Prognostic equation based on artificial neural network for quantitative rainfall forecast using numerical weather prediction model products / Wardah Tahir … [et al.]

Tahir, Wardah and Atan, Ismail and Abu Bakar, Sahol Hamid and Aminuddin, Ahmad Kamil (2012) Prognostic equation based on artificial neural network for quantitative rainfall forecast using numerical weather prediction model products / Wardah Tahir … [et al.]. [Research Reports] (Unpublished)


In Malaysia, there are two types of flood that normally occur namely, monsoon flood and flash flood. Floods associated with the monsoonal rainfall events are common occurrences on the eastern coast of Peninsular Malaysia during the northeast monsoon season. Every year tropical monsoon storms result in severe flooding and causes enormous economic damage, social disruption, and sometimes loss of lives. Extreme monsoon storm weather phenomenon is the most destructive natural disaster afflicting Malaysia with respect to the cost, damages to properties and the area of extent (Keizrul and Chong, 2002). Given the sparseness of ground based observations, missing records and uneven distribution of the existing raingauge network, there is no adequate and timely information about rainfall pattern in Malaysia. An alternative source of quantitative precipitation information is from the Numerical Weather Prediction (NWP) model products. However, the accuracy of quantitative precipitation forecast produced by the Malaysian Meteorological Department (MMD) is still lacking even though significant progress has been made on the technical aspects (Low, 2006) . The study examined the effectiveness of two high resolution Numerical Weather Prediction (NWP) models namely the Fifth Generation Penn State/NCAR Mesoscale (MM5) and Weather Research and Forecasting (WRF) in predicting Quantitative Precipitation Forecast (QPF) over a tropical region. In this study, Kelantan River Basin has been selected as the case study to evaluate the performance and accuracy of precipitation forecast produced by the NWP models for monsoon flood events in the catchment area. Hourly and daily total rainfall data in year 2009 had been analysed. The rainfall events were further classified into low, moderate and heavy rainfall by using Drainage and Irrigation Department (DID) Malaysia standard. The performance and accuracy of the NWP model outputs against rainfall amount was verified using Root Mean Square Error (RMSE) and correlation (r). Notably, the statistical verification shows that there is quite strong correlation for 24 hourly rainfall forecast and the RMSE values are smaller for short range forecast (hourly up to 24 hourly). It is also noted that the longer the rainfall forecast duration, the higher probability of detection (POD) and the lesser probability of the false alarm ratio (FAR).


Item Type: Research Reports
CreatorsEmail / ID. Num
Abu Bakar, Sahol HamidUNSPECIFIED
Aminuddin, Ahmad KamilUNSPECIFIED
Subjects: Q Science > QA Mathematics > Instruments and machines
Divisions: Universiti Teknologi MARA, Shah Alam > Research Management Centre (RMC)
Item ID: 23725
Uncontrolled Keywords: Prognostic equation; Artificial neural network; Rainfall


[img] Text

Download (48kB)

Actions (login required)

View Item View Item


Downloads per month over past year