On the missing data problem in rass wind profiler measurements: an algorithm based on functional differential equations

RASS (Radio Acoustic Sounding System).is usu- ally an option added to Wind Profiler to provide profiles of virtual temperature data. Rass-Wind Profiler belongs to me- teorological radars. Virtual temperature is uncompensated for humidity or pressure. Recovered data must be processed in a given period and are accepted if only they belong to a consensus window, whose amplitude is in general 3 m/s and only for those data it is suitable to make an average. Many difficulties rise during Rass operating mode: first the un- changing of data with the increasing of height in some con- ditions and the data loss because of interruption. To recover data, we use Genetic Algorithms as prediction and retrieval technique. to profiles of virtual temperature data. The RASS system, composed of four acoustic sources, one on each side of the profiler, transmits a vertically directed acoustic wave. RASS (Radio Acoustic Sounding System) uses acoustic wave emis- sion to measure atmosphere virtual temperature profiles. Vir- tual temperature is that one must have dry air to make equal humid air density at the same pressure. This variable is widely used because it is possible to study variations of vir- tual temperature instead of density ones. Thus we define: Tv = T(1 + 0.61Q) where T is absolute temperature and Q is specific humidity given by the ratio between water vapour mass and humid air mass containing water vapour (Q = Mw/(MW + Md)). Thus, a Wind Profiling Radar is a meteorological and atmospheric remote sensing instrument, gives information about a volume of the atmosphere at a distance without being physically located in the region. The profiler uses the acoustic wave as a target, receiving and processing the resulting backscatter and measuring the speed of propagation. The profiler can compute virtual tem- perature profiles because the speed of sound is easily related to air temperature (2). Raw temperature data are stored in the moment and spectral data files, but separated from wind data into consensus files. Table 1 shows an example of data file to obtain temperature profiles. The titles of each column are: HT height in km, T uncorrected temperature, Tc corrected temperature, W vertical component of wind, CNT consensus counting, and SNR signal-to-noise ratio. We see that there is no changing of T and Tc corrected with the increasing of height and the number of "good" data CNT decreases with height as well as SNR. The difficulty of recovering data af- ter a given height associated with the choice of "good" data, represents a limitation and is a source of data errors.