The Fabry-Perot Spectrometer at Arrival heights, Antarctica (77.8296 S, 166.6627 E, 190.3 m above sea level)
has been operated since 2002 by Gonzalo Hernandez of the University of Washington.  Magnetic apex coordinates
are about (-80, -34), so the location is in the polar cap, usually poleward of the auroral zone.
These are observations from the red line (630nm) of atomic oxygen, with a typical emission height peak 
between 210-300 km.  The cloud cover (cc) is usually in octas of the sky covered (0=clear, 8 overcast,
with 9=obscured by fog or snow).  Even overcast, the temperature is valid, but all winds were set to
missing (-32767 or -3276.7 for Wn) for cc>4.  Un is positive eastwards, Vn is positive northwards 
(or equatorwards), and Wn is positive upwards.  
  There is a cycle of 9 observation look directions (dir) that takes as long as 15*9=135 min, but is often
much shorter, especially for the winter season in June and July where there is total darkness and thus
24-h observations.  
  Doppler shifts (winds) are determined from the displacement of the line profile relative to the long-term
zenith observations, which are considered to have no long-term vertical Doppler shift.  (Long-term is
defined as months -observing season- of continuous observations.)  The vertical winds here are the deviation
from the long-term zero vertical winds.
  The glat and glon are estimated from the normal 20 degree elevation angle of the observations with a
spherical Earth assumption.  For an emission height of 250 km, and an elevation angle of 20 degrees, 
the distance between the station and any look direction is 587 km compared to a flat-Earth projection of 687 km.  The difference in degrees from an observation towards the south (geographic pole) and towards the north
(equatorwards) is 10.57 deg (1174 km) for a spherical Earth verus a flat Earth approximation of 12.37 deg of
latitude, where 1174km/10.57deg = 111.07km per deg.  A flat-Earth projection estimate was made to convert 
cos(45)(=0.7071)*587km = 415.07km /111.07 km/d = 3.74 deg lat.  The circumference of a circle at glat is
2piResin(90-glat) or 2piRecos(glat).  For glat=77.8296S,+3.74=81.57,+5.23=83.06,-3.74=74.09,-5.23=72.60,
cos(glat) = 0.2198,0.1466,0.1208,0.2741,0.2990, so 2piRe(=6376,40,061.6)/360deg(=111.07km/deg)*cos(glat) 
= 24.41km/dlon, 16.28km/dlon, 13.42km/dlon, 30.44km/dlon, 33.21km/dlon so 415.07 at 81.57S and 74.09S is
25.50dlon and 21.53dlon while 587 km at 77.83S is 24.05dlon.  The 9 locations become glat/glon pairs of:
N  72.60S,166.67E
NW 74.09S,145.14E
W  77.83S,142.62E
SW 81.57S,141.17E
S  83.06S,166.67E
Z  77.83S,166.67E
SE 81.57S,192.72E
E  77.83S,190.72E
NE 74.09S,188.20E
  The measurement uncertainties do not typically exceed 30 K and 10 m/s respectively for temperature and
winds.  The 'errors' given in the data are uncertainties of measurement, that is the statistically 
determined effect that noise in the measurement will cause in the final result.  This noise is inherent 
to the signal, since photons obey Bose-Einstein statistics.  These uncertainties are 1 sigma uncertainty 
of the deduced horizontal winds and temperatures.

 year dayn moda   uth  Tn (K)  err Tn  Un m/s  err Un  Vn m/s  err Vn  Wn m/s  err Wn cc    glat    glon dir