Rain attenuation predicted with a two‐layer rain model

The paper presents a new prediction model of rain attenuation in slant paths; it models the precipitation with two layers of constant vertical precipitation rate. The relationship between the rain‐rate in the first layer above the ground (layer A) and the apparent precipitation rate (defined in the paper) in the second layer (layer B, which models the melting layer) is theoretically derived: assuming a log‐normal probability distribution for the rain‐rate in layer A (measured at ground), the apparent rain‐rate in layer B is also log‐normal with the same standard deviation, but with its median value 3.134 times the median value of the ground rain‐rate. The rain‐rate statistical process in space is assumed to be also log‐normal with a correlation function determined from radar measurements. Assuming that the path attenuation (expressed in decibels) is also log‐normal, its average and standard deviation are then derived. The model is tested against the large CCIR data bank of slant path attenuation with or without concurrent rain‐rate, frequency ranges between 6 and 34.5 GHz, in the latitude range between 1.3 and 67.4 degrees. In the probability range 10−3‐10−3 the root mean square (RMS) log‐error of the Long Term data bank (single experiments of about 5 years) is about 26%, with average (<τ>) log‐error close to zero. For data banks of single experiments of shorter duration the RMS increases while <τ> decreases. The model is not significantly biased with attenuation, frequency of the experiment or latitude of the station. The distributions of the overall log‐error and that of the log‐error conditioned on given probabilities are both Gaussian. Sensitivity to different theoretical assumptions about the calculation of the specific attenuation kRα and to different numerical values of the snow density is also tested.