Lessons Learned from the Real-Time Implementation of the Spectral Bin Classifier Surface Precipitation Type Algorithm

During the winter of 2018-2019, the CIMMS/NSSL-developed Spectral Bin Classifier (SBC) algorithm was deployed on the experimental vMRMS system at NSSL as part of its transition into the operational MRMS product suite. The SBC ingests data from the High-Resolution Rapid Refresh model, determines the initial hydrometeor phase at cloud top, tracks phase changes as the particle falls, and finally determines surface precipitation type based on the phase path of the particle. The SBC does this across a discretized size spectrum of particles, allowing the presence of mixes of various precipitation types.

This paper presents results from this first real-time, operational test of the SBC, and will discuss some of the challenges encountered and additions to the algorithm based on this testing. The rationale for and implementation of a simple classification scheme where the model does not simulate clouds will be discussed, as will the potential pitfalls of using radar echo tops to correct for model errors in relative humidity. The SBC also has the ability to be run using rimed particles instead of pristine particles. This riming factor was combined with a prototype MRMS mosaic of the WSR-88D Hydrometeor Classification Algorithm to assess the impact of using rimed particles where the WSR-88D classifies graupel. The impact of changing the SBC particle-size distribution will also be shown. The impacts of these changes will be used to illustrate how changes in the thermodynamic profile dominate these other factors in producing changes in surface precipitation type. Finally, the potential to derive probabilistic precipitation type guidance from SBC output is used to show its potential to mitigate uncertainty and horizontal or temporal heterogeneity in the input thermodynamic fields.