This event is a good example of why these "precip-type" radar maps are often flawed. The indicated precip type is not based purely on radar data; they add supplemental model data to determine the likeliest precip type at the location of each radar observation. The dual-polarization observations, however, show the sleet/snow line from southern Carroll County northeastward into PA, where reflectivity (ZH) and differential reflectivity (ZDR) values are enhanced due to bright banding, and correlation coefficient is reduced due to a mix of melting particles with different shapes. SImilarly, the rain/freezing rain transition is where the correlation coefficient increases south of Anne Arundel County, with both ZDR and ZH values decreasing here as raindrops dominate the backscatter.
One other thing to note is the higher correlation coefficient and lower ZH and ZDR closer to the radar site in northern VA. The lack of a melting signature here is because the radar beam is closer to the ground at these locations and thus propagates beneath the melting layer; once the radar beam is higher off the ground at locations farther from the radar site, the polarimetric melting layer signature is visible. When the beam gets high enough, it samples above the melting layer and the signature disappears, as correlation coefficient becomes close to 1 and ZH and ZDR values decrease.