Additional notes on QC procedure:
The Chinese Meteorological Office released quality controlled
GTS resolution data at several SCSMEX sites.
Their QC flags ranged from 1 to 9 for each parameter where 1 was good and 9 was
absolutely not believable. We used this information in the following way. If
their QC flag was 1, we set ours to 1 (good), if theirs was from 2-4,
we set ours to 2 (questionable), if theirs was > 4, we set ours to 4 (bad).
On several occasions their QC indicated that the data were bad (e.g.,
high rh), but the data visually looked good. In these cases, during the
visual check we manually set the QC flag to 2.
An altitude value with a flag of 7 (or estimated) means that
that value was determined from a mean p(z) relationship
for that site.
Heights were computed by integrating the hydrostatic equation
(based on virtual temperature) starting from the lowest point
(typically the surface). Bad temperature
and humidity values were interpolated over before using these fields in
the integration procedure. If a questionable temperature data was used at
a given level, then the QC flag on z was set to (2 - that is,
questionable) at that level and all levels above it.