Known Problems

WISP04, NCAR Marshall Field Site
Feb/Mar/Apr 2004



There were several known problems with S-PolKa data or data quality during WISP04. This page presents a short summary of those problems, with links to further details. More routine issues concerning data quality can be found by referring to the appropriate sections of the on-line report (see the Table of Contents).

The WISP04 Project included the first deployment of the Ka-band radar add-on to S-Pol, creating the system know as S-PolKa. The Ka presented a whole series of problems and data quality concerns, some of which are not yet fully documented, nor, perhaps, fully understood. Insight is provided when possible, but characterization of the Ka must be considered to be on-going.

Further insight into these and other problems may be gained through review of the Operators' Log for this project.


Ka Reflectivity Calibration

The absolute Ka reflectivity calibration has yet to be determined. Comparisons of Ka reflectivity to S-band reflectivity are ongoing. Changing-out of the magnetron for the Ka on or about 24-Mar certainly affected any continuity in the calibration, as did other changes to the Ka system.

Additionally, it is known that temperature changes had significant impact on the Ka receiver gain, and that temperature also affected the Ka transmit frequency, a factor that could change apparent received power if the frequency drifted outside of the Ka receiver bandwidth (the center of the pass band was designed to automatically adjust to the changing transmit frequency, but the effectiveness of this scheme needs to be evaluated).


Tape recording glitches

There were differences in the performance of the data recording systems for the dual Radar Data Acquisition systems (RDAs). Generally, the secondary system experienced the most "glitches", and the primary RDA was used as the source of S-Pol archive data.

Only on one significant weather day was it considered to patch the primary data stream with data from the secondary data system. This was on 10-Mar-04, when problems with the S-Pol computer network caused several re-boots to the RDAs. For times between 18:44 and 21:03, care was taken to retrieve the most complete set of data possible between the two RDAs.

A separate document is available detailing the combination of the data sets.


Improper data reconstructions

The S-Pol and Ka data streams are initially two independent streams from two separate data processors. The data from each has to be combined, with beams carefully matched between the two systems. For this matching, the only housekeeping parameter common to the two data streams is GPS time, since the Ka lacks information on pointing angles.

Initial data combination performed well enough in real time, but problems were found during post-processing, where the effective data rate was very high. For those cases, the beam housekeeping information had a tendency to become separated from the corresponding gate data, resulting in odd results where ground clutter features might be shifted from known locations, or segments of scans might be introduced into scans at other fixed-angles.

The data reconstruction problem has apparently been corrected, but users should avoid incorporating the original, real-time data into any critical analyses. Instead, use the re-generated data.

Examples of this effect are provided, but only for reference since the problem is corrected.


Bad/redundant beam, repeating every 64th

Early in the project, it was noticed that every approximately 64th beam was bad. This was a result of an uninitialized/noise beam being created in the software that had the same timestamp as the preceeding good beam. The bad beam would over-write the good beam during data display, but the good beam still existed. A filter was added to processing software to remove this beam. This problem has re-appeared on occasion, suggesting that caution must be used when processing to ensure that the appropriate/corrected software version is used.

This problem manifested itself in the 19-Feb data, was characterized on 20-Feb, and the fix applied shortly thereafter.

For an example, see the Figure.

Note added by VanAndel on 15-Jun-04:

	OK, this problem was fixed on Feb. 25th.  The issue was that the
	ingest code wasn't properly handling the RHDR packets, which are sent
	every 64th beam (DWEL).

Ka/S-band beam mis-match

As noted in Data Reconstructions, above, there was a need to correlate the Ka and S-band data streams. Matching of closest beams was done reliably, but the sample dwells from the Ka and S-band may not over-lap exactly (the S-band and Ka-band sample volumes/beams may not start at exactly the same rotation angle). [Note: for WISP04, the matched times of the beams did agree very well, differing by less than 3 milliseconds for all events checked.]

A complete evaluation of Ka/S-band beam matching may be found in the Ranging and pointing section. Note that an error in conventions led to a chronic mis-match of one beam, as described in that section.

Note that the visual impact of any beam shift is significant during RHIs and PPIs, where the apparent sweep-to-sweep shift appears twice as large. Additionally, this effect can be amplified by display software, to create an alarming impression -- refer to SOLO Shifted Echoes for an example of apparent shifting due to the Solo display software.


SOLO Shifted echoes in PPIs and RHIs

[This section describes an artifact in SOLO that creates the impression of CC/CW or UP/DOWN shifting of radar echoes. This occurs even though data may be perfectly correct and aligned in azimuth/elevation.]


Ka Magnetron Change-Out and Calibration Change

The Ka magnetron failed part-way through the project, and change-out caused a change in the Ka band calibration. See the section on Ka cal for details. Don Ferraro's timeline and statements regarding the change-out are included here:

Ka Band worked until 3/5/05 @ 0130Z.  This was the first magnetron failure.

Ka Band was back in operation with repaired tube on 3/10/04.
Operation was not very stable until a hard failure on 3/11/04 at
0309Z.  Stability was suspect due to magnetron not firing on every
pulse.  The output power was fluctuating and the data quality would be
very poor, if useable at all.

Ka Band back in service on 3/24/04 with new magnetron from Radio
Research.  Pulse width was reduced from 840ns to 800ns.  The PRF was
the same 500 HZ.  Test data taken on 3/24/04.  From this point until
the end of WISP04 on 4/2/04, the Ka Band transmitter operated in a
stable mode with no further problems.  During this period, there were
some receiver failures, but these were corrected before data
collection and shouldn't have had any impact.

-- 
Don Ferraro
NCAR - ATD/RTF

--- Bob Rilling --- / NCAR Atmospheric Technology Division
Created: Mon May 17 13:29:08 MDT 2004
Last modified: