Laserfiche WebLink
<br />10.2.1 The PPS Algorithm <br /> <br />The PPS algorithm is described in section 3.3 of the Federal Meteorological Handbook No. <br />11 (1991), Doppler Radar Meteorological Observations, part C, WSR-88D Products and <br />Algorithms. The algorithm produces maps of accumulated precipitation (intended for rain, <br />not snow) for I-hour, 3-hour, and storm totals. Five basic steps are listed: 1) preprocessing, <br />2) rate, 3) accumulation, 4) adjustment, and 5) products. <br /> <br /> <br />1) Preprocessing: <br /> <br />The preprocessing algorithm prepares Ze data for input into the rate algorithm. Quality <br />control steps include: blockage correction, isolated bin check, outlier check, vertical echo <br />continuity check, and bi-scan maximization. The beam blockage correction adds 1, 2, 3, or <br />4 dBZ (or 2, 4, 6, or 8 to the biased dBZ) to radial data beyond ranges at which the beam is <br />estimated to be blocked by particular percentages. The occultation file has azimuth <br />resolution of 0.20 and is calculated from terrain data specific to each radar antenna site. <br />Isolated bin values of reflectivity are set to a code for "less than minimum detectable signal." <br />Bins with unreasonably large reflectivities are set to a near-minimum level or else to the <br />average of the eight adjacent bins. Bins that are completely occulted (>60-percent blockage) <br />are sometimes replaced by the values of adjacent bins having less blockage. An attempt is <br />made to remove spurious echoes in the lowest tilt ("tilt test") if such echoes tend to disappear <br />in the second tilt. A hybrid-sector file then specifies which tilt to use for precipitation <br />calculations. At far range, it is one of the two lowest tilts giving the greatest echo (bi-scan <br />maximization). The reflectivities are then ready for those further calculations. <br /> <br />The Snow Algorithm, presented here, uses a new hybrid-sector file, developed by Tim <br />O'Bannon of the OSF, which attempts to keep the bottom of the beam closer to the terrain <br />than previous versions while still maintaining 500 feet vertical separation. It does, not yet <br />attempt to use adjacent bins for substitution for beam blockages. It does not use the tilt test <br />because snow may occur from shallow clouds that are observed only in the lowest tilt at far <br />range. Though the preprocessing software coding in this version of the Snow Algorithm has <br />differences in style from that in the PPS version, most of the existing PPS routines are <br />acceptable for snow estimation. <br /> <br />Five significant differences are needed for work with snow: <br /> <br />(a) All radials with data, including overlaps, are stored in an array along with their <br />precise azimuth angles. (Each tilt of a volume scan typically has 367 radials of data, <br />including the overlap of the start and end of each tilt. The Snow Algorithm allows up <br />to 370.) No radials are averaged together when their integerized azimuth angles are <br />the same.' Such averaging requires the conversion of reflectivities to precipitation <br />rates before averaging. The integerization of azimuths may leave some angles with <br />no data even though no break occurred in the scan. Instead, a "nearest neighbor'~ <br />routine is used later when data for a particular azimuth angle are needed. That <br />routine avoids both averaging and gaps. <br /> <br />(b) Reflectivity values are kept in units of biased dBZ rather than precipitation rate. The <br />conversion back and forth between units adversely quantizes reflectivities at the low <br />precipitation rates typical of snow. Averaging of radials is no longer needed in the <br />Snow Algorithm preprocessing. The only other PPS preprocessing averaging is to <br /> <br />38 <br />