From vincent.gregorio at epfl.ch  Wed Apr 16 16:51:51 2008
From: vincent.gregorio at epfl.ch (vincent.gregorio@epfl.ch)
Date: Wed, 16 Apr 2008 16:51:51 +0200 (MEST)
Subject: statistical calculation
Message-ID: <1208357511.4806128763774@imapwww.epfl.ch>

Hello,

I have questions about the statistical calculation on great-er. In the
pre-processing step, i don't understand if is it the mean flow or the median
flow to enter in the .rna file( i have a lognormal distibution of flow values)?
What is the q_stdev calculate( does great-er transform the lognormal
distribution in a normal distribution? ) and how does great-er make monte-carlo
simulation (results of concentration are given in percentile, but is it for a
lognormal or normal distribution?)

is it possible to have more information about that?

Thanks to help me!

Vincent 



From frank.koormann at intevation.de  Thu Apr 17 07:05:35 2008
From: frank.koormann at intevation.de (Frank Koormann)
Date: Thu, 17 Apr 2008 07:05:35 +0200
Subject: statistical calculation
In-Reply-To: <1208357511.4806128763774@imapwww.epfl.ch>
References: <1208357511.4806128763774@imapwww.epfl.ch>
Message-ID: <200804170705.35894.frank.koormann@intevation.de>

Hallo,

Am Mittwoch, 16. April 2008 16:51:51 schrieb vincent.gregorio at epfl.ch:
> I have questions about the statistical calculation on great-er. In the
> pre-processing step, i don't understand if is it the mean flow or the
> median flow to enter in the .rna file( i have a lognormal distibution of
> flow values)?

The flow statistics are assumed to be lognormal distributed. The commonly 
available descriptive parameters are mean flow and percentiles, hence 
GREAT-ER requires the Q_mean and the Q_5.

> What is the q_stdev calculate( does great-er transform the 
> lognormal distribution in a normal distribution? ) 

The upload transforms the descriptive parameters of the lognormal 
distribution, since internally it is easier to derive MC-Shots from a 
(Mean, StdDev) distribution description.

> and how does great-er 
> make monte-carlo simulation (results of concentration are given in
> percentile, but is it for a lognormal or normal distribution?)

GREAT-ER generates series of random numbers depending on the given standard 
distribution of a parameter (constant, uniform, normal or lognormal) which 
are then transformed into the actual distribution of the parameter. 
The results are considered as lognormal distributed (GREAT-ER desktop Manual, 
section "Calculate River Csim X")
>
> is it possible to have more information about that?

Hope this helps.

Regards,

	Frank

-- 
Frank Koormann  |  ++49-541-335 08 30  |  http://www.intevation.de/
Intevation GmbH, Georgstr. 4, 49074 Osnabr?ck, DE | AG Osnabr?ck, HR B 18998
Gesch?ftsf?hrer: Frank Koormann, Bernhard Reiter, Dr. Jan-Oliver Wagner