RE: TLD Background subtraction

["Walker, Scottie" <swwalke@sandia.gov>]

The task does require some type of statistical software to ease the
calculation burden.  We use Statgraphics, but I am not soliciting its use
and I believe any similar program will do.

The basic steps we follow are to:
1.	Evaluate the stop and start dates for the dosimeters to determine an
appropriate use range and exclude outliers.
2.	Normalize the electronic channel data by dividing by the number of
days in the field (we have four chips in our dosimeter).
2.	Use a box and whisker plot to evaluate each dosimeter channel data
at the 95% confidence level and exclude outliers.
3.	Fit the remaining data with an appropriate distribution (we use the
lognormal, but you must determine based on your site).
4.	Adjust the fit iteratively to obtain the highest K-S
(Kolmogorov-Smirnov) statistic possible.  (We use this test because it is
not dependent on how the data is subdivided, as is the case with the
chi-square test).
5.	Obtain the average of the lognormal fit and add it to our historical
plot.
6.	Perform a linear regression of historical data.
7.	Determine the new background subtraction values for each dosimeter
channel based on the linear regression.

Our background evaluation method has not been published anywhere outside our
laboratories to my knowledge.  However, I would be happy to provide you and
any other interested parties with a copy of our technical basis document
that describes the method in detail and gives examples.  If you'd like this
information, please contact me offline at swwalke@sandia.gov.

Scottie Walker, CHP
External Dosimetry
Sandia National Laboratories
(505) 844-5241
swwalke@sandia.gov


-----Original Message-----
From: Scott Schwahn [mailto:schwahn@jlab.org]
Sent: Wednesday, February 03, 1999 9:55 AM
To: Multiple recipients of list
Subject: Re: TLD Background subtraction


I am intrigued, Scottie.  Perhaps if it wouldn't be too much effort, you
could provide a summary of your method here on RADSAFE?  Have you published
this method anywhere that we could access?

"Walker, Scottie" wrote:

> I would like to just mention an alternative to posting control dosimeters
at
> a designated location.  Our lab is fairly large and we do not have the
> luxury of establishing one badge rack for a building or even one technical
> area.  And, like many other R&D sites, many of our workers don't regularly
> enter radiological areas during the monitoring period.  So, we use
> statistical methods to remove the dosimeters having occupational doses and
> evaluate the remainder for the background subtraction.  What we're left
with
> is several thousand measurements of the background radiation levels during
> the monitoring period.

--
Scott O. Schwahn, CHP
Thomas Jefferson National Accelerator Facility
schwahn@jlab.org


************************************************************************
The RADSAFE Frequently Asked Questions list, archives and subscription
information can be accessed at http://www.ehs.uiuc.edu/~rad/radsafe.html
************************************************************************
The RADSAFE Frequently Asked Questions list, archives and subscription
information can be accessed at http://www.ehs.uiuc.edu/~rad/radsafe.html