1The karstic mountain landscapes of Crete have been the scene of human activity as early as the Mid-Holocene. Numerous dolines and poljes allowed widespread agriculture in pre-Christian times (Chaniotis, 2004) due to their flat topography and their capacity for accumulating large amounts of sediments. A lot of archaeological relics can nowadays be found inside of these depressions as well as in their vicinity. They help gain insights into the economic exploitation and the cultural evolution of corresponding regions, which is why karst landforms are of high interest for geoarchaeological research. Among others, L. Bruxelles et al. (2006) and A. Vött et al. (2009) demonstrated the great value of such terrestrial archives for reconstructing the environmental evolution in South France and Northern Greece. However, until recently, palaeoenvironmental research on Crete was mainly conducted in the lowlands on the basis of profound local records like fluvial or marine sediments. Several studies have been carried out in coastal sites, such as Malia (Dalongeville, 2001), Istron and Lato pros Kamara (modern city of Agios Nikolaos; Theodorakopoulou et al., 2008), proving continuous sea-land interactions of local civilisations from the onset of the Holocene. Above all, the geoarchaeological survey in the alluvial plain of Istron (gulf of Mirabellum), an area with continuous human activities from the Neolithic to the Byzantine period (Hayden, 2004), gave evidence for the evolution of the landscape during the Holocene. By correlating the results from pollen analyses and micropalaeontological studies on lagoonal sediments with the fluctuations of sedimentation rate, K. Theodorakopoulou et al. (2009) documented the palaeoenvironmental evolution of the broader area of Mirabello. In contrast to the lowlands, the karstified mountains have been neglected for a long time mainly because of the unknown suitability of dolines as terrestrial sediment archives. Only a few karst geomorphological studies have been carried out (e.g., Poser, 1976; Fabre and Maire, 1983; Bartels, 1991; Egli, 1993), which, however, largely focused on the development of landforms and their geoecologic setting. In terms of subsurface geometry and infilling, C. Siart et al. (2009a) were the first to investigate enclosed depressions in Central and Eastern Crete, while considering their geoarchaeological implications. The study at hand aims at continuing and advancing this work by applying new geoarchaeological techniques to karst landforms in the Dikti Mountains of Eastern Crete.
2The area of investigation, a complex of two deeply incised dolines at 278 m a.s.l., is located between the villages of Agios Nikolaos and Kritsa (35°10’21”N, 25°40’18”E; fig. 1). The geological setting is comprised of Mesozoic limestones and Neogene breccias, which are highly susceptible to chemical dissolution. Additional outcrops of ophiolites lead to a very heterogeneous distribution pattern of different petrographical units (Institute of Geology and Mineral Exploration, 1983). Due to the prevalence of carbonates, numerous karstic landforms can be identified, with sinkholes, uvalas and poljes being the most prominent features. In the northwestern part of the study area, archaeologists excavated the ancient settlement of Latô, which was occupied from Mycenian to Hellenistic times and experienced its cultural bloom under the Dorian rule (Evans, 1894, 1895-1896; Demargne, 1901). Since Latô pros Kamara (Modern Agios Nikolaos), the harbour of the Dorian city State, was prospering by the middle of the 2nd c. AD, it became the administrative centre, while the primary mountain settlement of Latô was abandoned (Blowsky, 1989). While the region was accessed and intensely exploited by humans during the first millennium BC, proof for subsequent occupation until medieval times is still lacking. Recent geomorphological studies were conducted by M. Ghilardi (2006) and M. Ghilardi and S. Kunesch (2008), who focused on a thorough description and mapping of karst landforms in the wider area. Similar to findings from other dolines on Crete (e.g., Siart et al., 2010), the Kritsa doline complex is of high geoarchaeological interest. Both the surface findings on the doline floor (e.g., pottery shards from Minoan to Modern times) and the subsurface structures, such as partly buried wall remains that were discovered during field surveying, provide evidence for ancient land use in and around the enclosed depressions. The use of local resources may have included, among others, the extraction of loamy limestone residuals for pottery, as suggested by the discovery of several ceramic ovens in Latô (Ducrey and Picard, 1996; Wurmser, 2008). Despite the fact that the Dikti Mountains represented an important part of the human ecoumene during parts of the Mid and Late Holocene, the exact relationships between man and his environment as well as further ways of economic exploitation and cultural valorisation of the region are fairly vague. Thus, the aim of the paper is (i) to contribute to a more detailed knowledge of the landscape evolution along with the corresponding landforms and (ii) to provide new insights into the human impact on the natural environment. Seen from a geomorphological perspective, the topics to be investigated relate to the subterranean topography, the amount of sediment fill and the potential of karst depressions for reconstructing palaeoenvironmental changes. A multi-method approach is carried out for defining the morpho-structural characteristics and analysing the geometrical, geophysical and geoarchaeologically relevant characteristics of loose substrates accumulated in the Kritsa doline complex. Ultimately, the potentials and benefits of combining new high precision techniques for investigations at the human-environmental interface are to be demonstrated.
Fig. 1 – Geological map of the study region (data source: IGME, 1981; Ghilardi, 2006).
Fig. 1 – Carte géologique du secteur d’études (source : IGME, 1981; Ghilardi, 2006).
The investigated doline complex (dashed rectangle) is located in a carbonated breccia and bordered by a northwest to southeast trending fault. 1: alluvium (Quaternary); 2: carbonated breccias; 3: flyschs (Oligocene); 4: limestone (Cretaceous); 5: dolomitic limestone (Jurassic); 6: ophiolites; 7: fault line; 8: karst depressions. The black star indicates the exact location of the archaeological site of Lato, black circles are modern cities or villages, and the dashed rectangle highlights the area of investigation (fig. 2A).
Les dolines étudiées (rectangle en pointillés) sont développées dans de la brèche à sédiment carbonaté, et sont généralement bordées par des failles de direction principale NW-SE. 1 : sédiments d’âge Quaternaire (remplissage de dépressions karstiques) ; 2 : brèche à ciment carbonaté (Pliocène) ; 3 : flyschs (Oligocène) ; 4 : calcaires (Crétacé) ; 5 : calcaires dolomitiques (Jurassique) ; 6 : ophiolites ; 7 : faille ; 8 : dépression karstique. L’étoile noire indique la localisation exacte du site archéologique de Latô, les cercles noirs représentent les localités (ville, village) et le rectangle en traits discontinus délimite le secteur d’étude (fig. 2A).
3In order to develop a holistic image and to better understand the local human-environmental systems, comprehensive data on both the subsurface and the surface structure of dolines are necessary. The former can be acquired by geophysical prospecting, e.g. electrical resistivity tomography (ERT), which has become increasingly popular for investigating karst features like sinkholes and cavities because of its reliable, easy-to-use and non-destructive properties (Ahmed and Carpenter, 2003; Gibson et al., 2004). Besides yielding valuable information on the development of the studied landform, this technique also allows detecting buried remains and deducing former land use practices by providing an insight into the underground and the thickness of sediment deposits (Gaffney, 2008; Hecht, 2009). Hence, the fieldwork was based on a multi-method approach, starting with geophysical studies on a mesoscale level. Electrical resistivity tomographies were measured in two-dimensional transects in order to capture the subsurface geometry of the dolines and to assess the depth to bedrock. A 100-electrode system (Geotom) was used, with currents ranging from 0.5-5 mA and different profile lengths of 25-200 m. Good ground contacts of electrodes (vertical metal rods), spaced at distances of 2 m (profile E1), 1.2 m (E2) and 0.5 m (E3), could be established due to the high penetrability of the pedo-sedimentary karst fills. Several transects were aligned rectangularly to produce a three-dimensional image of the entire underground structure. As documented by S. Hecht (2007), Schlumberger arrays with their high lateral resolution were used for the horizontal differentiation of loose sediments, while dipole-dipole arrays allowed specifying vertical subterraneous features, such as archaeological remnants. In the case of subsurface anomalies, pointing to specific areas of geoarchaeological interest, three-dimensional electrode configurations were applied (E4, pole-pole array, grid: 10x10 m; spacing: 1x1 m). Data processing and evaluation was carried out with RES2DINV and RES3DINV software packages using standard inversions without filtering. Calibration of resistivity values was conducted by comparison with ERT results from other karst geomorphological studies (e.g., Siart, 2010). Yet, geophysical prospecting in karstic areas can be very complicated due to considerable subsurface heterogeneities and a high level of geophysical noise (van Schoor, 2002; Terzic et al., 2007). As an integrated approach based on complementary techniques was indispensible to avoid ambiguous results and misinterpretation, ERT outcomes were gauged with percussion drilling results. Two sediment cores (liner tubes, 5 cm diameter) of a maximum depth of 5 m b.s. were obtained and analysed with regard to stratigraphy and granulometrical composition (Ghilardi and Kunesch, 2008; Siart et al., 2009a; for core location, see fig. 2a).
4In contrast to the underground, geomorphometrical information about the surface topography is mostly extracted from digital elevation models derived by radar and satellite imagery (Bubenzer and Bolten, 2008; Siart et al., 2009b). In fact, these data sets are often free of charge and ready to use, but their poor spatial resolution is insufficient for analysing meso- to microscale features like small terraces and gullies. Even though total station measurements and DGPS-tracking can be used alternatively to enhance the level of detail, both techniques require enormous efforts and bear the additional risk of being operator biased (selective sampling in the field; Armesto et al., 2009). Micro-relief objects could therefore quite easily be disregarded. For this reason, terrestrial laser scanning (TLS) was applied in the study region of Kritsa so as to both capture the doline complex in its entirety and to provide highly accurate topographical data on zones of special geoarchaeological interest. Data acquisition was carried out using a V-Line 3D time-of-flight scanner (Riegl VZ-400) with a narrow infrared laser beam and a fast scanning mechanism. A Nikon D 300(s) digital camera with a 10-mm fish-eye focus lense was mounted on top in order to provide high-resolution colour images. In total, 8 different scan positions were collected for maximum coverage of the study site (fig. 2). In analogy to S. Buckley et al. (2008), a best possible clear line of sight was ensured. The scanning process was based on either medium resolution for capturing the full extent of the karst depression (radius: 600 m; acquisition-rate: 40000 measurements/s) or high resolution in the case of geoarchaeological on-site studies (radius: 350 m; acquisition-rate: 122000 measurements/s). Shading effects caused by shrubs and trees were considerably reduced through full waveform analysis, which allowed separating several echoes from the emitted laser beam. Postprocessing included coarse registration or reflector-based matching of scans, as well as fine processing by means of multistation adjustment [iterative closest point algorithm (ICP)]. Among all scans an average point offset of only a few centimetres was achieved. As the geomorphological interpretation required unambiguous data sets, erroneous points caused by water surfaces were erased manually, while redundant details such as vegetation were deleted in a multistep approach using the RiSCAN Pro software: first, a coarsely triangulated elevation model was generated and compared with the original point cloud; second, all points exceeding a crucial threshold value in terms of distance between the point cloud and the triangulated DEM were eliminated. These steps were applied iteratively until satisfactory quality criteria were met. Third, the final DEMs (pixel sizes: 0.25 m and 2 m) were derived by meshing the improved point clouds. Subsequent to computing the volume and the contour lines of the enclosed depression, data was exported by converting the DEM into CAD-format. Additional morphometrical analyses were carried out in ArcGIS 9.3 (slope, aspect, hydrologic surface analyses). In order to reveal both the local geomorphodynamic processes and the characteristics of former land-use practices, the ERT data were georeferenced and superimposed on the DEM to identify the exact position of subterraneous findings and compare them with micro-relief features on the surface (e.g., the bottom of the doline).
Fig. 2 – Kritsa-Latô doline.
Fig. 2 – Doline de Kritsa-Latô.
A: The area of investigation as seen from above with orientation and direction of geophysical transects (white arrows) as well as laser scanning positions (dots) and drilling locations (x; background image by Google Earth 2010). B: In the embankments of two ponds, well arranged breccias limestone boulders are visible. C: At the southern margin of the doline, small wall remnants point to a tapped spring that recently dried out. For location of fig. 2B and fig. 2C, see black arrows in fig. 2A.
A : Cliché du secteur d’étude vu d’un point haut avec informations liées à l’orientation et la direction des transects géophysiques (symbolisés avec des flèches), ainsi que les endroits utilisés pour le scan laser (points) et les carottages (x ; image Google Earth, 2010 en arrière plan). B : Sur les bords de deux marres creusées pour des besoins d’approvisionnement en eau des troupeaux de caprins, des blocs bien taillés de calcaire bréchique sont visibles. C : Sur le rebord méridional de la doline, des murs de petites dimensions (vestiges archéologiques) sont reliés à une source tarie de nos jours. Pour localiser la fig. 2A et la fig. 2B, voir les flèches noires sur la fig. 2A.
5ERT profile E1, a W-E oriented cross-section covering the complete diameter of the doline, was measured on a total length of 200 m (Schlumberger configuration, spacing 2 m). Due to the large array, a high depth penetration could be achieved (33.5 m b.s.). As the tomography (fig. 3) displays large differences in the underground, the karst depression can be subdivided into several morpho-sedimentological zones. Its general three-part structure is comprised of mainly low resistivity values (R<25 Ωm) near the surface, which correspond to fine-grained sediments (e.g., residuals, soils). This result can be confirmed by the sediment cores, which almost exclusively consist of clayey to silty deposits without any macroscopic stratification. While gravels are totally absent, only small amounts of sand were identified over the total depth of 5 m (Siart et al., 2009a). In contrast, the deeper levels show a gradual increase of resistivity (~50-200 Ωm), caused by changing properties of the infilling. Similar to the findings by M. Roth et al. (2002), the existence of a transition zone made up of coarser detritus mixed with loose pedo-sediments must be assumed. The lowermost zone, which is indicated by high resistivity values (R>200 Ωm), corresponds to the basal limestone (Hecht, 2007). However, the depth to bedrock is subject to considerable variations, a typical phenomenon in buried karst terrains. While the solid rock crops out at the surface close to the doline margins, it dips towards the centre of the landform, resulting in the characteristic and concave subsurface morphology of sinkholes (Ford and Williams, 2007). In the middle of the transect, the loose overburden protrudes very deeply into the underground (buried sinkhole; Waltham et al., 2005), which is why the actual bedrock base could not be detected by ERT. Hence, a massive fill of at least 25 m of loamy sediment must have accumulated in the doline. As the interface between the intermediate shatter zone and the overlying sediments oscillates significantly, no sharp boundary can be detected. Superficial high resistivities at the edges of the doline are attributable to gravitational influx of coarse-grained material and must be considered as footslope deposits. In this context, a horizontal sorting of colluvial grain sizes can be observed with increasing transport distance (cf. Nemec and Kazanci, 1999). In total, profile E1 points to a very homogeneous subsurface setup with thick fills of loose sediments that lack any perturbation. The low RMS-error of 6.5% substantiates the reliability of all findings.
Fig. 3 – 2D ERT profiles crossing the doline of Kritsa-Latô.
Fig. 3 – Profils de Tomographie-Résistivité électrique (ERT) 2D effectués dans la doline de Kritsa-Latô.
The fig. 3A (west-east transect E1) displays the three-part segmentation of the subsurface as indicated by dashed lines. The loose sedimentary fill amounts to thicknesses of at least 25 m, while the epikarst zone resembles an inverted cone, showing a buried sinkhole (central part) bordered by potential pinnacles. Cross-section E2, which was measured perpendicularly to E1 (X: intersection point of profiles), supports this finding by highlighting the remarkably homogeneous setup of the fill that completely lacks stratification.
La fig. 3A représente le transect E1 (direction W-E) ainsi que les différentes unités (indiquées par des lignes en pointillés) du sous-sol de la doline de Kritsa-Latô. L’épaisseur des dépôts argilo-limoneux (partie supérieure) atteint 25 m tandis que l’épikarst (partie inférieure) ressemble à un cône inversé, bordé par de probables pinacles fossilisé par des sédiments de taille hétérométrique. Le transect E2, mesuré perpendiculairement à E1 (X marque l’intersection des deux profils), confirme les résultats de ce dernier et met en évidence le manque de stratification des matériaux argilo-limoneux.
6The ERT cross-section E2 (Schlumberger array) was measured perpendicularly to E1 to validate the geophysical results. It covers the entire bottom of the enclosed depression with a total length of 120 m. As for the underground structure, the results are in absolute accordance with E1, showing a massive accumulation of fine-grained sediments with almost no exception (R<25 Ωm; thickness >20 m; intersection with E1 at ~73 m). In the middle of the profile, the solid rock could not be identified because of insufficient depth penetration. Due to the short north-south extension of the doline, longer profiles were not possible (limestone outcrop at the footslopes). The bedrock dip towards the interior of the buried sinkhole can only be identified at the margins of the transect and therefore points to a relatively steep subsurface geometry. The two high resistivity areas at the lower edges of the tomography must be regarded as artificial anomalies. They were caused in the context of data acquisition or processing and do not correspond to real features. As in all the other slope-proximal parts, superficial detrital fans stretch towards the centre, causing resistivities of up to 250 Ωm. Even though ERT measurements are usually sensitive to sedimentological changes and provide insights into the alternation of different loose substrates (Hecht, 2009), almost no evident stratification of the doline fill can be observed.
7Besides the landform itself, a zone of special interest was identified in the central part of the doline, where small mounds of limestone boulders mixed with loose sediment were piled up. In immediate adjacency, the embankments of two water-filled ponds exposed blocks embedded within the doline fill, which seem to be distributed randomly in the upper parts but resemble well-constructed wall arrangements in the deeper levels. A high precision 2D profile (E3) was measured on a total length of 25 m (50 electrodes, 0.5 m spacing, penetration depth: 3.5 m) using a dipole-dipole configuration to reveal the subterraneous structure (fig. 4). The results are characterised by high resistivities at the surface, which always reach a salient and consistent maximum depth level of about 1 m b.s. Auger probing confirmed the existence of massive carbonate fragments. Underneath, the subsurface exhibits a very homogeneous composition, as indicated by extremely low and continuous ERT values (R<20 Ωm). While the deeper subsurface corresponds well with the results from E1 and E2, the near-surface zone is the very reverse of the other geophysical outcomes. In order to specify these findings, a 3D ERT grid (E4) was measured with a pole-pole configuration, overlapping with profile E3. The rectangle of 10x10 m, which is partly displayed horizontally and partly displayed vertically in fig. 4, helps place the 2D results into a spatial context (see also fig. 7). The uppermost level (0-0.7 m b.s.) is characterised by very heterogeneous resistivity values that can be ascribed to modern man-made perturbation of topsoil, e.g. by agricultural use or farming. While the depth levels between 1.5 m and 11.8 m equal those of E1, E2 and E3 (R<20 Ωm), the strata between 0.7 m and 1.5 m highlight the existence of two zones of higher resistivity surrounded by low values, which can be interpreted as a part of a superordinate linear structure. These anomalies are also verifiable in the vertical slices, where their distinct lower interface can be observed in 2 m b.s. Due to the good quality of the measurement (RMS-error: 7.5%), an artefact caused by data-immanent errors can be excluded.
Fig. 4 –2D ERT-profile E3 (above) and 3D ERT-grid E4 (below).
Fig. 4 – Le profil 2D ERT E3 (en haut) et a grille 3D ERT E4 (en bas).
The 2D ERT-profile E3 measured in the central part of the doline, reveals high resistivities near the surface that correspond to man-made remnants. The deeper subsurface displays very low values (fine-grained material) and appears quite uniform. Similar to E3, the 3D ERT-grid E4 shows increased values in the uppermost level (above: xy-plane, horizontal slices) caused by agricultural perturbation. Between 0.7 m b.s. and 1.5 m b.s., several anomalies are observable, which could be part of a linear structure such as an ancient wall or a channel (b; xy-plane, horizontal slice). Corresponding vertical slices (below: yz-plane, vertical slices) confirm this finding by the depthwise protrusion of high values at about 5-6 m profile length. Below 2 m, subsurface conditions are significantly uniform.
Le profil 2D E3 qui a été relevé dans la partie centrale de la doline révèle dans le sous-sol proche de la surface de fortes valeurs de résistivité qui correspondent à des artefacts d’origine humaine. La partie inférieure du profil indique de faibles valeurs de résistivité indiquant la présence de sédiments homogènes très fins. Comme E3, le profil 3D E4 montre des valeurs en augmentation vers le sommet qui peuvent être attribuées aux perturbations d’origine agricole (en haut : xy-sections horizontales). Entre 0,7 m et 1,5 m en-dessous de la surface, plusieurs anomalies sont observables et peuvent être interprétées comme des parties d’une structure archéologique linéaire, à savoir un mur ou un canal (b ; plan xy, sections horizontales). Les tranches verticales correspondantes (en bas : plan yz, tranches verticales) confirment la découverte archéologique et cela se traduit par des valeurs de résistivité élevées, à environ 5-6 m sur le profil. En-dessous de 2 m de profondeur, le sous-sol est uniforme et composé de matériaux très fins.
8Based on the TLS results, the study area can be analysed and described in high detail and visualised multi-perspectively for the first time. In this context, a theoretical doline model (fig. 5), which was established in order to define the relevant geometric parametres, allowed calculating crucial geomorphometric information using RiSCAN Pro and ArcGIS 9.3 (tab. 1). As displayed by the profiles in fig. 6, the enclosed doline has a diameter of 245 (west-east) and 153 m (north-south) at the bottom and a depth of 39 m (distance between the lowest point in the depression and the lowest point on the divide; Ford and Williams, 2007). Considering its geometry, the karst landform corresponds to a solution depression (cf. Sauro, 2003). While slopes are predominantly northwest- or southeast-facing, gradient values range between 0° and 29°. On the supposition that the subsurface geometry and volume of the sinkhole resemble an inverted cone of almost the same dimension (fig. 5) – a finding deduced from the geophysical outcomes in analogy to F. Sustersic (2006) – the subaerial capacity of the karst landform amounts to 2.37 Mm³, whereas the volume of the sediment fill roughly amounts to 0.245 Mm³. As with the ERT studies, particular attention must be paid to the zone of high geoarchaeological interest in the centre of the doline. After the GIS-based hydrologic analysis, both the micro-relief and the surface drainage pattern can be investigated. One of the most notable findings is the existence of a surface gully in the southern sector, which is oriented to the north-west of the doline (fig. 7). Even though it has been incised into the loose sediments less than half a metre, it represents the most prominent channel in the entire landform. It descends from the southern margin, where a well-structured accumulation of limestone blocks has been erected to capture an ancient spring that has dried up (fig. 1). However, the ditch is neither aligned with the water-filled ponds in the north, nor does it correspond to the subsurface findings that were indicated by ERT. The latter extend about 5-10 m east (fig. 7).
Fig. 5 – Theoretical doline model of Lato with relevant geometric parametres.
Fig. 5 – Modèle théorique de la doline de Latô figurant ses paramètres géomètriques.
Both the surface and the subsurface part of the karst depression must be considered in order to provide a holistic image of the landform. Based on the predefined and displayed geometric parametres, corresponding values, which were derived from ERT- and TLS-datasets, help access the entire geomorphometry of the Kritsa-Latô doline for the first time (for detailed description of attributes see tab. 1 and fig. 6).
La surface et le sous-sol de la dépression karstique doivent être intégrés simultanément afin de pouvoir livrer une image complète et précise des paysages aériens et souterrains. En se fondant sur les paramètres géométriques qui dérivent des résultats acquis en tomographie résistivité électrique et grâce aux relevés scan laser, la morphologie de la doline de Kritsa-Latô a pu être entièrement révélée (pour les détails concernant les détails des propriétés, voir tab. 1 et fig. 6).
Fig. 6 – TLS-based digital elevation models of the doline (2x2 m, upper left; 0.25x0.25 m, lower left) and topographical cross-sections (right).
Fig. 6 – Modèles Numériques de Terrain de la doline (résolution : 2 x 2 m, en haut à gauche; 0,25 x 0,25 m, en bas à gauche) et profils topographiques (à droite) réalisés grâce au scanner laser terrestre.
Both the constant inclination and the fairly smooth surface of outcropping breccia are highlighted. The geoarchaeological on-site area (lower left, see inset polygon in the 2-m model) displays the recent gully, which descends from the dried up spring and which is directed to the doline center (air-line distance A-A’: 508 m; B-B’: 430 m).
L’inclinaison constante et la surface de contact avec la brèche sont mises en évidence. L’aire d’étude géoarchéologique (en bas à gauche, résolution du MNT : 0,25x0,25 m, regarder dans le polygone de 2 m de résolution) se caractérise par la présence d’une ravine qui débute à la source tarie et se termine au centre de la doline. Distance directe A-A’ : 508 m ; B-B’ : 430 m).
Fig. 7 – Detailed block section of the on-site area in the Kritsa-Lato karst depression.
Fig. 7 – Bloc diagramme détaillé de la dépression karstique de Kritsa-Latô.
When superimposed on the TLS-derived DEM, the 3D ERT grid can be interpreted in a spatial context. The measured high resistivity values in a depth level between 0.7 m and 1.5 m are in remarkable alignment with the tapped spring (S) on the southern fringe and the wall remains in the recent pond, thus forming a potential water harvesting system (A). As revealed by the GIS-based hydrologic analysis, the current runoff (braided white lines) is not consistent with the ancient drainage pattern. A recent surface gully that is connected with the spring directs water to the northwestern sector of the doline bottom (R).
Superposé au MNT réalisé grâce au LST, le quadrillage 3D peut être interprété spatialement. Les fortes valeurs de résistivité obtenues entre 0,7 m et 1,5 m de profondeur se distinguent par un remarquable alignement depuis la source (S), située sur la marge méridionale de la doline, jusqu’à la mare située au centre de la dépression karstique ; l’ensemble forme un probable canal d’écoulement des eaux (A). Comme le démontre l’interprétation du réseau hydrologique via le SIG, les chenaux d’écoulement modernes (dessinés en blanc) ne se superposent pas avec le système de drainage antique. Une ravine, reliée à la source, présente une direction NW et draine les écoulements vers le centre de la doline (R).
Tab. 1 – Morphometric and geomorphologic properties of the investigated doline according to GIS-based analyses and evaluation of ERT-data.
Tab. 1 – Propriétés morphométriques et géomorphologiques de la doline étudiée. Les analyses ont été réalisées grâce au traitement des données TRE intégrées dans un SIG.
Parametres
|
Symbols and attributes
|
Values
|
Depth
|
DP1 (lowest point on doline bottom to lowest point on divide)
|
39 m
|
DP2 (lowest point on doline bottom to highest point on divide)
|
115 m
|
Diametre
|
D1A (west-east divide)
|
508 m
|
D1B (north-south divide)
|
430 m
|
D2A (west-east bottom)
|
245 m
|
D2B (north-south bottom)
|
153 m
|
Volume
|
V1 subaerial (based on depth DP1)
|
2.37 Mm³
|
V2 subaerial (based on depth DP2)
|
13.824 Mm³
|
Sediment depth
|
SD (R<25 Ωm)
|
25 m
|
Sediment volume
|
SV (based on sediment depth SD; V=1/3*(D2A*0.5)*(D2B*0.5)*π*25 m
|
0.245 Mm³
|
Total volume
|
TV1 (SV+V1)
|
2.615 Mm³
|
TV2 (SV+V2)
|
14.07 Mm³
|
Gradient
|
G1 (average inclination of doline bottom)
|
2°
|
G2 (average inclination of slopes based on DP1)
|
25°
|
G3 (average inclination of slopes based on DP2)
|
29°
|
Aspect
|
A (predominant aspect values)
|
NW, SO
|
For spatial arrangement and graphic illustration please refer to fig. 5.
Se référer à la fig. 5 pour consulter les données spatiales et les illustrations correspondantes.
9To this time, karst depressions in Eastern Crete have not been studied in detail. As a consequence, neither information about the surface and subsurface geomorphology, nor insights into the formation of corresponding landforms have been available. This lack of research also applies to the question of ancient land use within dolines, despite the well-known fact that they served as valuable areas for agricultural purposes ever since the Aegean Bronze Age (Siart, 2010). Among others, L. Hempel (1991) or G. Bartels (1991) mentioned that enclosed depressions in the Cretan mountains were filled with amounts of loose sediments, but did not specify this assumption. Recent investigations in the Psiloritis range southwest of Heraklion prove the existence of massive sediment accumulations in sinkholes and poljes, which occasionally amount to thicknesses of more than 35 m b.s. (Siart et al., 2010). These findings are in absolute accordance with the results yielded in the study at hand. As documented by the large sediment outcrop at the fringes of the ponds as well as all electrical resistivity measurements, the Kritsa doline complex exhibits huge sedimentary overburdens of at least 25 m in depth. Since the vertical penetration of ERT only extended to about 35 m b.s., even thicker accumulations must be expected, demonstrating the important reservoir function of karst depressions in the mountains of Crete. The heterogeneous subsurface relief displays many features typical of buried karst systems, in particular a very significant epikarst zone (Leucci and De Giorgi, 2005; Terzic et al., 2007), in which loose sediments intermingle with the heavily fractured bedrock and cause intermediate resistivity values. It must be considered as the first stage of authigenic weathering in terms of cryptokarst processes. Analogous to the findings of F. Sustersic et al. (2009), the sediment-filled dolines in the study area represent persistent landforms that outlast a continuous denudation of the basal breccia, whereas the initial bedrock geometry is completely reshaped. Moreover, the geophysical results point towards the existence of subsurface concavities in the parent rock (diameter of ~10 m). With reference to T. Waltham et al. (2005), such conical pits must be regarded as buried sinkholes. They serve as preferential drainage paths for the superordinated hydrographic system.
10Even though a comprehensive capture of the underground can only be achieved by additional geophysical methods, e.g. refraction seismics (Hecht, 2009), the investigated landforms serve as important sediment traps that prevent or reduce a further flux and loss of material. This fact constitutes the crucial sociocultural dimension of karst depressions, i.e. their economic value. In combination with archaeological findings, which indicate local land use ever since the first millennium BC (see following chapter), it suggests that the Kritsa doline complex represents a diachronic phenomenon. The absence of subsurface cavities in all geophysical tomographies, the homogeneous stratigraphy of vibra cores and the absence of suffosion processes point to a permanent accumulation phase without intermission. Given the predominantly fine-grained composition, the fill must have largely been formed and/or accumulated under rather stable and uniform geomorphodynamic conditions without significant changes in frequency and magnitude of geomorphic processes. Due to the fairly isolated location on top of a west-east trending crest only low energetic colluviation occurred in the doline complex. This fact is supported by the rather small catchment area (ca. 10 ha) that only allows short range transport. However, the enormous thickness of the infilling implies a long-term input of material or a great age of sediments. Since the formation of massive terra rossa accumulations in Mediterranean karst terrains only by limestone dissolution has been argued and widely disproved – the insoluble residues from Cretan carbonates generally only range between 0.5% and 3% by weight (Siart, 2010) – additional external input of material must be taken into account (Bronger and Bruhn-Lobin, 1997; Delgado et al., 2003). As demonstrated by numerous studies, the deposition of dust from North Africa significantly contributes to the soil formation in the Eastern Mediterranean (Ganor and Foner, 1996; Durn et al., 1999) and especially on Crete (Rapp and Nihlén, 1986; Pye, 1992; Rackham and Moody, 1996; Nihlén et al., 2002). Among others, the clay mineral composition can be regarded as an indicator for allochthonous material. Considering that Kaolinite is largely absent in parent rocks and that the currently moderate weathering conditions are unsuitable for its neoformation, the high amounts detected in the Kritsa doline fill can only be explained by palaeogenetic origin (Plio- to Pleistocene) or aeolian influx (Siart et al., 2009a). Thus, all findings thus point to the polygenetic properties of the loose sediments accumulated in the studied depression.
11In addition to the subsurface prospecting, topographical data obtained by terrestrial laser scanning helps complete the geomorphological investigation of the enclosed sinkhole. Regarding the morphometric properties as a whole, the TLS-based DEM clearly illustrates the funnel-like geometry typical of solution dolines (Ford and Williams, 2007). This fact is stressed by an average slope inclination below 30°, a characteristic of corresponding landforms (Gams, 2000). However, collapse processes that occurred during the initial stage of sinkhole formation cannot be excluded, as suggested by several escarpments in the uppermost parts of the slopes and the rim. The flat bottom is accentuated by a striking bend towards the slopes, while the margins are distinctly marked by an elliptical outline. The spatial orientation runs along a fault north of the doline, which can be verified in terms of a big scarp in the 3D data set (fig. 6). Evidently, local tectonics have had a major impact on the formation of the doline complex, analogous to the findings of M. Frelih (2003) and I. Gams (2005). Considering the topography of the slopes that is displayed in high detail by the DEM, terrain roughness is fairly smooth. Since the dip towards the bottom exhibits significant uniformity, surface runoff and colluviation are mainly directed towards the doline centre without being hampered or delayed by terraces and rims. This fact helps to explain the remarkable thickness of sediments.
12To conclude, the large extent and depth of the doline complex, as well as the thickness of its fill suggest a great age of the landform. Although no chronometric data are available so far, the onset of karstification must be dated back to the upper Pliocene or the Late Pleistocene, when the mountains of Crete experienced their first major phase of tectonic uplift (Fabre and Maire, 1983; Fassoulas, 2000). Local limestone weathering was triggered by orogeny and the relative subsidence of the underground karst watertable, respectively (cf. Papadopoulou-Vrynioti, 2004). According to F. Gabrovsek (2009), the amount of dissolution largely depends on the geometry of the rock surface. Increasing slope length of dolines consequently correlates with higher depth and volume (Sustersic, 2006), which in turn equals larger interfaces between parent rock and loose overburden (Gams, 2000). As dissolution under sediment cover is more intense than under aerial conditions, a positive feedback loop must be expected for the genesis of the Kritsa dolines, leading to a constant amplification of karstification over time.
13As shown by the geophysical results, the underground structure of the doline is largely homogeneous except for one anomaly in the southern part. This high resistivity zone, which has also been verified by auger probing, consists of limestone boulders in the near subsurface. Natural floaters, as described by T. Waltham et al. (2005), can be excluded since the blocks are neither the result of falling off or detachment from pinnacles, nor of mass wasting. Highly energetic colluvial influx is impossible precisely because of the good sorting and the absence of gravel and sand in the surrounding sediments. Due to the fact that the area is situated in the centre of the depression, while being embedded in a thick and extensive accumulation of fine-grained sediments, the blocks must be of anthropogenic origin. This finding is supported by the stone settings that are accessible in the embankments of the ponds and that correlate to the high resistivity clusters (fig. 4 and fig. 6). Quite likely, they form part of an ancient water reservoir, e.g. a walled cistern (Antoniou et al., 2006). Moreover, this result is consistent with investigations from the Central Cretan Ida Mountains, where karst depressions were also used by early settlers and modified for irrigation and drainage purposes (Siart et al., 2010). When combined with TLS data, the geoelectrical outcomes can be put into a spatial context that provides insights into the whole extent of this potential water harvesting system: The near-surface high resistivity values are exactly in line with both the northern wall remains in the ponds and the captured spring at the southern margin (fig. 7). However, as they only penetrate to a depth of about 1.5 m to 2 m b.s., the blocks point to a superficial way of land use. One can therefore suppose a prehistoric drainage channel that was constructed to direct surface runoff into the cistern. The high clay content of the doline fill must have favoured this irrigation technique. It might have been abandoned and subsequently buried under younger colluvium in consequence of decreasing precipitation or socioeconomic changes. The spring became active again at a later time, as indicated by a recent gully in the DEM. However, according to the GIS-based hydrologic surface analysis, it is completely disconnected from the ancient water harvesting system and shows no spatial correlation with the subsurface findings. It is the only deeper incision in the doline bottom, though, and therefore highlights geomorphodynamic stability in the more recent past.
14Above all, the ancient water supply network raises the question of age. With regard to the thin sedimentary overburden on top of the limestone blocks next to the ponds (ca. 0.5-1m), a near-surface occupation layer could be presumed, leading to the assumption of a relatively young construction. Instead, countless superficial sherds, which can most probably be dated back to Hellenistic times, indicate intense pre-Christian colonisation and land use in the study area. The nearby settlement of Latô, which was founded at the margins of another karst depression, as well as the lack of findings from Post-Dorian to medieval times suggest that the local occupation of the Kritsa doline took place during the first millennium BC. Evidentially, the availability of water was essential for subsistence in the karstified and, thus, arid mountains of Crete as early as the Bronze Age (Angelakis and Koutsoyiannis, 2003; Panagiotopoulos, 2007). Among others, A. Angelakis et al. (2007) refer to the sophisticated achievements of hydrotechnical engineering during Minoan and Mycenaean times, e.g. tapping of springs and irrigation techniques. G. Antoniou et al. (2006) additionally state that the settlement of Latô was supplied only by rainwater collected in cisterns since no springs existed in its immediate vicinity. The same holds true for the ancient city-state of Dreros, located about 10 km north of the Kritsa doline complex. All these findings demonstrate that the availability and the considerate distribution of water was indispensible for survival in the Dikti Mountains. In this context, the water harvesting system identified in the studied karst depression might represent an off-site water reservoir formerly associated with the Dorian city-state. Apart from that, the subsurface findings in the sediment fill could also be part of a drainage system constructed to prevent arable land from flooding. Since the flat bottoms of sediment-filled dolines were easy to use, extensive draining was common in Bronze Age Greece (Showleh, 2007; Koutsoyiannis et al., 2008). Even though dating the presumable water system still remains complex due to the absence of chronometrical data, the findings from the Kritsa doline complex provide first-time insights into the geoarchaeological history and the characteristics of ancient land use in the Dikti Mountain Range.
15The study at hand reveals new insights into the geometrical, geophysical and geoarchaeological properties of enclosed depressions in Eastern Crete. As shown by the results, comprehensive data sets are indispensible in order to understand the functional principle of karst landforms. Terrestrial laser scanning, which has not been used in karst research on Crete so far, proves to be highly suitable for non-invasively acquiring precise topographical information (cf. Lerma et al., 2010). In contrast to other DEMs derived by remote sensing techniques, micro-relief features become accessible for the first time due to area-wide capture of spatial information and, thus, can be considered for further interpretation. Moreover, results are independent of operator’s criteria for selecting surface points (Armesto et al., 2009). When implemented in GIS, TLS-based data can be used for both morphometric analyses that better the understanding of geomorphological landforms or processes and a combination with geophysical results. Hence, subsurface and surface findings can be linked in order to establish an integral image of karst depressions. With regard to the Kritsa doline complex, carbonate dissolution and neotectonics led to the formation of large funnel-shaped hollows during the Quaternary. Geoelectrical tomographies prove the complex and unpredictable properties of the buried karst system, particularly the irregular depth to bedrock and the heterogeneous subsurface. In addition, the multi-method approach allows a first-time quantity assessment of loose sediments accumulated within dolines. As has already been demonstrated for Central Crete (Siart et al., 2010), enclosed karst depressions in the Dikti Mountains are filled with thick deposits of up to 30 m, pointing to the great age of corresponding landforms. They must be considered as sediment reservoirs and important terrestrial archives, which have been subject to permanent influx of material. Supplementary vibra coring, an absolute prerequisite for calibrating the geophysical results (Terzic et al., 2007), highlights the complete absence of stratification and therefore points to permanent low energetic colluviation processes under rather stable geomorphodynamic conditions.
16Besides supporting geomorphological investigations, both TLS and ERT data provide valuable information for reconstructing the environmental evolution of the study area. With regard to ancient land use, several geophysical anomalies in the studied doline suggest a water harvesting system that dates back to Dorian times most likely. This result seems quite reasonable since the flat topography, the thick pedo-sedimentary infilling and the hydrologic favour of many Cretan dolines were of highest significance for economic exploitation ever since the Bronze Age (Siart et al., 2009a). Back then, like today, sufficient water supply was essential for agricultural purposes in the karstified mountains of the island, which is why the Kritsa dolines served as favourable locations for settlement and land use activities. Several cisterns unearthed in the proximate site of Latô substantiate these findings.
17However, a combination of morphometrical information with tomographical outcomes is obligatory in order to obtain detailed and comprehensive results. While high precision TLS data allows the uncovering of microscale features, which are generally invisible to the unaided eye, the actual presence and position of buried remains can only be identified through geophysical mapping. By integrating both data sets, the full extent as well as the differences between the ancient drainage system and the recent surface runoff pattern can be visualised for the first time. This fact highlights the importance of a multi-method approach to avoid ambiguous or incomplete findings. Even though further investigations are required to date the geoarchaeological findings and the geomorphological processes, the new methodical approach helps to better understand the landforms and the implications of ancient land use in the Dikti Mountains. In addition to palaeoenvironmental studies in the coastal areas, the study at hand offers first-time insights into the interactions between man and nature in the mountainous karst terrains of Eastern Crete. The results prove that sediment-filled dolines serve as valuable geoarchives, while fusing terrestrial laser scanning and geophysical mapping offers promising prospects for future geomorphological and geoarchaeological studies.
This paper is a contribution to the research project on the reconstruction of Holocene palaeoenvironmental changes on Crete (Geographical Institute, Heidelberg University, Germany) and the ANR (Agence Nationale de la Recherche, France) programme DIKIDA (Espace et Territoire, InSHS) directed by Daniela Lefevre Novaro (University of Strasbourg, France) under the supervision of Matthieu Ghilardi. The authors would like to thank Gerd Schukraft (Laboratory for Geomorphology and Geoecology, Heidelberg University, Germany) and Stéphane Cordier (University of Paris-Est Créteil Val-de-Marne, France) for their support during the fieldwork. Sincere thanks are due to Stefan Hecht for helpful advice on geophysical prospecting. Kosmas Pavlopoulos (Harokopeion University of Athens, Greece) and Irene Zananiri (IGME, Athens, Greece) are greatly acknowledged for facilitating and issuing the authorisation for the field campaigns. The central research pool of the Heidelberg University is greatly acknowledged for financial support. Moreover, special thanks are due to Alexandre Farnoux (University of Paris-Sorbonne, France), who is conducting a geoarchaeological programme on the Mirambello area since 2003, for providing precious information about the area. Two anonymous reviewers are greatly acknowledged for their valuable comments, which helped improving this paper.