A methodology is developed that objectively identifies areas of lower than expected precipitation estimates in regions within a radar network that are affected by beam blockage using spatiotemporal analysis of the radar data. This method of detection of beam blockage is novel because it relies only on a long-term precipitation climatology and does not require any knowledge of topography or known obstructions. At each radar, precipitation estimates are normalized by climatology and at several range intervals, Fourier series fits determine an expected precipitation value dependent on azimuth angle. Beam blockage is identified as radially coherent regions with normalized values that are significantly lower than the Fourier series function. Precipitation estimates sufficiently affected by beam blockage are replaced by values estimated using neighboring unblocked estimates.
The methodology is applied to the correction of Stage IV precipitation estimates originating from WSR-88Ds in the central and eastern United States. However, the methodology is flexible enough to be useful for most radar installations and geographical regions with at least a few years of data.