A class of single- and dual-frequency algorithms for rain-rate profiling from a spaceborne radar .2. Tests from airborne radar measurements

In Part I, four single-frequency (SF) algorithms and a dual-frequency (DF) algorithm for range profiling of the rain rate from a spaceborne radar were described and tested from numerical simulations. In Part II, performances of these algorithms are studied using data from a DF (X and K-a bands) near-nadir-pointing airborne radar. The data, gathered over ocean near Wallops Island in 1988, mimic future spaceborne radar measurements. Rain retrievals are performed for isolated and series of contiguous rain measurements within stratiform and convective rain regions overflown by the aircraft. General features and aspects specific to the experiment conditions are pointed out. The SF algorithms provide more or less scattered results according to their own sensitivities to uncorrected scaling errors. Improvement of the algorithms stability by constraining the total path-integrated attenuation from surface echo measurements is confirmed. The correlation between attenuation coefficients at both frequencies, which forms the theoretical basis of the DF algorithm, is verified from the data. Results from DF algorithm are likely more reliable than SF counterparts since they are globally corrected for scaling errors. Rain-rate thresholds above which ''attenuation'' algorithms should relay Z-R methods to avoid negative bias in rain-rate estimates are found near 12 mm h(-1) at X band and 1 mm h(-1) at K-a band for a 2.5-km rain depth. Coherent spatial structures of the rain rate within a vertical cross section of the storm are recovered from the ''attenuation'' algorithms. The data obtained with a cross-range resolution L(0) approximate to 1 km are also used to perform 2D simulations of beam-averaging, and nonuniform beam-filling (NUBF) effects are used for cross-range resolutions L = 2, 3, and 4 km. Degrading the resolution produces a smoothing of small-scale rain-rate structures and a lowering of the rain-rate estimation. Bias due to NUBF depend on the involved ''attenuation'' algorithm. It increases with L but remains below 15%, up to L = 4 km for mean rain-rate estimates at large horizontal scale (approximate to 100 km). The Z-R methods are weakly affected by the NUBF.