On August 1st, 2001 the South Pole Meteorology Department officially switched
to a GPS-based atmospheric sounding system.  The equipment now in use by the 
department consist of Vaisala RS-80 GPS radiosondes and a Vaisala Marwin 
reciever.  This replaces the former system that utlized A.I.R. Inc. model 5A 
radiosondes and an A.I.R. Inc radiotheodolite tracking antenna and receiver. 
The department continues to utilize the software developed by A.I.R. Inc to 
process the raw upper air data, though this is viewed as a temporary measure. 

**IMPORTANT - UPDATED 12/5/01**
The grid wind directions produced by this system are currently incorrect.  Due
to the location of the station at 90 degrees South latitude, wind directions 
are calculated using the prime meridian as north, the dateline as south, etc. 
The Vaisala system described above appears to be displaying a "mirroring" 
effect based off the grid East-West line.  As a result, winds directions along
that line seem to be accurate, but as the flight travels to grid north or 
south of that line, direction is calculated incorrectly, maximizing error at 
180 degrees off the correct direction when the flight tracks due north or 
south.  As soon as the problem is corrected this notice file will be updated. 
Please contact the South Pole Meteorology Department at met@spole.gov with any
questions.      

**IMPORTANT - UPDATED Lat Summer 2001**

> Additional Notes from South Pole Meteorology Office on radiosonde data:

Below is a little background on the grid wind direction errors with our
sounding system.  Most of this was written by Brad Halter, one of the summer
meteorologists whose extensive experience in studying atmospheric sciences
here at the Pole dates back to 1976.

At South Pole a grid system for defining wind direction is used, since east
and west directions would change by up to 180 degrees in moving around the
geographic point in its immediate vicinity if the true wind direction
system were used.  Winds arriving at the Station along the 0 degree, 90
degree East, 180th, and 90 degree West meridians, respectively, are taken to
be arriving from grid N, E, S, and W.

An error in the conversion of GPS winds from the true direction to the grid
system has caused a mirror image reversal in reported wind direction, with
the "mirror" being along the 90W - 90E line.  That is to say that there is
little difference between reported AIR sonde and GPS sonde wind direction
when the winds are arriving at the Station along the 90E or 90W meridians.
However, both comparisons of actual balloon sonde movement with reported GPS
winds and comparisons of winds from simultaneous AIR sonde and GPS sonde
flights indicate that, as the actual wind turns from easterly to become
increasingly northerly and northwesterly, reported GPS sonde winds turn in
the opposite direction, becoming more southerly and south-westerly, as
defined in the grid wind system.  This error in GPS winds caused unrealistic
low level wind shear to be reported on 17 NOV 2001.  At the surface, winds
from grid 002 degrees at 17.3 KT were reported by the GPS calculations to
have become from 197 degrees at 24.1KT within 28 meters of the surface.  The
actual winds, I feel confident, had turned not all the way to 197 degrees in
the space of 28 meters, but had in fact backed slightly from the surface
direction of 002 degrees to 343 degrees, the "mirror image" of the reported
197 degrees.  Note that the surface winds (2835 meters) are placed in the
sonde wind data report by the weather observer and retained in the final wind
report unmodified.  Winds above the surface are calculated from GPS data.

Below is the correction we applied to the GPS radiosonde wind directions to
get values in fair agreement with the comparison AIR radiosonde data obtained
last winter.  The correction is based on the assumption that the
current algorithm used by the MRS to calculate grid wind direction results
in wind directions from the NE and NW quadrants being assigned, respectively,
to the SE and SW quadrants, and vice versa.  The correction then simply puts
the wind directions back into the correct quadrants by calculating the
complement of a 180 degree angle as follows:

        IF UWD < 180, CWD = 180 - UWD

        IF UWD >= 180, CWD = (360 - UWD) + 180 = 540 - UWD

where UWD is the uncorrected wind direction
  and CWD is the corrected wind direction

The above correction applies to all flight files from Aug 1, 2001 0000Z
through and including the Jan 4, 2002 1200Z sounding.  A software upgrade
then resulted in the Jan 5, 2002 0000Z sounding through and including the Jan
8, 2002, 1200Z sounding displaying a constant 180 degree error.  As of the
Jan 9, 2002, 0000Z sounding the grid wind directions reported by our sounding
system are correct.  We are trying to devise a method for correcting the 5+
months of erroneous wind data but it is yet to be determined whether or not
this will be possible.

**UPDATED 10/01/02**
Corrections to the 5-plus months of flawed data files were completed in
August.  The corrected files were archived and sent to the AMRC.  This
project represented a major effort to maintain the quality of data produced
by the department.  It must be noted that completion of this project would
not have been possible without the assistance of Jon Berry, Sr.
Communications Technician.   
Mr. Berry was able to use his own software to write a script that applied
the correction devised by one of the summer meteorologists to all of the
sounding files from August 1, 2001 through January 8, 2002.   Once the raw
data files were re-processed by Mr. Berry and the corrections were confirmed
the meteorologists were able to run the files through the normal post-flight
processing routine.  After an entire month's files were re-generated in this
manner the monthly processing was replicated to produce the monthly upper
air files.