Known problems are broken down into two categories: S-band/general problems, and Ka-related problems. Overall, the control portion of S-Pol performed well, as did the S-band. Numerous problems are found in the Ka system, and these are documented to the extent possible.
Support information for some of these problems is found in the Operators' Log, and in the general report for S-Pol on this project (see the table of contents)
The S-band radar was uncalibrated until after the solar scanning done on 27-November. Corrections applied to the data set as a whole do not really apply to the time period prior to 28-Nov. The available calibration proxies for the period prior to 28-Nov have not been fully investigated to determine if they would allow a more accurate adjustment to the radar data for this early time period. Note that Zdr is very likely inaccurate for the early time period, as is the absolute reflectivity.
Please use S-Pol data prior to 28-Nov in a qualitative way, only!
The S-band process would freeze-up during unattended operation early in the project. Missing data periods are shown clearly in the Periods of Operation table (see blank spots in late November, early December; 25-Nov thru 4-Dec).
On occasion, the S-band transmitter would "fault", due to problems like overheating. Even though the transmitter might be off-line, the radar signal processor could still run, and continue routine production of radar beams. The net effect is a set of "empty" scans, where the radar receives/measures background noise of the environment. The periods of transmitter fault have not always been noted in the logbook, and are not flagged in the automatically created table of data availability.
An example of the transmitter experiencing multiple faults and corrections is provided.
Corrected during data set production.
Analysis of vertical pointing data, cloud droplet echoes, and clutter all indicated that a bias existed in S-band Zdr. The bias was negative, and a bias correction of 0.37 dB was added to all Zdr values during production of the final data set. This work is described in detail elsewhere. (link!)
Determined during summer of 2006, from comprehensive analysis by C. Knight.
Status: uncorrected in distributed data set. Correction left to the user.
Comprehensive first echo analyses conducted by Charlie Knight indicate that there is still a residual bias in S-band Zdr for RICO. That remaining bias is on the order of .05 dB, which should be subtracted from the Zdr values within the distributed data set (data set production timestamps of about January, 2006). Note that the additional .05 dB correction is within the stated limits of accuracy of +/- 0.10 dB for Zdr in the S-Pol distribution data set, and the data set itself is not likely to be corrected further to account for the residual bias.
Additionally, early project data (24-Nov-2004 thru 5-Dec-2004) show a residual bias in Zdr of approximately 0.25 dB (subtract .25 dB from the Zdr values in the distributed data set, same timestamps as noted above).
Corrected during data set production.
A bias of 1.4 dB was found in the S-band reflectivity (1.4 dB was added to all S-Pol horizontal co-polar reflectivity values).
Corrected during data set production.
The S-band VIRAQ radar processor did not re-initialize the year portion of the date at the start of 1-Jan-2005. Year initialization is done only on system start-up. The system was paused at about 16:27 UTC on 1-Jan, causing the year to re-initialize correctly. An offset of 31622400 seconds was added to all 1-Jan data prior to 16:27 UTC.
This is only an apparent problem.
Realtime scan definition at S-Pol was somewhat unreliable in that scan flags marking scan transitions were not always accurate. Scans sometimes included a single beam after the transition beams that was not marked as transitional. Scans displayed in solo or soloii will include that single beam, and "smear" the azimuthal beams that cover the gap between the last beams in a scan and that single beam.
This effect can be mitigated on display by setting the "angular fill" to 70% (this is in the "view" popup menu of soloii).
See the more complete analysis provided in e-mails to Rauber, extracted here.
The Ka radar antenna was located at the 7 o'clock position on the larger S-band antenna. This puts the radar on the lowest part of the S-PolKa antenna structure, just barely above the sea containers. As a result, there was some blockage of the beam in the lowest scans, with the 0.5 degree scan being the most affected. Images clearly showing this blockage in widespread stratiform precip are available:
[Note: lower panels in images show data from a period without precipitation.]
Note that there were several "discontinuities" in Ka hardware/software configuration that affect corrections applied to Ka data, and the use of those data. There are three different Ka periods: 24-Nov to 14-Dec (when Ka magnetron was changed); 15-Dec to 26-Dec (after which the VV and HH_K gates were shifted relative to each other); and 27-Dec to end-of-project.
At the time the S-Pol Ka system was fielded, the noise power of the system was uncertain. There was also concern that the noise power might drift with time. Since the noise power is required for application of a "noise power correction", it was essential that good information be acquired for both the HH and VV Ka-band noise power.
An approximation to system noise power was obtained through analysis of atmospheric background microwave noise. The approximation was used to apply the noise correction to Ka powers prior to calculating reflectivities. See the full report.
