118 Operational retrieval of radar refractivity with errors for numerical weather prediction

Thursday, 29 September 2011
Grand Ballroom (William Penn Hotel)
John Nicol, National Centre for Atmospheric Science, University of Reading, Reading, Berks., United Kingdom; and A. J. Illingworth, K. Bartholomew, T. Darlington, J. Ovens, and M. Kitchen

The refractivity technique has been installed at 5 of the C-band operational radars in the UK over the past three years. The principle of the refractivity technique is very simple: the time taken for the radar wave to travel to and from a fixed clutter target depends upon the refractivity of the air, and any change in the refractivity can be detected by a change in the phase of the return from a clutter target. In practice it is more complicated. The technique, first proposed fourteen years ago by Fabry, has excited considerable interest in the NWP community but is still not used operationally. We present a rigorous analysis which enables spurious data points to be recognized and removed so that the best estimate of the value of refractivity and its associated error can be derived over a particular region. Without such an error estimate, the data cannot be assimilated into NWP models By careful analysis of the data obtained over a long period we have found that four stages in the analysis must be carried out if reliable data are to be obtained.

a) The phase change at C-band is about 13degrees/km/N, where N is a change in refractivity in ppm; in the summer one ppm is approximately equivalent to a change in relative humidity of 1% or a change of 1K in temperature. Accordingly, the difference in the phase change between two targets on the same radial and separated by less than 1km must be used to avoid folding. This target pair must be chosen very carefully if accurate phases are to be retrieved. We define various criteria for accepting suitable targets. b) Very bright targets which spread over two or more adjacent radial gates will give an identical phase change at each gate and the inferred change in N will always be zero. These targets will bias the value of N low over the region. We outline methods by which such targets can be identified and removed. c) The positions of the range gates are defined by the delay between transmission and the sampling time of the received signal. Retrieval techniques assume that the target is in the centre of the gate. In reality this is not the case, so the path distance to the target is unknown by a random amount. When N changes this will introduce an additional random phase change proportional to the N change. For large N changes the retrieval will fail. We propose methods which can avoid this problem. d) Acceptable clutter targets couplets will yield noisy N changes, so averaging over a region, typically 2 by 2km, is needed. We show how this averaging can provide a best estimate of the N change and its associated error.

Finally we present an evaluation of the retrieved refractivity by comparing it with synoptic data and also compare the structure of the retrieved refractivity fields with those held in the operational NWP model.

- Indicates paper has been withdrawn from meeting
- Indicates an Award Winner