Abstract:
The quest for understanding the human brain and emulating its functions unfolded new versions in system analysis. The complexity and non-linearity of the neural networks in brain is best utilised in evolving the analogical Artificial Neural Networks (ANNs). In the last decade the ANNs have found in many hydrologic applications. The rainfall-runoff modeling is one such field where ANNs can be applied extensively. Runoff is the response of a catchment for a particular rainfall pattern under various hydrometeorological factors and the exact prediction of which is rather difficult but essential for planning of a water resources project for irrigation and water supply, hydropower, floodcontrol and navigation etc.
. This thesis deals with the application of ANNs for rainfall-runoff modeling of Paleru sub-basin (2928 sqkm) and Musi sub basin of Krishna basin in AP and Sanlakoi sub basin (787 sqkm) of Brahmani basin in Orissa with an available rainfall and runoff data of 18 years (monthly), 15 years (monthly) and II Illollsooll years (10 daily) respectively. In addition the Panevapotranspiration (PET) data is available in case of Samakoi sub basin for the above mentioned period.
The Linear Least Square SiMplex (LLSSIM) and Back Propagation algorithms are used for training the network. The results are compared with those obtained using Multiple Linear Regression (MLR) and analysed with the help of coefficient of correlation, sum of least squared errors and coefficient of efficiency...
