A G US T GUDMUND SSON, INGRID FJELD SKA AR & OT/LlE GJ ES DA L NGU -BU LL 4 3 9 , 20 02 - PAGE 6 1

Fracture-generated permeability and groundwater

yield in Norway
AGUST GUDMUNDSSON, INGRID FJELDSKAAR& OTILlEGJESDAL

Gud mundsson, A., Fjeldskaar, I. & Gjesdal, O. 2002: Fract ure-generated permeability and groundwate r yield in
Norway. Norges qeoloqiske undersoke/se Bulle tin 439 , 6 1-69.

The t ranspor t of groundwat er in bedrock is largely determined by interconnected fractu res and t heir apert ures.The
condit ions by w hich fract ures become interconnected are t hus of primary import ance in underst andi ng permeabil-
ity and groundwater yield. This applies in part icular to t he observed linear correlatio n betwee n th e current post-
glacial up lift rate and groundwater yield in th e bedrock of sout hern Norway.We present bound ary-element models
on t he form ation of interconnected fractur e pat hways w ith applicat ion to the coast al areas of West Norway. In t he
mod els, we useexte rnal tensile st ressand internal fluid overpressure; loading cond it ions th at are likely to have been
operative in large part s of Nor way duri ng t he Holocene.The ext ernal tensile st ress result s ind icate t hat many frac-
t ure pat hways form by t he linkin g up of offset j oint s t hro ugh t ransverse shear fract ures. Ot her pathways, however,
are fo rmed by linking up of j oints and contacts t hro ug h te nsile stresses associated w it h t he t ips of propagating
hydrofractures. How th e fractu re pat hways fo rm may affect t he subsequent permeability. In part icular, t ransverse
shear fractu res are likely to have relati vely small apert ures and limit t he grou ndwater t ransport.

Agust Gudm undsson, Ingrid Fje/dskaar & Otilie Gjesda/. Geolog ica/ lnstitu te, University of Berqen, Allegoten 41, N-5007
Berqen, Nor way (e-m ail: ag ust.g udm undsson@geol.uib.no)

Introduction extension fractures,w hich include many j oints and t ension

The observed linear and positive corre latio n between th e
current rate of postglacial uplift and yield of ground w ater
wells is one of th e basic hyd rogeologi cal relation s in
Norw ay.This corre latio n was init ially publi shed by Rohr -Torp
(1994), using five areas of different postglacial uplift rates
and includi ng a total of 1278 drilled wells in th e bedrock of
sout hern Norway. His results were suppo rted by th ose of
Morland (1997) who used a dat abase of 12,757 bedrock
wells,of w hich 8,726 are in Precambrian rocks.
This corr elation may be related t o increase in hydraulic
conductivity in areas of high upl ift rates compared wit h
th ose of low uplift rates (Gud mundsson 1999). High upl ift
rates would then coincide wit h areas of crustal doming
wher e associated t ensile st resses lead t o reactiv ation of old,
or form ation of new, wate r-conducting fractur es. How ever, a
fractur e syst em condu ct s wate r onl y if th e percolati on
threshold of th at system is reached, th at is, only if th e frac-
tu res form an inte rconnected cluster (Stauffer & Aharony
1994). Thus, in order to und erstand how uplift-gener ated
st ress field s contribute t o g roundwate r yie ld, we mu st also
und erstand how fractures link up into int erconn ected,
wat er-conduct ing syste ms.
This paper has thr ee prin cipal aims. First, t o present fi eld
examples of evo lvi ng small-scale fracture syste ms as an indi-
cati on of perm eabilit y development. All th e examples are
from th e island of 0 ygarde n ju st west of the city of Bergen.
Second, to present num erical models on how fracture sys-
tem s link up in j ointed and layered rock masses.The focu s is
on th e pro pagat ion of th e two main typ es of fractu res:
fractu res, and shear fracture s, i.e., faults.Third, to discuss th e
postglacial st ress fi elds in Norway in relatio n t o th e linking
up offract ures.
Fracture systems and faults
In many rocks, grou ndwater fl ow is primarily t hroug h frac-
ture syste ms, some of w hich eventually develop into large-
scale fault s.The fract ure networks may be primary str uctures
such as, for example, many joint systems in sediment ary and
ign eous rocks. Com monl y, th ese primary fractu res form
weaknesses from w hich inte rconnected tect oni c fract ure
systems and fault s develop.
As an example of a largely unconnected fract ure syste m,
one may consider exfo liation fract ures (sheet j oint s) in lay-
ered gneiss on th e island of 0yg arden (Fig. 1). These frac-
tures, subparallel wit h th e free surface, for m as a result of
rapid unloadin g of th e surface d ue to erosion and deg lacia-
tion. Re moval of th e overburden leads t o th e surface-parallel
compressive st resses exceeding t he vertica l st ress.The result
is fractur e developm ent parallel to th e maximum compr es-
sive st ress and, th erefore, parallel wi t h th e surface.The hor i-
zontal exfo liation fractures give rise to mechanical layering
w hich part ly foll ow s th e origi nal layering in th e g neiss, but is
partl y independent of th at layering (Fig. 1). Thus, th e rock
becom es divided into sheet s which, however, decrease
rapidl y in frequency wi t h depth (Fig. 1, cfJ ohn son 1970).
Exfoli ation fract ures t hat become interconnected may
cont ribute to bedrock permeability at shallow dep t hs.They,
like ot her fractu res, become int ercon nect ed through eit her
 NGU-BULL 439,2002 - PAGE 62 A GU ST GUOMUN O SSON, INGRJO FJEL O SKA A R & OT/LlE GJESOAL

Fig. 1. Suhorizonta l exfoliation

fractures in gne iss on t he island
of 0ygarden (Toft ey). West
Norway. Spacing of t he exfolia -
tion fractures increases rapid ly
wi th depth. View no rthea st; the
subver tical nor mal fault west of
t he person is th e on e in Fig. 3.

extensio n fractures or shear fractures . In an extensio n frac-

ture t he displacement is primarily perpe ndicular t o, and
away from, the fracture plane (Fig. 2).whereas in a shear frac-
ture t he main displ acement is parallel with the fracture
plan e (Fig. 3).

Fig.3. Steeply dipping no rmal fault interconnecti ng subhorizonta l exfo-

liatio n fractu res in th e g neiss of 0ygarden (Toft oy) (Fig. 1). Maximum
verti cal di splacem ent is 4 cm in t he cent re of t he fault. but decreases to
zero at the tip s.View northeas t; t he person s provide a scale.

Fig. 2. Linking up of exten sion fractures in gnei ss on th e island of
0ygarden (Toft ey). View northea st; t he fracture apertu re tends to
increase when meeting w it h the hori zontal exfoliation fractur es, at the
contacts wi t h the dark amph ibolite layers.The steel tape is 1 m long .
Extension fractures
Ext ension fractures (Fig. 2) are either tension fractures or
hydrofra ctures. Tension fractures form when the minimum
principal compressive stress (considered posit ive) is nega-
tive, that is, when there is an absolute tension in the crust.
 A G US T GUD MUN D SSO N, /NGRID FJ ELDSKAAR & O T/LlE GJESDAL NGU -BULL 439, 2002 - PAG E 63

Fig . 4. Well int erconnected,

orth ogo nal joint system in gneiss
on the island of 0 ygarden
(Toftey), View sout hwest; th e per-
son prov ides a scale.

They are mo stly limited to areas und ergoing acti ve exte n-
sion, such as areas of rifti ng and t hose of great postg lacial
upl ift such as in t he central part of Fenno scandia (M6rner
1980, Gudmund sson 1999).The maximu m depth to w hich a
tension fracture can propagate a.: from the surface (before
it chang es into a norm al fault) is giv en by (Gudm undsson
1992, 1999):
In contrast to tension fractures, hydrofractures can occur
at any crusta l depth. Hyd rofractu res are opened up by a flu id
pre ssure tha t is g reate r t han t he normal stress on the frac-
ture plane . Mo st hyd rof ract ures are extension fract ures
(Gudm undsson et al. 2001), in w hich case the norma l stress
on t he hydrofracture plane is th e min imum principal com-

dm" -R
- p,g (1)

where To is th e in sit u t ensile strength, P, th e de nsity of th e

host rock, and 9 th e accelerat ion du e to gravity.The joi nts in
Figs. 2 and 4 are in a crusta l layer with an average de nsity of
around 2500 kg rn' (Hansen 1998). lf th ese j oi nts are t ension
fractures, th eir maximum depths can be estim ated from Eq.
(1). Typical in sit u st rengt hs of bedrock are 0.5-6 MPa
(Amadei & Step hansson 1997), and t he acceleration due to
gr avity at th e surface is 9 = 9.8 m s' . Substit ut ing t hese val-
ues in Eq. (1), we obtain a.: as aro und 60 m (fo r To = 0.5 MPa)
and 700 m (for To = 6 MPa).
The presum ed tension jo ints in Figs. 2 and 4 are unlike ly
to reach th e above maxim um d epth because the hor izontal
exfo liat ion fractu res wou ld te nd to arrest th e joi nt t ip s. Also,
there is a corr elat ion bet ween ope ning of a tension fracture
at the surface and it s controllin g dim ension , defi ned as th e
smalle r of th e fractu re dip and strike dimensions
(Gudmundsson 2000). For a me chanically layered host rock,
th e controll ing dimen sion of a fracture is normally its di p
dim ension (Gud m undsson 1992, 2000). Thu s, tension frac-
tu res wi t h maximum surface openings of a few mill imetres
normally have di p (cont rolling) dimensions of several
metr es or, at most, a few te ns of metres.The interconn ected
j oint system in Fig.4 is th erefore likely to be limited to th e
upp erm ost t ens of metres or, at most , a few hundred met res,
Fig. 5. Dolerit e dyke, of Perm ian age, on th e island ofTysnesoy, south of
bot h as regards the depth of the exfolia tio n fractures as we ll Bergen, West Norway. View north east; th e dyke is 0.7 m thic k and dips
as the depth of the vertical tensionjoints. n ONw.
 NGU -BULL 43 9 , 20 0 2 - PAGE 6 4 A GUST GUDMUNDSSON, INGRID FJELDSKA AR & OT/ LlE GJESDAL

pre ssive stress.The normal condition for hydrofra cture for -

mat ion can t hus be given as (Jaeger & Cook 1979):

Pr"'O", + To (2)
wh ere P, is th e t otal fluid pressure, 0", is th e minimu m corn-
pressive princi pal stress (normal to the hydrofracture),and To
is th e in situ tensile streng t h of the host rock .
Because hydr ofractures can form at any crustal depth,
t hey are likely to be generally mo re im por tan t for t he devel -
opment of fracture pathways and permeability in rocks than
ten sion fract ures. For examp le, many joi nts (Figs. 2, 4) may
form as hydrofractu res w here the fluid disappears subse-
quent to the joint for mation. In genera l, where groundwater
or gas is th e fluid that dri ves a fracture ope n, it is difficu lt to
confi rm that t he fract ure was generate d by a fl uid. Oth er
fractures, however, are generated by flui ds or magma t hat
freeze in t he fracture subseque nt to its formation . Well
kno wn exam p les are dy kes, sills, inclined sheets and mineral-
filled veins. Com monly, mineral veins form net works of
ext inct geot hermal system s that provide important infor-
mat ion on permeabi lity gene ratio n t hroug h th eir em place-
ment (Gudm undsson et al. 2001). Some veins may reach
leng th s of several hund red metres, bu t most have lengths of
several met res or less (Gudmu ndsson et al. 2001L and are
thu s individually too small to have much effe ct on ground-
water flow.
Thick sil ls can have considera ble effects on groundwa te r
flow, bu t generally t he most important 'f rozen' hydrofrac-
t ures for groundwater flow are regional dykes (Fig. 5). In
Norway, dyk es are co mmo n along th e west coast and in the Fig.6 . Ext ension fractur es link ed up through transverse shear fractures,
Oslo reg ion (Thon 1985, Sundvoll & Larsen 1993, Torsvik et modelled in Figs. 7-8.View NNE; the lengt h of the steel tap e is 0.8 m.

al. 1997, Fossen & Dunlap 1999). Many dykes are of dense
basalt or dolerite wi t h low matrix permeability and act as
barr iers to t ransverse flow of groundwater. Ot hers, par ticu -
larly thick and fractured dykes, may be sources of ground wa-
te r (Sing hal & Gupta 1999).
Perhaps th e most important effect of dykes on gro und-
water flow in Norway, however, is that the basalt ic dyke rock
has nor mally very different mechanica l properties fro m it s
host rock (Fig. 5). It fo llo ws t hat stresses tend to conce ntrate
at t he con tact between the dy ke rock and the host rock and
generate fract ures th at may conduct groundwater. Dyke s
form th eir pathways in a similar way to t hat in th e mo de ls of
hydrof ract ure formation belo w. It follows that groundwater
condu it s at con tacts between dyke rock and host rock wi ll
foll ow th ese same path ways.
Fault s
Fault s may in itiate as small-scale shear fractu res (Fig. 3) or,
perhaps more commo nly, develop during the linking up of
small fract ures of various types (Fig. 6). In vertical sections,
shear fractures (like tension fractures ) norma lly have diffi-
culty in prop agat ing t hrou g h horizontal discontinuities such
as open exfol iation fractures (Fig. 3). At such discontinu it ies,
a shear f ract ure may eit her becom e offset or arrested.
In lateral sect ions (Fig. 6), shear fractures (and exte nsion
fractures ) normally grow by th e linki ng up of gradua lly
larger segm ents. These segm ent s may init ially be offs et
jo ints form ed early in the evolu ti on of th e ho st rock.
Alternat ively, lat er-formed tecton ic fractu res may propagat e
and link up into larger, segme nte d fract ures (Fig. 6). ln do ing
so, they form an int erco nnecte d system of fractu res or seg-
ment s that has th e potent ial of conducting g roundwater
along its entire leng t h.
The linking up of small-scale j oi nt s and fract ures into
larger shear fractures or fractu re systems is,in detai l, a com -
ple x process th at is sti ll only part ly un derstoo d (Cox & Scho lz
1988, Gudmundsson 1992, Acocella et al. 2000, Mansfi eld &
Cartwrig ht 2001). This process, however, largel y controls
w het her or not th e perco lat ion threshold of a fract ure sys-
tem is reached and, thereby, th e bed rock perme abilit y. Some
ge neral aspects of linking up of fractures into segmented
extension fractu res and shear fractu res are illustrate d by t he
fo llo w ing nu mer ical models.
Num erical models of fracture growth
All the models are made using t he bound ary-elem ent pro -
gram BEASY (1991Lth e method being descr ibed in deta il by
 AGUS T G U DM U N DSSON, ING RID FJEL D SKA A R & O T/ LlE GJ ESDA L
Fig. 7. Bound ary-element mod el show ing th e tensile st ress, in mega-
pascals as absolute values, between t he nearby t ips of two extension
fracture s of equal length subject to 6 MPa ten sile loading (indicated by
hor izontal arrows). In this model , the offset (horizontal distance) and
und erlapping (vertica l distance) between th e nearby t ips are bot h
equal to half t he fract ure length.
Fig.8. Bound ary-element mod el showi ng th e shear st ress, in meqa-pas-
cals,for the same fractur e config urat io n and loading as in Fig. 7.
Brebbia & Dom ingu ez (1989). The progra m makes it possi-
ble, for given bound ary condition s, to calculate t he stresses
and disp lacements at any points of interest inside the
model.The results are most conveniently presented as con-
to urs of equal stress magnit ude, g iven in mega-pascals, as
well as opening up of discontinuities such as fract ures and
contacts. The models are divided int o one or more zones
(correspondi ng, for examp le, to mechanical layers), each
wi th uniform mechanical prop erties; here Young's mo dulus
(also referred t o as stiffness) and Poisson's ratio .
The fir st mod els (Figs. 7, 8) indicate how offset fractu res
may link up to form an interconnected system of en echelon
segments (Fig. 6). In these mod els, the only loading on the
offset fract ures is a fracture-perpendicul ar t ensile stress of 6
MPa, a value simil ar to th e maximum in sit u tensile strength
N GU - BULL 439, 2002 - PA GE 65
of gneiss (Amadei & Stephansson 1997).This load is applied
as t ension perpendicu lar to the two bound aries of the
model that run parallel wi t h th e fract ures (Figs. 7, 8). A uni-
form Poisson's ratio of 0.25, used in all th e mo dels,is appro-
priate for gn eiss (Jumikis 1979, Bell 2000). The uniform
Young's modulu s used in th e first models, 10 GPa, is in the
lower range of laborato ry values of Young's mod ulus for
gn eiss (Jumikis 1979, Bell 2000). Because in situ rocks nor-
mally have much lower Young's moduli than small-scale lab-
oratory measurements of th e same rock typ es (Goodma n
1989), this value is appropriate. The fractures are modelled
as inte rnal springs, each wi th a stiffness of 6 MPa/m . The
units of fractur e st iffn ess differ from th ose of rock stiffness
(Young's modul us) because the stiffness of a solid rock is
determined from a stress-strain curve whereas that of a frac-
ture is determined from a stress-displacement curve
(Hudson & Harrison 1997).
The result s show zones of high tensile (Fig. 7) and shear
(Fig. 8) st resses between the nearby t ips of th e fractures . For
the given loading of 6 MPa t ensile stress, these zones gener-
ate potent ial t ensile stresses in excessof 11 MPa (Fig. 7) and
potential shear st resses in excessof 13 MPa (Fig.8).These are
the relevant values, but mu ch higher potential stre sses
occur at t he fract ure tips . A tensile stress of 11 MPa is around
double th e maximum in situ tensi le strength of gneiss
(Amadei & Stephansson 1997), so th at tensile fractur es
would be expected to form inside the high-stress zone.
Similarly, th e in situ shear strength of rock, which is com-
mo nly rough ly twice its tensile st rengt h (Jumikis 1979,
Farmer 1983), is also normally much less than 13 MPa, in
wh ich case shear fractures wo uld be expected to form in the
high -st ress zone.
The ESE-WNW-t rend ing and straight transverse fractures
connecti ng the main NNE-SSW-trending fractures in Fig. 6
are likely to be mostly shear fract ures. Shea r fractu res tend
to be st raig ht and form inside the zone of maximum shear
st ress (Fig. 8), w hereas most ten sile fractures generated in
such a high -stress zone are gentl y curving and for m hook-
shaped fractures (Gud mu ndsson et al. 1993, Acocella et al.
2000).
The ot her models (Figs.9-12) show how a hydrofract ure,
dr iven by an internal flu id overpr essure, ope ns up discon t i-
nuities (joint s and cont acts) in t he host rock ahead of the
hydrofracture tip. In all th ese mo dels, the hyd rofractu re is
mo del led as a vertica l ext ension (mo de I) fractu re wi th an
internal fluid overpressure th at varies from 10 MPa at the
bottom of t he fracture to 0 MPa at the fluid fro nt. The fluid
fro nt coincides with the fractur e ti p in th e mod els in Figs. 9-
11, but in Fig. 12 the flu id front is 0.1 units below the fracture
tip. Each model has a unit heig ht ; all the dim ensions are
given as a fraction of this unit. Only th e upp er half of each
hydrofracture is modelled, and the mode ls are fastened at
the lower corner s, so as to avoid rigid body t ranslatio n and
rotation .
As in th e earlier mo dels (Figs. 7, 8), a uniform Poisson's
 NGU -B ULL 43 9 , 2 0 0 2 - PAGE 6 6 AG UST GUD M UN DSS O N, IN GRID FJ ELDS KA A R s OT/Ll E GJE SDAL

Surface

Tension fractures

/ \'"

Fig. 9. Boundary-element mode l show ing how the ten sile st ress, in
mega-pascals, around the t ip of a vertical hydrofractu re opens up vert i- Fig. 10. Same model as in Fig. 9,except t hat a third jo int has been add ed
cal jo ints and a weak contact int o te nsion fract ures. Figs.8-12 show only (in the centre) and the weak contact is here tw ice as long as in th e pre-
tensile st ressesin th e range 1.5-10 MPa. vious (Fig. 9) mode l.

rat io of 0.25 is used and , for the first t wo models (Figs. 9-10), modulus of 5 GPa, is embedded in a layer A of 10 GPa, bo th
also a uniform Young's mo dulus of 10 GPa. In t he t hird and of w hich have a Poisson 's rat io of 0.25.There are two ver t ical
fourth models (Figs.11-12), ho w ever, th ere is a softe r layer of discon tinuit ies, each of lengt h 0.4 uni ts and present ed by
5 GPa embedded in th e stiffer layer of 10 GPa, rep resenti ng in terna l spring s of st iff ness 6 MPa/ m th at reach th e surface.
layering wi t hi n th e gneiss (Figs. 1-3).As before, the joi nt s and Fig. 11 show s tha t, even if t he diff erence in st iffn ess bet wee n
horizontal contacts in the roc k are m od elled as int ern al
spring s wit h a stiffness of 6 MPa/ m. Here, this stiffness may Surface
be ta ken as representative of a soft, but elastic , infil l in a joint
and a w eak sed imentary or soil material at a con tact. For A
comparison, sma ll-scale laborato ry samples of clay and
w eak mudsto ne have Young 's modu li so low as 3 MPa (Bell
2000). Nearly ide nt ical result s w ere obta ined in m odel run s ension fractu res
w here th e con tacts and joints w ere op en (em pt y) and th us B
wi t h zero material stiffness.
In th e fir st mod el (Fig. 9), a hydrofractu re t ip at a depth of
0.5 units belo w the surfac e approaches th ree discontinu -
iti es, presented by t wo vertical j oi nts and a hori zon tal con- A
tact, each w it h a length of 0.2 uni ts. Thi s m od el reflects, for
examp le, the mech anical effec t s of a verti cal hyd rofractu re
appro aching the sur face in Fig. 4. All th e th ree discontin u-
iti es ope n up, and t ensile stresses con cent rate around the
t ip s of th e cont act and th e low er t ip s of t he j oin ts. These
stress conc entrations indicate a tend enc y fo r th e j oint s and
t he con t act to link up, as is seen in the next model (Fig. 10).
w here a new centra l joint has also been added .The cont act
has the greatest open ing , followed by t hat of th e central
joint (Fig. 10) and, t hen, the joints to each side. All j oi nts have Fig. 11. Boundary-element model showing how the tensile stress, in
th e greates t open ing s in th eir deep er part s. mega -pascals, around the tip of a vert ical hydrofracture open s up verti -
cal joints into tension fractures.The joi nts are located in a soh layer B
The next t wo models (Figs. 11-12) show t he effects of
(Young's mod ulus 5 GPa) embedded in a stiffer layer A (Young's modu -
m echanical layering on discont in uiti es ope ni ng ahead of a lus 10 GPa). The soh layer reduce s the ten sile st resses associated wi t h
prop agat ing hydro fract ure. A soh layer B, w it h a Young 's the hydrofracture tip.
 AGUS T GUDMUNDSSON, I NG RID FJELDSKAAR & O T/Ll E GJESDA L
urface
A
Hydrofracture ~
Fig. 12. Same model as in Fig . 11 except th at t he t ip of th e hydrofra ctur e
here meets wit h th e bott om of th e soft layer B.
layers A and B is only by a factor of 2, th e tensile st ress
related to th e hydrofracture is much di min ished in th e soft
layer. There is a small ope ning of the vertical joints in t heir
deepest parts wh ile the near-surface part s remain closed.
When, however, the hydrofractur e has propagated an
additiona l 0.1 unit towards th e surface,th at is, to the bottom
of the soft layer B, th ere is considerable ope ning of t he
joints, the maximum being at the free surface (Fig. 12).Thus,
for th e given loading conditions, when th e hydrofractur e is
at 0.4 units below th e free surface,th e geomet ry of th e verti-
cal discontinuities changes from th e maxi mum opening
being in their deepest parts to its being at the surface. The
apertu re of an extension fract ure com monly becomes much
larger at a free surface, or at an op en, sliding di scontinuity,
because of th e lack of elastic constraint (Fig. 2).
Stress fields in Norway
The numerical mod els above consider t wo loading condi-
ti on s: exte rnal ten sile st ress and int ernal flu id overpressure.
Here we explore briefly how th ese loading condit ions match
with current ideas on the stress fie ld in Norway.
The present- day st ress fi eld in Norway has recently been
subject to a detailed st udy, using focal mech anisms, overcor-
ing and borehole breakout data (Feje rskov et al. 2000, Hicks
et al. 2000). For West Norway,t he main st ress is a WNW-ESE-
t rending compression offshore and a weaker NE-SW-t rend-
ing extension onshore.The offsho re horizont al compression
is wi dely att ributed t o ridge push (Fejerskov & Lindh olm
2000). Both t he onshore ext ension as well as th e offs hore
compression can, however, partly be exp lained in terms of
the inferred glacial erosion and postgla cial upli ft , depending
NGU - BULL 439 , 2002 - PAGE 67
on where the margin of th e up lifted crustal plate is taken to
be (Gud mundsson 1999).Thus,th e present -day st ressfiel d in
Norway de pends on its late-glacial hist ory.
There is considerable evidence that th e w hole of Norw ay
became degl aciated during th e period from 8000 to 12,000
B.P. The deglaciation began in the coastal areas (Mangerud
1991, Sejrup et al. 2000) and continued to th e highl and s
w here it was essentially complete some 8000 years ago
(Mangerud 1991, Nesje & Dahl 2002). In West Norway, the
postg lacial stre ss field generated neote ct onic activity
(Anundsen et al. 2002, Helle et al. 2000). The best docu-
ment ed neotectonic activity in Norwa y, howev er, is from th e
Lapland prov ince in the far no rth w here several reverse and
normal faults have been active in t he Holocene, some up to
the present (Olesen 1992, Dehls et al. 2000), and give n rise t o
hyd rological changes (Olesen et al. 2000).
During deg laciation in the coastal areas, tem porary t en-
sile stresses may have followed the ini tial comp ressive
st resses associated with t he formation of the ho rizontal
exfo liat ion fractu res (Fig. 1). Because rocks are much weaker
in tension than in compre ssion, the linking up of existing
joints and fracture s subject t o suita bly orient ed t ensile
stresses (Figs. 7, 8) is muc h more likely than for an equally
large compressive stress. Extensi on is still occur ring in some
coasta l parts of West Norway (Hicks et al. 2000). Postglacial
tensile stresses at shallo w crustal depth s are thus likely to
have been common, th ereby support ing the use of t ensile
loading in some of th e nu merical models (Figs.7, 8).
The ot her nu merical models use internal fluid overpres-
sure as th e only loadin g mechan ism (Figs. 9-12).This t ype of
loading emphasi ses the basic conditio ns for t he propaga-
tion of vertical hydrofra ctu res.The models in Figs. 9-12 also
indica te th at at least some of the horizontal exfoliation frac-
tures (Figs. 1-3) may have opened up dur ing flow of an over-
pressured groundwater associated wit h t he hyd rof ract ure
propagation. Similar suggestions have been made for hori -
zontal fractures in Sweden, some of w hich are thoug ht to be
partl y generated by flu id overpr essure beneath the retreat-
ing ice front during deglac iation (Talbot 1999).
Discussion
The relationshi p betwee n current postglac ial upl ift ratesand
yield of gro undwa ter wells (Rohr-Torp 1994) is of fund amen-
tal importance in the bedroc k hyd rogeology of Norway. To
unde rstand this relationship in mechanical t erms we must
explain how t he rate of up lift aff ects th e yield. The mo st
likely phy sical parameter t o be affected by postglacial uplift
and associated stresses is th e hydr aulic conductivity. In t he
bedrock of Norway, hydraulic conductivit y is com monly
almost entirely determined by interconn ected fract ures.
How th ese fract ures become int erconn ected and grow is
thu s perhaps the key to under standing this empirical rela-
tionship.
In this paper we use numerical models to indicat e some
of t he basic ways by w hich inte rconnected fracture systems
 N GU - BU L L 43 9 , 2 0 0 2 - PAGE 6 8 A GUST GUDMUNDSSO N , IN GRID FJELD SKA A R
may deve lop, using fract ures from the coastal areas of
Norway as examples (Figs. 1-4). How th ese fract ure systems
gro w affect s their effect iveness in transporting groundwa-
t er. For example, whe n the initially offset fractures link up
through transverse shear fract ures (Figs. 7, 8) th e transverse
fractures may limit the groundwater transport of the syst em.
This follows because the normal compressive stress on a
shear fracture is g reate r th an th e min imum compressive
princ ipal st ress so that the overpressure of ground water
flowing through a transverse fracture is usually less than
that of ground wat er flo wing through an exte nsion fracture.
Also, for a given controlli ng dimension of a fracture , t he
apertu re of such a tran sverse fracture is norma lly less than
th at of an eq ual-sized exte nsion fractur e. In min eral vein sys-
tems in Norway and elsew here, veins in shear fractur es are
norma lly th inner than veins in extension fractu res
(Gud mundsson et al. 2001).
When t he groun dwater path way is formed by propaga t-
ing hydrofractu res (Figs. 9-12), the direct ion of subsequent
groundwater flow along t he pathway depends on several
parameters. For example, once the hyd rofractu re reaches
the contact in Fig. 10, the grou ndwater could , theoretica lly,
flow into any of t he vertical joints. Although the open ing of
th e centra l joint is th e greatest, and wo uld thus normally
favour the ground water flo w, the pressure gradient also
influences th e flow directio n. For hydrofractures in an elastic
crust, th e pressure gradient depends on th e state of st ressin
th e host rock (Gudmundsson et al. 2001, Gudmundsson
2001). By contrast , if t he rock behaved in a rigid manne r sub-
seq uent to th e opening of t he discontinuit ies, th e event ual
flo w direction of groundwater would be partly determined
by the hydra ulic gradient.
In conclusion, the data and models presented in t his
paper indicate that rapid erosion and postgla cial upl ift are
likely to have gene rated stress fields suitab le for the link ing
up and growth of interconnected fract ure systems in variou s
parts of Norway. In particular, temporary ten sile stresses and
associated fracture gro wth may have significantly increased
the hyd raulic conduc tivity in th e bedroc k of Norway. The
poten tial ten sile stresses increase wit h increasing rate of
postgla cial upl ift (Gudmundsson 1999) so that th e fracture-
generated permeabil ity and associated yield of bedro ck
groundwater wells woul d be expected to follow em pirical
relatio nships similar to those proposed by Rohr-Torp (1994).
Ackno wledgements
We t hank Alvar Braath en and Helg e Ruistu en for he lpfu l co m m ents on
the manuscri pt. Thi s wo rk w as su pporte d by several grant s from t he
Norway Research Coun cil as w ell as a gr ant from t he Europ ean
Comm issio n (cont ract EVR1 -CT-1999-40002).
References
Acocella, v., Gudmund sson , A. & Fun iciell o, R. 2000: Interaction and li nk -
age of exten sio n fract ures and norm al fau lt s: exam ples from t he rif t
zone of Iceland . Journ al of Stru ctu ral Geology 22,1 233- 1246.
s OT/ Ll E GJESD AL
Am ad ei, B. & Steph an sson, O. 1997: Rock Stress an d its Measurement.
Cha pman & Hall, Lond on , 490 pp .
An undse n, K., Grim stv eit, L., Harsson , B.G., & Holsen, J. 2002:
Measuremen t s of neo tectonic movements in southern Norway :
implicat ion s for for m er ice t h ickn ess estima tes. Norsk Geolog isk
Tid skrift (in p ress).
BEASY, 1991: The Boundary Elem en t Analysis System User Guide .
Computation al Mechani cs, Bo ston.
Bell, EG. 2000: Engineering Proper ti es of Soils and Rocks, 4'" edn. Blackw ell,
Oxford, 482 pp .
Brebbia, CA & Doming uez, J. 1989: Boundary Elem en ts: an Introduction
Cou rse. Com p ut at ional Mec hani cs, Bo ston, 314 pp .
Cox, S.J.D. & Scholz, CH. 1988: On t he for mat io n and g rowth of fau lts : an
exper ime ntal st udy. Journal of Structural Geology 10,413-430.
Dehls, J.F., Olesen, 0 ., Olsen, L.& Blik ra, L.H. 2000: Neotecton ic fault ing in
northern Norw ay; t he Stuo rag urr a an d Nor dmannvikdalen post-
glacial fault s. Qua terna ry Science Reviews 19,1447-1 460 .
Farm er, I. 1983: Engin eering Behaviour of Rocks, 2"" ed. Chapman & Hall,
Lon d on , 208 pp.
Fej erskov, M . & Lind holm, C 2000: Crustal stresses in an d around
Nor wa y : an evaluati on of stress generating mechan isms. In:
Nott ved t, A. (ed.), Dynam ics of t he Nor w egian Ma rgi n. Geological
Socie ty of London, Special Publicat ion 167,451-467 .
Fej erskov, M ., Lindholm , C , My rva ng, A. & Bung um, H. 2000: Crusta I stress
in and around Norway:com pilation of in-sit u stress o bservations. In:
Nottvedt, A. (ed.), Dynam ics of the Norwegian Margin . Geological
Society of London, Special Publica tio n 167,44 1-449.
Fossen H. & Dunlap, WJ . 1999. On th e age and tecton ic sign ificance of
Perm o -Triassic d ikes in the Bergen-Su nnhordalan d region, south-
western Nor w ay. Nor sk Geologisk Tidsskrift 79, 169-1 77.
Goodman, R. E. 1989: Int roductio n to Rock Mechanics, 2"" ed. Wiley, New
York, 562 pp .
Gudmu nd sson, A. 1992: Formatio n an d g rowth of normal faults at the
di ver gent p late bo un dary in Iceland. Terra Nova 4, 464-47 1.
Gudmu nd sson, A. 1999: Postgla cial cru stal do m ing , stresses and fractu re
for m at ion wi t h application to Norway. Tectonoph ysics 307,407-419.
Gud m und sson, A. 2000: Fractu re dim ensions, displacements and flu id
t ran spor t. Journ al of Structural Geology 22, 1221-1 231.
Gudm und sson, A. 2001: Flu id overp ressure and flow in fault zones:fiel d
m easu remen t s and mod els. Tectonophysics 336, 183-197 .
Gudmundsson . A., Brynj o lfsson., S.& Jon sso n, M .T. 1993: St ruct u ral analy-
sis of a t ran sform fau lt-rift zone junction in North Icelan d.
Tecton oph ysics 220, 205-22 1.
Gud m und sson, A., Berg, 5.5., Lyslo, K.B. & Skurtvei t, E.2001 : Fracture net-
wo rks and fl uid t ranspor t in active fault zon es. Journa l of Structural
Geolog y 23,343-353.
Han ssen, T.H. 1998. Ro ck st resses and tecto nic act ivity. In:
Fj ell spreng ing stekn ikk, Berg m ekanikklG eot eknikk. Tron dhe im ,
No rw ay, Chapter, 1-24.
Helle, S.K., Rye, N., & Stabell, B. 2000 : Shorelin e dis plac ement and faul t
act iv ity in th e Hard angerfj ord, w este rn Norway, after t he deglacia -
t ion. Geon yt t, Norsk Geol og isk Forening 24, Norsk Geologisk
Vint erme te, Trondheim 6-9 Jan uary.
Hicks, E.C , Bungum, H. & Lind ho lm , C 2000: Stress inv ersion of earth-
qua ke focal mecha nism sol ut ions from onshore and offshore
Norway. Norsk Geolog isk Tidsskriit 80, 235-250.
Hud son, J.A. & Harrison, J.P. 1997: Engineering Rock Mechanics: an
Int roduction to the Principles. Pergam on, Oxford, 444 pp .
Jaeger, J.C & Coo k, N.GW . 1979. Fundam ent als of Rock Mechanics, 3'" ed.
Chapman & Hall, London, 593 pp .
Joh nson, A.M. 1970: Physical Processes in Geology. Freeman, Cooper &
Company, San Francisco, 577 p p.
Jum ikis, A. R. 1979: Rock Mechanics . Tran s Tech Pub licat ions, C1austhal.
356 pp .
Mang eru d, J. 1991: The Scandi navian Ice Sheet throug h t he last inter-
glacial/ glacial cycle. In: Frenzel, B. (ed.), Probl em s in th e Histor y of
th e Climate Dur in g the Last 130,000 Years. G. Fishe r, St uttgart, 307-
330.
Ma nsfield, C & Cart w right, J. 2001: Fault grow t h by linka ge : obse rvat ions
 AG US T GUDMUN D SSON, INGRID FJEL D SKA AR & O T/ Ll E GJE SD AL NGU -BU LL 43 9 , 200 2 - PAGE 6 9

and im p licat ions from an alo gu e m od els. Journal of Structu ral

Geology 23, 745-763
Mo rland, G. 1997: Petrology, Lith ology, Bedrock Structures, Glaciat ion and
Sea Level. Imp ortant Factors for Ground water Yield and Composition
of Norweg ian Bedrock Boreho les? PhD. Thesis, Montanun iversit at
Leo be n, 274 p p.
M6rner, N.A. 1980: The Fen no scandian uplift: g eological data and th eir
geodynam ic implicati on s. In: M6 rne r, N.A. (ed.), Eart h Rheol og y,
Isostacy and Eustacy.Wiley, New York, pp .251-284.
Nesje, A. & Dahl, 5.0. , 200 2. Lat e glaci al and Hol ocene glaciers and cli-
mate in sout hern No rway . Encyclopedia of Qua ternary Science (in
p ress).
Olesen, O. 1992:The St uo ragu rra Fault, evid ence of neotecto nics in t he
Precam brian of Finn mark, nor t hern Nor w ay. Norsk Geologisk
Tidsskrift 68,107-118.
Olesen, 0 ., Dehl s, J., Hilm o, B.O. & Olsen, L., 2000: Neot ecton ics and seis-
m ic pu mpi ng of gro un dw ater and hydrocarbon s in No rwa y.
Geonytt, No rsk Geo lo gisk Forenin g 24, Nor sk Geologisk Vinterm ote,
Tron dh eim 6-9 Janu ary.
Roh r-Tor p, E. 1994: Present u plift rate s and gr ou nd w ater potential in
Nor w egi an hard rocks. Norges geologiske und ersekelse Bulletin 426,
47-52 .
Sejrup , H.P., Larsen, E., Landvik, J., King, E.L., Hafiid ason , H. & Nesje, A.,
2000 : Quatern ary g laciat io ns in sourt hern Fenn o scandi a; ev ide nce
from so ut h-w estern Nor w ay and the north ern Nor th Sea regi on.
Qua ternary Science Reviews 19,667-68 5.
Singha l, B.B.5. & Gupta, R.P. 1999: Applied Hydrogeo/ogy of Fractur ed
Rocks. Kluwer, Lo ndon , 400 pp .
Stauffe r, D. & Ah aron y, A. 1994: Introduction to Percolation Theory, 2" ed.
Taylo r & Francis, Lond on, 181 pp .
Sundvoll, B. & Larsen, B.T. 1993: Rb-Sr and Sm-Nd relationsh ip s in dyk e
and sil l intru sion s in th e Oslo Rif t and relat ed areas. Norges qeolo-
giske undersekelse Bulletin 425, 25-42.
Talb ot , C. J. 1999: Ice ages and nuclear w aste isol ati on . Engineering
Geology 52,177-192.
Tho n, A., 1985:Th e Gullfj ell et Ophioli t e Com p lex and t he st ruct u ral evo-
lution of t he Maj or Bergen Arc, West Nor wegian Caled on id es. In:
Gee, D. G. & St ur t, B. A. (eds.), The Caledon ide Orogen - Scand ina via
and Related Areas, part I .John Wiley &Sons,Chichester,671-677.
Tor svik, 1.H., And ersen, 1.B., Eid e, EA & Walde rhaug, H.J. 1997: The age
and t ect on ic sig nificance of d olerit e d ykes in w este rn Nor w ay.
Journa l of the Geological Society, Lon don 154, 96 1-973.