Fig 1. S-Pol radar when temporarily on location in a wheat field near Wichita Kansas.
S-Pol (Fig 1) is a new S-band polarimetric weather radar developed at the National Center for Atmospheric Research (NCAR) to serve the weather research community. It replaces the NCAR CP-2 polarimetric radar. S-Pol was developed for the purpose of providing a state-of-the-art polarization diverse radar for cost effective world wide deployment. It is planned that S-Pol will be deployed during the mid-August to mid-November 1999 Special Observing Period of MAP 47km northwest of Milano, Italy near the town of Vergiate. It will be located approximately midway between the French RONSARD and Swiss Monte-Lema Doppler radars that are roughly 60 km apart. The following paragraphs briefly describe the meteorological measurements that can be made with S-Pol and the strengths and limitations of the measurements. The basic characteristics and parameters are also discussed.
Fig 1. S-Pol radar when temporarily on location in a wheat field near Wichita Kansas.
Doppler radar has made it possible to remotely measuring three-dimensional wind fields of the atmosphere with unprecedented time and space resolution. Air motions can be obtained in both the optically clear boundary layer and in precipitation. Such wind field measurements will be applicable to both the airflow dynamics and heavy precipitation objectives of MAP.
Within the mixed boundary layer and when temperatures are above 10C, there are generally sufficient numbers of insects being carried passively by the winds to obtain clear-air wind fields. Wind fields have recently been obtained from single Doppler radar using retrieval techniques that use a numerical model and its adjoint. Figure 2 is an example of such a retrieval in the vicinity of two colliding convergence lines. Three-dimensional wind fields using the RONSARD, Monte-Lema, and S-Pol radars will cover an area of roughly 100 x 100 km with a resolution of 1 km, and 200 km by 200 km with a resolution of 2-3 km. Accuracy of the wind estimates will vary with distance from the radar but are of the order 1 m/s. These three dimensional wind fields can be obtained every 5-10 min.
Fig 2. Horizontal wind field obtained from the WSR-88D using the adjoint retrieval
technique. The event occurred on 24 July 1996 as two convergence lines collided. Both
precipitation and clear-air echoes were present. Courtesy of J. Sun and A. Crook of NCAR.
S-Pol is equipped with 3 passive radar receivers synchronized to the S-Pol radar transmitter and can be placed in a pattern surrounding it at distances of 5 to 100 km. This bistatic radar network is designed to provide cost effective, full vector wind field retrieval over an extensive area near S-Pol. The receivers detect obliquely scattered signals from targets. This Òsecond lookÓ at the weather allows retrieval of full vector winds as in traditional multiple Doppler analysis. One important kinematic objective that can not be met by the S-Pol, RONSARD, and Monte-Lema radars is observation of air flow in most of the valleys because terrain will block the radar beam. A proposed remedy is to use at least one of the two Doppler on Wheels (DOW) radars owned and operated by the University of Oklahoma and NCAR. These X-band, scanning radars are mounted on trucks and can be driven into the valleys to observe the valley wind flow.
Fig 3. S-Pol precipitation accumulation map derived from specific differential phase
KDP. The gray scale shows accumulation in mm. The watershed which runs through the
community of Buffalo Creek is shown by the heavy black outline. Thin contours are terrain
height (MSL) at 0.25 km. Courtesy of E. Brandes of NCAR.
The primary use of radar for many years has been to describe the 3-dimensional structure of the precipitation field and to estimate precipitation rates and accumulation at the ground. Traditionally this has been done by measuring the power of the returned signal obtained at a single, fixed polarization and deriving a measure of radar reflectivity and, thus, precipitation intensity. Because of the importance of measuring rainfall, most of the past effort in radar meteorology has been directed toward optimizing the relationship between the power scattered back from rain (radar reflectivity) and the rainfall rate as measured by gages at the ground. However, there is not a unique relationship because the radar reflectivity depends on the sixth power of the raindrop diameter, whereas rainfall rate is roughly proportional to the third power. It is common that drop size distributions can very considerably, i.e., rain can consist of mainly very small drops or very large drops. Because of the drop size dependence, reflectivity measured for a few large drops, which cause a low rainfall rate could equal the reflectivity measured for many small drops, which cause a high rainfall rate. One of the reasons for excitement with polarization diverse radars is that they measure the oblatness of raindrops which is related to drop size. The differential reflectivity between the horizontal and vertical pulses (ZDR) measures this oblatness. Also, the specific differential phase shift (KDP), which is the range rate of change with of the differential phase (FDP) between the vertical and horizontal phases, is almost linearly related to rainfall rate. The KDP measurement provides a rainfall estimate nearly independent of that from radar reflectivity. In mountainous terrain, like the Alps, this has great potential since blocking of the radar beam will not effect the differential phase shift as it does the radar reflectivity. A variety of studies are underway to determine optimum techniques for combining the various polarimetric variables to estimate rainfall which should provide superior estimates than those from radar reflectivity alone.
Figure 3 is an example of rainfall accumulation estimates made in the mountains of Colorado for a flash flood producing storm that killed 2 people in the Buffalo Creek watershed. The estimates were obtained by S-Pol using KDP. In the watershed region beam blockage by terrain was as much as 50%. A bucket survey of rainfall amounts in the area of high rainfall accumulations showed excellent agreement with the radar estimates.
The accuracy of precipitation accumulation estimates via radar is a complex subject dependent on numerous atmospheric conditions. For MAP, over estimates caused by the enhanced reflectivity (bright band) that occurs just below the freezing level will be a common problem. Fortunately, with polarimetric radar the bright band is easily identified. The accuracy of the estimates also decrease considerably with range as the beam gets progressively higher and overshoots the precipitation. Again, depending on the meteorology, the range of quality precipitation estimates will vary. Typically the quality of the estimates will decrease rapidly beyond 150 km from the radars.
Fig 4. Vertical section (RHI) through an Oklahoma thunderstorm showing types of
hydrometeors as determined from polarimetric variables. LR - light rain, MR - moderate
rain, HR - heavy, HA - hail, H/R hail/rain mix, GR - graupl, DS - dry snow, WS - wet snow,
CR - snow crystals. Courtesy of A. Ryzhkov and D. Zrnic of NSSL.
Polarimetric measurements are influenced by the size, shape, orientation, and phase of the hydrometers; thus these measurements can be used to identify hydrometer characteristics. The two polarimetric variables mentioned above (differential reflectivity and differential phase), together with two additional ones called the linear depolarization ratio (LDR), and the cross correlation between the horizontal and vertical polarized reflectivities(rHV), can be used to estimate hydrometer characteristics such as type (rain, snow, hail, graupl), wetness, and snow crystal type. NCAR and NSSL are developing a fuzzy logic algorithm for S-Pol to recognize hydrometer types, snow and hail wetness, mixed phase regions, and rain and hail size. For MAP this algorithm will run in real-time. An example of what this product will look like is shown in Fig 4, which was obtained from data collected with a polarimetric radar at the NSSL in Norman, Oklahoma.
Knowledge of the accuracy of these polarimetric techniques to identify hydrometer characteristics is rapidly evolving and not yet well documented. It is expected that information like shown in Fig 4 will be very useful in developing understanding of how orographic influences affect cloud microphysical processes.
The entire S-pol radar including generators and radar operations control center is housed into six 20 ft. sea containers. These containers are of a standard size and can easily be handled in seaports. Site preparation and restoration costs are minimized because the antenna support structure is made from the shipping containers themselves. The transmitter is an ASR-9 based unit built by Westinghouse. It uses an air cooled Klystron and produces one megawatt of power and one microsecond pulses. The pulse repetition frequency can range from 325 to 1200 pulses per second. The 28 ft. antenna produces a 0.91 deg beamwidth with a -30 dB first sidelobe and -35 dB integrated cross polar isolation. The antenna can scan up to 18 deg/sec in a 30 m/s wind without use of a radome.
Further information on S-Pol can be obtained at www.atd.ucar.edu/rsf. Further reading regarding polarimetric radar can be found in a) Zrnic, D., 1996: Weather radar polarimetry-trends toward operational applications. Bull. Amer. Meteor. Soc., 77, 1529-1534; b) Meischner, P., C. Collier, A. Illingworth, J. Joss, W. Randeu, 1997: Advanced weather radar systems in Europe: the COST 75 action. Bull. Amer. Meteor. Soc ., 78, 1411-1430; and c) V.N. Bringi and A. Hendry, 1990: Technology of polarization diversity radars for meteorology, Chapter 19a, Radar in Meteorology, D.Atlas Ed., Amer. Meteor. Soc. 153-190.
