Databank

POSIVA Report 1997-8

Back

Name:

Regression Methodology in Groundwater Composition Estimation with Composition Predictions for Romuvaara Borehole KR10

Writer:

Ari Luukkonen; Juhani Korkealaakso; Petteri Pitkänen

Language:

English

Page count:

84

ISBN:

951-652-033-2; 1239-3096

Summary:

Working report: POSIVA-raportti POSIVA-97-08, 85 sivua
ISBN 951-652-033-2


REGRESSION METHODOLOGY IN GROUNDWATER COMPOSITION
ESTIMATION WITH COMPOSITION PREDICTIONS FOR ROMUVAARA
BOREHOLE KR10


ABSTRACT

Teollisuuden Voima Oy selected five investigation areas for preliminary site studies
(1987—1992). The more detailed site investigation project, launched at the beginning of
1993 and presently supervised by Posiva Oy, is concentrated to three investigation
areas. Romuvaara at Kuhmo is one of the present target areas, and the geochemical,
structural and hydrological data used in this study are extracted from there.

The aim of the study is to develop suitable methods for groundwater composition
estimation based on a group of known hydrogeological variables. The input variables
used are related to the host type of groundwater, hydrological conditions around the host
location, mixing potentials between different types of groundwater, and minerals
equilibrated with the groundwater. The output variables are electrical conductivity, Ca,
Mg, Mn, Na, K, Fe, Cl, S, HS, SO4, alkalinity, 3H, 14C, 13C, Al, Sr, F, Br and I
concentrations, and pH of the groundwater. The methodology is to associate the known
hydrogeological conditions (i.e. input variables), with the known water compositions
(output variables), and to evaluate mathematical relations between these groups. Output
estimations are done with two separate procedures: partial least squares regressions on
the principal components of input variables, and by training neural networks with input-
output pairs. Coefficients of linear equations and trained networks are optional methods
for actual predictions. The quality of output predictions are monitored with confidence
limit estimations, evaluated from input variable covariances and output variances, and
with charge balance calculations.

Groundwater compositions in Romuvaara borehole KR10 are predicted at 10 metre
intervals with both prediction methods. Predictions are done in two separate runs. At
first only the locality and planned drilling direction is known; all knowledge is based on
the situation in the spring of 1995. The second run is done with actual host type and
hydraulic conductivity information available in the autumn of 1996.

The two prediction methods are compared with each other, and the advantages and
drawbacks of the methods and data are identified. Finally, possible improvements to the
prediction methods are considered.

Keywords: Groundwater, geochemistry, hydrogeology, regression methods, principal
components, neural analysis



POSIVA-raportti POSIVA-97-08, 85 sivua
ISBN 951-652-033-2


REGRESSION METHODOLOGY IN GROUNDWATER COMPOSITION
ESTIMATION WITH COMPOSITION PREDICTIONS FOR ROMUVAARA
BOREHOLE KR10


ABSTRACT

Teollisuuden Voima Oy selected five investigation areas for preliminary site studies
(1987—1992). The more detailed site investigation project, launched at the beginning of
1993 and presently supervised by Posiva Oy, is concentrated to three investigation
areas. Romuvaara at Kuhmo is one of the present target areas, and the geochemical,
structural and hydrological data used in this study are extracted from there.

The aim of the study is to develop suitable methods for groundwater composition
estimation based on a group of known hydrogeological variables. The input variables
used are related to the host type of groundwater, hydrological conditions around the host
location, mixing potentials between different types of groundwater, and minerals
equilibrated with the groundwater. The output variables are electrical conductivity, Ca,
Mg, Mn, Na, K, Fe, Cl, S, HS, SO4, alkalinity, 3H, 14C, 13C, Al, Sr, F, Br and I
concentrations, and pH of the groundwater. The methodology is to associate the known
hydrogeological conditions (i.e. input variables), with the known water compositions
(output variables), and to evaluate mathematical relations between these groups. Output
estimations are done with two separate procedures: partial least squares regressions on
the principal components of input variables, and by training neural networks with input-
output pairs. Coefficients of linear equations and trained networks are optional methods
for actual predictions. The quality of output predictions are monitored with confidence
limit estimations, evaluated from input variable covariances and output variances, and
with charge balance calculations.

Groundwater compositions in Romuvaara borehole KR10 are predicted at 10 metre
intervals with both prediction methods. Predictions are done in two separate runs. At
first only the locality and planned drilling direction is known; all knowledge is based on
the situation in the spring of 1995. The second run is done with actual host type and
hydraulic conductivity information available in the autumn of 1996.

The two prediction methods are compared with each other, and the advantages and
drawbacks of the methods and data are identified. Finally, possible improvements to the
prediction methods are considered.

Keywords: Groundwater, geochemistry, hydrogeology, regression methods, principal
components, neural analysis



Keywords:

Groundwater; geochemistry; hydrogeology; regression methods; principal components; neural analysis

File(s):

Regression Methodology in Groundwater Composition Estimation with Composition Predictions for Romuvaara Borehole KR10 (pdf) (7.2 MB)


Back


Share article:
This website stores cookies on your computer. These cookies are used to improve our website and provide more personalised services to you.
Close

Cookies

To make this site work properly, we sometimes place small data files called cookies on your device. Most big websites do this too.

1. What are cookies?

A cookie is a small text file that a website saves on your computer or mobile device when you visit the site. It enables the website to remember your actions and preferences (such as login, language, font size and other display preferences) over a period of time, so you don’t have to keep re-entering them whenever you come back to the site or browse from one page to another.

2. How do we use cookies?

A number of our pages use cookies to remember your actions and preferences (such as login, language, font size and other display preferences.)

Also, some videos embedded in our pages use a cookie to anonymously gather statistics on how you got there and what videos you visited.

Enabling these cookies is not strictly necessary for the website to work but it will provide you with a better browsing experience. You can delete or block these cookies, but if you do that some features of this site may not work as intended.

The cookie-related information is not used to identify you personally and the pattern data is fully under our control. These cookies are not used for any purpose other than those described here.

3. How to control cookies

You can control and/or delete cookies as you wish – for details, see aboutcookies.org. You can delete all cookies that are already on your computer and you can set most browsers to prevent them from being placed. If you do this, however, you may have to manually adjust some preferences every time you visit a site and some services and functionalities may not work.

Close