Adaptive reconstruction of radar reflectivity in clutter-contaminated areas by accounting for the space-time variability

Abstract

Identification and elimination of clutter is necessary for ensuring data quality in radar Quantitative Precipitation Estimates (QPE). For uncorrected scanning reflectivity after signal processing, the removed areas have been often reconstructed by horizontal interpolation, extrapolation of non-contaminated PPIs aloft, or combining both based on a classification of the precipitation type. We present a general reconstruction method based on the interpolation of clutter-free observations. The method adapts to the type of precipitation by considering the spatial and temporal variability of the field provided by the multi-dimensional semivariogram. Six different formulations have been tested to analyze the gain introduced by each source of information: (1) horizontal interpolation, (2) vertical extrapolation, (3) extrapolation of past observations, (4) volumetric reconstruction, (5) horizontal and temporal reconstruction, and (6) volumetric and temporal reconstruction. The evaluation of the reconstructed fields obtained with the 6 formulations has been done (i) over clutter-free areas by comparison with the originally observed values, and (ii) over the real clutter-contaminated areas by comparison with the rainfall accumulations from a raingauge network. The results for 24 analyzed events (with a variety of convective and widespread cases) suggest that the contribution of extrapolation of past observations is not fundamental, and that the volumetric reconstruction is the one that overall adapted the best to the different situations.