I was out walking the dogs during a spectacular afternoon here in Maryland, and thinking about how incredibly corrupted the US temperature record is.
In the USHCN V1 docs, they show that TOBS (Time of Observation Bias) adjustments go flat after 1990.

ts.ushcn_anom25_diffs_pg.gif (650×502)
The USHCN V2 docs say that they use the same TOBS algorithm as V1, but their TOBS adjustment goes up exponentially after 1990. The same algorithm now produces warming, where it used to produce none. Junk science at its worst.

This net effect is about the same as that of the TOB adjustments in the HCN version 1 temperature data (Hansen et al. 2001), which is to be expected since the same TOB-adjustment method is used in both versions.
ftp://ftp.ncdc.noaa.gov/pub/data/ushcn/v2/monthly/menne-etal2009.pdf
But it gets worse. Now we find out that they have actually lost 30% of their station data since 1990.

How do you calculate TOBS on an observation which doesn’t exist?
Note in the graph above that 1990 was the year when they started rapidly losing station data. It was shortly after that when their divergence from measured data started increasing exponentially.


What this indicates is that their FILNET algorithm (used to generate non-existent data) is introducing a huge fake warming bias into the temperature set. It allows them to multiply bogus TOBS adjustments, by propagating them into into fake station data.