Introduction

High-quality data collection depends upon preparation and proper field procedures. Never-the-less, there follows a post-field evalution process that may result in modifications to the data set, or potential improvements to increase data utility. Changes made to the in-field S-PolKa data set are detailed, here.

Changes/Improvements to S-PolKa Data

Considerable energy has been invested in making the S-PolKa/DYNAMO data set one of the most accurate and usable S-Pol date sets, ever. Specific changes include:

• Noise power for all S-band and Ka-band power quantities has been re-estimated on a beam-by-beam basis; this impacts all noise-power corrections and mostly impacts reflectivity values and censoring/thresholding at low measured powers.
 
• Advanced techniques for data censoring have been developed, resulting in a more reliable elimination of bad data. See a complete description in Dixon and Hubbert, 2012.
 
• A merged data set has been created, incorporating S-band, PID, rain rates, and several of the most useful Ka-band fields within single cfRadial files. During creation, it was necessary to degrade Ka gate resolution from 75 m to 150m, to match the S-band gates. The merged data set was created for user convenience, and the original-resolution, quality-controlled, full-parameter, Ka data are avaialble in a separate data set. The merged data set should be considered the recommended, complete data set, except when the full-resolution Ka is required.
 
• Multiple S-band velocity fields have been created, reprocessed from the original phase data. The traditional velocity field is maintained, with two additional fields that provide less noisy data in low SNR regions and for non-meteorological echoes.
 
• PID uses local soundings to develop information on freezing levels. Reprocessing has allowed use of better time-aligned soundings for the entire data set.
 
• PID has been significantly altered to remove/mitigate erroneous classifications of graupel, make use of the noise-corrected RHOHV, and flag regions of likely receiver saturation (which led to errors in particle classification). Membership functions for the fuzzy logic algorithms have been changed; Scott Ellis has prepared a summary document.
 
• Some small tuning of the precipitation rate algorithms was performed (see this). Precip estimates were also improved by changes to censoring and recalculation of RHOHV.
 
• The test pulse was eliminated from the final data set. The test pulse had existed at extreme radar range. It was removed to avoid problems with any automated user processing and analysis of the data. (note that, at high elevation angles, the test pulse had already been eliminated as part of S-Pol's standard data reduction through truncation procedures)
 
• The S-band atmospheric attenuation was re-estimated using a technique detailed in Doviak and Zrnic (Doppler Radar and Weather Observations, 2nd ed., 1993, pp44-45). The attenuation-corrected reflectivity is included as the primary reflectivity field, while the non-attenuation-corrected field is also included in the merged data set.
 
• The original RHOHV field has been renamed to RHOHV_NNC_S, for NO_NOISE_CORRECTION. And the RHOHV_S field now has noise correction applied.
 
• The Ka reflectivity was corrected to account for the change-out of the Ka-band magnetron and to correct an ambiguity in the in-field calibration information.
 
• There were no changes to either the S-band power calibrations or the value of the Zdr-bias used during the field phase.
 

What Has Not Changed

EOL maintains a complete set of the original S-PolKa data. Should any problems be found in derivations or re-calculations, data can be reprocessed.

The original COVAR data set has not been changed. Derived power quantities (DBMHC, DBMHX, etc.) remain unchanged, except where censoring has flagged these values as "bad".

Implications for Data Users

Obviously, if you are using a data set derived from the real-time, in-field data stream, you should update your data set. While many parameters have not been changed, the censoring algorithm has been greatly improved, and use of the (version-1) final data set will ease any tasks associated with further data editing or censoring.

It is particularly important that you update any in-field data sets using the PID or precipitation rate variables. Both have been re-computed using the new censoring and the attenuation-corrected S-band reflectivity. The updated noise-power correction combined with a recalculated RHOHV, has resulted in better PID on the edges of echoes.