• Document: Rainfall Forecasting in Northeastern part of Bangladesh Using Time Series ARIMA Model
  • Size: 1.27 MB
  • Uploaded: 2018-12-08 06:59:39
  • Status: Successfully converted


Some snippets from your converted document:

Research Journal of Engineering Sciences _________________________________________E- ISSN 2278 – 9472 Vol. 5(3), 17-31, March (2016) Res. J. Engineering Sci. Rainfall Forecasting in Northeastern part of Bangladesh Using Time Series ARIMA Model Md. Abu Zafor1, Amit Chakraborty1, Sheikh Md. Muniruzzaman2 and Smriti Rani Mojumdar1 1 Department of Civil Engineering, Leading University, Sylhet 2 Department of English, Leading University, Sylhet cezafor@gmail.com Available online at: www.isca.in, www.isca.me Received 29th February 2016, revised 3rd March 2016, accepted 25th March 2016 Abstract To improve water resources management, time series analysis is an important tool. Bangladesh is a densely populated country and now facing shortage of drinking water. Rainwater harvesting is one of the major techniques to overcome this problem. For this purpose, it is very much important to forecast future rainfall events on a monthly basis. Box-Jenkins methodology has been used in this study to build Autoregressive Integrated Moving Average (ARIMA) models for monthly rainfall data from eight rainfall stations in the northeastern part of Bangladesh: (Sunamganj ,Lalakhal, Kanaighat, Sherupur, Chattak, Gobindoganj, Sheola, Zakiganj) for the period 2001-2012. Eight ARIMA models were developed forthe above mentioned stationsas follow: (2,0,2)x(1,0,1)12, (4,0,1)(1,0,0)12, (1,0,1)(0,0,2)12, (1,0,0)(2,0,0)12, (1,0,0)(2,0,0)12, (1,0,0)(2,0,0)12, (1,0,0)(2,0,0)12, (2,0,0)(2,0,0)12 respectively. The performance of the resulting successful ARIMA models were evaluated using the data year (2011). These models were used to forecast the monthly rainfall data for the up-coming years (2013 to 2020). The results supported previous work that had been carried out on the same area recommending the use of water harvesting in both drinking and agricultural practices. Keywords: Time series analysis, Monthly Rainfall forecasting, Box-Jenkins (ARIMA) methodology Introduction Box and Jenkins in early 1970's, pioneered in evolving methodologies for statistic modeling within the univariate case Winstanley1 reported that monsoon rains from Africa to India often referred to as Univariate Box-Jenkins (UBJ) ARIMA decreased by over five hundredth from 1957 to 1970 and modeling 6. expected that the long run monsoon seasonal rain, averaged over five to ten years is probably going to decrease to a minimum Methodology around 2030. Study Area Description and Datasets: Sylhet district is the north Laban2 uses time series supported ARIMA and Spectral Analysis eastern part of Bangladesh. It is sited on the bank of the river of areal annual rain of two same regions in East Africa and Surma. The total area of the district is 3452.07 sq. km. (1332.00 sq. counseled ARMA(3,1) because the best appropriate region indice miles). The district lies between 24º36’ and 25011' north latitudes of relative wetness/dryness and dominant quasi-periodic fluctuation and between 91º38' and 92030' east longitudes. around 2.2-2.8 years,3-3.7 years,5-6 years and 10-13 years. Data Acquisition: The time series data of rainfall for the Sylhet Kuo and Sun3 used an intervention model for average10 days region was collected from Bangladesh Water Development stream flow forecast and synthesis that was investigated by to Board (BWDB). Observed monthly rainfall of eight monitoring influence the extraordinary phenomena caused by typhoons and weather stations for the period 2001 to 2012 was collected to alternative serious abnormalities of the weather of the Tanshui river perform this study. basin in Taiwan. ARIMA Model Methodology: In 1976, Box and Jenkins, give a Chiew et al4 conducted a comparison of six rainfall-runoff methodology (Figure-2) in time series analysis to find the best fit of modeling approach to reproduce daily, monthly and annual flows time series to past values in order to make future forecasts6. The in eight unfettered catchments. They accomplished that time-series methodology consists of four steps: i. Model identification. ii. approach will offer adequate estimates of monthly and annual Parameters estimation. iii. Diagnostic checking for the

Recently converted files (publicly available):