Smoking Gun That The TOBS Adjustment Is Garbage

Our official US government data tamperers cool the past, and warm the present – based on the idea that observers used to read the thermometers in the afternoon, and now they read them in the morning.

If that were true, we would see a decline in raw minimum temperatures, but we see the opposite.

 

In other words, the official data tamperers are wrong at both ends. Their upwards adjustments are much too large, and their downwards adjustments are much to small.

About Tony Heller

Just having fun
This entry was posted in Uncategorized. Bookmark the permalink.

10 Responses to Smoking Gun That The TOBS Adjustment Is Garbage

  1. Me says:

    Yeah, is that their excuse to lower the past temps, making it look cooler, if so, it’s all GIGO now, or Appell’s and Oranges, 😆

  2. cb says:

    If I may suggest, for those of us who are unfamiliar with the temperature taking procedures: am I to understand that temperatures were-are taken only once a day? What is meant by “minimum temperatures” in this regard?

    Also, surely such a change would result in an error-band-based adjustment, and would have to be rather region (not to mention seasonally) specific? Or is this whole mess just that bad? Where the cumulative error-bands end up exceeding the numeric-values (of variability)?

  3. cb says:

    I realize my question may look stupid, but I don’t see how an afternoon measurement versus a morning measurement makes sense for either a min or max temp measurement. Unless the other measurement was taken at 23:59 in the evening?

  4. Thanks DirkH.
    It is unclear, Steve, or at least not intuitive… I haven’t read all of Karl to know exactly what is thought to be counteracted by the TOBS adjustment, but as you and I have both discussed before, it was quite clear that the desired result all along was to reset the device such that the actual low and high between 12:00am and 11:59pm was recorded. I understand how resetting before the min, after the min, before the max and after the max would introduce an error, and I think we both agree most observers were smarter than that. But even a smart observer could make a mistake. He could reset the min after the temperature started to rise during the day (say, 10:00 am), then only find out later that a massive cold front came through at 4:00 PM and crushed his min. Now, today, that might present a problem since the data has already been transmitted. But in the 1860’s, do we really suppose someone would be dumb enough to not update the min with the new value caused by the cold front? And what about the min after the min was reset, but before midnight, as the temperatures are cooling again for the day?

    If you know the underlying assumptions, it would make a good post. Either they knew to record the actual min or max, or we assume that they really dropped everything they were doing to go reset thermometers at an exact time of day, and disregard what would be recorded in newspapers of the day as “max” and “min”, when they knew this to be untrue. I would think even casual observers, if even remotely trained to record the objective, would already know that a bias would be introduced and would protest that the government statistics were BS, even way back then.

  5. Brian D says:

    I took a local home station from wunderground.com and looked at what readings would look like if they were done at 7a.m. and 5p.m., as this is what TOBS adjustments are based on. All a.m. readings, regardless of actual time, are adjusted based on 7a.m. criteria for the area. All p.m. readings are based on 5p.m. criteria. The monthly avg values for mid, 7a.m, and 5p.m. are as follows: (values in () see below)
    March 2012
    Mid: Max avg= 49.5F Min avg= 31.5F Avg=40.5F
    7a.m.: Max avg= 49.0F(49.5F) Min avg= 29.9F(30.1F) Avg=39.5F(39.8F)
    5p.m.: Max avg= 51.0F Min avg= 31.8F Avg=41.4F

    April 2012
    Mid: Max avg= 53.6F Min avg= 33.5F Avg= 43.6F
    7a.m.: Max avg= 52.9F(53.4F) Min avg= 32.5F(32.5F) Avg= 42.7F(43.0F)
    5p.m.: Max avg= 55.4F Min avg= 34.1F Avg= 44.8F

    If the morning reading would be put down for the previous day, the Highs would line up most of the time with the midnight readings. Lows would be mixed from either the previous day, or for that morning. No adjustments would be needed for the max.

    This is just 2 spring months, I don’t know what it would look like for the other months, and other times, and I may get around to that. But I know from reading this: http://www.isws.illinois.edu/pubdoc/MP/ISWSMP-81.pdf that it varies depending on the time of year, region, weather patterns in a given year, etc. Karl’s algorithm is more general than the specific nature such an adjustment demands (refer to PDF). Close enough for government work I guess, yet its gospel. It’ll average out, they’ll say. But when policy is being predicated on tenths to hundredths of degrees in trends, PROVE IT! Show us the monthly values from the various times(just like I did) from the 500 stations in VOSE et. al. that show the algorithm is truly valid.

    Another thing to consider is that all record highs and lows, number of days above or below a certain temp, etc. are from raw values.

Leave a Reply

Your email address will not be published. Required fields are marked *