Poldirad

POLDIRAD - Data Description

Martin Hagen

Data Types

File Names

Data Files

Format

Data types

For the polarimetric C-band Doppler radar of the DLR (POLDIRAD) various data have been collected during the 21 July 1992. All data which are of common use are available here in a special format to allow display as well as easy access to the physical values.

Radar scans are performed in a PPI (Plan Position Indicator) mode or a RHI (Range Height Indicator) mode. PPI scans can be performed at various elevation angles to scan a complete volume, similar RHI's can be performed at different azimuth angles.

Radar variables available from POLDIRAD are dependent on the recorded data type. For Doppler scans no polarisation switching is possible, only the horizontal polarisation is used. Polarimetric reflectivity measurements are based on linear horizontal and vertical polarisation.

Data type: Doppler

Reflectivity in dBZ
Doppler velocity in m/s
Spectral width in m/s (selected scans only)

Data type: Reflectivity

Reflectivity in dBZ
Differential reflectivity in dB
Depolarisation ratio in dB

Reflectivity is (among others) a measure of the amount of precipitation in the air. Values below 15 dBZ correspond to very light rain (0.3 mm/h), clouds and clear-air echos. Clear-air echos are observed frequently in the region close to the radar (up to 60 km range during summer time). Reflectivity values above 55 dBZ refer in most situations to graupel or hail.

When interpreting radar images, a few things have to be considered to avoid misunderstanding:

• Earth is a sphere: Even when looking horizontally, the radar beam will be at a considerable height above ground at far distances. At far distances radar echos close to the ground can not be seen.
• Ground clutter: Besides precipitation, mountains and other obstacles (buildings) reflect the transmitted power back to the radar. This is a significant source of radar echos at close ranges and low elevations.
• Attenuation: During the passage of the radar pulse through heavy precipitation a certain amount of the transmitted energy is dissipated, and no longer available for scattering at ranges further out. This results in sometimes easy-to-see gaps behind storm cells, as well as invisible precipitation areas at the rear of the squall line. However, not always attenuation is clearly visible. At the time, there is no stable algorithm available for the correction of attenuation. In composite images this problem can be overcome by other radars looking from the rear side.
  Differential attenuation in strong rain affects measurements of the differential reflectivity (Z_DR). This causes Z_DR to be negative beyond strong precipitation.
• Velocity aliasing: Due to the discrete sampling of the phase shift there is a maximum frequency or Doppler velocity, which can be retrieved. This maximum unambigious Doppler velocity (v_N) is given by the pulse repetition frequency (PRF) and the wavelength of the radar. For POLDIRAD v_N can be as high as ±16.35 m/s. Higher velocities are folded into this intervall (also termed Nyquist intervall). Folding can be easily located where the Doppler velocity jumps from one extrem to the other. Once the "true" velocity is found velocity in folded regions can be determined by subtracting or adding twice v_N.
• Beam blockage: When there are obstacles close to the radar, the radar beam is blocked, and no data are available beyond that point. This affects measurements at low elevations, because there are a few hills in the vicinity of the radar. A problem with POLDIRAD is, that the antenna is not sheltered by a radome. In situations with strong wind from behind, the radar beam is pointed into the ground (e.g. image at 2026 UTC). Also, precipitation south of the Alpine main ridge is normally invisible.
• Failure of the polarization network: During the CLEOPATRA experiment temperature problems in the polarization switch caused occasionally erroneous measurements of Z_DR (differential reflectivity) and LDR (depolarization ratio). In this situations Z_DR will be constant and very high. In other situations a drift of LDR towards higher values can be observed. This is not visible easily, however, LDR should be about -35 dB in slight rain and not higher than -10 dB in the hail-shaft or in the melting layer. Before further processing of the data, LDR should be corrected to be within this range.

File names

Files are organized in subdirectories, their names are constructed from the data type and the type of scan as well as the original storm number. During the experiment different storm numbers have been assigned to identify the purpose of the type of scan. File names are constructed from the radar variable, the time and the angle of the scan (elevation for PPI scans, azimuth for RHI scans):

sssdddnn/vhhmmaaa.ras

Example: The file ppidop03/r1240020.ras is a PPI scan in Doppler mode (storm number 3). The RAS image shows the reflectivity pattern at 1240 UTC at an elevation of 2°.

Radar data earlier than 1100 UTC have not been included in the database. This was necessary to reduce the amount of data. Measurements in the morning have been performed to study clear-air echos. Storms #4,6,7, and 8 are data in a special time-series mode. Data files are very large, and can only be processed with special software.

Data Files

Data Format

Radar data are originally sampled in polar-coordinates defined by the range from the radar, the azimuth and the elevation angle. The raw data are also compressed in a special format designed by the radar manufacturer. Accessing the radar data would therefore require portable special software for reading and displaying.

As an alternative the radar data have been interpolated to a cartesian grid and stored in a common image format (RAS). RAS images are uncompressed, this allows easy access to the value of destinct pixels as well as displaying the whole image with standard tools.

The size of the image pixels was choosen to be the same size of the rader range bins. Only if this would require images larger than 600 by 600 pixels, the size of the pixels was doubled. Interpolation was performed using a nearest-neighbour-technique for Doppler velocities, and averaging the radar sample volumes for reflectivities and spectral width.

The coordinates of the image (PPI) are defined by the dimension (in kilometres) in the x-direction (west-east) and y-direction (south-north) with the radar at the origin. xmin, xmax, ymin and ymax specify the four corners of the image. For a RHI scan y is the height above the radar.

In order to access the value of destinct pixels additional information on the dimension of the image and scaling of the pixel value is needed. RAS images do not provide space for additional information blocks; therefore a small part of the colour table was for this purpose. The RAS variable ras_maplength gives the total number of colours defined. Colour indices start with zero. Colour 0 is used to define the background colour of the image, colour 5 is used to identify areas where no usable radardata were collected. Colours 6 to (ras_maplength/3 – 1) are used for the radardata and equally distributed over the range of data defined by fmin and fmax. Colours 1 through 4 are used to identify the scaling of the image. According to the definition of the RAS format the intensities for the three RGB colours are stored separately, first all intensities of red, then all intensities of green and finally all intensities of blue. Intensities are 8 bit (= 1 byte) deep, ranging from 0 to 255.

For the scaling information 2 bytes have to be interpreted as one 16 bit integer word (short integer or integer*2). Distances are in kilometres, fmin and fmax are scaled by 100.

Example:

RAS header (hexadecimal):
59 A6 6A 95 00 00 01 86 00 00 01 AA 00 00 00 08
00 02 88 FC 00 00 00 01 00 00 00 01 00 00 02 6D

Colour table (hexadecimal):
red: 80 FF 35 FF 8D DC 00 00 00 00 00 00 00 00 00 00 ...
green: 80 00 27 00 87 DC 05 0A 0F 14 19 1E 23 28 2D 33 ...
blue: 80 F8 30 1F 40 DC F9 F4 EF EA E5 E0 DB D6 D1 CC ...

The last 32 bit word of the header is the length of the colour table, here the table is 0000026D_hex or 621 bytes long. Therefore 207 colours are defined, 201 colours are used for the radardata, colour 0 and 5 are gray background colours.

xmin is byte 1 and 2 of red which reads FF35_hex or -203 for a 16 bit integer word.
xmax is FF8D_hex or -115 km,
ymin is 0027_hex or 39 km, and
ymax is 0087_hex or 135 km.
fmin is F830_hex or -20.00 considering the scaling factor 100.
fmax is 1F40_hex or 80.00.
The image shows the reflectivity, therefore values range from -20 to 80 dBZ with a resolution of 0.5 dBZ.

Notes: Modifying the colour table by any kind of image processing can result in destroying the information associated to colours 1 through 4. When retrieving this information into 16 bit words byte swaping by PC processors has to be considered. If numbers are larger than 32767, one has to subtract 65536 to get negative numbers.

Poldirad