RADAR ESTIMATES OF RAINFALL RATES AT THE GROUND IN BRIGHT BAND AND NON-BRIGHT BAND EVENTS

The major error in estimating precipitation rates at the ground from r adar arises from the change in radar reflectivity (Z) with height. In stratiform precipitation Z rises in the 'bright band', where the snow is melting, and then falls steeply in the dry ice above. In regions of vigorous convection containing graupel, the bright band is absent and the fall in Z with height is less marked. If accurate estimates of su rface precipitation are to be made, based upon radar observations take n within or above the bright band, it is important to use a correction scheme which uses a different vertical profile of reflectivity (VPR) for the two precipitation types. Using a large radar data set, this wo rk addresses the issue of VPR variability by constructing average vert ical profiles and sorting them into snow and graupel categories using the value of the Linear Depolarization Ratio at the melting layer. The se observations show that convection is often embedded within stratifo rm precipitation, and some convective showers containing snow have bri ght bands. As a result, the use of an average profile over the whole r egion scanned by radar leads to only marginal improvement in rain esti mates. Instead, this work proposes that a bright band correction shoul d always be applied unless the value of Z at 1.5 km above the bright b and exceeds 30 dBZ, in which case a 'convective' vertical profile of - 2 dBZ km(-1) should be used. This Z threshold indicates that graupel o nly forms when the ice water content exceeds 0.35 gm(-3) which has imp lications for ice parametrization schemes in mesoscale models.