STOTEN-21393; No of Pages 14

Science of the Total Environment xxx (2016) xxx–xxx

Contents lists available at ScienceDirect

Science of the Total Environment

journal homepage: www.elsevier.com/locate/scitotenv

The impact of droughts and climate change on sinkhole occurrence. A

case study from the evaporite karst of the Fluvia Valley, NE Spain
Rogelio Linares a,⁎, Carles Roqué b, Francisco Gutiérrez c, Mario Zarroca a,1, Domingo Carbonel c,
Joan Bach a, Ivan Fabregat c
a
Departamento de Geología, Universidad Autónoma de Barcelona, E-08193 Barcelona, Spain
b
Àrea de Geodinàmica Externa i Geomorfologia, Universitat de Girona, Campus Montilivi, E-17003 Girona, Spain
c
Departamento de Ciencias de la Tierra, Universidad de Zaragoza, C/. Pedro Cerbuna 12, 50009 Zaragoza, Spain

H I G H L I G H T S G R A P H I C A L A B S T R A C T

• Sinkhole inventory reveals good chro-

nological correlation with drought pe-
riods.
• Environmental-related hydrodynamic
changes play a main role on subsidence.
• Sinkhole hazard probably will intensify
in a future due to the climate change.

a r t i c l e i n f o a b s t r a c t

Article history: This work introduces the concept that sinkhole frequency in some karst settings increases during drought pe-
Received 22 September 2016 riods. This conception is tested in a sector of the Fluvia River valley in NE Spain, where subsidence phenomena
Received in revised form 14 November 2016 is related to the karstification of folded Eocene evaporite formations. In the discharge areas, the evaporites behave
Accepted 14 November 2016
as confined aquifers affected by hypogene karstification caused by aggressive artesian flows coming form an un-
Available online xxxx
derlying carbonate aquifer. A sinkhole inventory with chronological data has been constructed, revealing tempo-
Editor: D. Barcelo ral clusters. Those clusters show a good correlation with drought periods, as revealed by precipitation, river
discharge and piezometric data. This temporal association is particularly obvious for the last and current drought
Keywords: starting in 1998, which is the most intense of the record period (1940–present). Climatic projections based on
Evaporite karst recent studies foresee an intensification of the droughts in this sector of NE Spain, which could be accompanied
Sinkhole hazard management by the enhancement of the sinkhole hazard and the associated risks.
Global change © 2016 Elsevier B.V. All rights reserved.
Hypogene karst

1. Introduction

The development of subsidence sinkholes in karst terrains involves

⁎ Corresponding author.
the simultaneous or sequential operation of two processes: subsurface
E-mail address: rogelio.linares@uab.cat (R. Linares). dissolution and the internal erosion and/or gravitational deformation
1
Serra Húnter fellow. of the overlying material (Gutiérrez et al., 2014; Gutiérrez, 2016).

http://dx.doi.org/10.1016/j.scitotenv.2016.11.091
0048-9697/© 2016 Elsevier B.V. All rights reserved.

Please cite this article as: Linares, R., et al., The impact of droughts and climate change on sinkhole occurrence. A case study from the evaporite
karst of the Fluvia Valley, NE S..., Sci Total Environ (2016), http://dx.doi.org/10.1016/j.scitotenv.2016.11.091
2 R. Linares et al. / Science of the Total Environment xxx (2016) xxx–xxx

Groundwater plays a major role in the development of both processes, and roughly controls the course of the Fluvia valley (Fig. 1A). (2) Neo-
as the agent responsible for the dissolution, and as a first order factor gene normal faults superimposed to the contractional structures with
that determines ground instability. In fact, recent reviews reveal that a NNW-SSE and NW-SE trends. These post-orogenic structures include
great proportion of the recent sinkholes have been induced by anthro- the east-dipping Camos-Celra Fault on the east edge of the study area,
pogenic changes in the hydrogeological systems (Waltham et al., which controls the western margin of the Emporda Neogene Graben
2008; Gutiérrez, 2016). Consequently, a sound knowledge on the (Saula et al., 1994) (Fig. 1B). A more detailed description of the geology
short- and long-term dynamics of the hydrogeological systems is essen- of the area can be found in Gutiérrez et al. (2016).
tial for sinkhole hazard assessment (e.g., Lamont-Black et al., 2005; In the middle reach of the Fluvia valley the Eocene bedrock is exten-
Yechieli et al., 2006; Acero et al., 2015; Parise et al., 2015). sively covered by Late Pliocene and Quaternary deposits, including
This work is focused on the potential relationships between natural mantled pediments, fluvial terraces and tufaceous limestones
hydrological changes, namely droughts and the associated declines in (Fig. 1B). These surficial formations mainly occur in the areas underlain
the piezometric level of karst aquifers, and the development of collapse by the less resistant evaporite rocks. Both the alluvial deposits and the
sinkholes. There are a large number of papers that document the impact lacustrine tufas locally fill basins N70 m deep generated by
of human-induced water table declines on sinkhole occurrence. The dissolution-induced synsedimentary subsidence, and typically show
groundwater level drops conductive to sinkhole formation are mainly abundant gravitational deformation and paleosinkholes. Additional in-
related to aquifer over-exploitation (e.g., Newton, 1984a,b; Dogan and formation of the morpho-stratigraphic sequence, including a detailed
Yilmaz, 2011; Aurit et al., 2013; Taheri et al., 2015; Youssef et al., geomorphological map, was presented by Gutiérrez et al. (2016).
2016) and dewatering for mining (e.g., Li and Zhou, 1999; Sprynskyy In the south-eastern sector of the study area there is a major ground-
et al., 2009). However, the impact of climate-related hydrological water discharge area associated with the so-called Banyoles-Besalu lake
changes on sinkhole hazard has been barely explored, largely due to basin (Julià, 1980; Sanz, 1981) (Fig. 1A). The majority of the lake de-
the lack of long records of hydro-climatic data, and especially represen- posits related to modern and Plio-Quaternary lacustrine environments
tative sinkhole inventories including chronological information. are concentrated in this area (Fig. 1B). The good spatial correlation be-
In the Spanish Pyrenean region the statistical climatic projections tween the distribution of lacustrine deposits and the evaporite bedrock,
foresee an increase in the mean temperature and a decrease in the aver- as well as the presence of paleolakes and present-day lakes associated
age monthly precipitation (10–25% for 2021–2050 with respect to with collapse sinkholes reveal the instrumental role played by
1971–2000) (Barrera-Escoda et al., 2014; Gonçalves et al., 2014). Ac- dissolution-induced subsidence phenomena in the development of the
cording to these authors, most probably the frequency and intensity of Banyoles-Besalu lake basin (e.g., Julià, 1980; Gutiérrez et al., 2016). In
droughts will increase in the near future. Studies based on paleoclimatic fact, the 2 km long Banyoles Lake is associated with multiple coalesced
data suggest the same trend (Cook et al., 2016). Climate change in the sinkholes related to the interstratal karstification of evaporites by up-
region may involve a wide range of environmental problems of biolog- ward groundwater flow derived from an underlying carbonate aquifer
ical and physical nature reviewed by Calbó et al. (2010). (e.g., Brusi et al., 1992; Canals et al., 1990).
This work proposes that the intensification of droughts may be ac- The catchment of the Fluvia River has a Mediterranean climate, with
companied by an increase in the sinkhole hazard and introduces an ad- subalpine features in the mountainous sectors, mainly located in the
ditional geohydrological impact associated with the current human- northern headwaters (Alta Garrotxa; Fig. 1). The mean annual precipita-
induced climate change: enhanced sinkhole activity during droughts. tion varies from 630 mm to 1165 mm and there are clearly differentiat-
This would be an innovative geomorphic effect with detrimental conse- ed humid (spring, autumn) and dry (summer and winter) seasons. The
quences of climate change, whose potential impact remains unex- average annual temperature is around 15.5 °C in the valley, and lower
plored. Previous investigations analyse the impact of the human- than 9 °C in the northern mountains (ACA, 2014). According to Sanz
induced temperature increase on the geochemistry of carbonate aqui- (1981), the actual evapotranspiration ranges between 58% and 80% of
fers by enhanced incorporation of CO2 from soils and dissolution the precipitation. During the period 1940–2008, the average actual
(e.g., Jeannina et al., 2016). Nonetheless, to our knowledge, this is the evapotranspiration in the Fluvia drainage basin reached 680 mm,
first time that this type of geomorphic change has been explored in re- which is the highest value estimated for the catchments of the area
lation to climate change. This potential trend should be taken in consid- (ACA, 2014).
eration in sinkhole hazard analyses performed in regions where an South of the Vallfogona Thrust, in the southern sector of the Fluvia
intensification of the droughts is expected. In those areas the use of River basin, the stratigraphy of the Eocene bedrock and its structure de-
spatial-temporal frequencies of sinkholes based on historical records termines the presence of a hydrogeological system comprising two hy-
may significantly lead to underestimated hazard values. draulically connected karstic aquifers mostly confined by low-
permeability sediments (Sanz, 1981; Bach et al., 1986) (Fig. 1A, C).
2. Geological setting and hydrogeology This aquifer system has a minimum estimated volume of ca. 60 × 106
m3. The lower karst aquifer (Lower aquifer) corresponds to the Eocene
The study area is located in the middle reach of the E-W-oriented limestones of the Penya and Girona formations. This hydrostratigraphic
Fluvia River valley, NE Spain. From the geological perspective, it is situ- unit is around 200 m thick in the middle reach of the Fluvia valley and
ated astride two major geological units related to the development of wedges out towards to south to b100 m (Fig. 1C). The overlying aquifer
the Alpine Pyrenean orogene. These are the Cadi Unit of the South Pyr- unit (Intermediate aquifer) includes two main evaporite units (Serrat
enean Zone to the north, and the Ebro Cenozoic Basin to the south, both Evaporites and Beuda Gypsum) (Carrillo et al., 2014). The Intermediate
separated by the south-verging Vallfogona Thrust (Fig. 1A, C). The bed- aquifer reaches a thickness of around 400 m in the middle reach of the
rock consists of an Eocene succession N 3 km thick including carbonate, Fluvia valley and wedges out progressively towards the south
evaporite and detrital formations that record the evolution of the South- (Fig. 1C). Due to its particular stratigraphy, this aquifer behaves as a
eastern Pyrenean Foreland basin; i.e., transition from marine to conti- complex multilayer system comprised by several units with very differ-
nental conditions and the partial incorporation of the basin fill into ent hydraulic conductivities. The horizontal and vertical connectivity
the thrust sheets (Puigdefàbregas et al., 1986; Mató et al., 1996; between the different units is largely controlled by the fracture system
Martínez et al., 2000; Carrillo et al., 2014) (Fig. 1B, C). These formations and karstic conduits, both intraformational and transtratal.
are affected by two main groups of tectonic structures (Fig. 1): (1) Paleo- The Alta Garrotxa massif in the northern sector of the Fluvia catch-
gene folds and thrusts with a dominant E-W trend related to the devel- ment, dominated by limestone outcrops, is the main recharge area of
opment of the Pyrenean orogene. These include the Vallfogona Thrust, the karstic system (Fig. 1C). Stable isotope data (18O) (Sanz et al.,
which is the sole thrust of the Cadi Unit (Clavell et al., 1988) (Fig. 1C) 1982) and tests carried out with chemical tracers (Vidal-Pardal, 1957)

Please cite this article as: Linares, R., et al., The impact of droughts and climate change on sinkhole occurrence. A case study from the evaporite
karst of the Fluvia Valley, NE S..., Sci Total Environ (2016), http://dx.doi.org/10.1016/j.scitotenv.2016.11.091
 R. Linares et al. / Science of the Total Environment xxx (2016) xxx–xxx 3

Fig. 1. Geological and hydrogeological setting of the study area (from Mató et al., 1996 and Martínez et al., 2000). (A) Synthetic hydrogeological sketch of the Fluvia watershed indicating
the main groundwater flow paths and the most relevant structures that control them. (B) Simplified geological map of the study area and shaded relief model showing the distribution of
evaporite bedrock and Plio-Quaternary tufaceous sediments deposited in paleolakes fed by groundwater discharge. (C) Simplified hydrogeological cross-section showing the geometrical
relationships of the main hydrostratigraphic units.

indicate that the majority of the infiltration occurs through outcrops of The groundwater that flows through the Lower aquifer has a rela-
the karstified Penya Fm. The main discharge areas of the aquifer corre- tively low ionic content as is close to equilibrium with respect to calcite
spond to the Espolla-Usall Platform (215 m a.s.l.), the Sant Miquel de (Sanz, 1981). In the discharge areas, when this water flows upwards
Campmajor valley (200 m a.s.l.), the Banyoles Lake basin (155 m through the overlying aquifer (intermediate evaporite aquifer), it rapid-
a.s.l.), and the middle reach of the Fluvia valley (155 m a.s.l.) (Sanz, ly causes the karstification of the evaporites and increases its minerali-
1981), where artesian springs are common (Fig. 1). The average dis- zation. This can be considered as a hypogene karst system in the sense
charge in the first zone is around 5 hm3/yr, and reaches 12 hm3/yr in of Klimchouk (2009), whereby the water responsible for the
Banyoles Lake (IGME, 1986). Probably, there is also unrecognised karstification flows upward from an underlying aquifer. Other authors
groundwater discharge through Quaternary fluvio-lacustrine deposits like Palmer (2007) restrict that term to situations where the chemical
in valleys close to Olot town (Fig. 1A). Some of the springs behave as aggressiveness of the waters is produced below the surface
trop-pleins functioning only during periods of high piezometric levels (e.g., incorporation of deep-sourced gases like CO2 or H2S). According
(e.g., Clot d'Espolla) (Sanz and Trilla, 1982). to Bischoff et al. (1994), the presence of dolomitic units within the

Please cite this article as: Linares, R., et al., The impact of droughts and climate change on sinkhole occurrence. A case study from the evaporite
karst of the Fluvia Valley, NE S..., Sci Total Environ (2016), http://dx.doi.org/10.1016/j.scitotenv.2016.11.091
4 R. Linares et al. / Science of the Total Environment xxx (2016) xxx–xxx

intermediate aquifer contributes to slightly accelerate gypsum dissolu- discharge zone of the aquifer, around 4 km south of the sinkhole area
tion due to dedolomitization reactions that deplete the concentration and considered to be representative of the hydraulic conditions of the
of dissolved calcium. The connection between the Lower and Intermedi- aquifers at different depths (Figs. 1B, 2). Groundwater withdrawal in
ate aquifers is most probably facilitated by the presence of carbonate the area is rather low and only affects the groundwater level of the sur-
units intercalated within the evaporitic formations, as revealed by ficial aquifer at a local scale. The piezometric levels in the lower aquifers
deep-borehole data (Carrillo et al., 2014), and tectonic faults. are controlled by the recharge related to precipitation episodes. The
The groundwater flow has a relatively rapid circulation. The tritium available reports indicate that these aquifers have an exploitation
(3H) content in the groundwater of the lower aquifer reveals a residence index of 0.07–0.1, which is the lowest of the Catalan region (ACA,
time between 11 and 18 months (Sanz and Trilla, 1982). The lag time 2014). The available groundwater resources in dry (70.1 hm3/yr) and
between heavy rainfall episodes in the recharge areas and the associat- average hydrological years (93.4 hm3/yr) are much higher than the av-
ed increases in the flow rate at the most distal springs is generally erage exploitation volume (6.81 hm3/yr). Moreover, according to the of-
around three months (Sanz, 1981). The hydraulic parameters of the sys- ficial inventories of groundwater exploitation points, our study area
tem are poorly known. The scarce data available indicate that the con- (Maia piedmont) is the one with the lowest number of pumping sites
fined karstic aquifers in limestone and evaporites have high hydraulic (ACA, 2006).
conductivity (N100 m/day) and storage coefficients between 0.1 and The characteristics of the borings and the data were provided by the
0.2 (ACA, 2006). The productivity of the wells change rapidly according Agencia Catalana de l'Aigua (ACA) and downloaded from Global Aquifer
to the seasonal fluctuations of the piezomentric levels, suggesting a Control (http://gac.cloudapp.net/aca). The 445 m deep IRYDA borehole
rapid emptying of the tertiary porosity related to large karst conduits provides the oldest piezometric data, between 1985 and 2011. The
(Sanz, 1981). The surficial formations generally have lower permeabili- screen is situated between 190 m and 441 m depth, and consequently
ty (1–10 m/day) and production rates. records average hydraulic conditions in the Lower carbonate aquifer
and in the lower part of the Intermediate evaporite aquifer (Fig. 2). In
3. Methodology November 2011 it was replaced by Banyoles-1 piezometer due to fail-
ure. Since 2009, the multiple piezometer Fontcoberta-1, -2 and -3, locat-
The construction of a detailed sinkhole inventory has been focused ed close to the Clot d'Espolla temporary spring, monitor the base of the
in the surroundings of Maia village, on the northern margin of the Fluvia surficial aquifer (38–53 m depth), as well as the upper (150–159 m
valley and in the footwall of the Camos-Celra Fault. In this area the depth) and lower sections (220–232 m depth) of the Intermediate
folded Eocene evaporitic bedrock is extensively covered by Quaternary evaporite aquifer (Fig. 2). Moreover, we have analysed the precipitation
gravelly pediments and terraces, as well as Plio-Quaternary tufaceous records (Servei Meteorologic de Catalunya, www.meteo.cat) in order to
limestones deposited in a paleolake. The selection of this zone was identify the drought periods, assess their severity, and analyse their re-
based on the following criteria: (1) This is one of the sectors of the mid- lationships with the piezometric records and the occurrence of
dle reach of the Fluvia valley with higher sinkhole density (Gutiérrez sinkholes.
et al., 2016). (2) A considerable number of sinkholes have been formed
or reactivated during the last 25 years. (3) Land use, mostly agriculture
and forestry, has barely changed over the last decades. Irrigation sys-
tems were implemented in some localized areas during a few years,
but the hydrological conditions have been governed by natural process-
es. (4) The area is located close to one of the main discharge areas of the Table 1
Characteristics and date of the images and maps used for the construction of the sinkhole
aquifer system (Fig. 1B). Commonly, the discharge of deep groundwater inventory.
flows from confined aquifers produces piezometric domes that experi-
ence significant head variations controlled by the temporal patterns of Type Date Scale/resolution

the recharge (e.g., Linares et al., 2010). Consequently, these zones are Greyscale aerial photograph 1946 ≈1:43,000
particularly favourable for exploring the effects related to piezometric Greyscale aerial photograph 1956 ≈1:32,000
Greyscale aerial photograph 1971 1:18,000
changes.
Greyscale aerial photograph 1974 1:18,000
Initially, we produced a preliminary cartographic inventory of sink- Greyscale aerial photograph 1980 1:18,000
holes and zones with probable subsidence signs through the interpreta- Greyscale aerial photograph 1988 1:18,000
tion of aerial photographs, orthoimages and topographic maps from Color aerial photograph 2002 1:22,000
different dates and with variable scales (Table 1). This inexpensive ap- Color aerial photograph 2006 1:22,000
Color aerial photograph 2010 1:22,000
proach, widely used in other sinkhole areas, allows bracketing the age Greyscale orthoimage 1946 100 cm/px
of the sinkholes formed or reactivated after the oldest image with suffi- Greyscale orthoimage 1956 50 cm/px
cient resolution and reconstruct the evolution of active sinkholes Greyscale orthoimage 1987 50 cm/px
(e.g., Gutiérrez et al., 2008a, 2011; Festa et al., 2012; Iovine et al., Color orthoimage 1993 2.5 m/px
Greyscale orthoimage 1996 50 cm/px
2016). Subsequently, the inventory was completed and refined by
Color orthoimage 2002 50 cm/px
means of field work, including: (1) direct inspection of the mapped Color orthoimage 2004 50 cm/px
sinkholes and acquisition of specific data (e.g., morphometry, signs of Color orthoimage 2005 50 cm/px
instability, relative age); (2) detailed reconnaissance of the areas with Color orthoimage 2008 50 cm/px
a higher density of subsidence signs in order to pinpoint the location Color orthoimage 2009 25 cm/px
Color orthoimage 2010 25 cm/px
of small sinkholes, depressions masked by dense vegetation and Color orthoimage 2011 25 cm/px
paleosinkholes exposed in outcrops; (3) interviews with local people Color orthoimage 2012 50 cm/px
to gather information on the spatial and temporal distribution of Color orthoimage 2013 50 cm/px
known and unrecognised sinkholes (e.g., sinkholes filled soon after Color orthoimage 2014 50 cm/px
Topographic map 1936–39 1:50,000
their formation).
Topographic map 1958 1:50,000
Once the sinkhole inventory was completed, we analysed the tem- Topographic map 1969 1:50,000
poral series of piezometric data in order to study the impact of the hy- Topographic map 1982 1:50,000
drological changes on the subsidence phenomena. We used the Topographic map 1988 1:25,000
piezometric records of five boreholes (IRYDA, Banyoles-1, and the mul- Topographic map 2006 1:5000
Topographic map 2014 1:5000
tiple piezometer Fontcoberta-1, -2, and -3) located in the main

Please cite this article as: Linares, R., et al., The impact of droughts and climate change on sinkhole occurrence. A case study from the evaporite
karst of the Fluvia Valley, NE S..., Sci Total Environ (2016), http://dx.doi.org/10.1016/j.scitotenv.2016.11.091
 R. Linares et al. / Science of the Total Environment xxx (2016) xxx–xxx 5

Fig. 2. Sketch showing the hydrogeological units investigated by boreholes and piezometers. The information on the subsurface stratigraphy is largely based on borehole logs and the
geological cross-sections of Martínez et al. (2000). The characteristics of the borings were provided by the ACA (Agencia Catalana de l'Aigua).

4. Subsidence processes and landforms in the study area the foundered material includes rounded gravels derived from allu-
vial levels initially located at higher elevation.
4.1. Paleosinkholes The depth of the karstification responsible for the paleosinkholes in
Maia area is unknown, although some of the subsidence structures are
In a previous work on the evaporite karst in the middle reach of rooted at depths higher than 20 m. In Besalu area there are large
the Fluvia River valley, Gutiérrez et al. (2016) documented a wide dissolution-induced basins filled by Quaternary alluvial gravels N
variety of dissolution-induced paleosubsidence structures. They re- 100 m thick. Also in this sector, cave studies indicate the presence of col-
corded the whole spectrum of subsidence mechanisms, including lapse chimneys (breccia pipes) N 50 m high that penetrate through thick
sagging, collapse, suffusion, and combinations of them. Moreover, caprock (Gutiérrez et al., 2016). Moreover, electrical resistivity surveys
subsidence may affect to the evaporitic bedrock, non-soluble cap- carried out across a large caprock collapse sinkhole in the vicinity
rocks and the alluvial cover. The distribution of the paleosinkholes Serinya, south of the study area, reveal collapse structures N 100 m
identified between Besalu town area and the Camos-Celra Fault are deep (Zarroca et al., 2016).
shown in Fig. 3. These karst-related gravitational structures provide
useful insight into the subsidence mechanisms involved in the devel- 4.2. Sinkholes
opment of sinkholes and the spatial distribution of the areas under-
lain by karstified evaporites. In the investigated area, covering 18.1 km2, a total of 58 sinkholes
In the Maia area, where the outcrops are mainly restricted to lim- have been inventoried (Fig. 4). The ascribed typology, the main mor-
ited artificial excavations, we have identified paleosinkholes in two phometric parameters and their age are compiled in Table 2. This
sectors (Fig. 4). (1) Incarcal quarry: Here, deformed Late Pliocene- cannot be considered as a complete inventory, since is not feasible
Early Pleistocene lacustrine calcisiltites and calcirudites are cross- to map all the sinkholes, especially the small ones, but should be a
cut by cylindrical collapse structures 5–30 m across. These representative sample of the sinkholes developed in the area. Thus,
paleosinkholes are filled by fossil-rich Early Pleistocene greenish the available data indicate a minimum density of 3.2 sinkholes/
grey clays deposited in relatively deep sinkhole ponds (Galobart km 2, and a minimum area affected by subsidence of 168,351 m 2 ,
and Agustí, 2003; Ros-Montoya et al., 2010). These are caprock col- which represents 0.93% of the study area. These are higher values
lapse paleosinkholes related to the interstratal karstification of evap- than those estimated for the middle reach of the Fluvia valley in a
orites, probably recording an old groundwater discharge area previous work (1.1 sinkholes/km2 and 0.56% of the area affected by
(Gutiérrez et al., 2016). (2) Puig Gras: In the cuttings of the N-260 sinkholes; Gutiérrez et al., 2016).
road and in nearby areas there are several paleosinkholes affecting The sinkhole typology has been ascribed on the basis of the morpho-
the Eocene sandstones and marls of the Coubet Fm., which consti- logical features and following the classification proposed by Gutiérrez
tutes the confining unit of the Intermediate evaporite aquifer. et al. (2008b) and Gutiérrez (2016). This genetic classification describes
These poorly exposed subsidence structures are 35–60 m across sinkholes according to the material affected by subsidence (cover, bed-
and consist of chaotically brecciated caprock juxtaposed on both rock and caprock) and the subsidence mechanisms (collapse, sagging
sides and with subvertical contacts against laterally continuous stra- and suffosion). Obviously, the reliability of the typological ascription
ta of Coubet Fm., with a general 25NE dip. Locally, the upper part of on the basis of the morphological features decreases with the age of

Please cite this article as: Linares, R., et al., The impact of droughts and climate change on sinkhole occurrence. A case study from the evaporite
karst of the Fluvia Valley, NE S..., Sci Total Environ (2016), http://dx.doi.org/10.1016/j.scitotenv.2016.11.091
6 R. Linares et al. / Science of the Total Environment xxx (2016) xxx–xxx

Fig. 3. Map showing the main geological, geomorphological and hydrogeological features of the eastern sector of the middle reach of the Fluvia Valley. The geological and geomorphological
data are derived from Martínez et al. (2000) and Gutiérrez et al. (2016), respectively.

the sinkholes due to the higher impact of the degradation processes of Eocene evaporites (Beuda Gypsum) and next to the Camos-Celra
(i.e., erosion on the edges and deposition in the bottom), especially for Fault. Most probably, the intense karstification in this sector is relat-
the sinkholes mapped with the 1946 aerial photographs, which are ed to higher permeability and lower mechanical strength of the
the oldest images with the lowest resolution. Essentially, the large evaporites caused by tectonic faulting. Sinkholes on outcrops of
diffuse-edged and pan-shaped basins have been interpreted as sag- sandstones and marls of the Coubet Fm. tend to occur close to the
ging sinkholes, whereas the well- and funnel-shaped depressions as- lower contact of this unit, where the caprock reaches a limited
cribed to the collapse category. The occurrence of cover suffosion thickness.
sinkholes is very unlikely since most of the cover deposits have a
considerable degree of cementation. According to our classification, 4.3. Temporal occurrence of sinkholes
a great proportion of sinkholes have been formed by collapse
(78%), and around 12% and 10% by sagging and a combination of The age of the sinkholes has been constrained with a variable de-
both processes, respectively. Around 10% of the sinkholes affect cap- gree of accuracy using the available topographic maps, aerial photo-
rocks (Coubet Fm. and Plio-Quaternary tufaceous limestones), 24% graphs, orthoimages and the data gathered during our field surveys
evaporitic bedrock, and 66% cover deposits. The latter may also affect (Fig. 4, Table 2). A limitation of the chronological data of the sinkhole
concealed bedrock or caprock. inventory is the lack of a precise age for a substantial part of the sub-
Sinkhole dimensions cover a broad range. The largest sagging de- sidence events: (1) 5 sinkholes out of 58 lack chronological data;
pression is 323 m long and around 61,694 m2 in area, whereas the (2) a maximum age (pre–1946 or pre–1956) is assigned to 36 (ca.
smallest collapse is 0.9 m across. The average area of the sagging sink- 60%); (3) the timing of 11 sinkholes is constrained with bracketing
holes and collapse sinkholes are 11,400 m2 and 450 m2, respectively. ages that may cover a span as large as 24 years, although in most
The average depth of the sinkholes is 2.5 m and reaches a maximum cases lower than 3 years; and (4) the date at month or year level is
of 6 m. These parameters represent minimum subsidence values, since only known for 6 sinkholes. Therefore, the sinkholes have been
sinkholes tend to reduce their depth through time by natural and an- grouped into different age ranges (pre–1956, 1956–1987, 1988–
thropogenic deposition. 1997, 1998–2015).
Regarding the spatial distribution, the mapped sinkholes show a Sinkholes formed before 1956. The aerial photographs taken in
high concentration north of Puig Gras, associated with an outcrop 1956–57 (American flight B) are the oldest images with good

Please cite this article as: Linares, R., et al., The impact of droughts and climate change on sinkhole occurrence. A case study from the evaporite
karst of the Fluvia Valley, NE S..., Sci Total Environ (2016), http://dx.doi.org/10.1016/j.scitotenv.2016.11.091
 R. Linares et al. / Science of the Total Environment xxx (2016) xxx–xxx 7

Fig. 4. Map showing the spatial and temporal distribution of the inventoried sinkholes in Maia area.

resolution. They were acquired when the extent and density of the tree work. In these photographs we have identified 41 sinkholes out of the
cover was much lower, and consequently allow identifying sinkholes 58 sinkholes of the inventory. The aerial photographs taken in 1946
that are nowadays masked by vegetation, but recognisable by field (American flight A) show 18 of those sinkholes, but the resolution of

Please cite this article as: Linares, R., et al., The impact of droughts and climate change on sinkhole occurrence. A case study from the evaporite
karst of the Fluvia Valley, NE S..., Sci Total Environ (2016), http://dx.doi.org/10.1016/j.scitotenv.2016.11.091
8 R. Linares et al. / Science of the Total Environment xxx (2016) xxx–xxx

Table 2 already existing in 1956 have experienced reactivations, indicating

Summary table of the sinkhole inventory indicating the ascribed typology, morphometric that subsidence is still active in those depressions. This is the case of
parameters and chronological data.
cover sagging and collapse sinkhole 57 (Figs 4 and 5A). This subcircular
Sinkhole Type Length-width Depth Area Age of sinkhole around 165 m in diameter can be recognised in the 1946 and
or diameter (m) (m2) formation/last 1956 photographs and shows a nested collapse 20 m across with a
(m) observed
pond. In the 1980 images the collapse has a diameter of 36 m and
activity/
shows a more fresh appearance. In 2014 this active sinkhole caused
1 Cover collapse 12 2.0 113 1980–1987
the opening of cracks in the track located on its western margin. The
2 Cover 27 1.0 573 Pre–1956
sagging-collapse nearby cover sagging sinkhole 32 has also experienced recent activity
3 Caprock collapse 8 3.0 50 Pre–1946 according to the land owner. The other two sinkholes formed before
4 Cover sagging 34–24 1.5 623 Pre–1956 1956 and with evidence of subsequent reactivation are located close
5 Cover 60–55 1.0 3100 Pre–1956 to the Camos-Celra Fault. In 2006, a collapse 1.5 m across occurred in
sagging-collapse
6 Cover 43–36 1.5 988 Pre–1956
the floor of the cover sagging sinkhole 28. To the south, the caprock col-
sagging-collapse lapse sinkhole 29, around 5 m deep and 58 m across, caused settlement
7 Cover collapse 5 1.0 20 2006–2008 on a track in 2013 and 2014.
8 Cover and caprock 31–18 5.0 359 Pre–1946 Sinkholes formed between 1956 and 1987 (Fig. 5B). During this time
collapse
interval, in addition to the growth of the nested collapse in sinkhole 57,
9 Cover collapse 10 1.0 79 2006–2008
10 Cover collapse 45–28 5.0 1050 Pre–1946 we have detected the occurrence of three sinkholes. The cover collapse
11 Cover collapse 31–22 2.2 1513 1956–1974 sinkhole 11 north of Maia is expressed as a ponded collapse 31 m across
12 Cover collapse 16 ? 201 Pre–1946 in aerial photographs from 1974. The cover collapse sinkholes 21 and 1,
13 Cover collapse 17 ? 227 Pre–1946 around 4 and 12 m across are first identified in images taken in 1980
14 Cover collapse 25–22 5.0 380 November 1998
and 1987, respectively.
15 Cover collapse 60–42 4.0 2043 Pre–1946
16 Cover collapse 3 1.0 7 2012–2014 Sinkholes formed between 1988 and 1997. No sinkhole occurrence
17 Cover collapse 10 ? 78 Pre–1956 has been documented in the study are during this period. Nonetheless,
18 Cover collapse 10–9 4.0 63 2005–2006 there are reports of sinkhole events formed within this time span in
19 Cover sagging 323–249 5.0 61,694 Pre–1946
other sectors of the Fluvia valley, notably four damaging collapses in
20 Cover collapse 1.5 1.0 2 2012–2014
21 Cover collapse 4 1.0 13 1956–1980
the Grup del Mont suburb of Besalu, occurred between 1988 and 1996
22 Cover collapse 16 ? 201 Pre–1946 (Gutiérrez et al., 2016).
23 Cover collapse 13 ? 133 Pre–1946 Sinkholes formed between 1998 and 2015. This is the period with
24 Caprock collapse 1.5 1.0 2 Pre–1956 the highest frequency of sinkholes, with a total of 11 subsidence
25 Caprock collapse 35–33 3.0 907 Pre–1946
events, yielding a spatial-temporal probability of 0.03 sinkholes/
26 Cover collapse 10–8 2.0 64 Pre–1946
27 Cover collapse 22 1.0 380 Pre–1956 km 2 /yr. The cover collapse sinkhole 14, locally known as Forat de
28 Cover 126–69 1.5 7504 Pre–1956/2006/ Can Cademont, appeared on November 11, 1998 (Barberà, 1998)
sagging-collapse (Fig. 5C). Three months after its formation, this subcircular sinkhole
29 Caprock collapse 58–40 5.0 1848 Pre–1946/2014
with overhanging walls was 8.5 m deep. The floor was inundated by
30 Cover collapse 10 2.0 78 Pre–1946
31 Cover collapse 35–32 5.0 3525 Pre–1946
phreatic water of the surficial aquifer, and its diameter was 12.5 m
32 Cover sagging 69–42 1.5 2495 Pre–1956/2014/ and 14.5 m at the surface and in the bottom, respectively (Barberà,
33 Bedrock collapse 84–59 5.0 4504 Pre–1946 1998; Miret et al., 1999). This depression is depicted in the Topo-
34 Cover sagging 57 2.0 2552 Pre–1956 graphic Map of Catalunya at 1:5000 scale (sheet Maià de Montcal
35 Cover sagging 55–32 2.0 1654 Pre–1946
257-7-6) produced with aerial photographs taken in 2000. In De-
36 Cover collapse 0.9 1.5 0.6 October 2015
37 Cover collapse 16 2.0 201 Pre–1946 cember 2009 the sinkhole had degraded margins inclined around
38 Bedrock collapse 12 2.0 113 Pre–1956 45°, the diameter had grown to 22 m long and the dry floor was
39 Caprock collapse 20 5.0 314 Pre–1956 5.5 m deep. In 2014 the depression was 25 m long and 4 m deep.
40 Bedrock collapse 15–10 3.0 139 Pre–1956
The comparison of multiple aerial photographs available in the view-
41 Bedrock collapse 24–21 5.0 599 Pre–1956
42 Bedrock collapse 2.5 0.5 5 Pre–1956
er of the Institut Cartogràfic de Catalunya (Table 1) indicate that the
43 Bedrock collapse 8 2.0 50 Pre–1956 sinkhole experienced continuous morphological changes between
44 Bedrock collapse 15–13 3.0 141 Pre–1956 1998 and 2006, involving the degradation of the scarped margins,
45 Bedrock collapse 20–11 4.0 194 ? the areal growth of the depression and its partial fill. From 2008,
46 Bedrock collapse 6 3.0 28 ?
the sinkhole shows a state of morphological equilibrium and contin-
47 Bedrock collapse 8 3.0 50 ?
48 Bedrock collapse 5 2.5 20 Pre–1956 uous vegetation growth. A collapse sinkhole 10 m across occurred
49 Bedrock collapse 8 3.0 50 ? 530 m to the east of the previous sinkhole between April 2005 and
50 Bedrock collapse 15–14 3.0 145 Pre–1956 June 2006 (sinkhole 18). In 2006, a collapse was reported within sag-
51 Bedrock collapse 3 1.0 7 ? ging sinkhole 28. In addition, at least 5 collapse sinkholes occurred
52 Cover sagging 122–105 1.0 9624 Pre–1956
53 Cover 76 0.5 4536 Pre–1956
between 2006 and 2008 in the study area. Three of them (sinkholes
sagging-collapse 54, 55, 56) are between 6 and 13 m in diameter and are filled by an-
54 Cover collapse 10 ? 78 2006–2008 thropogenic deposits. Collapse sinkholes 9 and 7 were 10 m and 5 m
55 Cover collapse 6 ? 28 2006–2008 in diameter, respectively, and are partially filled. Three small col-
56 Cover collapse 13 ? 133 2006–2008
lapses have been formed between 2012 and 2015 (sinkholes 16, 20
57 Cover 165–162 6.0 21,382 Pre–1946/2014
sagging-collapse and 36). The floor of sinkhole 29, developed before 1946, was
58 Cover sagging 254–150 2.0 31,491 Pre–1956 reactivated in 2013–14 and sinkholes 57 and 32 showed signs of ac-
Average 38–30 2 2903 tive sagging.
Total 168,350

5. Drought episodes and hydrodynamics

the images is much lower. Moreover, the 1:50,000 scale topographic
maps produced before or during that period (1936–39, 1958) do not de- In NE Spain, the most severe droughts documented during the last
pict any of the sinkholes. According to our data, four of the sinkholes 100 years occurred during the following periods: 1910–15, 1922–23,

Please cite this article as: Linares, R., et al., The impact of droughts and climate change on sinkhole occurrence. A case study from the evaporite
karst of the Fluvia Valley, NE S..., Sci Total Environ (2016), http://dx.doi.org/10.1016/j.scitotenv.2016.11.091
 R. Linares et al. / Science of the Total Environment xxx (2016) xxx–xxx 9

Fig. 5. Examples of sinkholes with documented activity in the study area. (A) Aerial photographs from 1945 and 1957 showing the enlargement of sinkhole 57. (B) Sinkhole 11, dry in 1956
and ponded in 1974. (C) Morphological evolution of sinkhole 14 (Forat de Can Cademont) between 1999 and 2009 and oblique aerial photograph taken in 2016.

1944–50, 1973, 1985–86, 1988–89, 1999–2003, and 2005–present gamma function; (2) the gamma function is transformed into a stan-
(ACA, 2007a, 2007b; Altava-Ortiz et al., 2008; Barrera-Escoda, dard normal variable Z with a mean of 0 and a standard deviation of
2008). One of the most intense, which affected the whole Iberian 1. Thus, the SPI is a representation of the number of standard devia-
Peninsula, was the 1944–50 drought, including the driest hydrolog- tions of a precipitation value from the mean of the considered period.
ical year of the 20th century (October 1944–September 1945). The Negative values indicate a precipitation lower than the average,
hydrological year 1972–73 was also a severe drought period, which potentially attributable to droughts with an associated probability.
mainly affected the headwaters of the drainage basins. A long Negative values indicate relatively humid periods. The SPI values cal-
drought started in 1999 and extends up to the present time, with culated for the time interval 1940–2013 indicate that the last and
the exception of 2001 and 2004. This period, and especially the current drought is the most severe one, both in terms of intensity
time interval 2004–2008, is the driest in our study area since 1916 and duration (Fig. 6B). The drought started in the hydrological year
(Barrera-Escoda, 2008). 1998–99 and includes the year with lowest SPI of the record;
The temporal distribution and intensity of the drought periods in − 2.19 in 2006–07 (ACA, 2014). An anomalous feature of this
the Fluvia River catchment can be analysed using the discharge and drought is that it includes a 9-years-long interval from 2005 to
precipitation data available since 1940 (Fig. 6). In this work we 2013 with continuous negative SPI values. The average annual dis-
have used the climatic data gathered in the Plan de Gestión de charge of the Fluvia River recorded at Esponella Gauging Station
Distrito de Cuenca Fluvial de Cataluña 2016–2012 (ACA, 2014), (Fig. 3) shows a good match with the SPI. The lowest mean discharge
which includes the records of 157 meteorological stations of the values (b 2 m3/s) occur within the current drought, well below the
Spanish Agency of Meteorology. The historic precipitation data has average of 7.1 m3/s since 1912–13 (Fig. 6A).
been analysed using the Standardized Precipitation Index (SPI), Piezometric data from deep aquifers also reflect the exceptional
which allows the identification of droughts and assessing their prob- nature of this drought period, showing the lowest values of the re-
ability (McKee et al., 1993). This index has been used in previous cord. The data from the IRYDA borehole, spanning from 1985 to
works to analyse sinkhole development in other regions (Iovine 2011, is representative of the hydraulic head in the Lower carbonate
et al., 2016). The calculation of the SPI involves a two-step process: aquifer and the Intermediate evaporite aquifer (Fig. 7A). It should be
(1) historical rainfall data is fitted to a cumulative probability noted that the location of the screen does not enable monitoring the

Please cite this article as: Linares, R., et al., The impact of droughts and climate change on sinkhole occurrence. A case study from the evaporite
karst of the Fluvia Valley, NE S..., Sci Total Environ (2016), http://dx.doi.org/10.1016/j.scitotenv.2016.11.091
10 R. Linares et al. / Science of the Total Environment xxx (2016) xxx–xxx

Fig. 6. (A) Average annual discharge of the Fluvia River at Esponella gauging station (see location in Fig. 3). Data provided by the Agencia Catalana de l'Aigua (ACA). The lack of data
between 1937 and 1942 is related to the Spanish Civil War and the flood of 1940, 1941 and 1942. (B) Standarized Precipitation Index (SPI) of the hydrologic years 1940–41 to 2012–
13 in the Fluvia River basin and droughts classified into four intensity degrees with their associated probability. The temporal distribution of the sinkhole events documented in the
study area and in other nearby sectors is shown with different symbols.

possible differences between the water heads of the Lower carbonate The lowering of the water table in the surficial aquifers
aquifer and the Intermediate evaporite multilayer aquifer. The limit- (e.g., Quaternary alluvium) may significantly affect the stability of
ed precipitation in the northern headwaters of the basin, which is the the cavities whose roofs are located within the oscillation
main recharge area of the aquifer (Alta Garrotxa), results in a general zone. Such water table drops cause an increase in the effective
decreasing trend since 1997, with a marked drop in 2005. The deep weight of the cavity roofs (loss of buoyant support), may favour
karst aquifers in Olot area, within the Fluvia watershed and 20 km internal erosion processes by the replacement of slow phreatic
to the W, show a similar behaviour (Bach et al., 2012). flow by more rapid downward percolation, and may reduce the
mechanical strength of the deposits by desiccation and contraction
(e.g., Lamoreaux and Newton, 1986; Gutiérrez, 2016 and references
6. Discussion therein). Additionally, our observations suggest that other mecha-
nisms, relating to the subsurface hydrodynamics, may be also
The constructed sinkhole inventory including chronological data playing a role in the subsidence processes.
allows the recognition of temporal clusters of collapse events, espe- The piezometric record of IRYDA borehole (1985–2011)
cially the period spanning from 1998 to 2015. The SPI values calcu- supports the concept that sinkholes in the area occur preferably
lated for the time interval 1940–2013 show a good correlation during prolonged periods of low hydraulic head in the deep
between the drought periods and those with higher frequency of aquifers. In contrast, sinkhole activity is very limited when the
sinkhole occurrence (Fig. 6). The last and current drought period, in- piezometric levels are high (Fig. 7A). This is also supported by
cluding a historical maximum in 2007–2008 (Fig. 6), coincides with a the data from Banyoles-1 borehole (2011–2015), although the
significant intensification of the subsidence phenomena in the area timing of the subsidence events for that period is not tightly
(Fig. 8). constrained (Fig. 7B).

Please cite this article as: Linares, R., et al., The impact of droughts and climate change on sinkhole occurrence. A case study from the evaporite
karst of the Fluvia Valley, NE S..., Sci Total Environ (2016), http://dx.doi.org/10.1016/j.scitotenv.2016.11.091
 R. Linares et al. / Science of the Total Environment xxx (2016) xxx–xxx 11

Fig. 7. Comparison between the piezometric records of IRYDA and Banyoles-1 boreholes and the periods of sinkhole formation and inactivity documented in the study area and
surrounding areas.

This temporal pattern is also observed in other discharge areas of the the roofs. It should be noted that the potentiometric amplitude dur-
hydrogeological system and during different periods. In Sant Miquel de ing the observation period reaches 20 m of water column, which is
Campmajor valley (see location in Fig. 1B) historical data indicate that equivalent to a hydrostatic pressure variation of ca. 200 kPa.
sinkholes tend to form during drought periods. This is the case of the Accordingly, our results suggest that the alternation of wet periods
collapse events documented in 1732, 1817, 1872, 1880, 1908 and and severe droughts periods plays a key role in the temporal patterns
1910 (Sanz, 1981; Abellán, 2010). In the surroundings of Banyoles of sinkhole development. On the one hand, the stability of the roofs of
Lake, and especially in the western sector of the lacustrine basin, the cavities that have propagated close to the surface are very sensitive to
data provided by Sanz (1981) and Brusi et al. (1987, 1990) for the peri- the water table variations of the Upper aquifer. On the other hand, the
od 1974–1989 also support the correlation between droughts and sink- heterogeneous hydraulic response of the different units during drought
holes (Fig. 6B). Moreover, the sole collapse sinkholes documented in the periods may cause the depressurization of the upper part of the Inter-
Sant Miquel de Campmajor valley in the 20th Century occurred within mediate aquifer. This situation may also lead to readjustments in the
that time period. Sanz (1981), in his pioneering hydrogeological work, stress state of cave systems located at higher depth, facilitating the for-
proposed that decreases in the hydraulic head within phreatic caves mation of collapse events.
could be a major triggering factor for the development of collapse sink- The potential impact of drought periods on the development of col-
holes in the region. lapse events should be also taken into account for the
The comparison of the time series recorded in the multiple piezom- paleoenviromental interpretation of mass movement processes record-
eter of Fontcoberta since 2008 allows recognising the hydrodynamic re- ed in karstic lakes. A significant proportion of the lakes selected in Spain
lationships among the different aquifers during wet and drought for paleoclimatic investigations are associated with collapse basins re-
periods (Figs. 2 and 9). During significant recharge periods, the hy- lated to evaporite dissolution, whose origin and evolution are largely
draulic head in the upper part of the evaporite aquifer (P2) is lower governed by the groundwater: Montcortés Lake (e.g., Corella et al.,
than that of the underlying lower part of the evaporite aquifer (P3) 2011; Gutiérrez et al., 2012), Estanya Lake (e.g., Morellón et al., 2009),
and the overlying Upper aquifer associated with the surficial forma- Arreo Lake (e.g., Corella et al., 2013), Banyoles Lake (e.g., Sanz, 1981;
tions (P1). This fact shows that the recharge of the system occurs Colomer et al., 2006). According to Valero-Garcés et al. (2014), the grav-
mainly through the Upper aquifer (P1) and the Lower aquifer (Eo- itational facies (mass movements) are more frequent in the lake basins
cene limestones), which is hydraulically connected with the lower with steep margins and high depth/surface ratio, mostly corresponding
part of the Intermediate evaporite aquifer (P3). Assuming a poor hy- to collapse sinkholes. The gravitational facies identified in the deposits
draulic connectivity between the surficial aquifer (P1) and the upper of Banyoles Lake are mainly ascribed to hydrogeological changes in
part of the Intermediate aquifer (P2) (Fig. 2), during wet periods, P2 the feeding confined aquifer (Höbig et al., 2012; Morellón et al., 2014).
is recharged by upward groundwater flows from P3 (I on Fig. 9B). Höbig et al. (2012) described four mass-movement events; one
Conversely, during drought periods like between November 2009 constrained at 20–35 ka and three occurred within the age range 42–
and February 2010, the highest hydraulic head is recorded in P2. 54 ka. The youngest event might be correlative with an arid period in
This hydraulic gradient causes the discharge of P2 towards P3. We the Pyrenees between 22 ka and 18 ka (González-Sampériz et al.,
propose that, during the depressurization of P2, some sectors of the 2006), which caused a drop in the level of the Isona karstic lakes
caves systems may suffer also an increase of the effective weight of (Pellicer et al., 2016). The other three events took place during the

Please cite this article as: Linares, R., et al., The impact of droughts and climate change on sinkhole occurrence. A case study from the evaporite
karst of the Fluvia Valley, NE S..., Sci Total Environ (2016), http://dx.doi.org/10.1016/j.scitotenv.2016.11.091
12 R. Linares et al. / Science of the Total Environment xxx (2016) xxx–xxx

Fig. 8. Airborne images from 1988 and 2010 illustrating the geomorphic effects of the last drought period (1998–present) in the vicinity of Can Cademont within the study area. The lower
image shows sinkholes 14 and 18, formed in November 1998 and in 2005–2006.

most arid phase recorded in northern Spain in the last glacial cycle, sit- of the drought periods associated with the climate change (Cook et al.,
uated between 40 and 43 ka and 57 ka, in the transition between MIS 3 2016) may result in an intensification of the sinkhole hazard in karst
and MIS 4 (Moreno et al., 2012). Moreover, Morellón et al. (2014) recog- areas. Within this situation in which resilience and adaptation to the
nise 17 homogenite layers in the recent sedimentary record of Banyoles new adverse conditions is of prime importance, it is highly advisable
Lake (last 7.6 ka) ascribed to fluidization processes and ascending to identify the most susceptible areas, analyse the potential impact of
turbiditic flows from the deeper sinkholes induced by increased climate change on sinkhole hazard, and design mitigation measure to
groundwater discharge (Serra et al., 2002, 2005; Soler et al., 2009; reduce the associated risks. The results of this work suggest that karst
Morellón et al., 2014). These data suggest that dry conditions enhance areas underlain by highly soluble evaporites that function as confined
collapse activity in the bottom of Banyoles Lake, causing the aquifers are particularly sensitive to changes in the precipitation pat-
oversteepening of the adjacent slopes and the development of subaque- tern. It would be desirable to conduct similar investigations in other
ous mass movements, whereas wet, high-recharge periods tend to be areas in order to assess whether this finding can be generalized.
recorded by homogenites. The observed correlation between drought periods and temporal clus-
ters of sinkholes should be also considered in the paleoenvironmental in-
7. Conclusions terpretation of gravitational events (e.g. subaqueous landslides,
sinkholes) recorded in the deposits of karstic lakes associated with collapse
This work, carried out in a Mediterranean area with a confined evap- sinkholes. Those events are commonly attributed to periods of higher
orite aquifer and hypogene karstification, reveals that sinkhole activity groundwater discharge and more humid conditions (e.g. Morellón et al.,
is intensified during drought periods, which are expected to increase 2014), but could be also related to drought periods (e.g., Pellicer et al.,
in the future according to the climatic projections. This is an innovative 2016).
and unexplored geomorphic effect with potential detrimental conse-
quences of the current anthropogenic climate change. This factor should Acknowledgements
be taken into account in sinkhole hazard assessments that generally as-
sume that the spatial-temporal frequency of sinkholes in the future will We would like to thank the three anonymous reviewers and the ed-
be steady and similar to that documented in the past. itor for their very constructive comments and suggestions on the earlier
In the context of the Mediterranean realm where the study area is lo- version of this manuscript. This work has been funded by project
cated, the predicted progressive increase in the duration and intensity CGL2013-40867-P (Ministerio de Economía y Competitividad, Spain).

Please cite this article as: Linares, R., et al., The impact of droughts and climate change on sinkhole occurrence. A case study from the evaporite
karst of the Fluvia Valley, NE S..., Sci Total Environ (2016), http://dx.doi.org/10.1016/j.scitotenv.2016.11.091
 R. Linares et al. / Science of the Total Environment xxx (2016) xxx–xxx 13

Fig. 9. (A) Piezometric time series of the period 2008–2013 recorded by the multiple piezometer Fontcobeta-1, -2, and -3, and precipitation data from the Alta Garrotxa meteorological
station. (B) Conceptual model showing the hydrogeological relationships described in the text, inspired by the diagrams proposed by Klimchouk (2000, 2009) for explaining
hypogenic development of caves. The figure illustrates the variation of the groundwater flow pattern related to the alternation between drought-to-wet (I) and wet-to-drought (II) stages.

References Barberà, M., 1998. Dictamen sobre l'esfondrament que s'ha produït als Camps de Can
Cadamont a Maià de Montcal (Garrotxa). Informe: AR-099-98. Servei Geològic de Ca-
Abellán, J.A., 2010. Històries, llegendes i curiositats del estanyols i brolladors del Pla de talunya. Institut Cartogràfic de Catalunya, Generalitat de Catalunya (6 pp.).
l'Estany. Revista de Banyoles 918, 26–33. Barrera-Escoda, A., 2008. Evolución de los extremos hídricos en Catalunya en los últimos
ACA, 2006. Masses d'Aigua Subterrània de Catalunya, Banyoles (08). Fitxa de 500 años y su modelización regional. (PhD thesis). Department of Astronomy and
Caracterització, Anàlisi de Pressions, Impactes i Anàlisi del Risc d'Incompliment. Meteorology, University of Barcelona, Barcelona (319 pp.).
Generalitat de Catalunya, Barcelona (35 pp.). Barrera-Escoda, A., Gonçalves, M., Guerreiro, D., Cunillera, J., Baldasano, J.M., 2014. Projec-
ACA, 2007a. La situació actual d'excepcionalitat en el context històric. http://aca-web. tions of temperature and precipitation extremes in the North Western Mediterra-
gencat.cat/aca/documents/ca/sequera/introduccio.pdf. nean Basin by dynamical downscaling of climate scenarios at high resolution
ACA, 2007b. Les sequeres a Catalunya. Principals episodis fins l'any 2003. http://aca-web. (1971–2050). Climate Change 122, 567–582.
gencat.cat/aca/documents/ca/planificacio/sequera/historic_sequera_fins2003.pdf. Bischoff, J.L., Julià, R., Shanks, W.C., Rosenbauer, R.J., 1994. Karstification without carbonic
ACA, 2014. Plan de gestión del distrito de cuenca fluvial de Cataluña 2016–2021. Anexo III. acid: bedrock dissolution by gypsum-driven dedolomitization. Geology 22, 995–998.
Inventario de recursos hídricos totales. Generalitat de Catalunya, Barcelona (60 pp.). Brusi, D., Pallí, L., Sanz, M., 1987. Caracterización de las inestabilidades gravitatorias en el
Acero, P., Auqué, L.F., Galve, J.P., Gutiérrez, F., Carbonel, D., Gimeno, M.J., Yechieli, Y., Asta, área de Banyoles. III Reunión Nacional de Geología Ambiental y Ordenación del
M.P., Gómez, J.B., 2015. Evaluation of geochemical and hydrogeological processes by Territorio, Valencia. 2 pp. 1021–1034.
geochemical modeling in an area affected by evaporite karstification. J. Hydrol. 529, Brusi, D., Bach, J., Sanz, M., 1990. Itinerari geològic de Banyoles. Eumo Editorial, Barcelona
1874–1889. (124 pp.).
Altava-Ortiz, V., Barrera-Escoda, A., Llasat, M.C., 2008. Monitoratge, classificació I Brusi, D., Maroto, J., Vila, X., 1992. L'estany de Banyoles. In: Pallí, L., Brusi, D. (Eds.), El medi
reconstrucció de Registres de Sequera a Catalunya. Primeres Jornades de natural a las terres gironines. Universitat de Girona, pp. 117–133.
Meteorologia i Climatologia de la Mediterrània Occidental, Barcelona. Calbó, J., Sánchez-Lorenzo, A., Cunillera, J., Barrera-Escoda, A., 2010. Projeccions i Escenaris
Aurit, M.D., Peterson, R.O., Blanford, J.I., 2013. A GIS analysis of the relationship between futurs. In: Llebot, J.E. (Ed.), 2n Informe sobre el Canvi Climatic a Catalunya. Grup d'Ex-
sinkholes, dry-well complaints and groundwater pumping for frost-freeze protection perts en Canvi Climatic de Catalunya. Generalitat de Catalunya i Institut d'Estudis Cat-
of winter strawberry production in Florida. PLoS One 8, 1–9. alans, Barcelona, pp. 183–239.
Bach, J., Brusi, D., Sanz, M., Trilla, J., 1986. Ejemplo de sistema kárstico “trop plein” (La Canals, M., Got, H., Julià, R., Serra, J., 1990. Solution collapse depressions and suspensates in
Garrotxa-Banyoles). Jornadas sobre el Karst en Euskadi Tomo I, 137–151. the limnocrenic lake of Banyoles (NE Spain). Earth Surf. Process. Landf. 15, 243–254.
Bach, J., Bach, A., Zarroca, M., 2012. Groundwater Monitoring in La Garrotxa Volcanic Zone Carrillo, E., Rosell, L., Ortí, F., 2014. Multiepisodic evaporite sedimentation as an indicator
Natural Park: Continuous Recording Since 1994. Volcandpark. 1st International Con- of paleogeographical evolution in foreland basins (South-eastern Pyrenean basin,
gress on Management and Awareness in Protected Volcanic Landscapes, Olot. early-Middle Eocene). Sedimentology 61, 2086–2112.

Please cite this article as: Linares, R., et al., The impact of droughts and climate change on sinkhole occurrence. A case study from the evaporite
karst of the Fluvia Valley, NE S..., Sci Total Environ (2016), http://dx.doi.org/10.1016/j.scitotenv.2016.11.091
14 R. Linares et al. / Science of the Total Environment xxx (2016) xxx–xxx

Clavell, E., Martínez, A., Vergés, J., 1988. Morfologia del basament del Pirineu oriental: McKee, T.B., Doesken, N.J., Kleist, J., 1993. The Relationship of Drought Frequency and Du-
evolució i relació amb els mantells de corriments. Acta Geol. Hisp. 23, 129–140. ration to Time Scale. Proceedings of the Eighth Conference on Applied Climatology,
Colomer, J., Casamitjana, X., Serra, T., Roget, E., 2006. Physical limnology in Lake Banyoles. Anaheim, California. American Meteorological Society, Boston, pp. 179–184.
Limnetica 25, 181–188. Miret, F., García, M., Alfambra, F., Borruel, X., Cardona, F., Mas, J., Porcel, E., 1999. Coves de
Cook, B.I., Anchukaitis, K.J., Touchan, R., Meko, D.M., Cook, E.R., 2016. Spatiotemporal drought l'Alta Garrotxa. Grup d'Espeleologia de Badalona, Barcelona (108 pp.).
variability in the Mediterranean over the last 900 years. J. Geophys. Res. 121, Morellón, M., Valero-Garcés, B., Anselmetti, F., Ariztegui, D., Schnellmann, M., Moreno, A.,
2060–2074. Mata, P., Rico, M., Corella, J.P., 2009. Late Quaternary deposition and facies model for
Corella, J.P., Moreno, A., Morellón, M., Rull, V., Giralt, S., Rico, M.T., Valero-Garcés, B.L., karstic Lake Estanya (North-eastern Spain). Sedimentology 56, 1505–1534.
2011. Climate and human impact on a meromictic lake during the last 6,000 years Morellón, M., Anselmetti, F.S., Valero-Garcés, B., Giralt, S., Ariztegui, D., Sáez, A., Mata, M.P.,
(Montcortès Lake, Central Pyrenees, Spain). J. Paleolimnol. 46, 351–367. Barreiro-Lostres, F., Rico, M., Moreno, A., 2014. The influence of subaquatic springs in
Corella, J.P., Stefanova, V., El Anjoumi, A., Rico, E., Giralt, S., Moreno, A., Valero-Garcés, B.L., lacustrine sedimentation: origin and paleoenvironmental significance of
2013. A 2500-year multi-proxy reconstruction of climate change and human activi- homogenites in karstic Lake Banyoles (NE Spain). Sediment. Geol. 311, 96–111.
ties in northern Spain: the Lake Arreo record. Palaeogeogr. Palaeoclimatol. Palaeoecol. Moreno, A., González-Sampériz, P., Morellón, M., Valero-Garcés, B.L., Fletcher, W.J., 2012.
386, 555–568. Northern Iberian abrupt climate change dynamics during the last glacial cycle: a view
Dogan, U., Yilmaz, M., 2011. Natural and induced sinkholes of the Obruk Plateau and from lacustrine sediments. Quat. Sci. Rev. 36, 139–153.
Karapinar-Hotamis Plain, Turkey. Asian Earth Sciences]–>J. Asian Earth Sci. 40, 496–508. Newton, J.G., 1984a. Natural and induced sinkhole development in the Eastern United
Festa, V., Fiore, A., Parise, M., Siniscalchi, A., 2012. Sinkhole evolution in the Apulian karst States. In: Johnson, A.I., Carbognin, L., Ubertini, L. (Eds.), Land Subsidence 200. Inter-
of southern Italy: a case study, with some considerations on sinkhole hazards. Journal national Association of Hydrological Sciences, Publ., pp. 323–331.
of Cave and Karst Studies 74, 137–147. Newton, J.G., 1984b. Sinkholes resulting from ground-water withdrawals in carbonate
Galobart, A., Agustí, J., 2003. Los micromamíferos de los yacimientos del Pleistoceno infe- terranes—an overview. America Reviews in Engineering Geology]–>Geol. Soc. Am.
rior de Incarcal (Girona, NE de la Península Ibérica). Paleontologia i Evolució 34, Rev. Eng. Geol. 6, 195–202.
63–67. Palmer, A.N., 2007. Cave Geology. Allen Press, Lawrence, Kansas (454 pp.).
Gonçalves, M., Barrera-Escoda, A., Guerreiro, D., Baldasano, J.M., Cunillera, J., 2014. Season- Parise, M., Ravbar, N., Živanovic, V., Mikszewski, A., Kresic, N., Mádl-Szonyi, J., Kukuric, N.,
al to yearly assessment of temperature and precipitation trends in the North Western 2015. Hazards in karst and managing water resources quality. (Chapter 17). In:
Mediterranean Basin by dynamical dowscaling of climate scenarios at high resolution Stevanovic, Z. (Ed.), Karst Aquifers - Characterization and Engineering. Professional
(1971–2050). Climate Change 122, 243–256. Practice in Earth Sciences. Springer, pp. 601–687.
González-Sampériz, P., Valero-Garcés, B.L., Moreno, A., Jalut, G., García-Ruiz, J.M., Martí- Pellicer, X.M., Corella, J.P., Gutiérrez, F., Roqué, C., Linares, R., Carbonel, D., Zarroca, M.,
Bono, C., Delgado-Huertas, A., Navas, A., Otto, T., Deboubat, J.J., 2006. Climate variabil- Guerrero, J., Comas, X., 2016. Sedimentological and palaeohydrological characteriza-
ity in the Spanish Pyrenees during the last 30,000 yr revealed by the El Portalet se- tion of Late Pleistocene and Holocene tufa mound palaeolakes using trenching
quence. Quat. Res. 66, 38–52. methods in the Spanish Pyrenees. Sedimentology 63, 1786–1819.
Gutiérrez, F., 2016. Sinkhole hazards. In: Cutter, S.L. (Ed.), Oxford Research Encyclopedia Puigdefàbregas, C., Muñoz, J.A., Marzo, M., 1986. Thrust belt development in the Eastern
of Natural Hazard Science. Oxford University Press, pp. 1–92. Pyrenees and related depositional sequences in the southern foreland basin. In:
Gutiérrez, F., Cooper, A.H., Johnson, K.S., 2008a. Identification, prediction and mitigation of Allen, P.A., Homewood, P. (Eds.), Foreland Basins. International Association of Sedi-
sinkhole hazards in evaporite karst areas. Environ. Geol. 53, 1007–1022. mentologists Special Publication Vol. 8, pp. 229–246.
Gutiérrez, F., Guerrero, J., Lucha, P., 2008b. A genetic classification of sinkholes illustrated Ros-Montoya, S., Madurell-Malapeira, J., Martínez-Navarro, B., Espigares, M.P., 2010. Late
from evaporite paleokarst exposures in Spain. Environ. Geol. 53, 993–1006. Villafranchian Mammuthus meridionalis (Nesti, 1825) from the Iberian Peninsula:
Gutiérrez, F., Galve, J.P., Lucha, P., Castañeda, C., Bonachea, J., Guerrero, J., 2011. Integrating Dentognathic remains from Incarcal (Crespià, Girona) and comparisons with the
geomorphological mapping, trenching, InSAR and GPR for the identification and char- Guadix-Baza sample. Quaternaire Hors-série 3, 47–48.
acterization of sinkholes: a review and application in the mantled evaporate karst of Sanz, M., 1981. El sistema hidrogeológico de Banyoles-La Garrotxa. Unpublished PhD the-
the Ebro Valley (NE Spain). Geomorphology 134, 144–156. sis, Universitat Autònoma de Barcelona, Barcelona (306 pp.).
Gutiérrez, F., Linares, R., Roqué, C., Zarroca, M., Rosell, J., Galve, J.P., Carbonel, D., 2012. In- Sanz, M., Trilla, J., 1982. Consideración sobre la dinámica del sistema cárstico de la
vestigating gravitational grabens related to lateral spreading and evaporite dissolu- Garrotxa. III simposio de Hidrogeología. pp. 475–483.
tion subsidence by means of detailed mapping, trenching, and electrical resistivity Sanz, M., Bach, J., Trilla, J., 1982. Aplicación del oxígeno-18 al sistema kárstico de
tomography (Spanish Pyrenees). Lithosphere 4, 331–353. Banyoles-La Garrotxa (Gerona). Reunión monográfica sobre karst de Larra.
Gutiérrez, F., Parise, M., DeWaele, J., Jourde, H., 2014. A review on natural and human- pp. 243–255.
induced geohazards and impacts in karst. Earth-Sci. Rev. 138, 61–88. Saula, E., Picart, J., Mató, E., Llenas, M., Losantos, M., Berástegui, X., Agustí, J., 1994.
Gutiérrez, F., Fabregat, I., Roqué, C., Carbonel, D., Guerrero, J., García-Hermoso, F., Zarroca, Evolución geodinámica de la fosa del Empordà y las Sierra Transversales. Acta Geol.
M., Linares, R., 2016. Sinkholes and caves related to evaporite dissolution in a Hisp. 29, 55–75.
stratigraphically and structurally complex setting, Fluvia Valley, eastern Spanish Pyr- Serra, T., Colomer, J., Gacia, E., Soler, M., Casamitjana, X., 2002. Effects of a turbid hydro-
enees. Geological, geomorphological and environmental implications. Geomorpholo- thermal plume on the sedimentation rates in a karstic lake. Geophys. Res. Lett. 29,
gy 267, 76–97. 2029.
Höbig, N., Weber, M.E., Kehl, M., Weniger, G.C., Julià, R., Melles, M., Fülöp, R.H., Vogel, H., Serra, T., Soler, M., Julià, R., Casamitjana, X., Colomer, J., 2005. Behaviour and dynamics of a
Reicherter, K., 2012. Lake Banyoles (northeastern Spain): a Last Glacial to Holocene hydrothermal plume in Lake Banyoles, Catalonia, NE Spain. Sedimentology 52,
multi-proxy study with regard to environmental variability and human occupation. 795–808.
Quat. Int. 274, 205–218. Soler, M., Serra, T., Casamitjana, X., Colomer, J., 2009. High sedimentation rates in a karstic
IGME, 1986. Proyecto de investigación hidrogeológica en los sistemas acuíferos de las lake associated with hydrothermal turbid plumes (Lake Banyoles, Catalonia, NE
cuencas de los ríos Ter, Fluvià y Muga, para su integración en la planificación Spain). Sediment. Geol. 222, 5–15.
hidrogeológica de la cuenca del Pirineo Oriental (85–86). Estudio de las cuencas Sprynskyy, M., Lebedynets, M., Sadurski, A., 2009. Gypsum karst intensification as a con-
hidrogeológicas de las cuencas altas de los ríos Ter, Fluvià y Muga. sequence of Sulphur mining activity (Jaziv field, Western Ukraine). Environ. Geol. 57,
Iovine, G., Vennari, C., Gariano, S.L., Caloiero, T., Lanza, G., Nicolino, N., Suriano, S., Ferraro, 173–181.
G., Parise, M., 2016. The “Piano dell'Acqua” sinkholes (San Basile, Northern Calabria, Taheri, K., Gutiérrez, F., Mohseni, H., Raeisi, E., Taheri, M., 2015. Sinkhole susceptibility
Italy). Bull. Eng. Geol. Environ. 75, 37–52. mapping using the analytical hierarchy process (AHP) and magnitude–frequency re-
Jeannina, P.Y., Hessenauer, M., Malard, A., Chapuis, V., 2016. Impact of global change on lationships: a case study in Hamadan province, Iran. Geomorphology 234, 64–79.
karst groundwater mineralization in the Jura Mountains. Sci. Total Environ. 541, Valero-Garcés, B., Morellón, M., Moreno, A., Corella, P., Martín-Puertas, C., Barreiro, F.,
1208–1221. Pérez, A., Giralt, S., Mata-Campo, M.P., 2014. The carbonate factory in karst lakes:
Julià, R., 1980. La conca lacustre de Banyoles-Besalú. Monografies del Centre d'Estudis sources, processes and depositional environments in Quaternary Iberian Lakes. Sedi-
Comarcals de Banyoles (188 pp.). ment. Geol. 299, 1–29.
Klimchouk, A.B., 2000. Speleogenesis under deep-seated and confined settings. In: Vidal-Pardal, M., 1957. La alimentación subterránea del lago de Bañolas. Resultados de los
Klimchouk, A.B., Ford, D.C., Palmer, A.N., Dreybrodt, W. (Eds.), Speleogenesis: Evolu- ensayos con fluoresceína. Solución al problema de la regulación del río Fluvia. Rev.
tion of Karst Aquifers. National Speleological Society, Huntsville, Alabama, Obras Públicas 2903, 105–107.
pp. 244–260. Waltham, T., Bell, F., Culshaw, M., 2008. Sinkholes and Subsidence. Springer, Chichester
Klimchouk, A.B., 2009. Morphogenesis of hypogenic caves. Geomorphology 106, 100–117. (382 pp.).
Lamont-Black, J., Baker, A., Younger, P.L., Cooper, A.H., 2005. Utilising seasonal variations Yechieli, Y., Abelson, M., Bein, A., Crouvi, O., Shtivelman, V., 2006. Sinkholes “swarms”
in hydrogeochemistry and excitation–emission fluorescence to develop a conceptual along the Dead Sea coast: reflection of disturbance of lake and adjacent groundwater
groundwater flow model with implications for subsidence hazards: an example from systems. GSA Bull. 118, 1075–1087.
Co., Durham, UK. Environ. Geol. 48, 320–335. Youssef, A.H., Al-Harbi, H.M., Gutiérrez, F., Zabramwi, Y.A., Bulkhi, A.B., Zahrani, S.A.,
Lamoreaux, P.E., Newton, J.G., 1986. Catastrophic subsidence: an environmental hazard, Bahamil, A.M., Zahrani, A.J., Otaibi, Z.A., El-Haddad, B.A., 2016. Natural and human-
Shelby County, Alabama. Environ. Geol. Water Sci. 8, 25–40. induced sinkhole hazards in Saudi Arabia: distribution, investigation, causes, and im-
Li, G., Zhou, W., 1999. Sinkholes in karst mining areas in China and some methods of pre- pacts. Hydrogeol. J. 24, 625–644.
vention. Eng. Geol. 52, 45–50. Zarroca, M., Linares, R., Roqué, C., Fabregat, I., Gutiérrez, F., Carbonel, D., Guerrero, J., 2016.
Linares, R., Rosell, J., Roqué, C., Gutiérrez, F., 2010. Origin and evolution of tufa mounds re- Aplicación de la tomografía eléctrica de resistividades (ERT) a la investigación de
lated to artesian karstic springs in Isona area (Pyrenees, NE Spain). Geodin. Acta 23, dolinas en el karst evaporítico del Valle del Fluvià (NE de España). In: Durán, J.J.,
129–150. Montes, M., Robador, A., Salazar, A. (Eds.), Comprendiendo el relieve: Del pasado al
Martínez, A., Samsó, J.M., Zamorano, M., Picart, J., Solà, J., Montaner, J., Mató, E., 2000. futuro. Instituto Geológico y Minero de España, Madrid, pp. 695–702.
Besalú (76–22). Mapa Geològic de Catalunya 1:25.000. Institut Cartogràfic de
Catalunya.
Mató, E., Saula, E., Picart, J., Solà, J., Montaner, J., Viñals, E., Samsó, J.M., Serra, J., Llenas, M.,
Agustí, J., Mallarach, J., 1996. Banyoles (76–23). Mapa Geològic de Catalunya 1:
25.000. Institut Cartogràfic de Catalunya.

Please cite this article as: Linares, R., et al., The impact of droughts and climate change on sinkhole occurrence. A case study from the evaporite
karst of the Fluvia Valley, NE S..., Sci Total Environ (2016), http://dx.doi.org/10.1016/j.scitotenv.2016.11.091