Read on Twitter
Preparing a geeky post-mortem radar analysis for @NWSKeyWest — some initial thoughts. Matching up radar w/ two distinct periods of heavy rain in #KeyWest yesterday, btwn 4-5PM & more significantly 9-10PM. Dual-pol instantaneous rates tell the whole story, 3-4" & 4-6" respectively
These are staggering instantaneous hourly rates - the latter of which was nearly stationary over the island for 30-45 min, resulting in an unbelievable ~4" of rain btwn 9-10PM at the EYW ASOS (possibly underestimated, as KEY WFO totaled a whopping 1.99" higher than the EYW total)
The radar determines these instantaneous rates from dual-pol data, the most important product being specific differential phase (or KDP). This shows where significant attenuation of the horizontal beam is occurring, revealing extremely high concentrations of liquid water.
Traditional horizontal reflectivity provides clearer evolution of #TD19/ #TropicalStormSally in this loop via #KeyWest radar, but the KDP loop above is better at pinpointing areas of heavy rainfall (especially true in warm core tropical systems as ice doesn't reach the low-levels)
KDP does a better job than just Zh b/c it's based on the cross correlation of the H AND V channels. CC is determined similarly, the dual pol variables are ALL related. Without the V attenuation for H to compare to, you can't have KDP. CC must also be high, as it's a quality check
Point I'm trying make is that dual pol is so much more than just detecting non-meteorological echoes. It's tough to understand deeply, but it's worth making use of ZDR & KDP in particular, because they provide quantitative info that relates to qualitative aspects of hydrometeors.
Dual pol quantitative precip estimates (and their MRMS counterparts) did a great job in estimating precip totals yesterday as a result of these fundamental relationships. We even know when ZDR & KDP signatures are "separated" that size sorting of hydrometeors is occuring.
This relates to higher SRH and is correlated well with tornadogenesis, a particularly useful signature when it comes to tropical cyclone tornadoes which are notoriously difficult to detect and predict with Z and V base moments alone. It's incredible to me what dual pol can do.
Anyway I'd like to inspire more operational meteorologists to leverage this amazing tool in real time during ops, because the AI systems that eventually replace us will definitely be doing so. Let's understand why these algorithms are so powerful and will only become more so.
