Diagnosing non-bright band precipitation in Met Office radar data using the linear depolarisation ratio

A major source of errors in radar-derived quantitative precipitation estimates (QPEs) is the inhomogeneous nature of the atmospheric vertical reflectivity profile (VPR). In areas where a beam samples close to the freezing level, the difference between the radar measurement and surface reflectivity can exceed an order of magnitude. Correcting for VPR is therefore an important step in obtaining accurate QPEs.

Different National Meteorological Services (NMSs) correct for VPR in a variety of ways. In most operational contexts a “global” VPR is determined by large scale spatio-temporal averaging of radar data, from hourly timescales up to periods of several months. These average profiles are then used to correct real-time measurements at each point on the radar domain.

Given the dominance of stratiform conditions, particularly at high latitudes, average VPRs tend to display a clear bright band. In convective conditions, however, bright bands are less intense or non-existent. In these cases, application of a climatological correction in the bright band region causes surface precipitation to be underestimated. This artificial reduction of the often intense rainfall measurements associated with convection is significant in an operational context, for flood forecasting, monitoring and impacts management.

The Met Office operational system uses an alternative, non-bright band VPR to correct reflectivity measurements in convective regions. A lack of bright band is diagnosed indirectly, via a minimum reflectivity threshold above the melting layer. This criterion requires reflectivity measurements at high levels, which are not always available even when the lowest scan is in the bright band region.

In this work we investigate linear depolarisation ratio (LDR) as diagnostic of bright band and non-bright band conditions, using range height indicator scans from the Met Office C-band research radar at Wardon Hill. The advantages of peak LDR as a bright band diagnostic are twofold. The UK operational scan strategy includes two LDR-mode PPI scans at the lowest two elevations, which means that a “lowest usable scan” reflectivity can be evaluated for bright band using the LDR from the same physical location. The second advantage is the low sensitivity of LDR peak values to broadening of the radar beam with range.

The large melting snowflakes associated with strong reflectivity bright bands also cause a sharp increase in LDR in the melting layer. We show that LDR peak values of -11dB to -15dB are typical in these cases. The higher density ice phase hydrometeors associated with convective conditions are both smaller and more spherical than snowflakes, and do not consistently produce strong LDR peaks. Smaller peaks can therefore be used to flag data that should not be corrected for bright band.