Vijay Balsane, Sandeep Deshmukh, Dipika Bhamare, Ravindra Bansod and Deepika Mavale
Abstract
Lane´s condensed disaggregation model was selected to represent the ten daily streams flows of the Godavari river river basin, Nasik (Maharashtra). Since, the disaggregation model for generating stream flows requires previously generated annual stream flow series, the annual stream flows series was modeled by using autoregressive (AR) model. The annual stream flow volumes and ten-daily stream flow volumes of Godavari catchment at Gangapur Dam site for 22 years (1979-2000) were used for developing mathematical model. The parameter of AR model and disaggregation model were estimated by the method of maximum likelihood and method of moments. The goodness of fit of autoregressive model was tested by Box-Pierce Portmanteau test and Akaike Information Criterion. The AR (1) model was selected to represent the annual stream flow volumes. The developed models were used to generate annual stream flow volumes of 1200 years data. The goodness of fit of disaggregation model was performed by checking the normality of residuals and by comparison of autocorrleograms of historical and generated ten-daily stream flow series. The developed models were used to generate ten-daily stream flow volumes. The suitability of disaggregation model was tested by comparing the statistical properties i.e. mean, standard deviation, skewness, kurtosis of historical and generated ten-daily stream flow series and those of historical aggregated annual stream flow series. The comparison of statistical characteristics of historical and generated data suggests that disaggregation model can be used to generate ten-daily stream flow data which preserve the mean, standard deviation, skewness coefficient and kurtosis along with the mean of aggregated annual stream flow series.