ABLE 915 MHz Radar Wind Profiler with RASS
The ABLE 915 MHz radar wind profiler with RASS: Three of these profiler systems are deployed within the ABLE region. Profiler locations are Whitewater, Oxford, and Beaumont, KS.
Other components in the system include four stationary acoustic sources located at the corners of the antenna, a receiver, an interface module, and a computer for data analysis and processing.
Radial components of motion along each pointing direction are determined sequentially. It takes, nominally, 30 - 45 seconds (dwell time) to determine the radial components from a single pointing direction. Thus, at the Beaumont and Whitewater sites the system cycles through five beams (South, North, East, West, and vertical) at low power, and then cycles the five beams again at a high power (longer pulse length) setting. Then the whole process is repeated. About five minutes elapse before the system returns to the beginning of its sequence. Within an averaging interval, the estimates from each beam-power combination are saved (11-12 in a 1-hr period) and these values are examined and compared at the end of the period to determine the consensus-averaged radial components of motion. The Oxford system is slightly different. When operating in multipower mode it averages in the low power mode for 25 minutes, then in the high power mode for 25 minutes.
Briefly, consensus averaging consists of determining if a certain percentage (e.g., 50%) of the values fall within a certain range of each other (e.g., 2 m/s). If they do, the median of those values is used to produce the radial wind estimate. The radial values are then combined to produce the wind profile. The results of this averaging process are what are reported in the ".cnw" data files produced by the ABLE data system. Included in these files are height, speed, direction, radial components, # values in consensus, and signal-to-noise ratio (SNR).
During a single time period during which the system operates in a single pointing direction (dwell time), the data that is produced in the ".mod" and ".spd" files is created. The system transmits pulses at about 1-10 kHz rate into the atmosphere. The backscatter from each transmit pulse is sampled at, for example, a 1 MHz rate. This results in 1 sample every 150 m in range. The samples at each range gate are averaged together (time domain integration) over some number (e.g., 100) of pulses to produce a phase value for input into a FFT. After (e.g., 64) values are produced, the FFT is performed (one for each range gate). This process takes on the order of 1 sec. A number (about 30) of these spectra are then averaged together during the dwell time. At the end of the dwell time we have produced a single averaged spectrum from each range gate along the designated pointing direction. The spectra themselves are placed in the ".spd" data files.
The spectra are analysed by the system before moving to the next pointing direction. This analysis produces estimates of the snr, the noise, the mean velocity (proportional to frequency), and the first moment (spectral width) at each range gate. This is the information that is stored in the ".mod" data files. Both the ".mod" and ".spd" data files thus have information at about (dwell time) intervals; however, the data sequences among pointing directions and output powers.
A note of warning about the mean values in the ".mod" files. The values are in % of full scale times 100, where full scale is the nyquist velocity of the spectrum. Thus, velocity estimates are determined by multiplying the "mdf" column times the full-scale bandwidth and dividing by 10,000.
RASS operation is essentially the same, except that the averaging time is about 10 minutes and only a single pointing direction (vertical) is used. Also, the atmosphere is "seeded" with a sound wave; the index of refraction changes created by the sound wave are the signal source. In order to sample both the sound wave (speed about 340 m/s) and the atmosphere (to remove air velocity from temperature estimates) a larger FFT is required (2048 points). This requires a smaller number of points for each time domain integration and increases the processor time required to calculate the FFT. The ".spd" files again are spectra; however, only a portion of the spectra are reported, namely a region near 0 Doppler shift to account for atmospheric motions and a region around the expected speed of sound. The ".mod" files now consist of moments and widths from both the atmospheric portion of the spectrum and from the acoustic portion (the main contributor to the temperature calculation). The ".cnt" files consist of profiles of temperature and number of consensus values.
In normal operation, temperature profiles are determined during the first 10 minutes of every hour and the wind profile is averaged over the remaining 50 minutes. To hear a sound clip of the RASS acoustic signal click here (264KB).
Return to ABLE Site Diagram
Return to ABLE Main Page