Rain retrieval algorithms used for the 13.8 GHz Precipitation Radar (PR) on board the TRMM satellite rely on the attenuation of the melting layer being estimated accurately. The attenuation not only affects the path-integrated attenuation estimated from the radar measurements but also the retrieved rainfall rates near the ground or sea surface. Several models have been proposed for predicting the attenuation in the melting layer. One such method is the Non-coalescence – Non-break up model (N N model), based on spherical melting particles made of dielectric composites. The model assumes the melting particle to be composed of a homogenous mixture of water, ice and air, and calculates its complex permittivity from Wiener’s theory. The calculation includes the parameter “form factor”, U, as well as the volume content of water, ice and air. The height variation of these parameters has been inferred fairly accurately in the past.

In this paper, we compare the predictions from the N-N melting layer model with ground based radar measurements at Ku-band. The measurements were made in February 1998, in Kashima, Japan, during a prolonged stratiform event. The radar used for these measurements operates at the same frequency as the TRMM-PR but has enhanced capability. It consists of a 2kW peak output power TWTA and a dual channel receiver for horizontal and vertical polarizations. The pulse width of 0.5 microseconds gives a range resolution of 75 meters and, moreover, over-sampling in range is possible so that the radar reflectivity can be derived at every 37.5 meters. Two data acquisition modes are available, namely, integral mode and all-hit mode. In the former case, the system records the radar reflectivity, the mean velocity and the spectrum width whilst in the latter, the pulse-by-pulse I/Q data are recorded, which enables the Doppler power spectrum to be derived. When operating in the vertically-pointing mode, the measured velocity spectra correspond to the fall velocity spectra of the hydrometeors.

The February 1998 measurements were taken in the vertically-pointing mode. The Doppler spectrum in the melting layer shows clearly the increase in the fall velocity and the broadening of the spectrum as the hydrometeors melt. The fall velocity spectra at specific heights were compared with the corresponding spectra extracted from the N-N melting layer model for the case of the best fitted rainfall rate. Results indicate that the bright-band thickness at Ku-band is a variable quantity that could be parameterized in terms of the rain intensity just below the melting layer, as reported previously by Fabry and Zawadski (1995). By incorporating an additional relationship between the melting rate and the particle diameter, as considered in some other melting layer models, it may be possible to improve the N-N model predictions. This option will be considered in the paper.