Detailed review shows that S-band and Ka-band radar beams on S-Pol match closely in space and time. Some problems need correcting, and will be fixed in the final data set. The preliminary data have a correctable mis-match of exactly one beam between S and Ka band beams, and one gate in range.
The S-PolKa band radar system is essentially two separate systems strapped together on the same radar pedestal. The system antennas, processors, and internal timers are independent, with individual outputs combined into a single data set after the fact. The Ka system has no pointing information available to it, and the only commonality that allows correlation of the data streams is the timestamp written on beams produced by the different data systems. Beams from the two systems are matched by a closest-in-time rule.
Matching of gates/beams between the two systems requires that
The independent radar system timestamps are based on GPS clocks combined with highly stable, disciplined oscillators slaved to the GPS time. This ensures that time can be accurately subdivided into units of tens of nanoseconds and that the clocks between the systems are in sync. For triggering of transmit pulses and matching of dwells (beams), it is required only that the two systems have pulse repetition times (or pulse repetition frequencies) that are integer multiples of each other, and that there is a known convention for calculating when a beam should start and end in time.
The Ka and S-band timestamps maintain their identity and are available for review after combination of the two data sets. Review of these timestamps shows that closest-matched beam times are consistently within 2.5 milliseconds of each other. Matched times were checked for subsets of data on every day of significant interest. Since both the Ka and S-band Horizontal PRF were 500/sec, and beams typically had 100 samples/beam with a rotation rate on the order of 5°/sec, there is an implied azimuthal match of +/- .025°.
This accuracy of beam matching was not bourne out in preliminary data review, as evidenced by a pronounced up/down shifting of Ka-band echoes during RHI scans (also CW/CW shifting during PPIs). Extensive analysis followed by review of radar system processor code determined that the problem was related to the S-band timestamp being applied at the start of a beam, whereas the Ka-band timestamp was applied at the end of each beam (see note). The net result is a two-beam shift in Ka band data upon changes in direction. This problem exists in the preliminary project data set, and will be corrected only after appropriate code has been written.
Further work was done in comparing azimuth and ranging using ground targets. The BAO and Fredrick towers provided good echoes with known azimuths and ranges. Careful review of specially collected slow and fast scan data, along with analysis of individual gate values indicates that:
Note that there is some suggestion that, independent of beam timestamp issues, the Ka antenna may be aligned about 0.1° to the right of the S-band antenna (this implies that the Ka echoes tend to appear to the left of S-band echoes). Further review should be done after the beam timestamp matching is fixed.
Vertical alignment of the Ka co-polar and S-band co-polar beams is within 0.2°, as deduced from antenna pattern determinations of 31-Mar-04. The antenna pattern plots show that the center of the Ka co-polar beam is slightly above the center of the S-band co-polar beam, by about 0.2°.
The following support information is provided:
[Actually, this info is still being compiled, but images are available in the subdirectory "pointing"]