Simulation of extreme rainfall events occurred in Bangladesh during 2007

Abstract

Heavy rainfall has potential damage to the life and properties of the people of Bangladesh. But forecasting of this weather phenomenon is the most challenging one in context of both location and time specific. Latest technologies like Doppler radar have enabled meteorologists to understand the characteristics and it is proven to be helpful in now-casting. Numerical modeling of this synoptic scale phenomenon was difficult in the past due to the non-availability of data with fine resolutions. Now-a-days, simulation facilities of these events have improved a lot due to the development of information technologies as well as conventional and non-conventional observations that provide data with high resolution. The simulation using meso-scale models would helpful to understand the underlying causes of this weather event. Most of the atmospheric prediction centres using global numerical models at present to predict the atmospheric variables including heavy rainfall or precipitation and regional models for higher spatial resolution. But regional models like MM5 are widely used to predict such type of weather system. Local modeling efforts have expanded dramatically with explosion of inexpensive, high-speed personal computers. Many National Weather Services (NWS) run the NCAR /PSU MM5 models using personal computers. In order to run regional or local models initial and boundary conditions are collected from the National Centers for Environmental Prediction (NCEP). In this regard it is essential to justify the prediction capability of MM5 model using existing facilities so that forecasters can use it in their day to day forecast. In this study fifth generation non-hydrostatic meso-scale model MM5 was run individually for five cases separately with the horizontal resolution of about 8.9 km (~ 9.0 km) and 2.97 km (~ 3.0 km) with the two selected domains D1 and D2 respectively. Case 1 (during 0000 UTC of 09 April to 0000UTC of 11 April 2007), Case 2 (during 0000 UTC of 21 April to 0000UTC of 23 April 2007), Case 3 (during 0000 UTC of 07 May to 0000UTC of 09 May of 2007) and Case 4 (during 0000 UTC of 10 June to 0000UTC of 12 June 2007) are run for the period of 48 hours and Case 5 (during 0000 UTC of 15 June to 1800UTC of 17 June 2007) is run for 66 hours with initial conditions collected from NCEP. The model predictions for these cases were having good indication about the events with respect to time and location. The amounts of simulated rainfall were comparable with the rain gauge value and TRMM estimation. The point or location specific simulated rainfall was comparable in some cases but for other cases the results were not satisfactory.