Using Intermediate Fields For Quality Control

GammaSpec creates several intermediate database fields during processing, rather than updating the same fields as processing progresses. These intermediate data fields are essential for effective quality control as described below. The intermediate database fields for typical processing are described in the following table. The "Field ID" column in the table is used to identify each group of fields in the discussion that follows.

Intermediate Database Fields Table

Field ID

Generic Aliases 

Description

1

Potassium Raw

Uranium Raw

Thorium Raw

Livetime-corrected window counts (no Nasvd applied)

2

Potassium Nasvd

Uranium Nasvd

Thorium Nasvd

Livetime-corrected window counts (with Nasvd applied)

3

Potassium Nas Corr

Uranium Nas Corr

Thorium Nas Corr

Fields (2) after all corrections are applied (backgrounds, stripping, height and sensitivity corrections), but no levelling.

4

Potassium Nas Final

Uranium Nas Final

Thorium Nas Final

Fields (3) after cross-over-tie levelling and microlevelling have been applied using either the Intrepid or Oasis Montaj processing systems. GammaSpec does not provide the functionality to apply levelling corrections.


The intermediate fields are used for quality control as follows:

  1. Images of the differences (2)-(1) should be random noise. Any coherent signal shows up areas where the NASVD has removed signal. So this aids the QA/QC of the NASVD noise removal. 
  2. A comparison of the images of (3) and (2) are useful for QA/QC on the background and height corrections. You can’t take differences here as the data are now at completely different levels. But residual long-wavelength, along-line features in (3) can indicate inadequate background removal. Short-wavelength, along-line features can indicate inadequate height correction. 
  3. Images of the differences (4)-(3) are an essential QA/QC tool for assessing the quality of the levelling (i.e. the combined cross-over-tie levelling and decorrugation/microlevelling). The differences are the levelling corrections – these should be long wavelength in the line direction only, and of limited amplitude. Signal is easily removed during decorrugation/microlevelling, so this intermediate dataset is essential.

Typically, the radioelement ratios Th/K, U/K and Th/U will also be calculated as the final stage of processing. The ratios are also a powerful QA/QC tool. Quite often both background estimation and levelling errors, not apparent in the final individual radioelement window data, show up in the ratios. Th/U is very sensitive to this. The ratios also tend to show up deficiencies in the NASVD noise removal – particularly uneven noise removal between NASVD groups, resulting in banding in the ratios.