Sensor Blackening

36 Comments

  1. Posted Oct 20, 2007 at 9:09 AM | Permalink

    Could we have a thread on Sensor Blackening? I have tried to raise this twice, but was off-topic both times, so no one rose to the bait. My latest attempt was at #144 on the TOBS thread, below:

    Sensor blackening is another station micro-issue that could have a big effect on measured temperature trends. On surveying the Circleville, OH station for Surfacestations.org last month, I noticed that the MMTS was coated with black soil buildup that must surely be raising its temperature from when it was installed. Closeup photos of the top and vane are posted in the Surfacestations gallery. A blowup of the crud on the top may be viewed by clicking here, and the vanes by clicking here.

    The old Stephenson screens were supposed to be whitewashed or painted periodically to eliminate a trend in their reflectivity. Anthony Watts has validly called attention to the fact that modern TiO2 latex paint has different heat properties than the old CaCO3-based whitewash. The plastic housings on the new MMTS sensors don’t require painting, so paint type is not an issue. However, shouldn’t they at least be cleaned occasionally??

    The SHAP adjustment is supposed to adjust station readings for recorded events like moves or instrument changes. A later adjustment (FILNET?) is supposed to “homogenize” the records by statistically detecting probable unrecorded events, and removing their effects. It seems to me that this might create a spurious warming trend if the gradual blackening of shelters or sensors is occasionally and abruptly offset by unrecorded repainting and/or cleaning. In fact, this shift merely corrects for the opposite trend that has built up over time, but the program might erroneously remove the cleaning or painting downshifts while leaving in the intervening blackening updrifts. Periodic cleaning might therefore create even more apparent GW than once-and-for-all blackening without cleaning!

    I would hope that any future station surveys would include similar close-up photos of the MMTS sensor in order to see how pervasive this blackening problem is. Even if the exteriors of the sensors were cleaned, would inaccessible soil buildup in the interior not also cause warming? Even if this soil is not in direct sunlight, it would still be receiving indirect reflected radiation off the vanes.

    Sensor blackening is admittedly a little off topic for this thread, but since this is an off-topic thread to start with, perhaps this digression might be excused. I tried to raise this issue last month on the “Second Look” thread (#2069), at #101, but no one took notice, as other events were popping at that time.

    The links to the Circleville Crud that didn’t come through in the above blockquote are here and here.

    Thanks!

  2. Anthony Watts
    Posted Oct 20, 2007 at 9:31 AM | Permalink

    Hu

    In the photos gathered thus far, sensor blackening appears to know no bounds. I’ve seen it in the Pacific NW, midwest and NE. Sometimes its mold, othertimes dust, or soot.

    Its gradual, and undetectable from a data analysis standpoint.

    There’s also another similar problem – roof blackening. Stevenson screens collect all sorts of crud on the roofs, and the roofs often get overlooked when repainting comes along. They then grey due to sun exposure. Again, gradual with no obvious signs in the data analysis.

    BTW I have two months now of Stevenson Screen data in my paint experiment, covering hot and cool weather patterns. I’ll release the data soon.

  3. Larry
    Posted Oct 20, 2007 at 9:43 AM | Permalink

    This is a similar issue to albedo change on ice/snow from particulates. The question is, how do we quantify it? You need some optical measurement for what is effectively the albedo of the surface of the enclosure. Then you need to quantify the effect of this on temperature. I’m sure this has been developed in a different context, but I have no idea where.

    Keep in mind also, that the Stevenson screen is irrelevant in a lot of modern installations that have a remote sensor. However, the albedo of the remote sensor housing may be an issue. If the remote sensor is fan aspirated, this may also be a negligible effect. We just can’t tell without the numbers. Maybe the aspirating sensor manufacturers have dealt with this issue?

  4. paul graham
    Posted Oct 20, 2007 at 10:53 AM | Permalink

    It’s an interesting point; and surly would affect urban sites more that rural. Are their guidelines on cleaning weather station, which should happem more often that repairing?

  5. Robert Marshall
    Posted Oct 20, 2007 at 11:22 AM | Permalink

    Perhaps Hu inferred this or perhaps I deduced it by mis-reading, but wouldn’t a cleaning routine also become yet another “quality of observation ” trap? A better solution might be to identify a neutral gray that represents long term deterioration of the whitewash or white paint and paint both sensor and shelter that tone. A one time correction would then re-benchmark future data. The same type of study that Watts was conducting initially would establish a correction factor.

  6. Posted Oct 20, 2007 at 1:16 PM | Permalink

    Paul Graham writes, #4,

    It’s an interesting point; and surly would affect urban sites more that rural.

    Actually, Circleville OH is quite rural. It’s upwind (W) from the town, and there is nothing further west but the Scioto River and then farmland. The blackening on it reminds me a lot of the black crud I got on my tomatoes here in a relatively clean suburb of Columbus this summer — I think it’s probably mildew, interacting with treesap, aphid goo, and/or pollen.

    Norwalk OH has similar black spots, but not quite as dark, even though that stuff in the background is exactly what it looks like! Click here for Norwalk.

    Bucyrus OH is relatively clean:
    Click here for Bucyrus,
    while Oberlin OH is yellowed with some black spots:
    Click here for Oberlin.

    The above are the only sites in Ohio with enough detail to see the state of the MMTS.

    Surely NOAA would have done a study of the effect of this on temperature records??

  7. steven mosher
    Posted Oct 20, 2007 at 1:40 PM | Permalink

    Blackened MMTS. The solution is very simple. We DO NOT need to worry about the
    past. You will never detect or adjust for the slow accumulation of crap on the MMTS
    Sheild. NEVER.

    Clean the sheilds. If it has had an impact over the years, then watch the temps drop.

    Clean the beehive. HOW?

    1. Long play: convince NOAA to make it a part of the calibration protocal.
    They calibrate. They clean.

    2. Short play. Call the operators. Suggest they clean the beehive with a simple blast of air or
    brake cleaner which removes warts from toads.

  8. Klaus Brakebusch
    Posted Oct 20, 2007 at 2:21 PM | Permalink

    … it’s the usual business …
    … when compares GISS dset0 vs. MeteoSwiss (Station Basel OR Basle (engl.))
    GISS still has not recognized, they have wrong datas for a whole year (1903),
    they have the 1904 data instead of 1903…

    data quality?

    meteoswiss data for 12 stations, mostly since 1864 is here:

    http://www.meteoswiss.admin.ch/web/en/climate/climate_since_1864/homogeneous_data.html

  9. Posted Oct 20, 2007 at 3:09 PM | Permalink

    Steve Mosher (#7) writes,

    Clean the beehive. HOW?

    1. Long play: convince NOAA to make it a part of the calibration protocal.
    They calibrate. They clean.

    2. Short play. Call the operators. Suggest they clean the beehive with a simple blast of air or
    brake cleaner which removes warts from toads.

    I think it’s best for us not to try to get them cleaned ourselves, since as I noted above this will be an undocumented event that will be detected by the computer and cause all later temperatures to be shifted up by the discontinuity. It’s better to let NOAA do it, on a documented schedule. The discontinuity then can be taken down out of the preceding data on a pro-rated basis since installation or the last cleaning.

    Meanwhile, there must be a study somewhere of this aspect of the MMTS that was done before it was introduced.

  10. steven mosher
    Posted Oct 20, 2007 at 3:46 PM | Permalink

    RE 9.

    there must be a study? Huh? Do you actually think they anticipated this? Look at the
    Original study of the MMTS versus the MinMax. You will see no rigor in doing side by side
    tests or calibration studies.

    Anthony has one of these beehives. I Would suggest that when he finishes his Paint study that
    he do some kinda crud on the beehive study.

  11. Posted Oct 20, 2007 at 6:13 PM | Permalink

    Robert Marshall: “A better solution might be to identify a neutral gray that represents long term deterioration of the whitewash or white paint and paint both sensor and shelter that tone.”

    Your proposed measure could eliminate the warming trend due to graying of white paint. However, then the temperature sensor would to a much greater extent be a “sunniness” sensor as well as a temperature sensor. (Its readings would be shifted upward much more on sunny days.) Then a trend in sun/cloud cover could counterfeit or hide a real temperature trend. Of course, this is all true to the extent that some of the screens are so dirty that they are already approaching this end-state level of gray.

  12. Anthony Watts
    Posted Oct 20, 2007 at 6:26 PM | Permalink

    “crud on the beehive study” Yeah that’ll make it into Nature

  13. Larry
    Posted Oct 20, 2007 at 6:35 PM | Permalink

    It’s rather pointless to talk about a new and improved Stevenson shelter, because the newer temperature sensors are remote and aspirated. The question here is, is there a reasonable way to figure out what this effect has been historically, and adjust historical data for it. Personally, I don’t think so.

  14. steven mosher
    Posted Oct 20, 2007 at 6:40 PM | Permalink

    tamino and Eli in seem to be in full meltdown mode. And all from a mere suggestion that their
    precautionary principal bears a striking resemblence to pascals wager.

    7 words set their thinning hair on fire: “Pascals wager. I’ll see you in church.”
    And hey run

    Tamino should have known better than to foist that oversimplified version of decision
    theory on people. I don’t know why he avoids the math details on his site. He could be
    an excellelant foil to CA, but he goes halfway. And its clear he’s not numerically challenged.

  15. steven mosher
    Posted Oct 20, 2007 at 6:44 PM | Permalink

    RE 11. The devices are supposed to be calibrated periodically. It’s a simple matter
    to replace the housing after X visits and the entire sensor package after Y visits.

    Ask Jiffy Lube how hard this crap is.

  16. Larry
    Posted Oct 20, 2007 at 6:46 PM | Permalink

    Oddly, some people who aren’t numerically challenged are logically challenged. I’ve seen it happen many times. But this belongs on unthreaded.

  17. Larry
    Posted Oct 20, 2007 at 6:47 PM | Permalink

    15, How many of these devices do you really think are periodically calibrated?

  18. Anthony Watts
    Posted Oct 20, 2007 at 7:06 PM | Permalink

    15, 17 There is NO field calibration for MMTS. NWS doesn’t have enough COOP personnel to go around to do maintenance. The only way to prove the issue is to:

    1) Go cleana bunch ourselves, or get observers to do it

    2) Setup some “crud on the behive” study separately

  19. steven mosher
    Posted Oct 20, 2007 at 7:13 PM | Permalink

    RE 18. Is the Ir shield plastic? does it get brittle over time or just crust up with junk?

    When I get back from China ( leaving in a couple hours) I’ll have a look at MMTS versus Other.

    Clayton could do this as well.

  20. Posted Oct 20, 2007 at 9:50 PM | Permalink

    RE #8, Klaus —
    I believe the Europeans now use a “beehive” sensor similar to our MMTS. Are any measures taken to clean them? Or is there any study of what happens to temperatures when they are not cleaned? I would think the Swiss would be particularly attuned to such details.

  21. Posted Oct 20, 2007 at 10:28 PM | Permalink

    Larry (#3) writes,

    Keep in mind also, that the Stevenson screen is irrelevant in a lot of modern installations that have a remote sensor.

    Even if the Stevenson screen has been replaced by an MMTS, the deterioration of its condition while it was still in use will be part of the historical trend. Furthermore, the adoption of the MMTS varies by state. In Ohio, most stations seem to be MMTS, while in California, most are still Stevenson, according to the Surfacestations.org gallery.

  22. Posted Oct 20, 2007 at 10:44 PM | Permalink

    On looking through the California Surfacestations gallery for closeups of MMTS sensors, I could find only 3 good closeups, in part because most stations still have Stephenson Screens. These might give some idea of how the MMTS does in a more “western” climate than the Ohio stations noted above.

    Chula Vista CA is not seen in great detail, but there is clearly a black stain coming down each vane from one of the four vertical bolts. I haven’t noticed this elsewhere.

    Adding insult to the injury of its siting, the Marysville CA MMTS is loaded with what appear to be clay deposits.

    Petulama CA is not just dirty, but is loaded with cobwebs.

    Note that in order to see these closeups in the Surfacestations gallery, you have to click on the little “Size” window and select the highest resolution. Most default to 640 pixels, but are viewable in 2-4 times this resolution.

    To judge from the Circleville OH station, the worst blackening is on the top of the sensor, where water doesn’t run off as freely. Unfortunately, the top doesn’t show up in most of the Surfacestations photos. Volunteers should make a special effort to get close-ups of both the top and the vanes in the future.

  23. aurbo
    Posted Oct 20, 2007 at 11:32 PM | Permalink

    In some 50+ years of being involved with surface observations from the instrumental to the data acquisition aspects, I have observed a collarary to sensor blackening that has impacted relative humidity measurements, some of which were progressively degraded for many month to more than a year.

    Basically, it deals with how dewpoints were derived from LIG thermometers in Stevenson screens. There were two thermometers used for current temperature measurements. One was the minimum temperature thermometer (used to ascertain current air temperature) and the wet-bulb thermometer used to determine humidity. This was a standard LIG thermometer with a cotton wick which had one portion over the bulb of the thermometer and the other in a small container of water. Capillarity of the wick kept the portion over the bulb well moistened. Evaporational cooling of the wick lowered the temperature sensed by the thermometer by an amount that was proportional to the percent of saturation of the ambient airmass. The observed wet-bulb temperature taken together with the dry bulb reading was used to determine the dew-point and relative humidity. Although empirically derived, there are thermodynamically formulated equations which define these relationships which are actually fairly complex.

    All that aside, there was a tendency over time for the wick to become contaminated by several elements ranging from deposited dust or smoke to absorbed salt products. All of these tended to raise the wet-bulb temperature (erroneously) which had the effect of raising the DP and RH. Many NWS stations we used in our daily analyses were obviously corrupted by these deposits and on numerous occasions we had to call the observer at the site to tell him his wick needed to be cleaned. (No scatological reference intended). I remember one consistently bad salty wick problem emanated from Salisbury MD. Most of the worst offenders were understandably those stations proximal to ocean areas or other salt water locations.

    The point here is that the slow, but steady degradation of observational instruments would occur over short to moderate time periods and if not corrected would create a monotonic trend in the observed elements that would last until some remedial action was taken.

  24. Larry
    Posted Oct 21, 2007 at 9:03 AM | Permalink

    21, most certainly MMTS have maintenance issues. They just wouldn’t suffer from this particular problem.

  25. Anthony Watts
    Posted Oct 21, 2007 at 1:53 PM | Permalink

    RE23, yes I’ve seen the wick issue as well. I’ve seen some setups where the wick has become basically brittle due to salt and dust absorption.

    The NWS is woefully bad at anything related to calibration or maintenance of the COOP network, it’s been pretty much left to the observers at this point.

    “High quality network”, indeed.

  26. Murray Duffin
    Posted Oct 21, 2007 at 2:47 PM | Permalink

    The sites I SURVEYED IN florida and south Georgia had more or less degree of mildew (very common in the southeast) on the Stevenson screens, and some but less on the MMTS. Screens near trees seem to be worse. Murray

  27. Posted Oct 21, 2007 at 4:20 PM | Permalink

    I surveyed one MMTS in Jennins, LA which had apparently been “egged” ( photo ) .

    In case “egging” is unknown elsewhere, it’s an act of throwing chicken eggs at some object, usually by teenagers and usually thrown at something of which they disapprove. I think this MMTS is simply an easy target, being close to a country road.

    Dried egg residue does not weather away very well. This stain could be around for a while, unless the curator cleans it.

  28. Clayton B.
    Posted Oct 21, 2007 at 7:11 PM | Permalink

    Per the request in #19

  29. Anthony Watts
    Posted Oct 22, 2007 at 8:39 AM | Permalink

    RE28 Clayton, thank you. FYI the graph has no meaning prior to 1985, since MMTS didn’t get deployed before them. Perhaps a reduced tie scale would be better?

  30. Posted Oct 22, 2007 at 8:58 AM | Permalink

    David Smith (#27) writes,

    I surveyed one MMTS in Jennins, LA which had apparently been “egged” ( photo ) .

    In case “egging” is unknown elsewhere, it’s an act of throwing chicken eggs at some object, usually by teenagers and usually thrown at something of which they disapprove. I think this MMTS is simply an easy target, being close to a country road.

    Dried egg residue does not weather away very well. This stain could be around for a while, unless the curator cleans it.

    Score one bullseye for Global Warming!

  31. Posted Oct 22, 2007 at 9:08 AM | Permalink

    Any visible darkening of the sensor would tend to elevate daytime high temperatures because of increased absorption of visible radiation.

    But what would be the effect on nighttime lows? I guess this depends on the emissivity of the dirt versus the white plastic in the long-wave IR spectrum, and might differ with mildew, soot, clay, rust, or egg yolk (per #27!).

    Does anyone have an informed guess? Or even actual knowledge?

  32. Neil Fisher
    Posted Oct 22, 2007 at 4:44 PM | Permalink

    Re: #13 & #24:

    Larry, you seem to believe that an aspirated sensor won’t suffer from contamination problems (or perhaps not as much), but I disagree. I would have thought that dust etc could easily collect on the sensor itself – open your PC and see how much dust collects on the CPU heatsink! IMO, this would be worse than a contaminated screen because you are essentially covering the sensor itself in a thick coating of unknown thermal properties, which could easily reduce max reading and/or increase min reading. If MMTS is aspirated AND remote reading (which AFAIK, it is), then the problem (if any) could go unnoticed for a LONG time!

  33. Posted Oct 22, 2007 at 6:12 PM | Permalink

    Anthony Watts writes, #12,

    “crud on the beehive study” Yeah that’ll make it into Nature

    A fitting sequel to your forthcoming “Bats in the Birdhouse”!

  34. Posted Oct 22, 2007 at 6:40 PM | Permalink

    31 Don’t forget the yellowing of the plastic that comes with UV exposure. That is one problem that has been noted by the AMS.

  35. steven mosher
    Posted Oct 22, 2007 at 7:23 PM | Permalink

    Talking about sensor blackening made me think about the whitewash issue again.

    That led me to compare MMTS against Minmax. Why, well before a station was switched
    to MMTS it was likley a MinMAX in a CRS ( stevenson screen)

    So you have a history of CRS stations. At some point they start switching to MMTS
    those that remain as MinMAX in a CRS will have the painting Issue. That is, after they
    switch paint on a CRS away from whitewash, we can expect a warming relative to MMTS
    All else being equal. Thus in the early station historys MinMax and MMTS should look
    the same, because MMTS is actually MinMax ( the current variable only has the end
    state of the site)

    I have to double check some dates with Anthony, But MinMAX and “MMTS” show no difference
    until the late 70s. then MinMAx warms relative to MMTS by .2C. change in paint may
    be more important than blackening sensor sheilds

  36. Posted Oct 23, 2007 at 9:04 PM | Permalink

    Captdallas2 writes (#34),

    31 Don’t forget the yellowing of the plastic that comes with UV exposure. That is one problem that has been noted by the AMS.

    Do you have a reference to an AMS publication on this?

Follow

Get every new post delivered to your Inbox.

Join 3,251 other followers

%d bloggers like this: