosher
    Posted Sep 21, 2007 at 11:21 AM | Permalink

    RE 57.

    I think the issue is this. UHI ( urban heat Island ) is known to Corrupt
    TMIN: the minimum temp f the day. So Assume TMax = 10 and Tmin = 5.
    TMEAN will be (10+5)/2 = 7.5

    Now assume that Microsite issues ( asphalt ) are a Subspecies of UHI.
    ( they are) So, lets say that the Ashphalt currupys TMIN. TMIN is raised
    3C

    SO TMEAN = (10+8)/2 = 9C. SO a 3C effect is diminished to a 1.5C effect.

    Then we also knw that UHI is strongest in warmer months ( more on this latter)

    So this 1.5C effect happens for 33% of the year. So this effect is Now .5C.

    Then we als knw that clunds and wind can modulate this… So we lose 50%
    of .5C.. and we are down to a .25C effect.

    THEN, that .25C effect shows up OVER TIME.

    I’m hinting at some work I’ve done with JohnVs model and differences between
    class 5 and class 1&2.. AS A FUNCTION of Season.. First cut only…

    Others are welcomed to look. Look fr microsite bias in seasonal differences between
    CRN12 and CRN5..

  57. steven mosher
    Posted Sep 21, 2007 at 11:45 AM | Permalink

    56.
    Just google MMTS and you’ll find the staudy that indiactes that MMTS
    registered Cooler temps, SO, it’s adjusted UP.

    THE ISSUE is this. was the MMTS study broad enough to capture other effects
    from switching to MMTS.. Namely moving close to buildings.

    The Point is the MMTS adjustment needs to be audited. IF switching to MMTS
    gives cooler temps for the same location ( as in the study) FINE. Adjust
    the record. But if the study was not broad enough, or long enough ( all seasons)
    then the MMTS study needs a revisit.

    Its Not the adjustment. Its the magnitude and the manner of the adjustment.

  58. Posted Sep 21, 2007 at 11:47 AM | Permalink

    ref 58 The microsite and shelter induced errors are generally greater in warm months with low wind velocities. Another factor like “latex” is that the MMTS shelters yellow with aging.

    This study by Hubbard et al shows the results of extensive test of MMTS, CRS and GILL shelters. A later study published in 2004 indicated that there is a 0.3 to 1.0 warming bias in the MMTS measurements. I can’t find the URL but it was an AMS published article. this error could be due to aging of stations (yellowing). Karl et al. published a paper (late 1990’s?)stating that a -0.3 to -0.7 degree error was associated with the MMTS system.

    The Hubbard study indicates to me that the error varies with region and season. One study that I cannot locate was by Thomas Blackburn. He maintained a fair number of in situ co-located MMTS and CRS sites. The only co-located site I could find was Ft. Collins CO.

  59. Scooternyc
    Posted Sep 24, 2007 at 11:00 AM | Permalink

    The photos made me laugh out loud at the stupidity of such a position! I wish someone would do a really good investigative report on this stuff, like John Stossel and he could show it up on 20/20 – it would be a-mahzing!

Follow

Get every new post delivered to your Inbox.

Join 3,208 other followers

%d bloggers like this: