Time of Observation Error (TOBs) in temperature maxima can be reliably measured from real data (rather than estimated from models)
Time of Observation Error (TOBs) in temperature
|Report Type||Working paper|
|Institution of Origin||University of Southern Queensland|
|Number of Pages||23|
|Place of Publication||Toowoomba, Australia|
TOBs is a phenomenon concerning the time of day at which measurements are taken, whereby some maximum or minimum temperatures are not recorded; instead, a faulty, but always high (for maxima) or low (for minima), value is recorded from the ‘detritus’ of a more extreme value the previous day. This paper explains why such a phenomenon should leave a detectable signature in the statistics of maximum and minimum temperature changes from day-to-day. The entire US unadjusted temperature data, over 200 million data points, is divided into yearly baskets and examined for average occurrences of certain day-to-day temperature change patterns whose probability and/or magnitude would be expected to change, if the TOBs hypothesis is true, under changes in measurement time of day at recording stations. Whereas official estimates of TOBs are made by inference from models or from pairwise homogenisation (a process with many severe critics, but beyond the scope of this paper), this paper obtains direct estimates of TOBs error in daily maxima (Tmax) from the real data, along with statistical reliability estimates. This method detects the systematic error that actually exists, rather than one inferred from modelling. We find that the official estimates of the errors due to TOBs are significantly over-estimated. We also assess the use of the same method to find the TOBs error in daily minima (Tmin).
|Keywords||temperature trend, TOBs, time of day, error estimation, US temperature data, time of observation bias|
|ANZSRC Field of Research 2020||370202. Climatology|
Unpublished USQ publication.
|Byline Affiliations||School of Agricultural, Computational and Environmental Sciences|
0views this month
5downloads this month