I-data modeling

[Adrien]

I have a question regarding a solid model I created of I-data. Ok, so I have contaminant concentrations that I have inserted in the I-data spreadsheet for approximately 80 wells. I know what the plume is supposed to look like (a long (almost a mile) and skinny (maybe 300 ft) plume, and although I can get the shape of the plume using the directional weighting algorithm, the concentrations are 1-tenth what they should be. For example, the concentration in one well with a 20-foot screened interval is 135 ppb, yet when the model is created the maximum concentration for the whole plume may only be 20 ppb. I've tried using the high fidelity option, but when I do the model looks unrealistic (the plume extend much further than it should and includes "swiss-cheese" looking holes at wells with non-detect contaminant concentrations. Can someone explain how the algorithm reduces the concentrations by an order of magnitude? Or, how best to model a plume with limited data that covers a large area?

[Molly]

Hi Adrien,

Good question. We may have to look at your actual data to give precise recommendations, but here are some generalizations:

1. Vertical spacing of the model nodes can affect how well your isolated highs are represented. If your node spacing is too large, you might not get the vertical resolution you need to represent localized high measurements. But, you're facing a dilemma here - with a large project area, tightening up nodes may do nothing but chew up a lot of time in modeling and 3D rendering.

2. The Logarithmic checkbox can help with modeling of anomalous highs. This, in combination with high fidelity, might help. This should also help to prevent the swiss cheese phenomenon.

3. If your goal is to generate a model that represents what you know the plume to look like, adding additional control points might be an option. This is not advisable if your goal is to best-represent observed data only.

4. If you need to constrain the shape of the plume, you can apply a polygon filter as a model constraint, to set outer nodels to a null value.

Feel free to zip and email your project folder to us for detailed suggestions. The support email is: tech@rockware.com.

Molly