Reference, Questions and Help > MARSSIM
MARSSIM Discussion Group
Rennhack:
I have started a new MARSSIM Discussion Group here. Feel free to exchange ideas.
Rennhack:
I have created a MARSSIM website, to help people find MARSSIM information and share resources.
If you have any suggestions for the new MARSSIM site, feel free to post them in this forum.
Check out the MARSSIM site here:
http://www.MARSSIM.com
alphadude:
In association with MARSSIM, VSP training should be taken also (two days B. Pulsipher).
COMPASS is very, very easy to use so not a lot of training is needed there.
"Managing Uncertainty with Systematic Planning for Environmental Decision Making" taught by Sebastian Tindal is an exellent training class to take in association with VSP. Both of these are free (VSP and Managing Uncertainty) and paid for by the DOE. An understanding on the management of the DQO process will give the MARSSIM user the background information and insight on the process that MARSSIM provides. Sebastian believes in the incremental sampling process and I think he does an excellent job on why it should be used for scoping and assessments.
Remember 59 samples is the magic number and 100 samples will work wonders!
Rennhack:
--- Quote from: alphadude on May 19, 2006, 10:04 ---Remember 59 samples is the magic number and 100 samples will work wonders!
--- End quote ---
That depends on your DCGL's and area factors. Many projects can take 12-15 samples.
Thank you for the insight on the training. I hope others read this thread and the MARSSIM discussion picks up.
Check out the MARSSIM site here:
http://www.MARSSIM.com
alphadude:
Wellll not really. Its more convoluted than the DCGL and area factors. Hence the hint on taking the additional training. If you notice after a while with MARSSIM techniques you will usually have no more than 25 samples for any MARSSIM project. 12-15 samples is typical, (seems a little low-look at the stdev of the scoping and MARSSIM results to see how they agree) and is not really dependent on DCGL, but the risk you wish to assume (alpha and beta), distribution design, variability, LLD, LBGR etc. The 100 samples will invoke the central limits theorm thus reducing the need for statistical schemes. 59 samples will provide a nice distribution showing that most sites are log normal and should be planned accordingly. The 12-15 samples you mention are the end of the road testing that results from limiting the decision unit size, adequate scoping surveys (lots of samples), limiting critical values, and a DCGL well below any level of concern for dose to the population.
Navigation
[0] Message Index
[#] Next page
Go to full version