CLARE 98 (Chilbolton, UK)

Data archive and quicklooks from the GKSS 95 GHz cloud radar MIRACLE


Contents of the archive:

Data (ASCII files)

Quicklooks (gif-format)

Quicklooks (PostScript format)


Some additional information:


Directory structure:

The data are archived in gzipped ASCII files. Reflectivities and Doppler velocities obtained from measurements in pulse-pair mode are supplied as function of time and height. During most of the aircraft overpasses, full Doppler spectra were recorded with a 256 point FFT algorithm, giving spectral amplitudes as function of time, height and velocity bin. For the pulse-pair measurements, quicklooks of height-time sections of reflectivity and Doppler velocity are available in gif- or gzipped PostScript format. The structure of subdirectories and filenames is as follows:

      ...clare98/gkss/data/octdd/dbz/octdd.hhmm.dbz.dat.gz
                                /vel/               vel
				/fft/               fft

      ...clare98/gkss/gif/octdd/dbz/octdd.hhmm.dbz.gif 
	                       /vel/           vel

      ...clare98/gkss/ps/octdd/dbz/octdd.hhmm.dbz.ps.gz 
	                      /vel/           vel           
(!! Please note that the y-axis title is slightly incorrect in all the 2D quicklooks: It should be ASL (above mean sea level) instead of AGL (above ground level) !! - 26 Feb 1999)

NEW (18 March 1999): Reflectivity profiles have been calculated also from several Doppler spectra recorded during the aircraft overpasses. These profiles can be found in the files octdd.hhmm.fdbz.dat (...fdbz.ps, ...fdbz.gif) in the corresponding .../dbz/... subdirectory.

Data file structure:

The first line of the ASCII data files gives the number of profiles and corresponding range gates, respectively, in this file. For the FFT files, it is number of profiles, corresponding range gates and corresponding FFT bins (always 256). For the ...fdbz.dat files it is just number of range gates. There are three rows (four in FFT files, two in ...fdbz.dat files): the first row gives decimal time (e.g., 15.7 for 1542 UTC), the second row gives height above mean sea level, the third row gives either reflectivity (in dBZe) or Doppler velocity (in m/s). In ...fdbz.dat files, it is only height and reflectivity. A value of -100.0 indicates "no cloud" (see also below). Negative Doppler velocities indicate downward motion. For the FFT files, the third row gives velocity bin (in m/s) and the fourth row gives spectral amplitude (normalized dB scale, see below). The temporal resolution is 10 seconds for the pulse-pair data and approximately 30 seconds for the FFT data. The vertical resolution is 82.5 meters for the pulse-pair data and 37.5 meters for the FFT data. The Doppler velocity resolution for the FFT data is 6.2 cm/s.

Calibration:

The accuracy of the given reflectivities is mainly based on the cross-calibration between the different radars which participated in the campaign (see also http://www.met.rdg.ac.uk/~radar/clare/quicklooks/), and should be within about +/- 2 dBZe at this point. The FFT spectral amplitudes are normalized in a way that the sum of the (linear!) amplitudes squared gives the total reflectivity. However, the FFT data have not been cross-calibrated yet, so the accuracy of these reflectivities may be lower than for the pulse-pair data.

Cloud identification :

The pulse-pair data in this archive was filtered with a cloud identification scheme to allow the determination of well-defined cloud boundaries. This algorithm basically follows Clothiaux et al. (1996), but has been improved taking into account the non-Gaussian behaviour of the receiver noise which was observed from time to time during the campaign. Comparisons with the unfiltered reflectivities (quicklooks at http://w3.gkss.de/~danne/clare/miracle/pulse_pair/) show that the procedure eliminates false alarms very well, and there is not very much elimination of obviously "real" clouds. Only in a few cases cloud tops seem to be underestimated or high, very thin cloud layers are lost.


If you have further questions, please contact Olaf Danne.


Link to the MIRACLE homepage at GKSS.


by Olaf Danne , danne@gkss.de