Good analysis tripwire!
Actually I was also looking for the data on the dispersion of
submunitions of rocket system, thank for providing the link.
Actually my simplistic view Circular Error Probable (CEP) was wrong
and I apologise to everyone here about it.
It is actually worst, remember the US General talk abt the
probabiity of hitting a stationary tow gun at range of 21km with a
pod of six rockets with a mean deviation of 50m with only 33% of
success!
A US MRLS rockets has abt 644 submunitions per rocket, which mean a
total 3,864 submunitions yet only 33% chance!!!
That is becos my understanding of the term CEP is wrong.I thought
that all rounds will hit within the CEP!!! 
Below if the quote from some arty expert from another
forum:
Yes it is a Gaussian distribution, but instead using 'Standard
Deviations' (ie almost everything will be within 3 times SD
distance) military ballistics used PEs, where almost everything
falls within 4 times the PE distance, with 50% falling within 1
PE.
Data quality for artillery is mostly very good, although there are
some development firing ranges in some countries (we won't name
them) where the quality control leaves something to be desired.
However, real-world conditions are not necessarily as well-managed
as they are on development ranges.
Let me give example to explain, the Astros 2 let say has a CEP
of 1% mean deviation at range 89km, that mean the mean error is
890m, an area of pi x 890sq =approx 2.5 sq km with probability of
50%!!!!
Based on the distribution, that is 32% chance of with deviation 2
times of the CEP( 1 CEP is actually 1 PE, probable error) that is
from 890m to 1780m deviation.And abt 14% chance with deviation from
1780m to 2670m(3 times of PE) and lastly 4% chance with deviation
from 2670m to 3560m!!!!
Below is the distribution chart:
To have a better understanding about arty ballistics, below is link
abt the science of it, a bit "chim" but very good:
http://155.217.58.58/cgi-bin/atdl.dll/fm/6-40/Ch3.htm#p7
I hope you will have a better understanding of arty
ballistics.
With this data, if the Astros 2 is to target Woodland at range of
89km, I think there is 25% chance of hitting JB itself and the
surrounding water!! 
CHARGE!!!
