These data
should replace any data for Hilo that may currently be in the
archive including any data received through the NDSC. Also
these data should not be forwarded to the NDSC. We take
care of that directly. The data currently at the NDSC was
submitted about a month ago and is nearly identical with this
data set but only begins in 1991 which is the period which
we cover for them.

I have attached a file with ozonesonde data for Hilo, Hawaii. These
data are for the period Sep. 1982 - Jun. 2002. These should replace
any data you have for Hilo. I will send similar files for South Pole
and Boulder in the near future.

The attached file contains the entire Hilo ozonesonde record
through June 2002.  Some further work has been
done to try and homogenize the record for various changes in
procedure over the years. The most notable is the change to the
2% unbuffered cathode sensing solution in April 1998. An
altitude dependent correction based on exposure to ozone was
applied to the data prior to this (the soundings that used 1%
buffered solution).

The data are normalized to the Dobson total ozone and use the
SBUV climatology for the residual above the balloon burst level.
Using the SBUV climatology gives a bias to this normalization
correction of about 3%. In other words the average correction
factor is 1.03. I believe this is because the climatology probably
has ozone amounts that are somewhat too low. If I use the
constant mixing ratio above burst to calculate the residual the
average correction factor is almost exactly 1.00. The primary
advantage of the SBUV correction is that it reduces the variance
of the correction factors significantly. For soundings with no Dobson
value or where they did not reach an altitude sufficient to do
the normalization an average correction factor of 1.03 was
applied. This will show up in the header as a correction factor
but there will not be a Dobson amount given.

The data file is in the earlier WOUDC comma separated format but
with added columns for the altitude (the next to last column) and the
humidity (the last column). I have also attached a file
(in a second message) that is in
our own internal format. This file contains some of the other
information that is now included in the extended CSV format
such as the background, flowrate, and pump temperature. The
only difference in the data is that the internal format does not
include data at the mandatory levels while the CSV format does
contain these values. Sometime we will get around to writing
a routine to submit the data in the extended CSV format. I
decided you would rather get the data than have me wait until
I got that done before submitting the data.

Some of the other data sets that you have gotten from
SHADOZ or the Langley GTE data bases will also undergo
the standardization process that has been applied to the Hilo
data, but we are still a ways from completing that. this process.
This is mostly important for folks doing trend analyses and most
of these sets are not long enough that people are looking at them
in this way yet. There are some important inhomogenities in these
sets, however, that need to be dealt with.