Agricultural Research Management Information System

Home
Research Summary
Search
About Us
- ARMIS
- Brochure
Contact Us
Report
User Request
Data Input
Help
- Operation Manual
  - PDF
  - Video
- Program Area & Commodity
We have reached 37600 number of research entries at this moment.

- Logout

Research Detail

Home
Research
Detail

A Dynamic Linear Model for Real Time Forecasting of Meghna River in Bangladesh

M.F.R. Mondal
Postgraduate Student

M.H. Khan
Department of Irrigation and Water Management, Bangladesh Agricultural University, Mymensingh-2202

Abstract/ Executive Summary:

A dynamic linear input-output model was applied to the Megha river system with view to developing a suitable tool for forecasting of discharge at Bhairab Bazar, using information of upstream tributary flows.The model simulated daily discharge values for forecast at Bhairab Bazar with one day lead, from the observed discharges at Bhairab Bazar itself and those of other three upstream gauging stations, namely Sylhet, Sheola and Jaria Janjail, respectively on the tributaries Surma, Kushiara and Bhogai Kangsa. The accuracy of simulation for any day was determined by calculating the deviation of the forecasted value from the observed one. The efficiency of the model was then calculated by comparing its performance against a no-model situation, in which -a forecaster assumes the discharge at any day to be a forecast for the next day. Using all the three tributaries as inputs, the model produced efficiencies of up to 52.1% in calibration and 35.6% in verification.

Keywords: Forecasting, Flood routing, Linear model

Location: Megha river system

Start Date:

End Date:

Program Area: Conservation and Biodiversity

Commodity: Water management

Objectives:

To develop a suitable tool for forecasting of discharge at Bhairab Bazar, using information of upstream tributary flows.

Methodology:

In the present work, a multiple input-single output dynamic linear model has been studied in flood routing context on the Meghna river in Bangladesh, for real time forecasting. The model was fitted in a calibration period and the performances were tested in a separate verification period. The efficiency of the model was determined by comparing its performance against a no-model situation. To run the model, the dynamic linear input-output model and parameter estimation equation were applied.

Source of Information: Bangladesh J. Agril. Sci. 25(1): 127-132; January, 1998

Article Link (Online Journal):

Funding Source:

Total Budget (Tk.) :

Achievement/Findings:

The model efficiency continued to increase with increasing number of MA parameters. However, the improvement in model fit was insignificant beyond 7 such parameters. The results show that when the model is run with one tributary at a time, it produces up to 47.8 per cent efficiency in calibration and 34.3 per cent in verification, the best results being obtained with the tributary Surma. With two tributaries taken together, an efficiency of up to 51.8 per cent is obtained in calibration using the tributaries Surma and Bhogai Kangsa. The highest efficiency in verification of 35.6 per cent is obtained with the tributaries - Surma and Kushiara. The best results in calibration and verification periods were obtained when the three tributaries - Surma, Kushiaa and Bhogai Kangsa were routed together. It was observed that, in all cases, the level of efficiency obtained in the calibration period does not persist when the model was tested in the verfication period. This study suggests that the dynamic linear model may be employed to obtain reasonably acceptable results in forecasting of discharge at Bhairab Bazar, from Upstream flow information of the three gauged tributaries- Surma, Kushiara and Bhogai Kangsa.

Knowledge Management: Journal