by A.R.Holt & D.H.O.Bebbington
University of Essex, COLCHESTER,
C04 3SQ, United Kingdom
Aim: To give a brief historical overview of the development of polarisation studies in weather radar.
Although various attempts were made to investigate sensing weather with polarisation-sensitive radars, the first quasi-operational studies were undoubtedly those connected with the Alberta hail project in Canada. The project, which had been commenced in the 1950s, was located in the region between the cities of Edmonton and Calgary where, regularly each summer, major hailstorms cause severe crop damage.
In 1968 an S-band, polarisation-diversity radar was installed at Penhold Alberta (ref. 1). Though capable of using any polarisation, it was operated using circular polarisation. The hail experiment was concerned with weather modification. The radar data, it was hoped, would enable hail to be identified aloft so that aircraft could, through cloud seeding, cause the size of the hailstones to be diminished. The theory behind this work was developed by McCormick and Hendry in a series of papers (refs. 2 & 3). It was recognised early on that propagation effects were important (ref. 4). But it was only in 1986, when the field project was wound up, that the way to take into account the effects of propagation was found (refs. 5 & 6).
In addition to the Alberta radar, which was fully scanning with a 1.1 beam width and a scan cycle of or 3minutes (for elevations to 8 or 20), McCormick and Hendry also built a non-scanning radar at 16.5GHz (eg. ref. 7). This was housed at the NRC, Ottawa, and allowed information to be gained over 8 consecutive gates of 150m. It was particularly designed for slant path propagation studies. Its frequency was later changed to 9.4GHz. This radar also initially used circular polarisation, though subsequently it also made measurements with linear polarisation (ref. 8).
Studies in linear polarisation really start from the seminal paper of Seliga and Bringi in 1976 (ref. 9), and a subsequent paper in 1978 (ref. 10). In these they discussed the use of first, differential reflectivity, and then differential phase, for the improvement of the measurement of rainfall. They pointed out that if one measures two parameters then one can obtain a two-parameter dropsize distribution. This led on to experiments to test this, which were carried out both in the States and also in the UK at Chilbolton. A number of papers resulted from 1979 to 1980 (eg refs. 11-13), setting out the early studies in this area. The idea here was that first of all horizontal polarisation is transmitted, and the power in the horizontal channel measured. Then the polarisation is switched, and vertical polarisation is transmitted and the power in the vertical polarisation channel is measured. One has then obtained both the horizontal reflectivity (ZH) and the difference in the reflectivities between horizontal and vertical polarisation. Clearly, when there is switching involved, there is a time delay, but this was not regarded, certainly in the early stages, as being a matter of great importance. With the simple model of the two-parameter exponential drop size distribution, one of these parameters, that in the exponential itself, is determined by the differential reflectivity, and then the total reflectivity will give the second parameter. Thus one can determine two parameters of a model drop size distribution for each of the rainfall resolution volumes in space.
The difficulty with the use of polarisation is that the antenna used for polarisations must be good so that the polarisation characteristics of each polarisation are very similar. Furthermore, the question of beam width comes into account, because the wider the beam width, the larger the resolution volume, and therefore the variables are being averaged over what can be considerable sized resolution volumes. The resolution volume is proportional in size to the square of its distance from the radar and to the antenna beam width. Most weather radars typically have a beam width of 1. One of the advantages of the Chilbolton antenna, which is 25M in diameter, is that the beam width is 1/4.
Almost all the linear polarisation studies in the 1980s were done with S-band 10cm wavelengths and they centred not only in the USA, but also in the UK. In the USA, the interest was on rainfall measurement, whereas the initial emphasis for studies in the UK at Chilbolton came from microwave propagation modelling. There, researchers were interested in predicting propagation characteristics at various frequencies, and the possibility of characterising the dropsize distributions in rain regions enabled better prediction of microwave attenuation. Later, the UK studies extended also to meteorological studies.
In the United States the radars used were first of all CP2, which belonged to NCAR, and then the CHILL radar was moved from Illinois to the Colorado State University at Fort Collins, polarised, and made portable. It has an air-blown radome, and studies were made not only in Colorado (ref. 14) but also in other parts of the southern States (eg ref. 15). Subsequently, the research radar at the National Severe Storms Laboratory in Oklahoma, was also used in many studies (eg refs. 16-7). Recently a new radar has been built by NCAR called S-POL. All these radars operate at S-band (typically about 2.82GHz) although there were some X-band studies made with CP2 in conjunction with the S-band, measuring just LDR, so it was measuring X-band LDR and S-band ZH and ZDR. In the UK the Chilbolton radar operates at a slightly higher frequency at 3.07GHz. Of recent years, it has been used by the Reading group for meteorological research. Its narrow beamwidth and the polarisation quality of its antenna make it ideal for studying small resolution volumes (eg refs. 18-9).
All these radars originally had limited polarisation capabilities. For example, initially the only polarisation parameter that the Chilbolton radar could measure was ZDR (by switching transmission between linear/vertical and linear/horizontal polarisation, and measuring the co-polar power). Then the system was developed so that they could measure the cross-polar return power as well as the co-polar return power, so that LDR as well as ZDR was measured. The measurement of total differential phase shift (DP), which requires Doppler capabilities, came at a later stage. Thus, throughout the 1980s and the early 1990s these systems were under development.
The first C-band radar was built for the German Space Agency, DLR, between 1984 and 1986, and was designed to be more flexible and to be able to use not only linear, but also circular, or any other polarisation. A number of studies have been done with this radar, which was shared between those working in radio propagation and those working in meteorology at the German Space Research Centre, Oberpfaffenhofen, near Munich), though the meteorology papers are more numerous (eg ref. 20). However, the effects of attenuation have proved to be very considerable, as was seen during the European funded PADRE and DARTH projects, and these attenuation effects have been seen in the shadow of the high reflectivity regions. They affect ZH, ZDR, and LDR. Similar effects have been seen at C-band in dual-linear radars operated in Italy by SMR (Servizio Meteorologico Regionale) in Bologna. Data from Bologna show strong negative differential reflectivity in the radial far side of the high intensity reflectivity regions (ref. 21). A C-band radar, C-POL has recently been installed at Darwin in Northern Australia, as part of the TRMM mission; similar attenuation effects have been seen. Another C-band radar is in use in Barcelona. The problem, of course, is to differentiate regions where ZDR becomes negative because of attenuation, and those where ZDR is negative because of hydrometeor type. However, polarisation at C-band is clearly going to have widespread use in the future. The German Weather Service has announced that it will be using linear C-band polarisation radars. In Canada approval has been given for the polarisation of the King City C-band radar, and the UK Met. Office has announced that it will be installing a C-band polarisation radar in Kent. Therefore, there is some priority in finding a way of reliably estimating the effects of attenuation, and hence correcting the values of ZH and ZDR. This is one of the matters of importance in the current CARPE DIEM project.