Sudhakar - Generate a number (100 or so) of synthetic traces for the desired
simulation period, such as 50 years. Then calculate the means, standard
deviations, and lag-one monthly correlation coefficients for each trace..
Compare these values with the same parameters determined from the historic
case. Graphical plots are helpful in assessing how good the simulated values
correspond to the desired statistical pattern.

For example, divide the January mean monthly for each trace by the historical
mean January flow. Plot these values versus trace number for the 100 or so
traces. .If they all have the value 1.0 you have an (impossible or at least
implausible) perfect score. They should plot on both sides of 1.0 within a
reasonable sampling error. If they plot all over the map you need to revisit
the model parameter determination.

You might want to make up a program to do these operations since there is a fair
amount of work involved.

Ron Kilmartin
Consulting Engineer
415 Turrin Drive
Pleasant Hill, Ca 94523
ronkilmartin@...
Ph 925 934 0378
Fx 925 945 7187
Cl 925 787 8342

dovari sudhakar wrote:

> Hi,
>
> I'd like to know the methods of model verification for
> Thomas-Feiring model by which I have generated
> synthtic series of reservoir inflows.
>
> Thanks
>
> Sudhakar
>
> __________________________________
> Do you Yahoo!?
> Find out what made the Top Yahoo! Searches of 2003
> http://search.yahoo.com/top2003
>
> **************************************************
>
> Post message: hydrologymodel@...
>
> Subscribe: hydrologymodel-subscribe@...
>
> **************************************************
>
>
> Your use of Yahoo! Groups is subject to http://in.docs.yahoo.com/info/terms/