ac

Sa

5 Ov

a r t i c l e i n f o

Article history:Received 30 July 2014Received in revised form 25 June 2015Accepted 15 July 2015Available online 17 July 2015

Keywords:

0). The extent of thedepth of the tunnelquifer characteristics.drogeological impacts

Engineering Geology 196 (2015) 158170

Contents lists available at ScienceDirect

Engineering

j ourna l homepage: www.e ls2008; Gutirrez et al., 2014). Active sinkholes resulting from human related to the drop of water table, such as the total or seasonal dryingup of streams, springs, wells and wetlands (Vincenzi et al., 2009,2014). This causes important damage on the environment andprocesses, but their increasing frequency is usually related to suddenchanges in the natural hydrogeological system induced by humanactivities like water pumping, quarry de-watering or tunneling(Newton, 1986; Tihansky, 1999; Waltham et al., 2005; Waltham,

table on the drilled aquifers (Raposo et al., 201water table depression is conditioned by thebelow the original water table, as well as the aFor this reason, tunneling usually produces hy1. Introduction

The development of cover-subsidence and cover-collapse sinkholesconstitutes the most common geohazard in karst landscapes. Thesephenomena can develop naturally due to a cluster of inter-related

activity have been widely reported (e.g. Gutirrez-Santolalla et al.,2005; Guerrero et al., 2008; Galve et al., 2012; Song et al., 2012).

Tunnels in karst areas may generate multiple engineering and envi-ronmental problems (Milanovic, 2004; Casagrande et al., 2005; Alijaet al., 2013). A tunnel works as a drain, producing a drop in the water Corresponding author.E-mail addresses: pvalenzuela@geol.uniovi.es (P. Valen

mjdominguez@geol.uniovi.es (M.J. Domnguez-Cuesta), m.m(M. Melndez-Asensio), mjimenez@geol.uniovi.es (M. Jimjoseantonio@saenzdesantamaria.es (J.A.S. de Santa Mara).

http://dx.doi.org/10.1016/j.enggeo.2015.07.0070013-7952/ 2015 Elsevier B.V. All rights reserved.expected to progress in the future. 2015 Elsevier B.V. All rights reserved.Pajares TunnelsCantabrian RangeSinkholeImpactRainfallrunoff modelGISa b s t r a c t

Two parallel base tunnels (Pajares Tunnels) were built from 2005 to 2009 through the Cantabrian Range(NW Spain), crossing an alternation of Paleozoic formations (shale, sandstone, quartzite and limestone)characterized by a complex geological structure. A section of the tunnels was built 450 m depth below AlcedoValley (Len, N Spain). Some evidence of collapse and swallow holes have been appearing from 2007 to presentat the bottom of the valley. Although the streamwas channeled in 2009 to control water inltration, the processcould not be avoided, constituting a good example of geomorphological impact caused by a base tunnel. Themanagement of hydrogeological, geomorphological and climatological information using a GIS allowedmappingthe affected area and estimating themeanwater volumeof inltration into the sinkholes, and the runoff decreasein the Alcedo Stream after the drilling of the tunnel. Precipitation data series (19702000) and four spatialvariables (outcrops, shallow deposits, slope and vegetation) were used to create a rainfallrunoff model.Presently, geomorphological evidence includes 4 main sinkholes (812 m long), 13 minor hollows, 7 swallowholes and a 120m2 areawith subsidence evidence,whichdevelopedover Quaternary deposits covering karstiedlimestone bedrock. These active swallow holes capture the surcial runoff of the Alcedo Stream throughout theyear. Because of that, the upper reach of the stream is isolated from the rest of the uvial network. The suddendevelopment and active growth of cover-collapse sinkholes is consistent with 1) the drop of the water table bytunnel drainage after excavation, 2) the increase in percolation from surcial runoff and 3) the internal erosionof the overlying Quaternary sediments by suffosion processes. The estimated mean water volume of inltrationinto the sinkholes is close to 308,903 m3 yr1, and the Alcedo Stream runoff in the natural base level hasdecreased by 35% throughout the year after the tunnel perforation. At present, the process is active and it isGEHMA Geologa y Geotecnia, S.L. C/ Prez de Ayala, 1, 3 C, 33007 Oviedo, SpainActive sinkholes: A geomorphological imp(Cantabrian Range, NW Spain)

Pablo Valenzuela a,, Mara Jos Domnguez-Cuesta a, MMontserrat Jimnez-Snchez a, Jos Antonio Senz dea Department of Geology, University of Oviedo, C/ Jess Arias de Velasco, 33005 Oviedo, Spainb Instituto Geolgico y Minero de Espaa (Ofce of Oviedo), C/ Matemtico Pedrayes, 25, 3300c ozuela),elendez@igme.es

nez-Snchez),t of the Pajares Tunnels

nica Melndez-Asensio b,nta Mara c

iedo, Spain

Geology

ev ie r .com/ locate /enggeoconicts with the regional population, creating problems related todrinking water supply, agriculture, tourism and other activities(Sjolander-Lindqvist, 2005; Chiocchini and Castaldi, 2011).

Drilled from 2005 to 2009, Pajares Tunnels are framed within theproject for the Pajares Railway By-pass, which aims to replace the

produced the ooding of the stream and the undermining and fractur-ing of the channel due to the development of new sinkholes. Presently,some of these active sinkholes keep on growing and drain all thesurcial runoff from the upper catchment of the Alcedo Valley, dryingup the Alcedo Stream throughout the year.

Some hydrogeological studies before and during the perforation ofthe tunnels have been undertaken, but very few hydrogeological datahave been published (lvarez Dez et al., 2009; Garrido Ruiz et al.,2009; Arnanz Gonzlez et al., 2009). Nowadays, works continue onthe tunnels and in the Alcedo Valley, constituting a controversial issue.

Themain goals of this paper are (i) the geomorphological character-ization of the sinkholes developed on the Alcedo Valley and (ii) theestimation of the mean water volume that drains into these sinkholesand the subsequent runoff decrease in the Alcedo Stream.

2. Description of the study area

The Alcedo Valley is a small watershed (6.87 km2) located on thesouthern ank of the Cantabrian Range (Sierra del Rozo, N of Len,Spain) with an altitude ranging between 1803 and 1160 m a.s.l.(Fig. 1). The area, characterized by continental high mountain weather,is under pluvionival conditions with average temperatures below 10 Cand mean precipitation between 10001300mm yr1, reaching a peakduring the winterspring period (Galn, 1990; Garrido Ruiz et al.,2009). Most of the valley preserves its natural vegetation of beech, oakand birch alternating with bushland and grassland. Due to its ecologicalinterest, this uninhabited area has been included in the UNESCO

Fig. 1. Geographical location of the Alcedo Valley.

159P. Valenzuela et al. / Engineering Geology 196 (2015) 158170current railway line over the Pajares Mountain Pass by a new high-speed line between Asturias and Len (NW Spain) (Fig. 1). The tunnelscross theCantabrian Range, amountainous area characterized by a com-plex geological structure and a great lithological variety, which can begrouped in three main kinds of Paleozoic materials (Mguez Bailo,2005): (i) shale and shalesandstone (San Emiliano, Oville, Formigoso,Huergas, La Vid, San Pedro, Ermita and Subhullero Fms.); (ii) sandstoneand quartzite (Herrera, Barrios, San Pedro, Oville and San EmilianoFms.); and (iii) calcareous materials (Lncara, La Vid, Alba, Portilla,Santa Luca, Barcaliente and Valdeteja Fms.).

Pajares Tunnels constitute a complex underground structure withthe characteristics of the biggest high-speed railway base tunnels.Their layout between Pola de Gordn (Len) and Telledo (Asturias)has a NNWSSE orientation and reach a maximum depth of 1100 m.It bridges an altitude difference of 414.6 m, showing a continuouslongitudinal 16.386 gradient descending toward the Asturian side.The structure consists of two parallel single track tubes with a lengthof 24.9 km and an interior free diameter of 8.5 m in a free circular sec-tion of 51.3 m2. Both tunnels show a separation among axis of 50 mand are connected to each other bymeans of perpendicular by-pass gal-leries of 41.25 m long at every 400 m. Its central sector is connected tothe outside with two evacuation tunnels: Buiza intermediate accessadit (5.5 km long and 6% gradient) and Folledo intermediate accessadit (2.1 km long and 13% gradient) (Mguez Bailo, 2005; MguezBailo et al., 2007). The characteristics of these tunnels comply with theaerodynamic and safety conditions required in a high-speed railwayline. In this context, the groundwater inow into the tubes poses a seri-ous challenge, taking into account that the railway line was designed toallow speeds over 250 kph (Mguez Bailo, 2005).

Different drilling methods were used during the construction ofthe Pajares Tunnels. The perforation of the two main tunnels wasperformed through the use of TBMs (Tunnel Boring Machines). Thesupport of the tubes consists of high-strength concrete precast ringswith a thickness of 5060 cm. Each ring is divided in 7 segments.Concrete with different characteristic strength (between 40 and110 MPa) was used to fabricate the rings depending on the structuralrequirements in each section of the tunnels (Segura Prez andMartnez Daz, 2009; Arlandi Rodrguez et al., 2009). The Buiza gallerywas also drilled with a TBM, but the Folledo gallery and the by-passgalleries were drilled by using NATM (New Austrian Tunnelingmethod). In those cases, different support methods were used in eachcase taking into account the lithology variations (Arroyo Cedrn et al.,2009). The complexity of the project made it necessary to divide itinto four different contracts. The southern sector of the tunnels, underthe study area, was included in the Contract 1, awarded to the Spanishcompanies FOMENTO DE CONSTRUCCIONES Y CONTRATAS S.A. andACCIONA INFRAESTRUCTURAS S.A. In this section, the main tubeswere perforated through the use of two single shield TBMs: a NFM-Wirth single shield in the east tube and a Herrenknecht single shieldin the west tube (Mguez Bailo, 2005).

Apart from the aforementioned geotechnical and constructiveaspects, the construction of the Pajares Tunnels remained a majorchallenge in other aspects, such as the management of the extractedmaterial (Ferreras Gonzlez et al., 2009; Campomanes Snchez, 2009;Cayn Martnez et al., 2009) or the design of a pumping system andthe pumping water treatment (Arnanz Gonzlez et al., 2009; DezCadavid and Luengo Troitio, 2009).

In March 2007, a section of the tunnels (within the Contract 1) wasdrilled 450 m below the Alcedo Valley (Len, N Spain) (Fig. 1). In thefollowing summer, two sinkholes appeared at the bottom of the valley,affecting the main stream, and another six sinkholes appeared over thefollowing 14 months; due to this situation, in April 2008, the AlcedoStream was losing 40% of its ow (43 l s1) (lvarez Dez et al., 2009).Although the stream was channeled with concrete along a 370 m-longsection in 2009 to prevent water inltration, water loss into the under-

lying aquifer was not avoided. In 2010, an intense rainfall period Biosphere Reserve Alto Bernesga (Rodrguez Fernndez, 2011).

Geologically, the Alcedo Valley is located in the Cantabrian Zone(NE Iberian Massif), the outer zone of the Variscan orogen in theNW of Spain (Lotze, 1945; Julivert et al., 1972), and more specicallywithin the Somiedo Unit (Alonso et al., 2009). During the earlyCambrian-late Emsian, tectonic deformation affected the Paleozoicrocks in the study area, giving as a result a complex structure charac-terized by an alternation of lithologies with very different mechani-cal and hydrogeological behaviors (Fig. 2). From the bottom to thetop, the geological succession consists of the following formations:Lncara Fm. (limestone, dolomite and red limestone), Oville Fm.(green shale, gray shale and sandstone), Barrios Fm. (quartzite),Formigoso Fm. (black shale interbedded with sandstone), SanPedro Fm. (red ferruginous sandstone) and La Vid Group (dolomite,limestone, marlstone and shale) (Fig. 2). The Paleozoic bedrock iscovered by Quaternary uvialcolluvial deposits and screes derivedfrom the Barrios Fm. (Marqunez et al., 1990).

From a tectonic point of view, the study area is characterized by athin-skinned structural style with low internal deformation (Alleret al., 2004),which essentially consists ofWNWESE oriented imbricatethrusts and their associated folds with the main detachment locatedwithin the earlymiddle Cambrian limestone of the Lncara Fm.(Julivert, 1971). The subsequent extensional and compressionalepisodes (late Variscan and Alpine tectonics) affected the originalVariscan structure, producing thrusts inversion, folds compression andthe formation of reverse faults (Alonso and Surez Rodrguez, 1990;Aller et al., 2004; Alonso and Rubio, 2009). Below the Alcedo Valley,the main tectonic issue is an out-of-sequence thrust (Collado de Alcedothrust) (Alonso and Surez Rodrguez, 1990; Toyos et al., 2009), leadingto the development of an imbricated structure involving Lncara andOville Fms. (Toyos et al., 2009) (Fig. 2).

Concerning the hydrogeological setting, there are three formationswhich have a key role: Lncara Fm., Barrios Fm. and San Pedro Fm.

ubi

160 P. Valenzuela et al. / Engineering Geology 196 (2015) 158170Fig. 2. Geological map and geological section of the study area, modied from Alonso and R

taken from lvarez Dez et al., 2009.o, 2009. Location of the aquifers SA-8, SA-8b and SA-9 and the boreholes SR-10 and SR-11,

25/10/2013), by means of portable instruments (HANNA HI9025 pHmeter and HANNA HI9033 conductivity meter). Stream and spring

schedule. This made it possible to interpret the variations in thepumped ow as variations in the water inow into the tunnels.

161P. Valenzuela et al. / Engineering Geology 196 (2015) 158170Lugeon test conducted during the project phase of the tunnelsresulted in the only hydraulic conductivity data available: in therange of 108106 m s1 for Lncara limestone and in the widerange of 109103 m s1 for Barrios and San Pedro sandstone(lvarez Dez et al., 2009). However, the same authors suggest thatthese data are too low and unrepresentative of the real hydrological be-havior, considering fracturing, karst and alteration processes developedinto the bedrock. Therefore, three types of aquifers in the study area arerelevant for this investigation: (i) karst aquifers developed in Lncaralimestone and characterized by generalized micro-karstication;(ii) secondary porosity aquifers related to sand-alteration of Barriosquartzite and San Pedro sandstone due to fracturing and hydrothermalprocesses; and (iii) small porous aquifers developed inside Quaternarydeposits with variable permeability values depending on their lithologyand particle size (lvarez Dez et al., 2009; Garrido Ruiz et al., 2009).Limestone and dolomite of La Vid Group are also interested by karstprocesses, but their outcrops are far away from the sinkholes area,which make them irrelevant for this research. On a large scale, thegeometry of the aquifers in the valley bedrock is controlled by thepresence of low permeability materials and tectonic structureswhich act as permeable barriers (lvarez Dez et al., 2009; GarridoRuiz et al., 2009) (Fig. 2).

Although all the aforementioned aquifers have a reduced storagecapacity (between 0.03 and 6.1 hm3 yr1), the groundwater resourcesare used for drinking water supply and farm activities (Galn,1990; lvarez Dez et al., 2009; Garrido Ruiz et al., 2009). Alsoaccording to lvarez Dez et al. (2009), average annual rechargeis 130 mm yr1, reaching higher values in areas where there aresandstone and quartzite outcrops.

Under natural conditions, the recharge of the aquifers depends onthe surface water inltration and the discharge takes place throughthe drainage network and the existing springs (Garrido Ruiz et al.,2009). Nowadays the natural hydrogeological conditions have beenaltered in coincidence with the construction of the tunnels belowthe study area. According to lvarez Dez et al. (2009), a maximumow rate of 520 l s1 was measured in the southern entrance ofboth tunnels on 21 March 2007 during the perforation works. Untilthat moment, more than 9 km had been drilled and over half of thislength corresponded to permeable materials from 9 different aqui-fers. However, there is no published data about the water inowfrom each aquifer into the tunnels. The same authorsmention a com-plete drainage of Barrios and Lncara Fms. (dened by them as SA-8and SA-9 aquifers respectively, Fig. 2) within the study area for theperiod 20072008, with the exception of the northern sector of Bar-rios Fm. (SA-8b, Fig. 2), which does not show signicant changes inits hydrogeological behavior. Piezometric level control carried outfrom July 2006 to November 2008 in borehole SR-10, drilled intoLancara Fm. (SA-9 aquifer, Fig. 2), shows a temporal relationship be-tween the beginning of the tunnel perforation below the valley, the70 m drop of the water table (within a year, from March 2007 toMarch 2008) and the development of the rst eight sinkholes. Ac-cording to lvarez Dez et al. (2009), Lncara limestone (SA-9 aqui-fer, with a storage capacity of 0.4 hm3 yr1) constitutes an inowpreferential zone into the tunnels, where some karst features wereobserved during the perforation. Piezometric data from boreholeSR-11 (Arnanz Gonzlez et al., 2009) (Fig. 2) also show a 500 mdrop of the water table coinciding with the perforation, from 600 mto 100120 m above the tunnel. Unfortunately, the scarce availabili-ty of quantitative hydrogeological data prior to the drilling of thetunnels makes it impossible to accurately quantify the magnitudeof the change.

3. Material and methods

This research has been developed through two main tasks:

geomorphological analysis and GIS data management. An approachPiezometric data were also corrected taking into account the angle ofthe boreholes and expressed in meters above sea level. A temporalcomparison was performed with all these information together withdischarge measurements were performed through tracer dilution.A sudden-injection method using sodium chloride as a tracer waschosen (Graves, 2007). Measurements were taken between 1418 m downstream the injection points. The same tracer was usedin a tracer test (Graves, 2007). The purpose was to establish a rela-tionship between the sinkholes and the springs in the study area.The inject point was located upstream the sinkholes and the mea-surements were taken downstream the dry reach of the AlcedoStream and the springs. Both experiments were performed with aSalinoMADD device.

Moreover, ADIF (Administrador de Infraestructuras Ferroviarias-Spanish Railway Infrastructures Manager) and FOMENTO DECONSTRUCCIONES Y CONTRATAS S.A. and ACCIONA INFRAESTRUCTURASS.A., the contractors for the Pajares Tunnels-Contract 1, provided dailypumping records of the water inow into the tunnels, piezometriclevel records measured in boreholes SR-10 and SR-11 and informa-tion relative to the lithology and the stratigraphic formations drilledevery day during the perforation works under the Alcedo Valley.Pumping records were corrected taking into account the coolingwater ow rate of the two TBMs used in Contract 1 (40 l s1 in TunnelE and 25 l s1 in Tunnel W during the drilling) and its daily operatingto the new hydrological regime developed in the Alcedo Streamand to the appearance of the sinkholes was also performed throughhydrogeological analysis.

3.1. Geomorphological analysis

The geomorphological analysis of the study area was carried outthrough interpretation of aerial photographs and eld surveys.The sinkholes and the other features developed in the Alcedo Valley(swallow holes, evidence of collapse and scarps) were measured inthe eld, determining their position, length, width and depth byusing a GPS and a laser measurement device. Based on a 2011orthophoto provided by the IGN (Instituto Geogrco Nacional National Geographic Institute of Spain), all these features and thechanneled section of the Alcedo Stream were plotted on a 1/10,000scale map. Geomorphological maps and aerial photos prior to 2011do not show evidence of sinkhole in the study area. In addition, theshallow deposits existing in the valley were represented in another1/10,000 geomorphological map. Different data were taken overthe main deposits during eld activities in order to determine theirsedimenthological characteristics: matrixclasts ratio, sandclayratio, internal structures and clasts lithology, size and morphologyData were taken according to the common methodology followedin geomorphological eld surveys (Dackombe and Gardiner, 1983).Both maps were digitized and analyzed by using GIS.

3.2. Hydrogeological analysis

The hydrogeological analysis of the study area included severalmethodologies, aimed at characterizing the development of the sink-holes and establishing its inuence over the surcial hydrogeologicaldynamics of the Alcedo Stream and its relation to the perforationof the Pajares Tunnels. Physico-chemical parameters (temperature,pH and conductivity) were measured in the stream and the springswithin the study area in the course of two surveys (21/04/2013 andprecipitation data also provided by the contractors.

cients were used to calculate cell-by-cell the catchment area likely togenerate surcial runoff. This made it possible to calculate the amountof rainfall in each cell that results in surcial runoff and, consequently,to estimate the average annual andmonthly volumes of water collectedby the Alcedo watershed on its natural base level on the basis of a31 years dataset. It was also possible to calculate the volume of waterthat drains into the swallow holes, taking into account only thecatchment area of sinkholes.

4. Results

4.1. Geomorphological analysis

The 1/10,000 geomorphologicalmap allowed carrying out a physicaland hydrogeological analysis of the Alcedo watershed and understand-ing the collapse and inltration phenomena, taking into account thedifferent surcial deposits outcropping in the Alcedo Valley (Fig. 4A).

The valley slopes are mainly covered by two types of shallow de-posits: (i) extensive NE-facing active rockfall screes (0.2 km2 surface,covering 3.4% of the catchment), derived from the erosion of quartziteand sandstone outcrops (Barrios and San Pedro Fms.), and (ii) rankersoils less than one meter thick, composed of angular clasts with asandy-organic matrix, which cover up to 75% of the watershed(5.3 km2 surface) (Fig. 4B). In the lower part of the slopes (0.5 km2

surface, 7.6% of the catchment), these ranker soils grade into colluvialdebris deposits with sandy matrix up to one meter thick that behaveas surcial aquifers. Several landslide deposits were found in NE-facing slopes, associated with alluvial plain deposits. Also, a small

162 P. Valenzuela et al. / Engineering Geology 196 (2015) 1581703.3. GIS data management

The use of GIS allowed not only mapping the affected area and themain features of the phenomenon but also estimating the hydrologicalbehavior of the Alcedo Streambefore the construction of the tunnels de-spite the lack of hydrogeological information over this previous period.Datamanagementwas performed using ArcGIS 9.3 software (ESRI). Theapplied process can be divided into four main steps:

3.3.1. Creation of digital data layersLandscape characteristics which condition the surcial hydrological

behavior of the catchment were considered. Three vector data layerswere created by using GIS: outcrops, vegetation and shallow deposits.Geological data were compiled from Alonso and Rubio (2009)and from the Sheet n. 103 of 1/50,000 scale National Geological Map(Alonso et al., 1991). The vegetation map created was based on1/50,000 scale Map of Crops and Land uses of Spain 20002010(MARM, 2009). Two Digital Terrain Models (DTM) were created:a Digital Elevation Model (DEM) with a pixel of 10 m (built fromthe topographical information provided by the National GeographicInstitute) and a Digital Slope Model (DSM), reclassied in twocategories (slope b 3 and slope 3) determined by a differenthydrogeological behavior, according to Ferrer (1993).

3.3.2. Calculation of runoff coefcientsThe runoff coefcient of each cell (100 m2) was calculated for

each land category of the Alcedo Valley following the methodologydescribed by Fernndez Rodrguez et al. (2002), based on the USA SoilConservation Service equation (USDA-SCS, 1972):

C Pd=P0 1 : Pd=P0 23 Pd=P0 11 2

where C is the runoff coefcient, Pd is the daily precipitation and P0 isthe runoff threshold.

The P0 estimation for the different land categories was performedbased on the values tabulated by the USA Soil Conservation Serviceand adapted by Ferrer (1993), taking into account the climatic andgeographic characteristics of Spain. The runoff threshold correctioncoefcients and the daily precipitation values were taken from MOPU(1990) and Ministerio de Fomento (1999).

3.3.3. Rainfall interpolation modelsA rainfall database was elaborated, including monthly values,

covering a period of 31 years (19702000), with data gatheredfrom 10 weather stations of the AEMET (Agencia Estatal deMeteorologa-Spanish Meteorology Agency of Spain) (Fig. 3). All ofthem are located within a 21-km radius around the Alcedo Valleyat altitudes ranging from 945 to 1280 m a.s.l. The softwareCORTREST under the application HIDROBAS 3.0 (Ortiz et al., 2001)was used to ll the gaps of the data series by normal correlation.The achieved correlation coefcients (R) were always equal orhigher than 0.79.

Monthly precipitation data series were statistically analyzed toobtain information about the precipitation regime in the study area.The following values were extracted: annual precipitation in eachweather station, average annual precipitation in the study area(considering all the stations), annual precipitation of the yearswith the highest and lowest precipitations for each station, monthlyprecipitation of the rainiest month of the data series (taking intoaccount all the stations), average monthly precipitation for eachmonth in every station and average monthly precipitation in thestudy area.

Sixteen precipitation models were generated through the interpola-tion of the precipitation data calculated in the previous step by using

ArcGIS 9.3: average, maximum and minimum annual precipitations,maximum monthly precipitation and average monthly precipitationfor each month. Taking into account the limited availability of rainfallunderlying data, geostatistical interpolation methods were discarded.Models were calculated by using the deterministic method of InverseDistance Weighting (IDW) (Shepard, 1968).

3.3.4. Calculation of the volume of inltrated waterDifferent rainfall raster data layers (annual and monthly data) were

combined with the runoff coefcient raster data layer. Runoff coef-

Fig. 3. Location of the AEMET weather stations.nivation moraine was mapped in the head-waters of the catchment.

163P. Valenzuela et al. / Engineering Geology 196 (2015) 158170The most extensive Quaternary deposits, also formed by quartz andshale debris (Oville and Barrios Fms.), are located in the middlesector of the watershed, covering 0.2 km2 (2.9% of the catchment).These heterogeneous deposits, so-called valley polygenic deposits,mantle the valley bottom and consist of surrounded-angularcentimetricmetric quartz boulders in a matrix with signicantclay content (90%matrix/10% boulders) (Fig. 4C, Level 1). They alter-nate with other beds consisting of rounded decimetric clasts in asandy matrix (20% matrix/80% clasts) (Fig. 4C, Level 2). The twolevels described above have colluvial and alluvial origins, respective-ly. Several erratic blocks with glacial striation were found over thevalley polygenic deposits, which prove a glacial inuence in itsdevelopment (Fig. 4D).

Seventeen sinkholes weremapped in themiddle reach of the AlcedoValley (Fig. 5A and B): Eleven sinkholes are located at the bottom of thevalley, near the channeled section of the stream; the remaining sixsinkholes are located in the western slope, about 5 m above the stream(Fig. 5A). The length of the sinkholes ranges between 1.5 and13.1mandthe depth between 0.8 and 5.5 m (Table 1). In general, the depressionsshow subcircular edges, round shapes and cylindrical or conical proles.The graphical representation shows that the lengths ofmajor andminordiameters are very similar in most cases (Fig. 6). Graphic also showsthat, in general, sinkholes with the greatest area are deeper than thosewith a smaller area. At the bottom of the valley, collapses grew andjoined, with the development of large irregular sinkholes along thestream bed. The eld surveys conducted in different dates revealed

Fig. 4. A. Map of shallow deposits and outcrops of the Alcedo Valley; B. Ranker soils; C. ValleyD. Erratic blocks on the valley polygenic deposits.that some sinkholes grew signicantly: sinkhole n increased itsdiameter from2 to 4.1m in 3months (Fig. 5C andD). A 120m2 area sur-rounding the sinkholes with evidence of active subsidence (averagedrop in the terrain of 30 cm) was mapped at the bottom of thevalley (Fig. 5A).

All the sinkholes developed on the Quaternary deposits previous-ly described: uvial, colluvial and glacial deposits with signicantclay content that cover the bottom of the valley and the westernslope. The characterization of these deposits in the eld alloweddetermining the matrix clay content (2050% of the matrix), whichis higher in the materials covering the slope. However, at the bottomof the valley, there are some beds with sandy matrix (99% sand/1%clay), due to a greater inuence of uvial processes (Fig. 4C).The growth of the sinkholes undermined the eastern slope andthe channeled stream, causing its fracturing in two points and thedevelopment of minor landslides with small lateral displacement(Fig. 5E and F).

4.2. Hydrogeological analysis

The presence of exhumed endokarst features in Lncara outcropssuggests that there is an important karst development in Lncaralimestone. Indeed, most of the sinkholes located along the channellevel are connected with narrow karst conduits. However, not all thesinkholes behave as ponors or swallow holes on a continuous basis.In the course of the investigation, inltration was observed in 7 of the

polygenic deposits with a colluvial deposit (Level 1) overlying a uvial deposit (Level 2);

164 P. Valenzuela et al. / Engineering Geology 196 (2015) 15817011 sinkholesmapped at the bottom of the valley. Their activity is relatedto stream ow seasonal variations, so that water inltrates throughall the ponors simultaneously only during ood events. The controlcarried out in the study area proved that these ponors drain all the

Fig. 5.A. Karst features in theAlcedoValley: ponors (1, 2, 3, 4, 5, 6 and 7), sinkholes (a, b, c, d, e, f,and the sinkhole catchment; C. Sinkholen (07/01/2013); D. Sinkhole n (13/04/2013); E. Fracturelandslide scarps in the middle section of the Alcedo Stream.surcial ow, even during the maximum discharge events, drying up a100 m channeled reach of the Alcedo Stream throughout the year(between 277816X-4754615Y and 277673X-4754869Y, UTM 30T).The stream discharge measurement performed upstream ponor 1

g, h, i, j, k, l,m, n, o, p and q) and springs (A and B); B. Natural base level of theAlcedoValleyd channel over the sinkhole j; F. Sinkholes (b, e, g, h and j), ponors (1, 2, 3, 4, 5, 6 and 7) and

provided a ow rate of 180 l s1. Themeasurementwas performed dur-ing themelting period (21/04/2013) and because of that data can be es-

Table 1Dimensions of the sinkholes mapped in the Alcedo Valley.

Major diameter (m) Minor diameter (m) Maximum depth (m)

a 8.2 7.7 3.4b 12.9 9.4 1.7c 2.6 2 1.3d 2.3 2 1.6e 1.8 1 1f 2.6 2.8 0.6g 1.7 0.9 0.8h 13.1 4.2 1.3i 1.5 1.2 2.6j 12.6 11.1 5.5k 8.7 1.9 1.4l 1.6 1.6 1.3m 3.1 2.1 0.8n 4.1 3.8 2.1o 3.6 2.9 1.4p 4.9 3.9 3q 8.9 7.7 2.3

165P. Valenzuela et al. / Engineering Geology 196 (2015) 158170timated as a maximum ow rate for this year.Due to the development of sinkholes in the valley oor, the upper

part of the watershed (with a 3.9 km2 surface) is drained by the ponorsand isolated from the rest of the uvial network (Fig. 5B). The streamow is the result of the precipitation over this catchment locatedupstream from the sinkholes. A signicant part of precipitation givesthe stream runoff as a result, but the other part is stored into theaquifers present in this catchment. No stream ow contributions fromkarst aquifers have been observed in this catchment. The largest aquiferis Barrios quartzite (2.5 km2 surface), which constitutes 62.5% of thebedrock in the watershed drained by the ponors, but the most relevantfor this research is Lncara karst aquifer, which only constitutes 1.4% ofthis catchment (54,774m2 surface) and is related to the development ofthe sinkholes. However, 99.1% of the area is covered by surcial forma-tions that conditionwater inltration processes. Themost important areseveral colluvial debris deposits that behave as surcial aquifers due totheir high sand content (95%) and cover 9.2% of the watershed drainedby the ponors (0.6 km2).

Two springs, arising from karst conduits in Lncara limestone,supply water to the Alcedo Stream 100 m downstream of thesinkholes area: spring A (277815X-4754669Y, UTM 30T, 1336 m a.s.l.)and spring B (277792X-4754675Y, UTM 30T, 1339 m a.s.l.) (Fig. 5A).Springs discharge measurements resulted in ow rates of 146.1 l s1Fig. 6. Graphical representation of the size of the sinkholes.in spring A and 26.9 l s1 in spring B (21/04/2013). Both springsshow a seasonal behavior. Physico-chemical parameters measured atthe springs and at the stream upstream of the sinkholes area(277696X-4754811Y, UTM 30T, 1349 m a.s.l.) are presented inTable 2. Tracer test performedgave a negative result. Therefore, the con-nection between the swallow holes and the springs downstream fromthe sinkholes was not proved.

The temporal comparison of hydrogeological records andgeotechnical information achieved during the works under theAlcedo Valley (Fig. 7) shows a coincidence in time between thebeginning of the perforation of Lncara limestone in both tunnelsand the increase of the pumping rate in 250 l s1 with respect tothe previous average rate (b100 l s1). This great water inowconrms the presence of a relevant karst development in Lncaralimestone and its drainage due to the construction of the tunnels,causing the drop of the piezometric levels observed in boreholesSR-10 and SR-11. After the drilling of the Lncara limestone, thepumping rate does not return to the previous values, maintainingan average ow of 260 l s1. This fact, coincident with the develop-ment of the two rst sinkholes appeared on the Alcedo Stream inthe summer of 2007, is interpreted as a result of the inltration ofsurcial runoff. Precipitation in the study area over the same perioddoes not shows appreciable effects, causing small scale variations inthe water inow into the tunnels.

4.3. GIS modeling

The characterization of the study area through the joint analysis offour digital layers (outcrops, shallow deposits, slope and vegetationcover) was carried out by the use of GIS. Different categories weredened for each variable to facilitate their combination and analysis(Fig. 8A, B, C, and D).

In the Alcedo Valley, the predominant lithology is represented byBarrios quartzite, which constitute 49.4% of the watershed bedrock(3.4 km2), followed by Oville shale and sandstone that constitute29.3% of the area (2 km2). San Pedro sandstone also shows importantoutcrops totalling 4.2% of the watershed bedrock (0.3 km2). Lncaralimestone only constitutes 2.6% of the valley bedrock. All the otherformations (Formigoso Fm. and La Vid Group) constitute the remaining14.4% of the catchment (1 km2). Taking into account that 94.1% of thestudy area is covered by shallow deposits, lithology is only consideredfor the modeling in the areas with bedrock outcrops. Therefore, theoutcrops map has been reclassied in two kinds of categories: lowpermeability and permeable outcrops (Fig. 8A). The shallow depositlayer was divided into 8 categories: active rockfall screes (3.4%), colluvi-al debris deposits (7.6%), landslide deposits (2.5%), valley polygenicdeposits (2.9%), alluvial plain deposits (0.1%), nivation moraines(0.2%), ranker soils (where surcial deposit thickness b 1 m; 77.4%)and bedrock outcrops (without surcial formations; 5.9%) (Fig. 8B).A slopemapwas created from the reclassication of the DSM in two cat-egories: slope b 3 and slope 3, showing that only 0.2% of the studyarea has slopes under 3 (Fig. 8C). The vegetation cover layer was alsoreclassied, giving as a result the denition of 6 categories: forest(most common vegetation, covering 37% of the study area), dense bush(14.7%), scrubland (32.3%), grass (7.8%), soil without vegetation (associ-ated with active rockfall deposits; 3%) and rock without vegetation (as-sociated with bedrock outcrops; 5.2%) (Fig. 8D). On the basis of thedifferent data layers and their classes, 34 different land categories withtheir corresponding runoff coefcients were calculated (Table 3).

The completed monthly precipitation series allowed estimating themaximum and minimum values of precipitation in the study areaconsidering a period of 31 years and the average annual rainfall(1094 mm). The rainiest year is 1978, with an accumulated annualprecipitation of 1463mm, and the least rainy year is 1986,with an accu-mulated annual precipitation of 764mm. The calculation of the average

monthly precipitation in every station in the study area provided

information about the rainfall distribution through the year, showingthat December and August are respectively the most and least rainymonths in the majority of the stations. December 1987 was the rainiestmonth of data series,with an averagemonthly rainfall of 423mmtakingin account all the stations.

The combined analysis of sixteen annual and monthly precipitationmodels and the layer of runoff coefcients allowed estimating thevolume of water that is expected to inltrate into the sinkholes underdifferent precipitation scenarios. Moreover, rainfall modeling allowedinferring the volume of water runoff in the Alcedo Stream natural baselevel before the drilling of the tunnels and its decrease due to thecapture of the Alcedo Stream by the ponors. The mean estimatedwater inltration volume is 308,903 m3 yr1, the maximum is409,100 m3 yr1 and the minimum is 213,477 m3 yr1 considering aperiod of 31 years. A maximum monthly volume of inltration of114,013 m3 per month was estimated by using December 1987data. These inltration volumes reach a peak in spring and autumn(Fig. 9A). Considering the natural inferred hydrogeological conditionsbefore the perforation of the tunnels, with an 877,395 m3 volume ofyearly water runoff on the natural base level, the decrease in the AlcedoStream runoff is estimated at 35%. The analysis of monthly mean watervolumes shows that the ow decrease is constant throughout the year,reachingmaximum andminimum average values close to 39,600m3 in

5. Discussion

A joint analysis of the mapped sinkholes and subsidence evidencetogether with the geological and geomorphological maps was per-formed. This analysis showed that all the observed features are locatedin a 160 m-long section of the Alcedo Valley where the stream and thetrajectory of the west tunnel are practically overlapped. The tunnel inthis area crosses a tectonic imbricated structure composed by karstiedLncara limestone covered byQuaternary deposits (Fig. 10). The tempo-ral coincidence between the perforation of the tunnels 450m under theAlcedo Valley and the beginning of the inltration, together with thespatial relationship between the sinkholes and the karst conduits,suggest a close relationship between both phenomena. This relation issupported by the increase of the water inow into the tunnels recordedduring the perforation of Lncara limestone, the consequent drop ofthe water table in SR-10 and SR-11 and its correlation with thedevelopment of the rst sinkholes in the summer of 2007 (lvarezDez et al., 2009). Consequently, the perforation of the tunnels induceda generalized drainage, not only in the karst aquifer, but also in theoverlying Quaternary deposits, triggering a set of surcial and internalerosion processes.

Water table decline caused different effects within the Quaternarydeposits and the karst aquifer: (i) the loss of buoyant support of the

Table 2Physico-chemical water parameters measured at the springs A and B and at the stream.

Water parameter Stream Spring A Spring B

21/04/2013 25/10/2013 21/04/2013 25/10/2013 21/04/2013 25/10/2013

T (C) 10 7.8 9.5 6.8 9.9pH 7.2 5.8 7.4 6.6 7.5 6.7Electrical conductivity (S/cm) 30 28.5 36 67.8 34.5 44.6

166 P. Valenzuela et al. / Engineering Geology 196 (2015) 158170December and 9000 m3 in August (Fig. 9B).Fig. 7. Temporal comparison between water inow records from both tunnels, piezometricformations drilled each day during the works under the Alcedo Valley and precipitation recThe diagram about the drilled formations must be understood not as a geological cartographydata have been provided by ADIF, FOMENTO DE CONSTRUCCIONES Y CONTRATAS S.A. and ACsupercial deposits, (ii) the increase of percolation from the superciallevel records from the boreholes SR-10 and SR-11, information related to the geologicalords in the area over the same period. The average level of the tunnels is of 845 m a.s.l.but as a temporal diagram of progress of the works. The hydrogeological and geotechnicalCIONA INFRAESTRUCTURAS S.A.

167P. Valenzuela et al. / Engineering Geology 196 (2015) 158170runoff, and (iii) the increase of groundwater ow through the karstconduits previously lled with water or endokarst sediments. All theabove has enhanced the internal erosion of the cover through a rangeof processes designed as suffosion (Tihansky, 1999; Waltham et al.,2005; Gutirrez et al., 2014). Suffosion consists of the downwardmigra-tion of the overlying unconsolidated Quaternary deposits toward theopenings in the top of the bedrock. The result is the creation of voidswithin the shallow sediments which size and growth depend on thesoil cohesive strength (Waltham et al., 2005).

The signicant amount of clay of the Quaternary sediments in thestudy area suggests a cohesive behavior for these materials. Underthese conditions, suffosion process begins with the formation ofephemeral cavities over the karst conduits which migrate upwardsby progressive roof spalling toward the surface and eventually createsmall sudden collapses with sub vertical walls (Tihansky, 1999;Waltham et al., 2005). After the collapse, sinkholes are degraded bywalls failures and small landslides, growing in size and changingtheir original shape until the development of conical proles. Thesedegradation processes leave debris blocking the karst conduits ontheir oors (Waltham et al., 2005; Gutirrez et al., 2014). Therefore,the cylindrical and conical proles observed in the Alcedo Valleysuggest the operation of the processes described above. On thisbasis, Alcedo Valley sinkholes were classied as cover-collapse sink-holes according to Tihansky (1999), Williams (2004) and Walthamet al. (2005) classications.

Fig. 8. Thematic maps of the Alcedo Valley. A. Outcrops;Presently, in the streambed, the intermittent stream ow related toprecipitation events and the consequent oods are enhancing theinltration and erosion processes, giving as a result the developmentof great sinkholes by coalescence; where the stream was captured,sinkholes developed active ponors on their oors. Also at the bottomof the valley, a big area shows subsidence evidence, which can beexplained by the compaction of the Quaternary sediments and thesubsurcial erosion. This evidence sometimes precedes sinkholesformation (Newton and Hyde, 1971). The channeled section sufferedfrom fracturing and undermining processes, therefore this channeldoes not fulll its function.

Thehydrogeological characterization performed in theAlcedoValleyallowed understanding the new hydrological behavior of the AlcedoStream. On the one hand, the karst aquifer drainage and the wholesurcial ow inltration lead to the alteration of the stream hydrologi-cal regime. At present the streamhas an inuent behavior and theupperreach of the valley is isolated from the rest of the uvial network, whichconstitutes a serious environmental impact. However, the absence ofhydrogeological data from before the excavation of the tunnels onlyallows estimating its previous behavior and the magnitude of the changethrough a rainfall-runoff modeling. On the other hand, two springs, locat-ed in Lncara limestone, supply water to the Alcedo Stream downstreamthe sinkholes area. Despite the similarity between the stream inltrationow rate measured upstream the ponor 1 (180 l s1) and the ow ratemeasurement reached in both springs (146.11 + 26.9 l s1), the tracer

B. Shallow deposits; C. Slope; D. Vegetation cover.

cehyd

cate

w d

all sall sialialialialialialpopopopopoal pal pal pal plidelidelidelidelideionr sor sor sor so

168 P. Valenzuela et al. / Engineering Geology 196 (2015) 158170Table 3Land categories and runoff coefcients calculated in the Alcedo Valley and their equivalenconditions (Ferrer, 1993) refer to different characteristics of the terrain which condition its(ii) Straight row= vegetation following the steepest slope.

Slope% S.C.S. modied categories (Ferrer, 1993) Land

Soil group Cover type Hydrologic condition Shallo

3 A Bare soil Straight row Rockfb3 A Bare soil Straight row Rockf3 B Woods Thick Colluv3 B Woods Medium Colluv3 B Woods Low-density Colluvb3 B Woods Thick Colluvb3 B Woods Medium Colluvb3 B Woods Low-density Colluv3 C Woods Thick Valley3 C Woods Medium Valley3 C Meadows Very good Valleyb3 C Woods Thick Valleyb3 C Meadows Very good Valley3 C Woods Medium Alluvi3 C Meadows Very good Alluvib3 C Woods Medium Alluvib3 C Meadows Very good Alluvi3 C Woods Thick Landsb3 C Woods Thick Lands3 C Woods Medium Lands3 C Meadows Very good Landsb3 C Meadows Very good Lands3 B Woods Medium Nivat3 B Woods Thick Ranke3 B Woods Medium Ranke3 B Woods Low-density Ranke3 B Meadows Very good Ranketest performed has failed to demonstrate the connection between thesetwo points. Both springs (A and B) (Fig. 5A) show a seasonal behaviorand neutral or acidic pH values. Water temperature values show impor-tant seasonal variations between the two measurement dates. Low con-ductivity levels (ranging from 34.5 to 67.8 S/cm) do not suggest acalcareous inuence on the water chemistry, despite the location ofboth springs. lvarezDez et al. (2009) observed in the same area conduc-tivity levels lower than 100 S/cm and pH values ranging from 6.3 to 6.9in the water from Barrios quartzite. The above-mentioned observationssuggest that water of both springs comes from siliceousmaterials, proba-bly from Barrios quartzite, or directly from rainfall, showing a very shortcirculation time through the calcareous formation. Therefore, there areno evidences of a relationship between the springs and the inltrationinto the sinkholes toward the karst aquifer. Disconnectionmay be relatedto the existence of compartmentalized aquifers due to the presence oftectonic structures that behaves as permeability barriers, according tolvarez Dez et al. (2009).

Geomorphological and hydrogeological evidences, together withdata published by some authors (lvarez Dez et al., 2009; GarridoRuiz et al., 2009; Arnanz Gonzlez et al., 2009) and provided byADIF and the contractors, are consistent with the inltration of sur-cial runoff through the drained karst aquifer toward the tunnels,which represent a new articial baseline for the isolated area of theAlcedo catchment.

There is a variety of methods to estimate stream ow in moun-tainous ungauged catchments as the Alcedo Valley. The analysis ofobserved precipitation records and physical properties of drainagebasins using GIS (Ko, 2004) allows developing rainfallrunoffmodels as a function of precipitation and parameters determined

b3 B Woods Thick Ranker sob3 B Woods Medium Ranker sob3 B Woods Low-density Ranker sob3 B Meadows Very good Ranker sob3 Imp. rock Low perm3 Imp. rock Low perm3 Perm. rock Permeablwith the S.C.S. land categories modied by Ferrer (1993). Notes: (i) the term Hydrologicrologic behavior, mainly related to the development of the vegetation or the type of crop;

gories dened in the Alcedo Valley

eposits/outcrops Vegetation cover Land categories Runoff coef.

crees 1 0.285crees 2 0.197debris deposits Forest 3 0.003debris deposits Dense bush 4 0.065debris deposits Scrubland 5 0.147debris deposits Forest 6 0.003debris deposits Dense bush 7 0.065debris deposits Scrubland 8 0.147lygenic deposits Forest 9 0.085lygenic deposits Dense bush 10 0.171lygenic deposits Grass 11 0.171lygenic deposits Forest 12 0.085lygenic deposits Grass 13 0.137lain deposits Dense bush 14 0.171lain deposits Grass 15 0.171lain deposits Dense bush 16 0.171lain deposits Grass 17 0.137deposits Forest 18 0.085deposits Forest 19 0.085deposits Dense bush 20 0.171deposits Grass 21 0.171deposits Grass 22 0.137moraine Dense bush 23 0.065ils Forest 24 0.003ils Dense bush 25 0.065ils Scrubland 26 0.147ils Grass 27 0.027by basin properties: drainage basin network, surcial formations,land use and slope (Ko and Cheng, 2004; Cheng et al., 2006). In thisresearch, the methodology was adapted taking into account thestudy area specicities and the available information. Thus, themodeling carried out shows several limitations. The rst limitationwas the scarcity of complete meteorological data series in the studyarea, which limited the use of more sophisticated geostatisticalmethods. Taking into account data series characteristics, the deter-ministic method of Inverse Distance Weighting (IDW) (Shepard,1968) was chosen for spatial interpolation. The fundamental as-sumption of this method is that the areas closest to a weather stationhave more similar precipitation values than the most distant areas,giving more weight to the neighboring stations through the interpo-lation (Tobler, 1970). Secondly, the runoff modeling performed donot take into account subsurcial ow, water storage time on thesurcial aquifers or inltration through the existing fractures in thevalley bedrock. In addition, pluvionival precipitation conditions,which prevail in winter in the study area, were not taken into ac-count in the calculation of monthly surcial runoff that inltratesthrough the sinkholes. All these hydrogeological and climatic condi-tionsmay give rise to runoff variations. Finally, the limited number ofow rate measurements performed in the Alcedo Stream during thisresearch made it difcult to calibrate the proposed model. The owrate measurement (180 l/s in April) was not used in the calibrationbecause it is a punctual daily measurement (21/04/2013) estimatedas a maximum ow rate during a melting period. However, themodeling results are annual or monthly average values. Both dataare not comparable. Nor were daily pumping records used becausethe water comes not only from the Lncara karst aquifer but also

ils Forest 28 0.003ils Dense bush 29 0.065ils Scrubland 30 0.147ils Grass 31 0eability outcrops 32 0.727eability outcrops 33 0.882e outcrops 34 0.802

Alcedo Valley, can be useful for further works which face this kind of

Fig. 9.A.Monthlymeanwatervolumeof inltration in theAlcedoValley sinkholes. B. Estimated

169P. Valenzuela et al. / Engineering Geology 196 (2015) 158170from all the aquifers drained by the tunnels. On this basis, the mean,maximum andminimum annual andmonthly stream ow values ob-tained for the period 19702000 in the Alcedo Stream should be un-derstood as an estimation of average stream ow lost due to the

Alcedo Stream runoff in the natural base level before and after the tunnel perforation.inltration through the sinkholes. This value constitutes a good

Railway By-pass. The authors also thank to F.J. Torrijo, L. Gonzlez de

Fig. 10. Spatial relation between ponors, sinkholes, tunnels, stream, shallow deposits andbedrock aquifers in the Alcedo Valley.Vallejo and two anonymous referees for their comments, who greatlyhelped to improve themanuscript, and to C. Lpez Fernndez, A. Blanco,geomorphological and environmental problems:

The natural hydrogeological regime of the Alcedo Valley has beenaltered. The generalized drainage related to the perforation ofthe Pajares Tunnels produced the drop of the water tablenot only in the Lncara karst aquifer but also in the overlyingQuaternary deposits.

From 2007 to present 25 sinkholes have been developing inthe Alcedo Valley (8 sinkholes between 2007 and 2008 and 17sinkholes as of 2010). Their origin is interpreted as a result ofthe suffosion processes affecting Quaternary sediments with acohesive behavior due to a signicant amount of clay. Moreover,the increase of percolation and groundwater ow across thekarst aquifer enhanced the subsurcial erosion of the Quaternarydeposits, increasing the size of the sinkholes.

Currently, the Alcedo Stream behaves as an inuent stream, losingall the surcial ow by inltration through 7 active ponors devel-oped at the stream bed. For this reason, a 100 m-long reach of theAlcedo Stream is dry throughout the year and the upper part ofthe watershed (with a 3.9 km2 area) is nowadays isolated fromthe rest of the valley, which constitute a serious environmental im-pact. The channeled section suffered from fracturing andundermining processes, therefore this channel does not fulll itsfunction.

Themethodological approach applied to this study allowed estimat-ing the pre-existing hydrological regime in the Alcedo Streambeforethe drillingworks and the current volume of inltration through thesinkholes.

Due to the capture of the upper reach of the Alcedo Stream by theswallow holes, annual and monthly ow rates estimated in the nat-ural base level of the Alcedo Stream prior to 2007 decreased about35% throughout the year, oscillating from 9000 m3 in August and39,600 m3 in December.

The estimatedmeanwater volumeof inltration is 308,903m3 yr1,with a maximum of 409,100 m3yr1 and a minimum of213,477 m3 yr1 for the period 19702000. In addition, a maxi-mum monthly volume of inltration of 114,013 m3 per monthwas estimated

At present, the process is active and it is expected to continue inthe future.

Acknowledgments

The authors gratefully acknowledge ADIF, FOMENTO DECONSTRUCCIONES Y CONTRATAS S.A. and ACCION INFRAESTRUCTURASS.A. for the hydrogeological and geotechnical data provided. Specialthanks to A. Gutirrez Blanco, ADIF General Director between 2012and 2014, and to J.M. Jimnez Snchez, current director of the Pajaresapproach in a case study for which there is no data from before theobserved hydrogeological alteration.

6. Conclusions

The present case study constitutes a good example of surcialimpacts caused by the drilling of base tunnels through karst aquiferswithin the phreatic zone. These impacts mainly consist on the develop-ment of sinkholes and the associated alteration of the hydrogeologicalregime. The following conclusions, derived from the research in theD. Moral, F. Mrquez, L. Pelez and F. Valenzuela for the assistance.

ng Geology 196 (2015) 158170References

Alija, S., Torrijo, F.J., Quinta-Ferreira, M., 2013. Geological engineering problems associatedwith tunnel construction in karst rock masses: the case of Gavarres tunnel (Spain).Eng. Geol. 157, 103111. http://dx.doi.org/10.1016/j.enggeo.2013.02.010.

Aller, J., lvarez-Marrn, J., Bastida, F., Bulnes, M., Heredia, N., Marcos, A.,Prez-Estan, A., Pulgar, J.A., Rodrguez-Fernndez, L.R., 2004. Zona Cantbrica.Estructura, deformacin y metamorsmo. In: Vera, J.A. (Ed.), Geologa de Espaa.Sociedad Geolgica de Espaa-Instituto Geolgico y Minero de Espaa, Madrid,pp. 4247.

Alonso, J.L., Rubio, A., 2009. La estructura geolgica de la seccin del Tnel de Pajares. In:Pando, L., Lpez Fernndez, C., De la Rubia Mir, L. (Eds.), Jornadas tcnicas de laVariante de Pajares. Comunicaciones presentadas. Universidad de Oviedo, Oviedo,pp. 2939.

Alonso, J.L., Surez Rodrguez, A., 1990. Tectnica, in: Rodrguez Fernndez, R.L. (Ed.),Memoria explicativa de la Hoja 103 (Pola de Gordn) del Mapa Geolgico de Espaa(E.1:50.000). Instituto Tecnolgico GeoMinero de Espaa, Madrid, pp. 7989.

Alonso, J.L., Surez Rodrguez, A., Rodrguez Fernndez, L.R., Farias, P., Villegas, F., 1991.Hoja 103 (Pola de Gordn). Mapa Geolgico de Espaa (E.1:50.000). InstitutoTecnolgico GeoMinero de Espaa, Madrid.

Alonso, J.L., Marcos, A., Surez, A., 2009. Paleogeographic inversion resulting from largeout of sequence breaching thrusts: The Len Fault (Cantabrian Zone, NW Iberia). Anew picture of the external Variscan Thrust Belt in the Ibero-Armorican Arc. Geol.Acta 7 (4), 451473. http://dx.doi.org/10.1344/105.000001449.

lvarez Dez, J., Vinding Fallesen, J., Garrido Ruiz, M.S., Lombardero Barcel, M., MarcosBermejo, F., Monge Cristbal, J.C., Serrano Martn, L., 2009. Hidrogeologa de losTneles de Pajares. In: Mguez Bailo, R. (Ed.), Tneles de Pajares. ADIF, pp. 167208.

Arlandi Rodrguez, M., Domnguez Gonzlez, M.L., Pelez Gonzlez, M., 2009. Gnesis delproyecto de Pajares. Estudios previos, Proyecto bsico y Proyectos constructivos. In:Pando, L., Lpez Fernndez, C., De la Rubia Mir, L. (Eds.), Jornadas tcnicas de laVariante de Pajares. Comunicaciones presentadas. Universidad de Oviedo, Oviedo,pp. 3550.

Arnanz Gonzlez, J., Garca Aldaya, D., Carballo Flrez, L.B., 2009. Back pumping system on15 km long descended tunnels. In: Pando, L., Lpez Fernndez, C., De la Rubia Mir, L.(Eds.), Jornadas tcnicas de la Variante de Pajares. Comunicaciones presentadas.Universidad de Oviedo, Oviedo, pp. 221227.

Arroyo Cedrn, J.C., Gonzlez del Tnago Chanrai, J., Ramos Gmez, J., 2009. Galerastransversales en los Tneles de Pajares. In: Mguez Bailo, R. (Ed.), Tneles de Pajares.ADIF, pp. 279306.

Campomanes Snchez, P., 2009. Extraccin dematerial a lo largo del tnel. Caractersticas,funcionamiento y rendimiento de la cinta de tnel y tren de cintas exteriores. In:Pando, L., Lpez Fernndez, C., De la Rubia Mir, L. (Eds.), Jornadas tcnicas de laVariante de Pajares. Comunicaciones presentadas. Universidad de Oviedo, Oviedo,pp. 203210.

Casagrande, G., Cucchi, F., Zini, L., 2005. Hazard connected to railway tunnel constructionin karstic area: applied geomorphological and hydrogeological surveys. Nat. HazardsEarth Syst. Sci. 5 (2), 243250.

Cayn Martnez, A., Ramos Gmez, J., Fueyo Castan, L.E., Llamazares Fernndez, V.,2009. Estudio de ubicacin y estabilizacin de vertederos para el desescombro detneles de gran longitud. In: Pando, L., Lpez Fernndez, C., De la Rubia Mir, L.(Eds.), Jornadas tcnicas de la Variante de Pajares. Comunicaciones presentadas.Universidad de Oviedo, Oviedo, pp. 373380.

Cheng, Q., Ko, C., Yuan, Y., Ge, Y., Zhang, S., 2006. GIS modeling for predicting river runoffvolume in ungauged drainages in the Greater Toronto Area, Canada. Comput. Geosci.32, 11081119. http://dx.doi.org/10.1016/j.cageo.2006.02.005.

Chiocchini, U., Castaldi, F., 2011. The impact of groundwater on the excavation of tunnelsin two different hydrogeological settings in central Italy. Hydrogeol. J. 19, 651669.http://dx.doi.org/10.1007/s10040-010-0702-1.

Dackombe, R.V., Gardiner, V., 1983. Geomorphological eld manual. George Allen &Unwin, London.

Dez Cadavid, L., Luengo Troitio, D., 2009. Medio ambiente en los tneles de Pajares. In:Mguez Bailo, R. (Ed.), Tneles de Pajares. ADIF, pp. 209234.

Fernndez Rodrguez, F.J., Menndez Duarte, R., Riera Valds, R., 2002. Clculo de CaudalesPunta y diseo de Sistemas de Drenaje. Modelos Digitales de Flujo Corregido (MDFC).Rev. Obras Pblicas 3419, 1525.

Ferrer, F.J., 1993. Recomendaciones para el clculo hidrometeorolgico de avenidas. Centrode Estudios y Experimentacin de Obras Pblicas, Ministerio de Obras Pblicas, Madrid.

Ferreras Gonzlez, R., Higueras Higuero, J., Muz Rodrguez, G., Campomanes Snchez,P., 2009. Ejecucin del depsito controlado de residuos inertes de La Cortina. In:Mguez Bailo, R. (Ed.), Tneles de Pajares. ADIF, pp. 483508.

Galn, L., 1990. Hidrogeologa, in: Rodrguez Fernndez, R.L. (Ed.), Memoria explicativa dela Hoja 103 (Pola de Gordn) del Mapa Geolgico de Espaa (E.1:50.000). InstitutoTecnolgico GeoMinero de Espaa, Madrid, pp. 120121.

Galve, J.P., Gutirrez, F., Guerrero, J., Alonso, J., Diego, I., 2012. Application of risk,costbenet and acceptability analyses to identify the most appropriate geosyntheticsolution to mitigate sinkhole damage on roads. Eng. Geol. 144145, 6577. http://dx.doi.org/10.1016/j.enggeo.2012.07.002.

Garrido Ruiz, M.S., Serrano Martn, L., Mguez Bailo, R., Lombardero Barcel, M.,Fernndez-Serrano Snchez, M., lvarez Dez, J., 2009. El Estudio Hidrogeolgico delos Tneles de Pajares. In: Pando, L., Lpez Fernndez, C., De la Rubia Mir, L. (Eds.),Jornadas tcnicas de la Variante de Pajares. Comunicaciones presentadas. Universidadde Oviedo, Oviedo, pp. 7786.

Graves, C., 2007. Hydrological methods. In: Goldscheider, N., Drew, D. (Eds.), IAH:International Contributions to Hydrogeology: Methods in Karst Hydrogeology 26.Taylor & Francis/Balkema, pp. 4564 (Robins, N.S. (Series Ed.)).Guerrero, J., Gutirrez, F., Bonachea, J., Lucha, P., 2008. A sinkhole susceptibility zonationbased on paleokarst analysis along a stretch of the MadridBarcelona high-speedrailway built over gypsum- and salt-bearing evaporites (NE Spain). Eng. Geol. 102(12), 6273. http://dx.doi.org/10.1016/j.enggeo.2008.07.010.

Gutirrez, F., Parise, M., De Waele, J., Jourde, H., 2014. A review on natural and human-induced geohazards and impacts in karst. Earth-Sci. Rev. 138, 6188. http://dx.doi.org/10.1016/j.earscirev.2014.08.002.

Gutirrez-Santolalla, F., Gutirrez-Elorza, M., Marn, C., Desir, G., Maldonado, C., 2005.Spatial distribution, morphometry and activity of La Puebla de Alndn sinkholeeld in the Ebro River valley (NE Spain), applied aspects for hazard zonation. Environ.Geol. 48, 360369.

Julivert, M., 1971. Dcollement tectonics in the Hercynian Cordillera of NW Spain. Am.J. Sci. 270, 129.

Julivert, M., Fontbot, J.M., Ribeiro, A., Nabais-Conde, L.E., 1972. Mapa tectnico de laPennsula Ibrica y Baleares a escala 1:1.000.000 Memoria explicativa. InstitutoGeolgico y Minero de Espaa, Madrid.

Ko, C., 2004. Storm runoff volume estimation in the Oak Ridges Moraine Area using GISand remote sensing techniques (Doctoral thesis) University of York, Toronto, Ontario.

Ko, C., Cheng, Q., 2004. GIS spatial modeling of river ow and precipitation in the OakRidges Moraine area, Ontario. Comput. Geosci. 30 (4), 379389.

Lotze, F., 1945. Zur Gliederung der Varisziden der Iberischen Meseta. Geotekt. Forsch.6, 7892.

MARM, 2009. Mapa de Cultivos y Aprovechamientos de Espaa 20002010 (E. 1:50.000). Ministerio de Agricultura, Alimentacin y Medio Ambiente. Availableonline in: SIGA, 07/03/2014, http://www.magrama.gob.es/es/cartograa-y-sig/publicaciones/agricultura/mac_2000_2009.aspx.

Marqunez, J., Menndez Duarte, R.A., Jimnez Snchez, M., 1990. Geomorfologa, in:Rodrguez Fernndez, R.L. (Ed.), Memoria explicativa de la Hoja 103 (Pola de Gordn)del Mapa Geolgico de Espaa (E.1:50.000). Instituto Tecnolgico GeoMinero deEspaa, Madrid, pp. 7479.

Mguez Bailo, R., 2005. The Pajares Tunnels. Rev. Obras Pblicas 3460, 732.Mguez Bailo, R., Benito Alonso, A., Angona Sopea, A., 2007. The Folledo Access adit to the

Pajares Tunnels. Rev. Obras Pblicas 3474, 3354.Milanovic, P.T., 2004. Water resources engineering in Karst. CRC Press, Washington D.C.Ministerio de Fomento, 1999. Mximas Lluvias diarias en la Espaa Peninsular. Secretara

de Estado de Infraestructuras y Transportes. Direccin General de Trco, Madrid.MOPU, 1990. Instruccin de Carreteras 5.2-IC Drenaje supercial. Boletn Ocial del

Estado 123, 23-5-1990. Ministerio de Obras Pblicas y Urbanismo.Newton, J.G., 1986. Natural and induced sinkhole development in the eastern United

States. In: Johnson, A.I., Carbognin, L., Ubertini, L. (Eds.), Land Subsidence: Proceedingsof the Third International Symposium on Land Subsidence 151. IAHS Publication,Venice, Italy, pp. 549564.

Newton, J.G., Hyde, L.W., 1971. Sinkhole problem in and near Roberts Industrial Subdivi-sion, Birmingham, Alabama a reconnaissance. Geol. Surv. Ala. Circ. 68 (42 pp.).

Ortiz, J., Iglesias, D., Hernndez, R., De Mera, A., Gmez, M., Lpez, J., Ballester, A., 2001.Desarrollo de programas para el tratamiento de series hidrolgicas y evaluacin deaportaciones hdricas subterrneas (HIDROBAS v. 3.0). VII Simposio de Hidrogeologa23. Asociacin Espaola de Hidrologa Subterrnea, Murcia, pp. 333341.

Raposo, J.R., Molinero, J., Dafonte, J., 2010. Quantitative evaluation of hydrogeologicalimpact by tunnel construction using water balance models. Eng. Geol. 116 (34),323332. http://dx.doi.org/10.1016/j.enggeo.2010.09.014.

Rodrguez Fernndez, B., 2011. Alto Bernesga. In: Cantos Mengs, F.J. (Ed.), Guide toSpanish Biospheres Reserves, Man and nature in harmony. Spanish Authority forNational Parks, pp. 185190.

Segura Prez, P., Martnez Daz, J.F., 2009. Hormigones de alta resistencia en el tramo II delTnel de Pajares. In: Mguez Bailo, R. (Ed.), Tneles de Pajares. ADIF, pp. 153166.

Shepard, D., 1968. A two-dimensional interpolation function for irregularly spaced data.Proceedings of the 23rd National ConferenceAssociation for Computing Machinery.Association for Computing Machinery, Las Vegas, pp. 517524.

Sjolander-Lindqvist, A., 2005. Conicting perspectives on water in a Swedish railwaytunnel project. Environ. Values 14 (2), 221239.

Song, K.I., Cho, G.C., Chang, S.B., 2012. Identication, remediation and analysis of karstsinkholes in the longest railroad tunnel in South Korea. Eng. Geol. 135136,92105. http://dx.doi.org/10.1016/j.enggeo.2012.02.018.

Tihansky, A.B., 1999. Sinkholes, west central Florida. In: Galloway, D., Jones, D.R.,Ingebritsen, S.E. (Eds.), Land Subsidence in the United States: U.S. Geological SurveyCircular 1182. USGS, pp. 121140.

Tobler, W.R., 1970. A computer model simulation of urban growth 80 in the Detroitregion. Econ. Geogr. 46 (2), 234240.

Toyos, J.M., Surez, M.A., Rodrguez Fernndez, L.R., Serrano, L., 2009. Perl geolgico a lolargo del trazado de los tneles de Pajares. Jornadas tcnicas de la Variante de Pajares.Comunicaciones presentadas Universidad de Oviedo, Oviedo, pp. 4152.

USDA SCS, 1972. National Engineering Handbook, Section 4: hydrology. U.S. Departmentof Agriculture-Soil Conservation Service, Washington, D.C.

Vincenzi, V., Gargini, A., Goldscheider, N., 2009. Using tracer and hydrological observationsto evaluate effects of tunnel drainage on groundwater and surface waters in theNorthern Apennines (Italy). Hydrogeol. J. 17, 135150. http://dx.doi.org/10.1007/s10040-008-0371-5.

Vincenzi, V., Gargini, A., Goldscheider, N., Piccinini, L., 2014. Differential hydrogeologicaleffects of draining tunnels through the Northern Apennines, Italy. Rock Mech. Rock.Eng. 47 (3), 947965. http://dx.doi.org/10.1007/s00603-013-0378-7.

Waltham, T., 2008. Sinkhole hazard case histories in karst terrains. Q. J. Eng. Geol.Hydrogeol. 41, 291300.

Waltham, T., Bell, F.G., Culshaw,M.G., 2005. Sinkholes and subsidence. Springer-Praxis, Berln.William, P., 2004. Dolines. In: Gunn, J. (Ed.), Encyclopedia of Caves and Karst Science.

Fitzroy Dearborn, New York, pp. 304310.170 P. Valenzuela et al. / Engineeri

Active sinkholes: A geomorphological impact of the Pajares Tunnels (Cantabrian Range, NW Spain)1. Introduction2. Description of the study area3. Material and methods3.1. Geomorphological analysis3.2. Hydrogeological analysis3.3. GIS data management3.3.1. Creation of digital data layers3.3.2. Calculation of runoff coefficients3.3.3. Rainfall interpolation models3.3.4. Calculation of the volume of infiltrated water

4. Results4.1. Geomorphological analysis4.2. Hydrogeological analysis4.3. GIS modeling

5. Discussion6. ConclusionsAcknowledgmentsReferences