• To participate in the 911Metallurgist Forums, be sure to be Logged-IN
  • Use Add New Topic to ask a New Question/Discussion about GeoMetallurgy or Geology.
  • OR Select a Topic that Interests you.
  • Use Add Reply = to Reply/Participate in a Topic/Discussion (most frequent).
    Using Add Reply allows you to Attach Images or PDF files and provide a more complete input.
  • Use Add Comment = to comment on someone else’s Reply in an already active Topic/Discussion.

Soil Assay Data (7 replies)

8 months ago
Sturmbann 8 months ago

I'm looking for some suggestions on how to normalize soil assay data; it comes from a similar environment but from different labs with different LDL any suggestions?

Marshal Meru
8 months ago
Marshal Meru 8 months ago

Different labs, different LDL and probably different digestions, analysis, calibration, etc. I suggest keeping the analytical data separated and plotted separately although you can tile the plots and keep the colour schemes the same, etc. It’s also likely that analytical data from the same lab (and same method) but with a long time separation should be split out.

David Kano
8 months ago
David Kano 8 months ago

I will treat the soil data per "labs" population. Then, for each dataset, calculate the background for each element assayed. Once background is calculated, I recommend you to calculate the ratio (value of assay of an element divided by background of this element). Then distribute your ratio on your map using the percentile distribution. The map will show the different labs population (using different symbol and applying a color scale showing the percentile distribution). Then, if assays from different labs where done correctly, you should pick-up anomalous areas.

8 months ago
Unterstarm 8 months ago

Is the data from an overlapping spatial area, and did you insert standards that would allow you to assess whether there is any analytical bias to the data? Is the underlying geology the same in each survey? Also, was the digestion used during analysis equivalent between the two methods used?

I ask the above questions because it is not good practice to level data without some reason to do so... as you may be levelling away real variation i.e. inadvertently hiding the very anomalies you are hunting for or creating false positives resulting in a waste of money in follow-up.

Zander Barcalow
8 months ago
Zander Barcalow 8 months ago

Take a look at the articles on my website on stream sediment sampling and what I do to normalize this type of data. Same procedure can be applied to soils and rocks. http://is.gd/vQT1cZ

Tony Verdeschi
8 months ago
Tony Verdeschi 8 months ago

I understand that merging different datasets (due to different labs, different assay methods...) sounds like heresy but, if I had to merge two datasets anyway and admitting that my datasets slightly overlaps, would there be any statistical method more suitable than another to do so?

8 months ago
OberstGruppen 8 months ago

If you have to use data obtained in different labs, possibly using different methods the best method to harmonize the results is that a set of test samples are analysed in each laboratory. The samples should be real samples from the covering the concentration ranges and matrix types of the test field. Then select one laboratory you expect to be the most reliable/experience and regress the results of the other laboratories against this laboratory. If there are systematic differences you can then convert all result to match the reference lab results. I used this method when I was working as an analytical chemist in a mineral exploration project.

Marshal Dienes
8 months ago
Marshal Dienes 8 months ago

Subset the samples from the overlapping area and create a QQ plot, if all the quintiles plot on the first bisector (45degreees) then the different lab data is sufficiently similar to treat as the same. The q-q plot will highlight any biases.