SGP/SONDE/C1 - Mixed use of RS92-SGP and RS92-KL radiosondes

Because of a manufacturer's production problem, we will be using both RS92-SGP (GPS 
windfinding and RS92-KL (Loran windfinding) radiosondes at the SGP/CF starting February 10, 
2006.  The radiosonde type used for the each sounding may be determined by looking the 
metadata variable "launch_status" in the netCDF data files.

• wstat
• lon
• Mean Wind Speed(wspd)
• Wind Direction(deg)
• Northward Wind Component(v_wind)
• lat
• Eastward Wind Component(u_wind)

SGP/SONDE/C1 - Soundings done as part of dual-sonde flights

The following soundings done during the subject time period were flown in dual-sonde mode 
in which the standard ARM RS92 radiosonde was attached to the same balloon with a NASA 
ATR (atmospheric temperature reference) radiosonde.  The radiosondes were separated by a 6' 
styrofoam bar.  To accommodate the additional weight, we used 1200g balloons rather than 
350g balloons.  Accordingly, we want to alert users that these soundings may reach 
higher altitudes than normal for ARM soundings and that the ascent rate may be somewhat 
higher.  The following soundings are affected:

Sequence  Date    UTC launch time
    1     2/22        0530
    2     2/22        1730
    3     2/22        2330 <---- No data: launch failure    
    4     2/23        0230
    5     2/23        0530
    6     2/23        1730    
    7     2/23        2030
    8     2/23        2330
    9     2/24        2330
   10     2/25        0215
   11     2/25        0509    
   12     2/25        1730
   13     2/25        1930
   14     2/25        2300

• Dewpoint Temperature(dp)
• Dry Bulb Temperature(tdry)
• lon
• Mean Wind Speed(wspd)
• base_time
• asc
• Pressure, atmospheric, constant value of model level, RUC model output(pres)
• Wind Direction(deg)
• Northward Wind Component(v_wind)
• time_offset
• lat
• alt
• wstat
• Humidity, relative(rh)
• Eastward Wind Component(u_wind)

SGP/SONDE/C1 - No Wind/Lat/Lon Data in Sonde Files

The wind finding hardware failed so that no wind or latitude/longitude data was collected 
on the launches for these days.

• wstat
• lon
• Mean Wind Speed(wspd)
• Northward Wind Component(v_wind)
• lat
• Eastward Wind Component(u_wind)

SGP/SONDE/C1 - RH values offset due to error in ground check

Background: After installation of the latest software update , we began performing formal 
ground checks on RS92 radiosondes.  By comparing the
radiosonde readings with reference values in a controlled environment, the
ground check is intended to ensure that the radiosonde sensors are in
calibration before launch.  The radisonde values during the ground check
process are recorded automatically; the reference values are entered by the
operator.  If the radiosonde readings vary from the reference values, the
system 'corrects' the radiosonde values.  In the case of RH, the correction
is applied as an offset (true RH = indicated RH + correction).

Problem: Some operators misunderstood the ground-check procedure and entered the 
radiosonde values of RH rather than the reference value (assumed 0%RH), thus, some soundings done 
during this period have RH values that are different from what they would be if the 
ground check procedure had been done correctly.

Consequence:  Because the RH corrections generally are small (averaging
about 0.5% RH) and the precision of the reported RH data during this time
period is 1% RH, some values will not be affected at all; others will be off by about 1% 
RH (still within the stated accuracy of the sensor). 
Furthermore, because entering the radiosonde value as the reference value
results in a near zero 'correction,' one may view these soundings as being
done without the ground check correction.  Also note that the dew point
value will be suspect as well.

I am attaching below a listing of the soundings that were affected during the indicated 
time period.  Users can see the magnitude of the problem and choose whether or not to apply 
an offset in RH.  Note that application of the offset to the processed RH data (in the 
netCDF file) will not have the same effect of applying the correction to the absolutely 
raw data (before processing), but the differences should be insignificant.  Also note that 
the corrections for each sounding are included in the netCDF header information.  These 
attributes are named 'humidity_correction' and 'humidity_correction_2'.  The values of RH 
indicated by the radiosonde while it is in the ground check set are in attributes 
'humidity_gc_sonde' and 'humidity_gc_sonde_2.'  There is an RH reading for each of the two RH 
sensors used in the RS92 radiosonde.

Here is the list - RH1 and RH2 are the radiosonde readings in the ground
check set.  Ref is the reference value entered by the operator.  The
correcte reference value should be 0%.  To obtain an approximation to the RH value that 
would have been recorded if the ground check had been done
properly, subtract the Ref value from the indicated RH.

 YYYYMMDD  HHMM  SerialN      RH1     RH2     Ref
 20060827  2332  B2624322    1.00    1.00    1.00
 20060828  1731  B2634078    1.00    1.00    1.00
 20060828  2332  B2624380    3.00    3.00    2.00
 20060909  2331  B2615266    1.00    1.00    1.00
 20060915  1759  B2914012    1.00    1.00    1.00
 20060919  2350  B2825128    1.00    1.00    1.00
 20060920  1800  B2825088    1.00    1.00    1.00
 20060920  2346  B2815633    1.00    1.00    1.00
 20060921  1728  B2834732    1.30    1.30    1.30
 20060921  2330  B2825042    1.50    1.50    1.50
 20060922  0527  B2825043    1.70    1.80    1.70
 20060922  1128  B2825031    1.60    1.50    1.60
 20060922  1729  B2834437    1.50    1.50    1.50
 20060922  2330  B2825040    0.40    0.50    0.50
 20060923  0529  B2834733    0.30    0.30    0.30
 20060923  1128  B2825026    0.40    0.40    0.40
 20060923  1729  B2825064    0.60    0.70    0.60
 20060923  2330  B2825041    0.40    0.50    0.50
 20060924  0529  B2825039    0.30    0.30    0.30
 20060924  1729  B2825025    0.20    0.30    0.20
 20060924  2329  B2825032    0.30    0.40    0.40
 20060925  0529  B2824234    0.20    0.30    0.20

• Dewpoint Temperature(dp)
• Humidity, relative(rh)

SGP/SONDE/C1 - Increased precision for alt, dp, pres, rh, tdry

The output precision of the following variables was increased 20060928

alt  changed from 1m to 0.1m
dp   changed from 0.1degC to 0.01degC
pres changed from 0.1 hPa to 0.01 hPa
RH   changed from 1%RH to 0.01%RH
tdry changed from .1degC to 0.01degC

Note that the increase in precision does not necessarily imply an increase in accuracy.  
Users are cautioned to continue to use their best scientific judgement when using these or 
any other data.

• alt
• Dry Bulb Temperature(tdry)
• Dewpoint Temperature(dp)
• Humidity, relative(rh)
• Pressure, atmospheric, constant value of model level, RUC model output(pres)

SGP/SONDE/C1 - RH Sensor failure

The differences between the two RH sensors on this sonde are unreasonably high (~20%).  As 
a result the reported RH values oscillate by this amount during the times the sensors 
are alternately heated.  After the heating stops, the reported RH comes from the sensor 
reading the higher value.  There is no way of knowing if this is the correct value.

• Dewpoint Temperature(dp)
• Humidity, relative(rh)

SGP/SONDE/C1 - Suspect RH profile from 150-350 m above ground level

Several radiosondes were noticed to have a very stark dry layer between 150 m and 350 m, 
where the RH dropped from approximately 60-80% (which I would take as close to the ambient 
conditions) to less than 2%.  The character of these dropouts look like instrument 
issues, as it only a handful of radiosondes that appear to be affected.  The quicklook images 
from the Raman lidar does not show any very dry layers, thus supporting the hypothesis 
that the RH measurements from these radiosondes (at least in this small ~200 m level) are 
in error.  I should note that the temperature profiles look normal.

• Dewpoint Temperature(dp)
• Humidity, relative(rh)