Prediction of High Impact Precipitation Events over Bangladesh and its surrounding areas using NWP Models

Abstract

Bangladesh is a disaster prone country. Almost every year, the country experience disasters of one kind or another - such as tropical cyclones, storm surges, coastal erosion, hoods due to heavy rainfall and droughts - causing heavy loss of life and property and jeopardizing the development activity. Because of south-westerly wind, high amount of moisture come to the Bangladesh region from the warm surface of the Bay of Bengal and heavy rainfall occurs by convection in our region. Tropical cyclone also forms over the warm oceans and ravage life and property especially over the coastal belt due to extremely strong winds and associate storm surges at the time of landfall. The Bay of Bengal is highly vulnerable to tropical cyclogenesis. To save the life and minimize the damage it is necessary to make advance warning and prediction of both the heavy rainfall and tropical cyclone quite ahead of time. Therefore a study has been conducted to investigate the formation and vertical structure of the heavy rainfall events and structure, intensity and track with landfall time of the tropical cyclone events. In the present study, two state-of-the-art mesoscale models MM5 and WRF-ARW have been used to evaluate their performances in the simulation of heavy precipitation events and tropical cyclone (TC) and its impact on Bangladesh and its surrounding areas. Three test cases for the heavy rainfall events are considered and the cases are during 1-3 May 2009, 1-3 July 2008 and 9-14 June 2007. First one is the pre-monsoon seasonal case and last two are the monsoon seasonal test cases. On the other hand, three test cases for the TCs are considered and the cases are TC Aila, 23-27 May 2009; TC Sidr, 11 - I 7 November 2007 and TC Rashmi, 24-28 October 2008. For TC prediction, one nested domain is used inside the mother domain with horizontal resolution mother and nested domain are 90 and 30 km respectively. On the other hand, for heavy rainfall events, two nested domains are used inside the mother domain with the resolution of mother domain is 90 km that for two nested domains are 30 and 10 km respectively. For MM5 model, in this present study Medium Range Forecast (MRF) PBL scheme, Kain - Fritisch (KF) cumulus parameterization (CP) scheme, Dudhia Simple Ice microphysical Scheme for moisture anticipation, Cloud Radiation Schemes for radiation calculation and 5- Layer Soil model to predict soil temperature is used as model physics. Model equations in the surface flux form and solved on Arakawa B grid. Leapfrog time integration scheme with time splitting technique is used in model integration. For WRF model, the modified Kain - Fritisch cumulus parameterization scheme, WRF Single-Moment (WSM) 3-class simple ice scheme, Translèr Model (RRTM) scheme, Dudhia scheme short wave scheme, Yonsci University Scheme (YSU) Planetary Boundary Layer (PBL) parameterization which is the next generation MRF—PBL is used as model physics. For the heavy rainfall events Mean Seal Level Pressure (MSLP), wind with rain, wind with humidity, rainfall, vertical structure of vertical velocity, divergence, relative vorticity, relative humidity and mixing ratio have been simulated and analyzed to understand the convective activity of the precipitation system by both the models. Both of the models would simulate the convective activity of the precipitation events fairly well. Amount of precipitation are more than those of the observed by TRMM in monsoon and than pre-monsoon. On the other hand, for the TC events MSLP, wind (vector, radial, tangential, vertical wind), vorticity, temperature anomaly, relative humidity, mixing ratio, rainfall and tracks have been analyzed to understand the structure and behavior of the cyclones by the models. Both the models would simulate the cyclonic nature at the lower and anticyclonic nature at the upper levels. Simulations of intensity are more or less than the observed but realistic to observed except for TC Sidr. Amount of precipitation are more than those of the observed by TRMM. To understand the knowledge about the sensitivity of various physics options of the MM5 model, model was run for various sensitivity cases: sensitivity study on PBL with CP is both for heavy precipitation events and TC events, sensitivity study on microphysics, radiation are only for TC Aila. In case of sensitivity on PBL with CP for heavy precipitation events, it is observed that some schemes have overestimated the rainfall and someone has underestimated that. The options for simulations of heavy rainfall events with MM5 have been found dependable on resolution and location of the area. According to our study, no single option may be considered as the most suitable among the 10 options for the assessment of rainfall over Bangladesh but it can be understood that AK CP is better for daily rainfall prediction and KF2 is better for total rainfall prediction. More case studies are necessary to make final commend for our region. The PBL option with CP play very vital role on the track and intensity of TC. No one combination plays the best in the three TC cases but PBL MRF with KF CP plays better among the 10 combination in the three TC cases. Sensitivity test on microphysical parameterization option and radiation parameters options is done only for TC Aila. It is seen that they have their own impact on the simulation of track and intensity of TC Aila. The present study has investigated only one cyclone, and more cases should be examined to supplement these results. It is suggested that it would be desirable to make sensitivity experiments with all possible combinations of the schemes of the physical processes. Other sensitivity tests like horizontal resolution, bogussing vortex may be done to improve the performance of the two models. Then the two models would be used as an operational purpose for the simulation of all kind of heavy precipitation events.