Laserfiche WebLink
<br />phase of the sine wave gives the azimuth of the winds. The amplitude of the sine wave, <br />divided by the cosine of the tilt, gives the horizontal wind speed. The fitting is done in <br />successive iterations; outlier values are discarded between iterations. <br /> <br />The Snow Algorithm uses nearly,the same processes for determining the vertical wind profile <br />and the same VAD subroutines for the fitting itself. The first difference, apart from coding <br />style, is that more data points are used. Four velocity measurements reside within each <br />reflectivity bin of I-kIn length. The Snow. Algorithm uses the velocity measurements within <br />0.5 kIn of the calculated range. That distance increases the data density for the fitting <br />routine and should thereby improve the fit. Secondly, the 30-kIn nominal range has been <br />changed to 10 kIn, a value used by some radar sites. The radar beam is therefore sma~ler <br />when it passes through each 1000-foot interval. Comparing the 10-kIn with the 30-kIn results <br />shows that the latter are systematically biased toward greater speeds. The top of the beam <br />normally samples greater speeds than the bottom under typical wind profiles. The taller <br />beam at 30 kIn sees a greater spread of velocities than the shorter beam at 10 kIn. It. is <br />therefore expected that the 10-kIn wind profiles are more accurate for use in the advection <br />scheme of the Snow Algorithm. The wind profile is not used in the PPS algorithm for rain <br />because rain is not advected as severely as snow. <br /> <br />~ <br /> <br />11. Functional Description of the Snow Algorithm <br /> <br />The Snow Algorithm, RADAR7.F (seventh in a series of development programs), performs the <br />calculations for the algorithms leading to the integration of snowfall, based on radar <br />reflectivities and wind velocities. The geometry of the Snow Algorithm is nested cylindrical <br />coordinates. For each tilt (elevation angle), the view is 230 kIn by 3600 in 1 kIn and 10 <br />resohition for most data. The Algorithm was adapted to a Sun (Solaris 2.4) UNIX <br />environment. Rather than use the complexities of the PPS shared memory, this version <br />makes use of direct access files. This usage generally affects only coding style and execution <br />speed but not the functioning of the Snow Algorithm. The Snow Algorithm retains a shared <br />memory, /STORE/, for the data used most frequently in the V AD wind and reflectivity <br />calculations. The Snow Algorithm will be rewritten to conform with NEXRAD program <br />standards. The Snow Algorithm is. in modules, some of which are very much like those <br />already in use in the PPS algorithm and with similar names. Others can be readily <br />converted. An abundance of internal comments have been placed in the FORTRAN code so <br />that programmers can understand what the Snow Algorithm is doing. ' <br /> <br />Three terrain-based files are needed that are site-specific. The programs generating these <br />files are described in appendix E. The programs themselves are listed in appendixes F, G, <br />and H. The first program simply trims the occultation file of its first four bytes and all of its <br />trailing bytes in order to create a file that is easy to read by direct access. The second <br />program takes the new hybrid sector file, which is an attempt to have the bottom of the radar <br />beam be at least 500 ft above the terrain, and greatly rearranges its structure. The result <br />is a simple direct access file that indicates which elevation tilt is to be used for each <br />azimuth/range location. The third program generates a file in radar coordinates indicating <br />the ground elevation in meters. That terrain file is needed in the advection routine to help <br />determine where falling snow should land on the ground. <br /> <br />40 <br />