Inverse Modeling, Large Uncertainty

[toupis12]

Dear Mr. Parkhurst,

I have an issue with my inverse model similar to what most people post here about, but I was not able to use other people's solutions to apply to my situation.

Could you please look over my script for one of my samples and see why I only get models for uncertainty of 1?

I use rainwater from an area as the initial water and a stream sample for the second solution and want to see how much of which phase dissolves/precipitates to explain that water chemistry.

Thank you very much in advance,

[dlparkhurst]

In general, you need an independent source/sink for each element included in the mole balance. If you include 15 element, usually you will need at least 15 minerals. I added the following:

      kaolinite
      O2(g)
      Goethite
      Strengite

which produced one minimal model. 

That said, it is going to be difficult to include all of trace elements, mostly because one needs well defined stoichiometry of the phases. Trace elements tend to be trace constituents in other minerals, or involved in ion exchange or surface complexation, for which stoichiometries are hard to quantify.

So, I would concentrate first on the major dissolved constituents: Ca, Mg, Na, (K), Cl, and C, plus major elements in the minerals that you chose. The minerals would probably add Si and Al, possibly Fe.

Depending on your system, you may need to add O2 or CH2O as reactants to account for redox reactions

To check for plausibility, i'd like to see a weathering sequence that dissolves primary minerals, like feldspars, and makes secondary minerals like kaolinite, calcedony, or other clays. Also, dissolution and precipitation that is consistent with the saturtion indices. So, in your system, you would not want to precipitate calcite because it is undersaturated.

Finally, I would not add the tiny amounts of elements to rain; just use zeros. You may want to charge balance your rain (presumably on Cl). For simplicity, I would probably just start with pure water, unless you want to consider the effects of evapoconcentration.

[toupis12]

Thank you so much for the quick reply! When I ran the model without some of the elements to keep the list of minerals as short as possible, I kept running into the issue of the sample not being charge balanced, so I then ended up keeping my whole analysis but I'll keep playing with that then.

And thank you for the other points!