Numerical simulation of a prolonged rainstorm in Hong Kong on 22 - - PowerPoint PPT Presentation

Numerical simulation of a prolonged rainstorm in Hong Kong on 22 July 2015

Ira, Chan Yan-chun Wong Wai-kin 27 July 2016, WSN-16

 Prolonged intense rainfall on 22 July 2015.  Amber Rainstorm Warning lasted for 11 hours 55 minutes,

the longest record in the past decade.

 Daily rainfall 191.3 mm at

HKO, highest among 2015.

 More than 1000

cloud to ground lightning

• ver the territory.

 Waterspout was reported

near Kau Yi Chau.

Background

Hong Kong >30mm/hr rainfall

• ver widespread area

Daily rainfall map on 22 July 2015

 Uneven rainfall distribution  Rainfall confined over southern part of Hong Kong.  Weather relatively stable over northwestern part of Hong Kong,

with much less rainfall.

Summarizing contributed factors in order to help forecaster to predict intense precipitation

Experiment of high resolution numerical model - WRF

Radar horizontal resolution Model physics - Cloud microphysics and boundary layer scheme Orographic effect experiment

Performance of forecasting models

Global model Meso-scale model Short-term forecasting products

Synoptic and meso-scale analysis

Investigation of cause and mechanism contributed to rainstorms

Research direction

A B C

 Intense thundery showers moved

from SW to NE crossing HK and brought heavy showers in the morning.

 Mid level winds changing

gradually from SW to W/NW. Showers easing off in the afternoon with more stable atmosphere.

 Thundery showers advanced to

the S of HK. Developments persisted on a meso-scale convergence line over the coastal water of Guangdong in the afternoon and at night.

Background information on 21 July (1 day before)

A broad trough of low pressure moved N, towards inland Guangdong progressively Active southwesterlies continued to affect the south China coastal areas. Convective clouds developing over south China coastal areas

2015-07-21 00:00 UTC 2015-07-21 12:00 UTC 2015-07-22 00:00 UTC

500 hPa 700 hPa

A A’ A A’ A A’

Upper-air anticyclone (A) stayed around western Guangdong. Coupling with another anticyclone (A’) over the Luzon Strait, the westerly trough moved E slowly and HK was located near the tail of the 500hPa westerly trough. Anticyclone (A’) edged W into the northern part of the South China Sea gradually, making the westerly trough stayed between two anticyclones, favorable for development and unsettled weather.

Radar situation on 22 July

Waterspout near Kau Yi Chau at around 8am on 22 July

Kau Kau Yi Chau Yi Chau During the time with intense echoes, positive vorticity observed from surface winds, matching with the location of the reported waterspout. Analysis of wind field retrieved from dual Doppler radar radial winds even showed local-scale vortex.

Favorable conditions in meso-scale

Based on 2015.07.21 00Z Based on 2015.07.20 12Z Based on 2015.07.20 00Z

Model forecast the anticyclone at 500hPa over western Guangdong would weaken, and the mid level westerly trough would continue to move E. The westerly trough would be E of HK at 18Z 21 July. Streamline at mid level even showed slight ridging flow. As such, the forecast rainfall area confined over the eastern coast of Guangdong, and less rainfall in HK. In fact, the forecast position of the westerly trough was more E than the actual.

Model performance (ECMWF)

Based on 2015.07.20 12Z Based on 2015.07.20 00Z Based on 2015.07.21 00Z

With more later model runs, model gradually corrected the weakening trend of the anticyclone over western Guangdong and the moving speed of the westerly trough. Forecast of 00UTC 21 July showed that the tail of the westerly trough close to HK, but the strength and the intensity of rainfall were still relatively weak.

Model performance (NCEP)

Based on 2015.07.20 12Z Based on 2015.07.20 00Z Based on 2015.07.21 00Z

Model run of 00UTC 21 July forecast better the evolution of the anticyclone over the western part of Guangdong. Comparing with the model forecast of 12UTC 20 July, the weakening speed of the anticyclone was slower, which is closer to the actual

• situation. However, the short

wave feature was unstable between model runs and conclusion for heavy rain was not confident.

Model performance (JMA)

The forecast rainfall area by Meso-NHM is close to ECMWF and NCEP.

HK

Model performance (meso-NHM / rapids-NHM)

Using Meso-NHM as boundary condition, RAPIDS-NHM has a finer resolution. The assimilation included

• bservation such as radar data

and AWS from HK and

• Guangdong. There were some

members capturing development of rainband over the southern water of HK. However, intense rainfall still located E of HK in earlier runs. The intensity was also lower than actual and persisted in shorter time.

Short-term forecast performance (SWIRLS)

WRF parameter Version 3.4.1 Horizontal resolution 2km Grid point 305×305 Vertical layer 51 layers Upper level 21 km Surface to 2km 11 layers

 Purpose  Aim to reproduce the rainfall event  Understand the related dynamical and physical process

Use forecast from ECMWF 12UTC 20 July as boundary condition Downscale to 2km resolution for initial condition 3DVAR – Assimilation including GTS, AWS, radar and wind profiler. Run the model forecast

High resolution numerical model experiment - WRF

WDM-6 WSM-6

Res esoluti tion o

• f

radar ar d data: a: 4 k 4 km Res esoluti tion o

• f

radar ar d data: a: 2 k 2 km

Comparison of 24-hour rainfall amount between models and estimation from radar (QPE)

For model forecast using 4km resolution radar data, low level moisture flux convergence showed more N-S fluctuation near southern waters

• f HK with time. The

intense rainfall region (red region >100mm) covered larger area than using 2km resolution radar data.

Effect of horizontal resolution of radar data and cloud microphysics

Clou

• ud mic

microphys ysic ics p proc

• cess

WDM-6 +YSU PBL WSM-6 +ACM-2 PBL WSM-6 +YSU PBL

Effect of cloud microphysics and boundary layer physics

WDM-6 considered the changing in concentration of various

• hydrometeors. The represented

cloud microphysics is more

• complete. There is positive effect

for simulating cloud physics and triggering development of

• convection. It is even more

effective to simulate the rainband

• ver coastal waters in early

development and also the distribution from W to E.

Development of meso-scale vortex

WSM-6 +YSU PBL WSM-6 +ACM-2 PBL WSM-6 +ACM-2 PBL modif ifie ied t terrain in

B A C X Y Z

Terrain height Modified terrain height

WSM WSM-6 + 6 + ACM ACM-2 2 PBL PBL WSM WSM-6 + 6 + ACM ACM-2 PBL PBL modi

• difie

ied t d terra rrain in

w > 0 q*V v > 0 S X Y X Y X Y X Y

Orographic effect

In the modified terrain experiment, the northerly component at around 3 to 4 km is more S than the original terrain. It also represented more S of the associated mid level westerly trough tail.

modi

• difie

ied t d terra rain

• rig
• rigin

inal t l terr rrain in T + 0 min T + 0 min T + 270 min T + 270 min

X Y X Y X Y X Y

Comparing with the evolution of the vertical cross section from initial time, the initial field of the moisture transport with

• riginal terrain matched with the intense echoes over southern

waters of HK. Conversely, under the situation without local terrain, the moisture transport was more evenly distributed.

X Y

v > 0 v > 0

modif modified d te terrai ain

• rigi
• riginal

te terrai ain

B A X Y

WS WSM-6 + + ACM ACM-2 PBL WS WSM-6 + + ACM ACM-2 PBL modif modified d terrain in

 Performance of models 

Global model could not give a good indication of heavy rain, but meso-scale model with high enough resolution could give some signature.



Models seem to overestimate the weakening trend of the anticyclone and could not well predict the quasi-stationary feature of the westerly trough.

 WRF Model 

From the sensitivity test discussed, rainfall pattern is more sensitive to the terrain height.



Relatively less sensitive to model physics, but the result is more matching with the radar

• bservation using the scheme of WSM-6 and ACM-2.

 Overall mechanism 

The axis of the quasi-stationary westerly trough stayed around HK. Coupling with various favorable factors such as low level moisture flux convergence and jets and enhanced by

• rographic effect, thundery showers continued to develop S of HK, leading to prolonged

intense rainfall.

 Hindsight 

Bear in mind when heavy rain is highly sensitive to the location of system.



Nowcasting technique and meso-scale model could give an earlier indication to rainstorm.

Conclusion and outlook

Thank you!!

• Dr. Tin

HKO’s Mascot