In June 2003, as a result of a user query about 6-hourly forecast data, (itself a rather confusing area), I ran some tests on the air.2m, tmin.2m and tmax.2m surface flux parameters.
I was somewhat bemused to find that, in many cases, the air.2m value (for a particular cell and timestep) fell outside the range implied by the relevant tmin.2m and tmax.2m values.
Let's recap what these parameters are supposed to represent:
|Parameter||Definition for Time = 00.00|
|air.2m||6-hour forecast valid at 06:00 (instantaneous)|
|tmin.2m||minimum temperature over 00:00 to 06:00|
|tmax.2m||maximum temperature over 00:00 to 06:00|
So it looks as though air.2m values should never be less than tmin.2m or greater than tmax.2m values - as you'd intuitively expect.
How severe is this (apparent) problem? Well here are some examples:
For the year 1962, 3000615 values are 'out of bounds' and 3797145 are OK.
The air.2m value falls below the tmin.2m value on 1580889 occasions,
and above the tmax.2m value on 1419726 occasions.
The worst excession is 4.3 degrees (above tmax.2m).
Numbers are similar for all years, and available on request. The issue persists into the daily data, albeit much less frequently.
Whilst of course harbouring an insecurity that I've just misunderstood something fundamental, and that everything's alright really, on balance I think it's better to let users know. Please feel free to examine the data yourself.
NOAA are aware of the issue, and have marked it for investigation.
Ian Harris, 10th December 2003