If you have ever used a min/max thermometer, you are aware of the time of observation bias problem. It is not a subtle problem.
Suppose you reset your thermometer at 3 pm, and the temperature is 60F. At 4 pm a cold front comes through and drops the temperature to -10F. The next day you bundle up in your Arctic expedition gear and go to read the maximum temperature. It reads 60F. You realize that the number is ridiculous, because it hasn’t been above 0F all day.
The solution is obvious, you reset the thermometer at night before you go to bed. I figured out the solution to the TOBS problem when I was seven years old, and I am sure just about every other station owner did too. You would have to be a complete idiot to not figure it out.
That being said, the actual TOBS “adjustment” being done by USHCN software doesn’t even vaguely match their documentation. The whole thing is complete BS.
Steve, I asked this in response to your WUWT post on this subject, so I thought I would ask it here.
Do they apply TOBS to every day’s readings regardless of the actual weather?
We do not exerience such vast swings in temp you describe in the UK but we can get about 10 degrees C changes.
Those instructions S Mosher pointed me to suggest 4 off 6 hourly readings, is that realistic?
4 fairly equally spaced readings in 24 hours?
TOBS is supposed to be a statistical adjustment based on what USHCN believes the time of observation was at a particular station. It doesn’t have anything to do with the actual weather conditions,
That is why Evan M. Jones and Anthony Watts are going through the data individually and ‘Correcting’ for Tobs based on the original data and notes for Anthony’s up comming US Surface Station paper.
I sure hope that paper come out soon!
Obamascience:
1. 97% of the people use the thermometers correctly
2. Of the 3% who use it wrong, half get measurements too high and half get them too low.
3. Therefore, we need to lower all readings from the 1930’s and 1940’s, and raise all readings after 1980
That’s how you Obamadjust a thermometer
Using your example, with a consistent TOBS, you’ll have the same anomalies in the opposite direction when a warm front comes charging through.
Thus, in the long run, you may shift some record temperatures into the wrong day, yet your long term averages won’t be horribly off.
And the only station operators who could be such idiots would be government
parasitesemployees who take the reading at 3:30 PM just before going home…. That should be fairly easy to sort out….A max temperature leaves a mark at the high temperature. So when a warm front comes through after the yesterday’s reading, the mark rises to the new higher temperature. When a cold front comes through, the mark stays at yesterday’s high temperature until reset.
A demonstration at this time would be helpful; a time-series of ‘numbers’ representing the ambient air temperature and the value indicated on the min-max thermo until reset. Cases for warm and cold fronts and a sufficiently long enough series to show one effect of the other …
But, the Min temp will be accurate and the average of the 2 readings (the Avg temp that is recorded for all posterity) will roughly balance out the “warm front” scenario where the Max temp will be accurate with the Min temp then being off. Ironically, the actual (absolute) temp will be biased but the Avg temp when averaged over a longer period will probably offset the errors. Not sure if/how it will affect the Avg of the “anomaly”, which I don’t like because a lot of time in the past they only recorded temps to the nearest degree that introduced errors greater than that of the warming trend that you are trying to detect.
Probably the only global temperatures that are reliable are satellite data, and then only when verified by monitored ground truth (not anecdotal).
Shazaam you are absolutely correct, as long a a station is consistent in the time of day of the reading over a long period of time the reading will adjust itself. Thus is two stations close together have different reading times the average temperature read maybe shifted by 24 hours in one station but this shift would be consistent over time no adjustment is needed. As for the cold front/warm front moving in how would that make a difference, should the temperature reading be adjusted or should the reader wait until the front has moved thru and what about a front that stalls just before it gets to the station. I have written it before, the bias will average out over time if the reading station is consistent. The read bias in the temperature reading is the over location in the east and the Stevenson’s box locations and the switch from white wash paint to latex paint.
Just another excuse to tamper with data. Only leftists think they can perfect past weather observations, and weather.
with first order stations, the take a reading every hour, thus this should not be an issue. But in any event, I would think those kinds of occurrences should take place very infrquently and should be obvious at the time.
While we are at it, I don’t like it when they move a station and then consider it part of the same temp record instead of treating them as 2 distinct weather records. Here in Denver there have been demonstrable differences in the records from Downtown, at Stapleton and at DIA after they closed Stapleton, when the overlapping periods were compared to Buckley Air Base in Aurora. DIA further removed from the foothills to the plains show much hotter and drier data now that they have been spliced together to form the Official Denver record.
Jim,
I agree. This TOBS business makes no sense to me. Kind of like “saving daylight”.
I think it has something to do with using the totally unnatural thing called a month instead of physically definable concepts such as day or year. But I have not found anything on the net telling me what the adjustment does with actual data.
The entire rationale for TOBS adjustment is one paper, by Thomas Karl. Despite criticisms of the paper over the years, it’s been the philisophical basis for all such adjustments. Wouldn’t be because the paper recommended adjusting the past lower and the present warmer, would it?
Put another way, would it be the seminal paper on the subject if it recommended either doing nothing, or adjusting in the opposite direction?
Who made the decision to make the TOBS wholesale over that part of the data it was applicable? One man, the same man who wrote the paper? Was their internal or external discussion or review at the time prior to the decision to commit to this methodology?
“The solution is obvious, you reset the thermometer at night before you go to bed.”
NWS Coop observers have to reset at consistent times agreed with NWS. They can’t make undocumented decisions like that. What if some do and some don’t? NWS can’t report reliably if it doesn’t know how the readings are made.
Different observers take readings at different times. Your suggestion that it is uniform is incorrect.