Bifurcation of a Summer Convective Thunderstorm over Beijing, Part 1: Observations and uWRF Simulations

This study investigates Beijing urban-area bifurcation of a summer evening convective thunderstorm (TS) precipitation event.  The case was observed during the Study of Urban Impacts on Rainfall and Fog/Haze (SURF) Project IOP on 22 July 2015.  The convective TS formed in associated with an 850 hPa synoptic vortex and a surface frontal zone.  Radar precipitation reflectivity showed urban bifurcation of the NW to SE moving storm.  Available AWS 2-m weather observations showed only a weak (0.8 K) UHI, and thus it was anticipated that building-barrier bifurcation would be the dominant urban impact on the TS.  The AWS observations also showed that one hour before the storms reached Beijing, surface winds over the city were from the SW and had bifurcated around the city.  When the storms reached Beijing, the surface winds were now from the NW, and also bifurcated around the city. 

The IUM urbanized WRF model (uWRF) simulation used the following options: multi-level urban scheme (BEP), but without the now available building energy scheme (BEM); Thompson microphysics; and Boulac PBL/TKE.  The ICs and lateral BCs were from the IUM BJ-RUC operational weather forecast model (with a 9 km grid), while uWRF was run on a nested 3 km grid.  The simulations started at 14 LST and ran for 10 hours.  Two simulations were carried out: (a) Urban, which used 2010 land use data and (b) No-Urban, which converted all urban Beijing grid cells into cropland, the dominant surface type surrounding Beijing. 

Results showed that uWRF correctly reproduced a weak UHI, but underestimated its magnitude by 0.5 K (because of the missing BEM module).  Both the pre-TS and storm bifurcating flow directions were correctly simulations, as was the strong urban induced deceleration, i.e., the no-urban case average speed of 4 m/s (over the missing city) was reduced to 2 m/s with the city.  This latter result compared well to observed value of 1.5 m/s.  The model, however, generally overestimated rural wind speeds by a factor of two to three.  While uWRF overestimated the peak bifurcating precipitation total accumulation around the city (60 vs. 20 mm), it did however correctly simulate its timing and location around the city, a first correct simulation of a bifurcating summer convective TS!

Additional analyses of WRF outputs were carried out to gain insights into the thermodynamic mechanisms leading to storm bifurcation.  Preliminary results and conclusions are discussed in the next paper.