Chilbolton 94 GHz Galileo radar data

Robin Hogan
11 April 2002


This dataset consists of effective radar reflectivity factor (Z) measured by the 94 GHz Galileo cloud radar situated at Chilbolton, England (51.1445N, 1.4370W). The instrument is a bistatic system that operates continuously round the clock in a vertically-pointing configuration. It was developed for the European Space Agency by Officine Galileo, the Rutherford Appleton Laboratory and the University of Reading.

This document is a revision of the original documentation of 26 February 2001, and describes the recalibrated data that was released to BADC in spring 2002. If you have used the reflectivity values of the original Galileo data for anything quantitative then you should definitely download the newer data, as there was a steady and significant loss of sensitivity during the 18 month duration of this dataset, that we were previously unaware of.

The characteristics of the radar are as follows:


Frequency:                              94.00 GHz
Antenna diameter:                       0.6 m
Peak power:                             1.6 kW
Pulse width:                            0.5 s
Pulse repetition frequency (PRF):       6250 Hz
System noise figure:                    10 dB
Beamwidth:                              0.5
Range resolution:                       60 m

History of operations

The radar was initially operated in summer 1996 for a few months, but a problem with sensitivity was found. This was rectified when the radar was Dopplerised, and from September 1999 until February 1999 the instrument was operated from the `Receive Cabin' at Chilbolton. It was then mounted on the side of the main 25 m antenna at Chilbolton to permit scanning with the 3-GHz radar, although did not operate round-the-clock until the end of April 1999. On 17 October 2000 it was removed from the side of the main dish to solve the problem of gaps in the data whenever the 25 m antenna was scanning through precipitation. During this time the output power from the instrument dropped steadily; the calibration section below discusses the implications of this for sensitivity and calibration. Doppler velocities measured during this time were unreliable and so have not been included in the dataset. The data described here are from 1 May 1999 until 17 October 2000.

From January 2001 to March 2002 the radar was operated from the ground at Chilbolton and a new data acquisition system installed which permitted measurement of the mean Doppler velocity and Doppler spectral width. Operations from the ground meant that the data stream was not interrupted every time the 3-GHz radar was scanning, but unfortunately it also meant that direct cross-calibration with the 3-GHz radar was not possible. Nonetheless, it is hoped that these data can be released on BADC in the future.

Processing that has been applied to the data

The following stages of processing have been applied to the raw data (in this order):

The result of this processing is that the minimum-detectable signal at a range of 1 km was around -50 dBZ in May 1999, increasing to around -38 dBZ in October 2000. The sensitivity decreases with range according to the inverse square law; therefore the minimum-detectable signal at 10 km is 100 times (20 dB) higher than at 1 km. The `chil2nc' program was used to process the data and convert it to NetCDF; source code for the latest version may be downloaded from

Data format

The data is provided as daily NetCDF files containing the following variables:

The following global attributes are present:

The following global attributes were added when the data were recalibrated:

Simple programs to read NetCDF files of Chilbolton data into Matlab, IDL and PV-WAVE can be found at

Interpretation of radar reflectivity factor and calibration issues

It is taken for granted that if quantitative use is to be made of the Z values then the user has a fairly good understanding of the concept of radar reflectivity and how it is related to the particle size distribution. In this section therefore we concentrate on calibration issues, although the problem of attenuation at 94 GHz is also discussed since it is much stronger than at lower frequencies.

Our approach to calibration of the 94 GHz radar is by reference to the 3 GHz `CAMRa' radar at Chilbolton. CAMRa can be calibrated absolutely by exploiting the non-independence of the radar parameters Z, ZDR and KDP in heavy rain, as described by Goddard et al. (1994). However, to compare the two radars directly requires a Rayleigh-scattering target which is both near enough for attenuation to be small at 94 GHz, yet far enough that near-field and ground-clutter effects are small at 3 GHz. We correct for gaseous attenuation at 94 GHz using the temperature and humidity from either the ECMWF or Met Office models, and make the small correction for the near-field effect at 3 GHz using the expression of Sekelsky (2001). The results are expressed in the figure below in terms of the noise-equivalent reflectivity at 1 km (see the definition of nez above):

The error bars indicate the 13 calibration events and their estimated uncertainty, and the numbers indicate the Chilbolton tape number. A steady deterioration in sensitivity is evident, which is due to a loss of transmit power from the tube. Given the uncertainties in the intercomparison of the two radars, the error in the resulting 94-GHz reflectivity factor after calibration is estimated to be around 1.5 dBZ.

It is important to understand the convention used in the intercalibration of radars of different frequencies because of the temperature dependence of the |K|2 parameter of liquid water at millimetre wavelengths. We have calibrated our radars such that Rayleigh-scattering liquid water droplets at 0C produce the same reflectivity factor at all frequencies. For example, a population of 100m droplets with a concentration of 106 m-3 at 0C would have a Z of 0 dBZ at all frequencies. Hence a radar at frequency f (after calibration and correction for attenuation) will report an effective reflectivity given by

Zf = Integral from D=0 to D=infinity { (|Kf|2/|Kf,0|2) n(D) D6 Mf(D) dD },
where Mf is the Mie/Rayleigh backscatter ratio. |Kf,0|2 is the dielectric parameter of liquid water at 0C, and is 0.93 at 3 GHz, 0.877 at 35 GHz and 0.668 at 94 GHz. Because the |Kf|2 of liquid water varies with temperature at 94 GHz, if the example above were repeated at 20C, a 94 GHz radar using this calibration convention would report a Z of +0.82 dBZ while a 3 GHz radar would still report 0 dBZ. Formulae for the dielectric constants of ice and liquid water can be found in Liebe et al. (1989).

Attenuation by both atmospheric gases and liquid water is much stronger at 94 GHz than at lower frequencies. At 10C, 1013 mb and 100% humidity, the one-way attenuation due to gaseous attenuation at 94 GHz is 0.636 dB km-1. In summer the typical two-way gaseous attenuation to top-of-atmosphere is 2 dB. If the temperature and humidity profile is known with some degree of accuracy (such as from a model or a radiosonde ascent) then gaseous attenuation can be corrected for. However, when low clouds are present then the liquid water attenuation can easily exceed the gaseous attenuation, and of course the profile of cloud liquid water content is generally far more uncertain. This makes quantitative use of the reflectivity data in ice difficult if there is any low cloud present. At 10C and 1013 mb, the one-way attenuation of 1 g m-3 of liquid water is 4.34 dB km-1. Attenuation by rainfall is even greater, and in moderate and heavy rain can extinguish the signal completely. Wetting of the radomes of the radar also introduces a large attenuation; see point 4 in the next section.

Known problems with the data

  1. Gaps in the data are present due to:
  2. A 50 kHz interference was present in the raw data which, if unchecked, resulted in anomalous horizontal `lines' of cloud every 3 km in processed time-height plots of reflectivity. This problem has been tackled in the following ways: Removing this interference has unavoidably compromised the sensitivity to some extent. Also, some anomalous echos may still be present in some of the data, although they are fairly easy to locate subjectively. The removal of pairs of pixels (indicated by the options attribute containing the string `-doubleclean') obviously will remove any genuine cloud that is only two range gates thick.
  3. The reflectivity values in the lowest gates are affected by:
  4. Wetting of the radomes covering the two antennas of the radar has been found to cause a two-way attenuation of 9-14 dB. The data should therefore not be used quantitatively during rain events, which can be identified using the Chilbolton rain-gauge data also provided by BADC. Other radars at 94 GHz presumably also suffer from this problem.

Conditions of use

If data from the Galileo radar is used in any publication or report then acknowledgement must be given to RCRU at the Rutherford Appleton Laboratory for providing the data. The acknowledgement should be of the form:
We thank the Radiocommunications Research Unit at the Rutherford Appleton Laboratory for providing the 94 GHz Galileo radar data. The Galileo radar was developed for the European Space Agency by Officine Galileo, the Rutherford Appleton Laboratory and the University of Reading, under ESTEC Contract No. 10568/NL/NB.

Who to contact

If you have any problems obtaining the data, please contact the British Atmospheric Data Centre. If you have problems, queries or comments regarding the data themselves that are not covered adequately by this document, or would like to know if any data was recorded on specific dates outside the period available on BADC, please contact Charles Kilburn ( and Robin Hogan ( Even if you use the data and have no difficulties at all, we are very interested in knowing the uses to which our cloud radar data is being put, so please contact us!

See also