The effect of serial correlation on reservoir capacity using the modified Gould's probability matrix method (MGPM)

The capacities of 6 potential reservoir sites in Kenya were determined using the modified Gould's probability matrix method (MGPM) and the effect of serial correlation of inflows on these capacities were examined using different months for the starting date of the analysis. Average capacity ratios of 1.02, 0.91, 0.85 and 0.76 were obtained at the 30, 50, 70 and 90 per cent draft levels, respectively. Serial correlation had an effect for demands greater than 30 per cent of the mean flow and the effect increased with draft. If the demand was 30 per cent greater than the mean flow, then the starting month of analysis by MGPM should be that which gave the minimal annual serial correlation coefficient (ASCC). Since the ASCC for 5 of the 6 sites were quite considerable, the use of the starting month of analysis (SMA) giving the minimal ASCC might not prevent an under-design of reservoir capacity. There was a requirement for the routing period used to be long enough to give an insignificant correlation coefficient with an SMA applicable to the period. A modified MGPM was proposed for the simulation. With 5 of the 6 sites, a 2 year period gave a correlation coefficient of adequate insignificance.