Rainfall runoff model for Uma-Oya basin

Abstract

The Rainfall-Runoff model presented here could be used to predict the monthly runoff values at the unengaged site of Uma Oya confluence, "and hence monthly runoff values at the Mnipe anicut in the Mahaweli river.'These estimated, monthly runoff values are needed for the operation of the control gates of Minipe right bank and left bank main canals in order to divert water to system B,C and E respectively. Although there are long periods of rain fall records over the Uma Oya basin the availability of runoff records is limited and are available for the two upstream gauging stations at Welimada and Talawa kanda across the Uma Oya . As the location of the Uma Oya confluence is different from the locations of the river gauging stations the model parameters need to be renationalized so that the runoff at any location of the Uma Oya river could be estimated. A deterministic black-box regression model for rainfall-runoff simulation in the two sub-basins at Welimada & Talawakanda is developed. The model is mathematically expressed as; Qp.t = nt [∑_(1=0)^k▒〖a 〗t2l Pip1t] I = 01I………K t = 1.2 …….12

The model structure depends on two key parameters which are; 1. An Order parameter- K (an integer, greater than or equal to 1) 2. A Memory parameter- nt ( t = l,2,3........12) An order parameter K, characterizes the runoff behavior of the Das in .ie. When K=l, the catchment is linear and for K >1, the catchment is non-linear. A memory parameter nt characterizes the memory of the rainfall-runoff process The order parameter K is determined from monthly rainfa 11-runoff functional relationships and the model coefficients 0t,1 are estimated for the same functional relationships by the least-square technique for both sub basins. Finally the model coefficients are regionalized over the Uma Oya basin, so that the model can be used to estimate monthly runoff at the unqauged site of Uma Oya confluence. These estimated runoff values have been compared with the runoff values of observed differences between . Rantambe and Randenigala of Mahaweli river, before application of the model.