Simulation of heavy rainfall event during 30/10 - 01/11/2008 over hanoi by wrf model

Kỷ yếu Hội nghị: Nghiên cứu cơ bản trong “Khoa học Trái đất và Môi trường” DOI: 10.15625/vap.2019.000136 289 SIMULATION OF HEAVY RAINFALL EVENT DURING 30/10 - 01/11/2008 OVER HANOI BY WRF MODEL Nguyen Van Hiep * , Nguyen Xuan Anh, Nguyen Duc Nam, Dang Hong Nhu, Nguyen Tien Manh, Pham Le Khuong, Pham Xuan Thanh, Hoang Hai Son Institute of Geophysics (IGP), Vietnam Academy of Science and Technology (VAST) * Email: hiepwork@gmail.com ABSTRACT In urban areas, flooding can be very serious due to high population density and important infrastructure. As rainfall data at high temporal and spatial resolutions are required as input for urban flooding forecast, in this study, the Weather Research and Forecasting model (WRF) with four nests, two-way interaction is used to simulate the heavy rainfall event in Hanoi from 30/10/2008 - 01/11/2008 to investigate model ability on generating rainfall data on high temporal and spatial resolutions. The horizontal resolutions of the four nets are 27 km, 9 km, 3 km and 1 km. Two model experiments are performed including: (1) model inputs are poorly from reanalysis data only (CTRL) and (2) model input in CTRL is modified with JMA satellite radiance data in deep convective cloud regions (WSAT). The results showed that without satellite data assimilation, event with initial conditions from the best available analysis data, the WRF failed to capture observed heavy rainfall in many cases. With enhancement of initial conditions by satellite data, although there are some overestimated rainfall, the model can reasonably reproduce observed heavy rainfall amount over Hanoi region. Keyword: WRF model, heavy rainfall, satellite data. 1. INTRODUCTION Heavy rainfall is one of the main causes of flooding in urban areas in Vietnam. In urban areas, flooding can be very serious due to high population density and important infrastructure. Real-time flood forecasts are based on rainfall input data, so the accuracy of flood forecast is limited by precipitation forecast accuracy. Recently, there has been an increase in number and severity of urban flooding events causing serious losses to human and property lives worldwide (Bisht et al., 2016; Sandink, 2016; Smith et al., 2016; Zope et al., 2016). These losses can be minimized by improving the forecast quality of heavy rainfall that often causes flooding (Smith et al., 2016). Shortening lead time for flooding forecast is necessarily to protect people and property. The best way to reduce the forecast lead time and increase quality of the hydrological model forecast is improving the quantitative rainfall forecast. Javier et al. (2007) performed flood forecasting for the metropolitan area in Baltimore using a hydrological modelling and rainfall estimation using the high-resolution Doppler (WSR-88D) High-resolution Weather Surveillance Radar (Javier et al. 2007) suggested that better input rainfall is important for flood forecasting. The surface heating in urban areas can change the wind field and low-level temperature, humidity to enhance convection and cause heavy rainfall (Ryu et al., 2015). Hardy et al. suggested that improving high-resolution quantitative precipitation forecasts (QPFs) is important for flood forecasting (Hardy et al., 2016). As rainfall data at high temporal and spatial resolutions are required as input for urban flooding forecast, in this study, the Weather Research and Forecasting model (WRF) is used to simulate the heavy rainfall event in Hanoi from 30/10/2008 - 01/11/2008 to investigate model ability on generating rainfall data on high temporal and spatial resolutions. Hồ Chí Minh, tháng 11 năm 2019 290 2. METHOD WRF model was developed with the cooperation of the Atmospheric Research Division of the National Center for Atmospheric Research (NCAR/MMM), National Center for Environmental Prediction in the Oceanic Atmosphere Research Institute, USA (NOAA/NCEP). The WRF model is widely used for both operational and research purposes in weather forecasting, climate prediction as well as climate projection in research in many countries around the world. The model configuration to simulate the heavy rainfall in Hanoi includes four nests with two- way interaction. The horizontal resolutions of the four nets are 27km, 9km, 3km and 1km, with the number of grid points of 174 × 156, 331 × 325, 379 × 304, 271 × 238, respectively. The number of vertical levels is 38 for all nests (Fig. 1). The physical parameter options include: Thompson microphysics scheme (Thompson et al, 2008), Betts-Miller-Janjic convective parameterization scheme (Janjic, 1994), RRTM longwave radiation scheme (Mlawer, 1997), Duhia short-wave radiation scheme (Dudhia, 1989), Yonsei University planetary boundary layer scheme (Hong et al, 2006). Figure 1. The four nest domains employed in the research The heavy rainfall event caused severe flooding in the Hanoi capital from 30/10/2008 - 01/11/2008 was selected to simulate in this work. Two model experiments are performed including: (1) model inputs are poorly from reanalysis data only (CTRL) and (2) model input in CTRL is modified with JMA satellite radiance data in deep convective cloud regions (WSAT). 2. RESULTS AND DISCUSSION 2.1. Evolutions and losses From 30/10/2008, in the North and North Central provinces of Vietnam experienced a record heavy rainfall in more than 100 years. Heavy rainfall is observed in a range of 100-500mm. In some places, rainfall is over 700mm. The heavy rainfall event caused widespread flooding. Both the inner city and the suburbs are flooded up to 2-3m deep, causing severe damage to winter crops. Many residential areas, irrigation system were flooded affecting production, business, services, infrastructure works, electricity, traffic jams at many street hot spots (Vietnam Annual Hydrometeorology Report, 2008). 2.2. Simulation results Figure 2 shows that on 31st Oct both CTRL (Fig. 2, b) and WSAT (Fig. 2, c) runs reflect the trend of shifting the heavy rainfall areas to the North as in station observation and in the GSMaP estimation (Fig. 2, a). In detail, the CTRL run significantly underestimate rainfall at the heavy Kỷ yếu Hội nghị: Nghiên cứu cơ bản trong “Khoa học Trái đất và Môi trường” 291 rainfall region. GSMaP slightly underestimates rainfall in the southern stations, whereas the SWAT run overestimates rainfall in the Northern region. In comparison with the CTRL run, the SWAT run shows advantages of simulating heavy. In the SWAT run, simulated rainfall over Hanoi is almost equivalent to that in GSMAP. The values are in the range of 75-300mm, which is slightly lower than observed rainfall at raingauge stations (Fig. 2). It needs to note that the GSMaP rainfall is estimated from cloud radiance, in some cases, the GSMaP (satellite estimated) rainfall is significantly deferent from raingauge station data. Although GSMaP quality is not as good as raingauge station data, it is a good reference source of observed rainfall on grid. Overall, with the advanced technique of satellite data assimilation into high resolution WRF model, the SWAT run can reasonably simulate heavy rainfall regions during late October to early November, 2008. One of the most important advantages of the SWAT simulation to GSMaP and station data is that it can generate high temporal resolution (10 minute to hourly) data on grid which are crucial input by urban flooding model. (a) (b) (c) Fig. 2. 24-h accumulated rainfall ending at 12:00UTC 31/10/2008 for GSMAP estimation (a), CTRL (b), and WSAT (c), The circular dots indicate raingauge stations. Color scale at raingauge stations is identical to that of the grid rainfall. 3. SUMMARY AND DISCUSSION The Weather Research and Forecasting model (WRF) with four nests, two-way interaction is used to simulate the heavy rainfall event in Hanoi from 30/10 - 01/11/2008 to investigate model ability on generating rainfall data at high temporal and spatial resolutions. The horizontal resolutions of the four nets are 27km, 9km, 3km and 1km. Two model experiments were performed including with and without satellite data. Without satellite data, event with initial conditions from the best available analysis data, the WRF failed to capture observed heavy rainfall in many cases. With enhancement of initial conditions by satellite data, although there are some overestimated rainfall, the model can reasonably well reproduce observed heavy rainfall amount and location in Hanoi region. Data from with satellite data run with temporal resolution of 10 munutes to one hour, spatial resolution of 1-km can be used as input data for urban flooding model to simulate significant flood events in Hanoi. Biển Đông Biển Đông Biển Đông Hồ Chí Minh, tháng 11 năm 2019 292 REFERENCES [1]. Bisht D. S., Chatterjee C., Kalakoti S., Upadhyay P., Sahoo M., Panda A., 2016: Modeling urban floods and drainage using SWMM and MIKE URBAN: a case study. Nat. Hazards, 84, pp 749-776. [2]. Javier, J. R. N et al., 2007: Flash Flood Forecasting for Small Urban Watersheds in the Baltimore Metropolitan Area, Weather and Forecasting, 22(6), pp.1331-1344. [3]. Janjic, Z. I., 1994: The step-mountain eta coordinate model: Further development of the convection, viscous sublayer, and turbulence closure schemes. Mon. Wea. Rev., 122, 927–945. [4]. Hardy, J. et al., 2016: A method for probabilistic flash flood Forecasting. Journal of Hydrology, 541, pp. 480-494. [5]. Sandink, D., 2016: Urban and land-related flooding homes in Canada: an overview. Flood Risk Manage. 9 (3), pp. 208-223. [6]. Smith B. K., J. Smith, M. L. Baeck, 2016: Flash flood-producing storms in a small urban watershed. J. Hydrometeorol. 17, pp. 2631-2647. [7]. Thompson, G., P. R. Field, R. M. Rasmussen, and W. D. Hall, 2008: Explicit forecasts of winter precipitation using an improved bulk microphysics scheme. Part II: Implementation of a new snow parameterization. Mon. Wea. Rev., 136, 5095–5115. [8]. Vietnam Annual Hydrometeorology Report, 2008. National Centre for Hydro-Meteorological Forecasting (NCHMF), Vietnam Meteorological Hydrological Administration (VMHA). [9]. Zope, P. E., T. I. Eldho, 2016: Impact of Jothiprakash land use change and Urbanization on flooding: a case study of the Oshiwara River Basin in Mumbai, India. Catena, 145, pp. 142-154.