Synthetic Flow Generation with Stochastic Models

Abstract

A first-order Markovian model was proposed based on the Three-parameter Log Normal distribution. This model was comparatively used with the Thomas-Fiering models, one formulated using Normal distribution and the other with Gamma distribution. Primary objectives of the study were: (1) to examine the satisfactory applicability of the proposed model, and (2) to investigate the importance of preserving the skewness in streamflow generation. In the methodology, each model was used to generate sequences of annual mean flows for six watersheds. Average estimates of the statistical parameters were computed for the generated sequences and then compared with the historic values. To determine the effect of skewness, the probable maximum annual flows were estimated for three return periods using generated sequences and the magnitudes due to different models were compared. The results of the study indicate the importance of preserving the skewness in flow generation. In addition, it reveals that the proposed model can be satisfactorily used.