Laserfiche WebLink
<br />- -----~~~.=-~"<T,,..., _.:,.~,,-';"'~~~~~~;;-,,~~~:.~~t.'~,~-:~ :"Z;".~\~~~.~;J~-~-r-.'1''i''':::':'~:~~!\~:;~:, ~~~ <br /> <br />APRIL 1978 <br /> <br />MELVIN J. SCHROEDER AND GERARD E. KLAZURA <br /> <br />501 <br /> <br /> <br />IF-C. <br /> <br />[d1t/ColIt_ <br />(step 1) <br /> <br /> <br />IF -HC* <br />Records <br />Corrected <br />ReIIort <br /> <br />IttpOrts <br />......ts <br /> <br /> <br />ItDort <br />Ilenerl. <br />(Stitt t) <br /> <br />etI1crof1cllt Ind Ill,,*," <br /> <br />FIG. 2. Flow 'diagram of automatic data processing procedures <br />for HIPLEX digital radar data. <br /> <br />data for attenuation and partial beam filling. Hilde- <br />brand (1978) calculated that storms with reflectivity <br />maxima of less than about 50 dBZ are not seriously <br />attenuated at 5 cm wavelengths, provided the storms <br />are not too extensive. <br /> <br />a. Raw data edit <br /> <br />. <br /> <br />Original data can contain an assortment of errors:, <br />date and/or time may be set or incremented incorrectly; <br />azimuth, elevation and record count may be erroneous; <br />preset variables such as sampling mode, pulse repetition <br />frequency and azimuth increment may have been reset <br />by accident; and radar characteristics may change, <br />signaling problems. When any of these conditions exist <br />they are flagged and displayed in the header edit report. <br />Radials are automatically deleted for reasons such as <br />azimuth and elevation angles not being within accepted <br />limits, azimuth skipping and noise in all header entries. <br />A complete history of the antenna sweep sequence <br />(elevations, times and azimuthal degrees swept) is <br />another useful output from this step. <br />To conserve computer processing time, the "noise" <br />or "lower threshold" level is saved in the compressed <br />data file. This number is the result of searching 100 <br />consecutive bins in a radial for a value which occurs in <br />at least 55 bins, i.e., the modal value. The search begins <br />with the first 100 recorded range bins. If a modal value <br />is not found, the first 10 range bins are deleted from the <br />l00-bin string and the next 10 bins (bins 101-110) are <br />added. This procedure is continued until a modal value <br />is gleaned from 100 consecutive bins of data, which <br />mayor may not contain echoes. It is highly improbable <br />that a signal (echo) region could be involved in the <br />computation of the modal value. During the search the <br />minimum DVIP value is saved. It is subtracted from <br />the modal value and the difference is then added to the <br />modal value plus a guard value of 3 DVIP units, i.e., <br />modal value+(modal value-minimum value}+3= <br />"noise" or "lower threshold" DVIP value. Addition <br />of the 3 DVIP units (1 dB) accounts for the small <br />noise fluctuation of the radar system and adds reli- <br />ability to the precipitation data because system noise <br />is not as apt to be identified as returned echo data. <br />Before a DVIP value is converted to meteorological <br />units, it must be larger than the "lower threshold" <br />DVIP value. This step, done at records 1, 5401, 8701 <br />and 11501, also provides a check on the accuracy of the <br />manually set threshold l~vel of the "blue-sky" elimina- <br />tion feature in the hardware. The procedure described <br />above was arrived at after inspection of data spanning. <br />two operational seasons. Nevertheless, even with the <br />1 dB guard value added on, the noise band occasionally <br />extends slightly beyond the computed value. When <br />this occurs an additional guard value of 2 or 3 DVIP <br />units is added for subsequent processing when minimum <br />threshold reflectivities are required. <br /> <br />. <br />