Hourly Cloud Radars Quicklooks

The following links lead to hourly updated cloud radar quick look archives:

See also the Fog Examples from the TeFiS-Campaign in Munich and from Iqaluit. This page also describes the VAD and ETHI algorithms used for getting time height cross sections of wind and reflectivity from the RHI and PPI scans.

Some of the radars are operated continuously, others are only used for field campaigns. Therefore, data is not available for all times.

In case the radars are operated in vertically pointing mode each hour a set of time-height cross section images showing the reflectivity, Doppler velocity, peak width, and unfiltered SNR of the last 4 hours (adjustable) is generated and copied to this web server. The png images are sorted in directories for different devices and dates where they can be located as on an ftp sever in the following way:



<dev>: Device, moli=DWD; mpi=MPI; fzk=KIT ...

<YYYY>: Year

<MMDD>: Month and date

<hhmm>: Hour and minute in UTC of the end of the time interval shown in the image

<type>: Type, see below “Data products ...”

Some of the radars are making PPI and or RHI scans routinely. In this case images of each scan are generated resulting in a large number of images. Webinterfaces are provided which allow animating sequences of the images as a movie (see the links "scan movies"). Alternatively, the PPI and RHI images can be found by clicking to the directories.

Data products (<type>) shown in the different images:

  1. The intensity of the returned radar signal (usually only plotted for the co channel). Sometimes, it is plotted as signal-to-noise ratio SNR which is the 0th moment calculated from the spectrum divided by the noise power. The minimal SNR value depends on the FFT size and the number of spectral averages, which is typically -23 dB. Sometimes the intensity is converted to reflectivity Z which is preferred by radar meteorologists. Z ~ SNR * H^2 where H is the measuring height. In case the targets are droplets which are small compared to the radar wavelength (Rayleigh scattering) Z is proportional to the density of droplets multiplied by the droplet diameter D^6. Therefore, few large droplets cause more signal than many small droplets and generally rain causes stronger Z-values than clouds. A few insects or other advected particles cause even larger signals which are denoted as plankton. Plankton can typically seen in the region which is warmer than 0 C. Unfortunately, the plankton covers the signals caused by clouds and rain. Therefore, efforts are made to filter the plankton signals. In the melting layer, where snow flakes are melting to droplets, there is an enhanced reflectivity which often can be recognized as horizontal border between the ice and the water phase of hydrometeors.

  2. The Doppler velocity VEL of the returned signal (also only plotted for the co channel). From the fall velocity the diameter of droplets can be deduced. Larger droplets have larger fall velocities and snow flakes and cloud droplets have small fall velocities. Normally, the size of rain or drizzle droplets increase while they are falling through clouds as they catch small cloud droplets. While the droplets fall through non-cloud regions their size and fall velocity decreases.

  3. The Doppler spread RMS. Increased Doppler spreads are caused by turbulence and drop size distribution.

  4. The Linear De-polarization Ration LDR is the ratio of the reflectivities measured in the cross channel and the co channel. Round droplets cause only reflections which have the same polarization as the transmitted signal. Therefore, LDR of droplets is small. In this case the observed LDR values are between -28 and -35 dB which is caused by the polarization decoupling of the antenna. In contrast, ice crystals or plankton may turn the polarization of the transmitted signals which may cause larger LDR values. Large LDR values are observed in the signal caused by the melting layer, plankton, and sometimes by clouds above the melting layer.

For filtering the signal from insects (plankton) the peak separation and classification algorithms described in the conference paper “Target Separation and Classification using Cloud Radar Doppler-Spectra” are used. The main improvement achieved by these algorithms is that the hydrometeor and the plankton signals are separated and the plankton signals are filtered.

The following images are generated:

<type> = “9pz1”: SNR, un-filtered

<type> = “ldr”: LDR, un-filtered

<type> = “v1”: VEL, un-filtered

<type> = "r1": RMS of all peaks classified as hydrometeors (filtered)

<type> = “z1”: Reflectivity Z in dBZ units of all peaks classified as hydrometeors (filtered)

Note, that the velocities give some information about the turbulence in the boundary layer. Therefore, the velocity of the unfiltered data is plotted.

There are two black lines in the 9pz1-plots. One of them represents the melting layer as deduced from the external temperature measurements. When a melting layer was detected due to an increased LDR then the melting layer height used by the peak classification is overwritten by the detected height. In these cases the wrinkled black line shows the melting height deduced from the radar data.