Real of fake? Follow-ups on signatures observed on US radar composites

At the last conference, we presented some results from the analysis of 16+ years of US radar composites and showcased the value of such analysis to investigate new or unexpected signatures. Three unexpected results were presented: a possible convection minimum in summer afternoons in the Mississippi Valley, a signature of the effect of cities on convection, and a possible difference between weekday and weekend precipitation maximized downwind of the industrial North East. Additional work was performed on these three unexpected results to determine their basis and their statistical significance: were these signatures real, or are we the victim of limited data sampling?

* Mississippi convection minimum: Real. The data signature is very clear, and numerical simulations show that the weak orography around the valley encourages leads to reduced stability and triggers convection preferably in these areas compared to the valley itself.

* Signature of cities on convection: Probably real. While it is very difficult to identify for certain the signature of individual cities and the mechanism for their impact, the combination of data from 12 cities show a weak but definite fingerprint on both radar and lightning data over a ~25*20 km area slightly downwind of the city, compatible with its expected local impact. Significance calculations are however difficult to make.

* Weekday-weekend precipitation difference: Probably fake. Though the 16 years data shows a ~15% difference between the probability of precipitation on a weekday than on a weekend, data resampling experiments suggest that such large differences can be obtained by natural variability and are not yet significant enough to be declared a reality.