Alice Springs recreational dam hydrology report project 6

Jackson, D.; Paige, D.

E-Publications; E-Books; PublicationNT; Report no. 12/1979

1979-10-01

Date:1979-10

English

Northern Territory Government

Darwin

Report no. 12/1979

application/pdf

Attribution International 4.0 (CC BY 4.0)

Northern Territory Government

https://creativecommons.org/licenses/by/4.0/

https://hdl.handle.net/10070/228346

https://hdl.handle.net/10070/674275

Technical Report WRD79012 Viewed at 00:02:46 on 18/02/2010 Page 113 of 153. I I I I I I I I I I I I I I I I I I I I 20 Rainfall and runoff points of equal probability from these extended frequency curves were then plotted on the correlations shown in Figures 9 and 10. At low flows (See insert ~igure 9), the historic points give a more realistic correlation than the ranked points. This is because at low rainfalls, storm distribution is not uniform over the catchment, losses are high and rainfall at Alice Springs is less indicative of flow in the catchment than for the higher rainfalls. The accuracy of the correlations shown in Figures 9 and 10 was tested using the Kolmogorov - Smirnov test (Spiegel 1965). This test was done on the historic flows at the damsite and the generated synthetic flows. For both seasons they were found to be within the 5% confidence limit. Using the rainfall-runoff correlation, 105 years of synthetic flows were generated on a monthly basis. 3. 7 Statistical Streamflow Generation Two methods were investigated to generate stochastic streamflow sequences based on streamflow statistics. One method generates log normal annual flows having zero serial correlation. i.e. log Q = log Q + t x (log Q) , where t is a random variate log Q is the mean of the log of the annual flows and (log Q)' is the standard deviation. These statistics are based on the 105 years of synthetic flows. An arbitary cutoff point is applied to the generated annual flows to account for low flow years. The statistically generated annual flows are ranked and the original annual flows are ranked. Monthly temporal patterns corresponding to the equivalent rank to the original flows are fitted to the statistically generated flows. The other method assumes correlation between months. It is based on the Thomas and Fiering seasonal model (McMahon & Mein 1978) applied to the 18 years historical data at the damsite. To allow for the large number of zeroes in each monthly sequence, Beard's Artificial Negative Analysis Method, Beard (1973), is used.