Comparison of rain-profiling methods from ARMAR data in TOGA COARE with a view to a possible use with the TRMM radar

摘要

Rain measurements of the airborne Tropical Rainfall Measurement Mission (TRMM) radar simulator Airborne Rain-Mapping Radar in a typical event of Tropical Ocean and Global Atmosphere Coupled Ocean-Atmosphere Response Experiment are used to compare near-nadir rain-rate profiles retrieved from a set of deterministic attenuation-compensating algorithms. This set includes generic algorithms,such as the Hitschfeld-Bordan (HB) estimate and the surface reference (SR) methods constraining the total path-imegrated attenuation (PIA), and two hybrid algorithms of which one uses the same principle as the "standard" TRMM radar-profiling algorithm. In absence of reference rain data for validating the retrievals, the study is based upon an intercomparison of the results and analyzes their features in relation with theoreretical predictions. The rain data sequence is the same as used previously in a companion paper, which allowed the authors to get rain relations better adjusted to the observed rain system than those associated to a Marshall-Palmer (MP) drop size distribution. These two sets of rain relations are used here tO study subsequent changes in the various rain retrievals. How comparing the results among the generic and/or the hybrid algorithms may help identify the physical assumptions and error sources in the various methods is pointed out. With the MP relations, all methods provide almost similar retrievals.in stratiform light rain. In convective heavy rain. the retrievals are largely scattered; erroneous HB-derived PIA estimates are responsible for a downward collapsing effect in the HB and the TRMM-like radar algorithms, which prevents them from recovering a credible cell-like structure. With the adjusted relations, the rain estimates from all methods are generally increased and in much better agreement. A mirror image algorithm performing PIA profiling from the surface is also exploited. The direct estimate of the surface backscatter coefficient, obtainable below very light rain only, agrees with the value measured in clear air. The one-way PIAs, immune to errors in the radar calibration and the rain relations, are found to be well correlated with those derived from SR algorithms over a 5-dB range, provided that a bulk correction factor involving these two error types be adjusted in the latter case.