loading page

Formation vs meteoric water: characterization using principal component analysis
  • Dasapta Erwin Irawan,
  • Prana Ugi,
  • INArxiv
Dasapta Erwin Irawan
Dept of Groundwater Engineering, Bandung Institute of Technology
Author Profile
Prana Ugi
Dept of Statistics, University of North Sumatra

Corresponding Author:[email protected]

Author Profile


Formation water is one of the problem in oil fields, in terms of production as well as environmental. When it reaches the surface or shallow groundwater system, then it will interact with main water supply in the area. In this paper we discuss a robust principal component analysis method to characterize both water bodies and how its interactions. This multivariable approach will reduce the number of variable in to fewer set for easier analysis and interpretation.
We re-use with permission a nice dataset from Po Plain area Italy, as a ”toy” dataset in the analysis. Here we also use some geothermal water data in the analysis to show the contrast. We use R, an open source statistical package to run the model. We also compare the PCA model from two packages: Pcamethods and FactomineR to test the consistency.
Aside to the trace elements like Sr, Br, and I, we can also differentiate formation water from shallow groundwater using alkalinity (Alk) major elements Ca, Mg, Na, K, Cl, SO, H, O, C, and trace elements Sr, Br, I, Fe. PCA technique can amplify the contrast of concentration of major elements to show the differentiation. In the biplot we can see that formation water samples from oil fields are fall in the region of Cl and SO4, while geothermal water show stronger control of HCO3. The shallow groundwater however, fall randomly in the area of Ca and Mg due to the natural water-rock interaction processes.
The PCA model presented in this paper, can be used as an example to finger print water system based on the hydrochemistry. The biplot can also help environmental division to do a fast and robust water quality classification in the field.
Key words: formation mater, meteoric water, R statistical package, PCA