SGP/MWR/B4/C1 - LOS cycle skipping

When MWR software version 4.12 was installed at the SGP, it was observed 
that the MWRs at CF and BF4 skip line-of-sight (LOS) observing cycles. In LOS mode, the 
software begins an observing cycle at 0, 20, and 40 seconds after the minute to provide 3 
LOS cycles per minute. If a cycle is delayed so that it takes more than 20 seconds to 
complete, then the next start time is missed, the cycle is skipped, and the data that would 
have been acquired are lost.  

It was demonstrated that the interaction with the IRT at CF slowed the MWR observing cycle 
noticeably and contributed significantly to the LOS cycle skipping. The IRT was removed 
from the MWR on 5 November 2002 and the LOS cycle skipping at the CF was resolved.  The 
IRT data are now available in a new separate datastream: sgpirtC1.a1 (and soon 
sgpirt2sC1.a1).

The BF4 cycle skipping may have resulted from a combination of an additional air 
temperature sensor on this instrument and the use of a fiber optic cable.  However, the cycle 
skipping on this instrument appears to have abated without modifications to the instrument 
configuration.

• bbn23
• bb23
• tkxc
• wet_window
• tnd_nom23
• sky23
• tnd_nom31
• Mean 31.4 GHz sky brightness temperature(tbsky31)
• tnd23
• tc31
• 23.8 GHz sky brightness temperature(tbsky23)
• Sky infrared temperature(sky_ir_temp)
• Water, liquid, total along line-of-sight path(liq)
• tkbb
• tnd31
• sky31
• tc23
• tknd
• time_offset
• bb31
• bbn31
• Total water vapor along LOS path(vap)
• tkair

SGP/MWR/C1 - Intermittent Negative Sky Brightness Temperatures

Several related and recurring problems with the SGP MWRs have been
reported dating back to 1999.  These problems were due to the
occurrence of blackbody signals (in counts) that were half of those
expected. The symptoms included noisy data (especially at Purcell),
spikes in the data (especially at Vici), negative brightness
temperatures, and apparent loss of serial communication between the
computer and the radiometer, which results in a self-termination of the
MWR program (especially at the CF).

Because these all initially appeared to be hardware-related problems,
the instrument mentor and SGP site operations personnel (1) repeatedly
cleaned and replaced the fiber optic comm. components, (2) swapped
radiometers, (3) sent radiometers back to Radiometrics for evaluation
(which has not revealed any instrument problems), and (4) reconfigured
the computer's operating system.  Despite several attempts to isolate
and correct it, the problem persisted.

It became apparent that some component of the Windows98 configuration
conflicted with the DOS-based MWR program or affected the serial port
or the contents of the serial port buffer. This problem was finally
corrected by upgrading the MWR software with a new Windows-compatible
program.

• Mean 31.4 GHz sky brightness temperature(tbsky31)
• 23.8 GHz sky brightness temperature(tbsky23)
• Total water vapor along LOS path(vap)
• Water, liquid, total along line-of-sight path(liq)

SGP/MWR/C1 - Incorrect min and max values

The values of valid_min and valid_max applied to fields tkxc and tknd were incorrect. They 
should be 303 and 333, respectively.

• tknd
• tkxc

SGP/MWR/C1 - no air temperature signal

When the new blower was upgraded by Radiometrics and reinstalled on the MWR, the air 
temperature sensor failed to properly report. It was determined that the wires carrying the 
signal to the analog board did not conform to the standard expected by the upgraded blower. 
The problem was corrected by changing the wiring.

• tkair

SGP/MWR/C1 - Wet window flag "on" more frequently than expected

The wet window flag was set "on" more frequently than expected during the time periods 
specified.  This indicates the heater has been running more than necessary.  In most 
instances the moisture sensitivity was adjusted at the end of these periods.

• wet_window
• Mean 31.4 GHz sky brightness temperature(tbsky31)
• 23.8 GHz sky brightness temperature(tbsky23)
• Total water vapor along LOS path(vap)
• Water, liquid, total along line-of-sight path(liq)

SGP/MWR/C1 - Instrument problem

The MWR mixer temperature, blackbody temperature, and moisture flag are incorrect. This 
began when the instrument was returned to service after the analog board was temporarily 
removed to check the presence and absence of certain resistors. The board must have been 
accidently damaged during this process.

• tkbb
• tnd31
• tc23
• tknd
• tkxc
• wet_window
• tnd_nom23
• tnd_nom31
• tnd23
• tc31
• tkair

SGP/MWR/C1 - thermal instability

The analog board was replaced with a spare (D041001.3) with a reference temperature that 
was set too low (306 K) so that during periods of high ambient temperature, the instrument 
became thermally unstable. The problem was corrected when the temperature setting was 
increased (to 311 K).

• Mean 31.4 GHz sky brightness temperature(tbsky31)
• 23.8 GHz sky brightness temperature(tbsky23)
• Total water vapor along LOS path(vap)
• Water, liquid, total along line-of-sight path(liq)

SGP/MWR/C1 - Reprocess: wrong retrievals

When the computer and core configuration were upgraded, retrieval coefficients for BF1 
were accidently included in the configuration file.
The correct coefficients for CF1 were applied when the configuration file was updated.

• Total water vapor along LOS path(vap)
• Water, liquid, total along line-of-sight path(liq)

SGP/MWR/C1 - spare instrument

The spare MWR, S.N. 04, was installed while the production instrument, S.N. 10, was 
returned to the manufacturer for upgrades.

• bbn23
• tkxc
• wet_window
• East longitude(lon)
• tnd_nom31
• tc31
• 23.8 GHz sky brightness temperature(tbsky23)
• Water, liquid, total along line-of-sight path(liq)
• tkbb
• North latitude(lat)
• sky31
• tc23
• time_offset
• bb31
• Total water vapor along LOS path(vap)
• bbn31
• tkair
• bb23
• tnd_nom23
• sky23
• Mean 31.4 GHz sky brightness temperature(tbsky31)
• tnd23
• base_time
• Sky infrared temperature(sky_ir_temp)
• tnd31
• tknd
• Altitude above mean sea level(alt)

SGP/MWR/C1 - REPROCESS - Revised Retrieval Coefficients

IN THE BEGINNING (June 1992), the retrieval coefficients used to derive the precipitable 
water vapor (PWV) and liquid water path (LWP) from the MWR brightness temperatures were 
based on the Liebe and Layton (1987) water vapor and oxygen absorption model and the Grant 
(1957) liquid water absorption model.  

Following the SHEBA experience, revised retrievals based on the more recent Rosenkranz 
(1998) water vapor and oxygen absorption models and the Liebe (1991) liquid waer absorption 
model were developed.  The Rosenkranz water vapor absorption model resulted a 2 percent 
increase in PWV relative to the earlier Liebe and Layton model.  The Liebe liquid water 
absorption model decreased the LWP by 10% relative to the Grant model.  However, the 
increased oxygen absorption caused a 0.02-0.03 mm (20-30 g/m2) reduction in LWP, which was 
particularly significant for low LWP conditions (i.e. thin clouds encountered at SHEBA).

Recently, it has been shown (Liljegren, Boukabara, Cady-Pereira, and Clough, TGARS v. 43, 
pp 1102-1108, 2005) that the half-width of the 22 GHz water vapor line from the HITRAN 
compilation, which is 5 percent smaller than the Liebe and Dillon (1969) half-width used in 
Rosenkranz (1998), provided a better fit to the microwave brightness temperature 
measurements at 5 frequencies in the range 22-30 GHz, and yielded more accurate retrievals.  
Accordingly, revised MWR retrieval coefficients have been developed using MONORTM, which 
utilizes the HITRAN compilation for its spectroscopic parameters.  These new retrievals 
provide 3 percent less PWV and 2.6 percent greater LWP than the previous retrievals based on 
Rosenkranz (1998).

Although the MWR data will be reprocessed to apply the new monortm-based retrievals, for 
most purposes it will be sufficient to correct the data using the following factors:

PWV_MONORTM = 0.9695 * PWV_ROSENKRANZ
LWP_MONORTM = 1.026  * LWP_ROSENKRANZ

The Rosenkranz-based retrieval coefficients became active as follows (BCR 456):
SGP/C1 (Lamont)     4/16/2002, 2000
SGP/B1 (Hillsboro)  4/12/2002, 1600
SGP/B4 (Vici)       4/15/2002, 2300
SGP/B5 (Morris)     4/15/2002, 2300
SGP/B6 (Purcell)    4/16/2002, 2200
SGP/E14(Lamont)     4/16/2002, 0000
NSA/C1 (Barrow)     4/25/2002, 1900 
NSA/C2 (Atqasuk)    4/18/2002, 1700
TWP/C1 (Manus)      5/04/2002, 0200
TWP/C2 (Nauru)      4/27/2002, 0600
TWP/C3 (Darwin)     inception

The MONORTM-based retrieval coefficients became active as follows (BCR 984):

SGP/C1 (Lamont)     6/28/2005, 2300
SGP/B1 (Hillsboro)  6/24/2005, 2100
SGP/B4 (Vici)       6/24/2005, 2100
SGP/B5 (Morris)     6/24/2005, 2100
SGP/B6 (Purcell)    6/24/2005, 1942
SGP/E14(Lamont)     6/28/2005, 2300
NSA/C1 (Barrow)     6/29/2005, 0000 
NSA/C2 (Atqasuk)    6/29/2005, 0000
TWP/C1 (Manus)      6/30/2005, 2100
TWP/C2 (Nauru)      6/30/2005, 2100
TWP/C3 (Darwin)     6/30/2005, 2100
PYE/M1 (Pt. Reyes)  4/08/2005, 1900**

** At Pt. Reyes, the original retrieval coefficients implemented in March 2005 were based 
on a version of the Rosenkranz model that had been modified to use the HITRAN half-width 
at 22 GHz and to be consistent with the water vapor continuum in MONORTM.  These 
retrievals yield nearly identical results to the MONORTM retrievals.  Therefore the Pt. Reyes 
data prior to 4/08/2005 may not require reprocessing.

• Total water vapor along LOS path(vap)
• Water, liquid, total along line-of-sight path(liq)

SGP/MWR/C1 - New software version (4.15) installed

A problem began with the installation of MWR.EXE version 4.12 in July 2002. The software 
had been upgraded from a "DOS" to a "Windows"-compiled program to address an earlier 
problem.  The software upgrade corrected the earlier problem but introduced a new one that 
caused line-of-sight observing cycles to be skipped, a 15% reduction in the number of tip 
curves, and saturation of CPU usage. Software versions 4.13 and 4.14 also produced these 
problems.

The new MWR software, version 4.15, was installed on 08/04/2005. As a consequence of this 
upgrade, the tip curve frequency increased. The tip cycle time decreased from ~60s to 
~50s.

• bbn23
• bb23
• tkxc
• tnd_nom23
• tnd_nom31
• sky23
• Mean 31.4 GHz sky brightness temperature(tbsky31)
• tnd23
• tc31
• 23.8 GHz sky brightness temperature(tbsky23)
• Sky infrared temperature(sky_ir_temp)
• Water, liquid, total along line-of-sight path(liq)
• tkbb
• tnd31
• sky31
• tc23
• tknd
• bb31
• bbn31
• Total water vapor along LOS path(vap)
• tkair

SGP/MWR/C1 - Missing data

Data are missing and unrecoverable.

• bbn23
• tkxc
• wet_window
• East longitude(lon)
• tnd_nom31
• 23.8 GHz sky brightness temperature(tbsky23)
• tc31
• Water, liquid, total along line-of-sight path(liq)
• tkbb
• North latitude(lat)
• sky31
• tc23
• time_offset
• bb31
• Total water vapor along LOS path(vap)
• bbn31
• tkair
• bb23
• tnd_nom23
• sky23
• Mean 31.4 GHz sky brightness temperature(tbsky31)
• tnd23
• Sky infrared temperature(sky_ir_temp)
• base_time
• tnd31
• tknd
• Altitude above mean sea level(alt)

SGP/MWR/C1 - IRT bias

Measurements of sky temperature, especially those <250K, were positively biased compared 
with AERI because there was no compensation for the reflectivity of the gold mirror. The 
problem was corrected by changing the IRT emissivity from 1.000 to 0.987.

Use sky temperature measurements less than 230K with caution. Biases as large as 30K have 
been observed when the actual temperature, as measured by AERI, is less than 210 K.

• Sky infrared temperature(sky_ir_temp)