The Use oF Linear Perturbation Model for Real Time Flow Forecasting of the Blue Nile River System at Khartoum

Abstract

Flood is one of the most damaging natural disasters worldwide. Flood problems lead to significant damages. In addition to loss of lives, floods result in damages to ecosystem, properties, and poses risk to public health. The unexpected occurrence of floods results in bigger amount of damages and higher costs of rehabilitation than the predicted flood event. Thus, the need for flood forecast especially flood early warning systems is vital. Along the Blue Nile River the floods have been the most common form of natural disaster. In recent years the frequency of natural disasters associated with extreme floods events were increased (e. g. 1988, 2003, and 2007). The floods have negative impacts to Khartoum State. This results mainly in loss of lives and properties. To combat or reduce such impacts, two broad approaches can be used namely structural and non structural. This work falls in the non structural approach. The main objective of this study is to develop a model that is capable of issuing forecast flows at Khartoum based on measured flows at Sennar, Dinder, and Rahad with a suitable lead time. The Linear Perturbation Model(LPM) was used as substantive model. The model was calibrated using 11 years of data and verified using 5years of data. The model was used in two forms namely parametric and non parametric as well as in single input- single output and multiple input-single output forms. The calibrated models are then updated using both linear transfer function (LTF)and autoregressive(AR) models with lead time of up to six days. The model is then used for real time forecasting after calibration and updating. iii Based on Nash – Sutcliffe efficiency criteria, the model performance is judged and found very satisfactory in all its forms and stages of uses. During the calibration period the model efficiencies are 94.2% and 95.5% for the cases of single input- single output and multiple inputs – single output for Non-Parametric LPM respectively. The model efficiencies during the verification period (R2) are 93.0% and 91.6% for the cases of single input- single output and multiple inputs – single output for Non-Parametric LPM respectively. For the model in parametric form, (R2) is 99.73% during the calibration period for single input case and 99.72% for multiple input. While during the verification period of the parametric form the R2 are 99.76% and 99.75% for single and multiple input cases respectively. The calibrated models in updating mode for six days lead time provide efficiencies ranging from 99.5% to 96.5% for single inputsingle output for non parametric case and 99.5% to 96.for multiple input- single output for non parametric case. For the Parametric case the model efficiencies ranging from 99.5% to 96.5% and 99.5% to 96.7 % for single input and multiple inputs respectively. It can be concluded that the flow at Khartoum can be forecasted to an excellent degree using the flow at upstream station. It is also showed that the effect of the Rahad and Dinder on flows at Khartoum is negligible.