Stochastic simulation experiment to assess radar rainfall retrieval uncertainties associated with attenuation and its correction

摘要

As rainfall constitutes the main source of water for the terrestrialhydrological processes, accurate and reliable measurement and prediction ofits spatial and temporal distribution over a wide range of scales is animportant goal for hydrology. We investigate the potential of ground-basedweather radar to provide such measurements through a theoretical analysis of someof the associated observation uncertainties. A stochastic model of range profilesof raindrop size distributions is employed in a Monte Carlo simulation experimentto investigate the rainfall retrieval uncertainties associated withweather radars operating at X-, C-, and S-band. We focus in particular on theerrors and uncertainties associated with rain-induced signal attenuation andits correction for incoherent, non-polarimetric, single-frequency, operationalweather radars. The performance of two attenuation correction schemes, the (forward)Hitschfeld-Bordan algorithm and the (backward) Marzoug-Amayenc algorithm, is analyzedfor both moderate (assuming a 50 km path length) and intense Mediterranean rainfall(for a 30 km path). A comparison shows that the backward correction algorithmis more stable and accurate than the forward algorithm (with a bias in the orderof a few percent for the former, compared to tens of percent for the latter), providedreliable estimates of the total path-integrated attenuation are available. Moreover,the bias and root mean square error associated with each algorithm are quantifiedas a function of path-averaged rain rate and distance from the radar in order to providea plausible order of magnitude for the uncertainty in radar-retrieved rain rates forhydrological applications.