The Ka band horizontal reflectivity measurment was subject to poorly characterized (and perhaps, uncharaterizable) variations during WISP04. Using the S-band reflectivity as a comparision "standard", it has been seen that the Ka reflectivity could vary from the S-band by more than 7 dB under conditions that were not obviously due to Ka attenuation. None of this is a surprise during a development effort, and recent analyses indicate that there are reasonable corrections that can be applied to the Ka data to remove a significant fraction of the difference between Ka and S-band reflectivity; any remaining differences might then be explained by more detailled meteorolgical analysis.
The Ka band had a magnetron changed-out on about 11-Mar-04, with subsequent impacts on Ka system characteristics. This was the only significant hardware change to the system during the project. However, other system maintenance tasks may have affected the system calibration unintentionally by altering connections between system components, or in other uncertain ways. It is not clear how such inadvertant changes can be tracked, and will not be addressed here.
Frank Pratte has done as complete a characterization of the Ka band calibration and sensitivity as possible. System parameters as summarized by Pratte are provided:
Nominal KaBand Radar Parameters 02 April 2004 Predicted Blue Sky Noise Power in 1MHz Bandwidth dBm -112.1 Carrier Frequency GHz 34.9 Pulse Width microseconds 0.8 Peak Power Watts 11980.0 Antenna Gain dB 45.2 Half-Power Beamwidth degrees 1.0 Fixed Losses (filter mismatch) dB 2.8 Meteorological Radar Constant (|K2|=0.93) 66.0 MDZe at 30 km dBZ -16.6 The measured peak transmit power was 6340 Watts with the first magnetron.
Pratte's parameters agree with those found in the data headers for the Ka system during the latter portion of the WISP04 project. Especially, the radar constant used for realtime Ka calculations varies from the Pratte value by only .2 dB. For the earlier portion of the project, and substituting the appropriate peak power and a bandpass filter loss of 2.4 dB, it is determined that the wrong radar constant was used (data header value was 67.4, whereas calculations yield 68.6). Efforts are ongoing to recover the earlier parameter tables and analyze differences.
Following is a list of range (km) versus minumum detectable blue sky reflectivity (dBZ) assuming standard conditions, also from Pratte. Loss at Ka band at our altitude appears to be the order of 0.1 dB/km in precipitation.
Range MDBZe (km) 5.0 -32.12 10.0 -26.10 15.0 -22.58 20.0 -20.08 25.0 -18.14 30.0 -16.56 35.0 -15.22 40.0 -14.06 45.0 -13.04 50.0 -12.12 55.0 -11.29 60.0 -10.54
During engineering and characterization of the Ka system, engineers noted that the Ka system receiver gains (one for co-polar quantities, one for cross-polar) were sensitive to changes in operating temperature. Injection of a non-varying test pulse signal resulted in a time-varying output. The Ka system was configured for operations using a test pulse signal, and variations in that test pulse can be used as a first-order approximation for removal of receiver gain changes (and therefor, correction of Ka reflectivity). While the operational test pulse signal is known to be somewhat variable, its variation is much less than that of the receiver gains, allowing corrections to be made.
The Ka band test pulse has been analyzed for all meteorologically significant project periods. In one case, the Ka co-polar test pulse power was compared to a recovered, non-noise corrected "blue sky" measure of background noise. The test pulse power and the blue sky noise tracked reasonalby well, with the amplitude of the blue sky variations about three-quaters of the variation of the test pulse (blue sky amplitude, ~3 dB; test pulse amplitude, ~4 dB). This is considered good agreement, given the crudeness of the blue sky calculations. This result supports the use of the test pulse power for correction of the Ka (co-polar) reflectivity.
This work will be continued, with supporting figures prepared. Software will be written to use the Ka test pulse power in a semi-automated, post-processing scheme to correct the Ka co-polar reflectivity. Engineering work will continue on stabilization of the Ka receivers, so such characterizations and corrections are not necessary in the future.