Hydrological features of Matese Karst Massif, focused on endorheic areas, dolines and hydroelectric exploitation

ABSTRACT This study presents an original mapping of hydrological karst features of the Matese massif (southern Italy), whose relevance is given by large basal springs supplying millions of people and the hydroelectric exploitation of the major endorheic areas. We mapped dolines and endorheic areas from 1- and 5-meter Digital Elevation Models (DEM) using Geographic Information System (GIS) tools and techniques. Instead, ponors, caves, and karst springs were mapped mainly based on cartographic and literature analyses. We identified 321 endorheic areas occupying 31% of the massif area and 489 dolines, distinguished in (i) solution (N = 433) and collapse dolines (N = 56), the latter located in the discharge zones of the massif and connected to ascendant flows of CO2- and H2S-rich groundwater. The map shows the hydrological features of a karst massif from a more detailed and new perspective, and it can be helpful in water management, groundwater resource protection, environmental safeguarding, and ecological development.


Introduction
The hydrology and landforms of carbonate karst areas arise from a combination of two main factors: high rock solubility and well-developed fracture networks (Ford & Williams, 2007).Carbonate karst absorbs significant amounts of meteoric water, which is then transferred underground, stored, and discharged at the surface through the springs.
In karst areas, the groundwater systems are in a close relationship with surface hydrology and landforms, and the mapping of surface features represents a key in any hydrological analysis or study, including water balance, hydrological modelling for assessing surface runoff and groundwater recharge (Fiorillo et al., 2015), delineation of protection zones for wells and springs (Goldscheider, 2010), simulation of point source contamination, and making artificial water tracer tests (Goldscheider et al., 2008).
Several landforms, whose role is to discharge water into the underground, such as dolines, poljes, ponors, and sinking streams, characterize a karst area.These landforms are normally located inside endorheic areas, which are closed catchments where the internal runoff is completely absorbed by one or more ponors, or swallow holes (White, 2002), representing fissures in the karst massif through which water sinks underground (Bonacci, 2004).Endorheic areas are important recharge zones of the aquifers and are generally hydraulically connected to one or more springs.Outside these areas, runoff can escape from the karst area and does not infiltrate underground, especially during intense rainstorms (Fiorillo et al., 2015).
In this study, we mapped the main karst hydrological features of the Matese massif (southern Italy), including endorheic areas, dolines, ponors, caves, and karst springs.Such landforms arise mainly from processes involving the karst surface, even if the groundwater dynamic also has a role in the genesis of specific karst landforms, in some cases.
In general, studying hydrological features have a fundamental role in the quantification and protection of the water resource of a karst massif (Goldscheider, 2010).In this study, in particular, attention was given to endorheic areas, because of their importance in the (i) recharge processes of the huge groundwater resources of the Matese massif, which supply millions of people in southern Italy by many aqueducts, and (ii) their hydroelectric exploitation since the beginning of the last century.

Study area
The Matese massif is located in the median sector of the Apennine chain, between the Campania and Molise regions, and has an extension of 540.3 km 2 .Steep slopes and high and ground elevations, up to 2050 m a.s.l of Miletto Mountain peak, characterize the massif.The ground elevation follows a normal distribution, with a maximum between 1000 and 1100 m a.s.l, and more than 50% of the massif lies above 900 m a.s.l., as shown by histograms in the main map.
A high permeability limestone and limestone-dolomite sequence of Jurassic-Miocene age, with a thickness of 2500-3000 m, built the Matese massif.Faults have prevalent NW-SE and NE-SW trends and dissect the sequence.The massif is tectonically joined by thrust fault to low permeability argillaceous complexes (Paleocene) and flysch sequences (Miocene) along the northern and eastern sectors.Normal faults bound the Matese massif along the southern and western sectors and divide the massif from the Volturno alluvial plane.More specific insight into the geology of the area can be found in the Geological Map of Italy of Istituto Superiore per la Protezione e la Ricerca Ambientale (ISPRA) (n.d.) and Vitale and Ciarcia (2018), while a description of the geological sequence is given in Robustini et al. (2003).Further geological and geomorphological information on the Matese massif is provided by Aucelli et al. (2013).
The climate is Mediterranean.Climate graphs in the main map provide monthly mean precipitation and temperature values from two local weather stations.The maximum monthly rainfall occurs in November, and the minimum in July.The temperature and potential evapotranspiration patterns are almost opposite to rainfall: maximum temperature occurs in July-August and minimum in December-February.Because rainfall primarily occurs during the non-warm season, its distribution allows for the highest recharge into the aquifer (Fiorillo, 2009;Fiorillo et al., 2015;Fiorillo & Pagnozzi, 2015).In the high-elevation zones (above 1000 m a.s.l.), precipitation is generally snowy during winter.Annual mean precipitation and temperature charts in the main map provide further details on the climate of the Matese massif.The precipitation and temperature spatial distributions were estimated based on the paper of Fiorillo and Pagnozzi (2015).The authors found two distinct altitude-rainfall regressions, indicating that an up-wind sector (wetter) and a downwind sector (less wet) can be distinguished for the Matese massif.
The Matese massif represents a huge karst aquifer feeding numerous small springs as well as large basal springs.As in most of the karst areas of the world where the morphology of underground karst features is unknown (Bonacci & Andric, 2015), the actual catchments of the Matese springs are roughly known.For this reason, we did not map spring catchments on the main map.However, it must be outlined that faults, aquitards, aquicludes, and/or weakly-karstified rocks bound the spring catchments, which also include one or more endorheic areas at the surface (Bonacci & Andric, 2015;Fiorillo & Pagnozzi, 2015).

The karst springs of Matese massif
Karst springs distribute between about 50 and 1450 m a.s.l., as shown by the main map.Some of them are located inside endorheic areas or in high elevated zones, and their discharge is lower than that of basal springs.
Table 1 shows the major springs of the massif.Among them, basal springs are the largest ones and are located along the southern (Torano and Maretto springs) and northern sides (Boiano springs) of the massif and in the extreme south-eastern sector (Grassano-Telese spring group).
The central sector of Matese massif (Civita, 1969) feeds the Torano spring (201 m a.s.l.), which has a mean annual discharge of 2.04 m 3 /s (Fiorillo & Guadagno, 2012).Details on the hydrological behaviour of this spring can be found in Fiorillo and Doglioni (2010) and Fiorillo (2011).The Maretto spring (170 m a.s.l.) is located at the foot of a limestone-dolomite slope in Piedimonte Matese village, and has a mean annual discharge of 1.0 m 3 /s (Fiorillo & Guadagno, 2012).Both Torano and Maretto springs were tapped during the 1960s by the Campano aqueduct, which supplies Naples metropolitan area.
The Grassano and Telese springs discharge along the southern side of Montepugliano relief and are made by large fresh and sulfureous thermal springs closely located.The mean discharges are 4.5 and 0.2 m 3 /s for fresh and sulfureous springs, respectively (Fiorillo et al., 2019).CO 2 − and H 2 S-rich waters characterize the Telese thermal springs, and the area surrounding this spring group is the most important non-volcanic thermal area in Campania region (Rufino et al., 2020).Other sulfureous springs are Table 1.Main springs of the Matese massif (modified from Fiorillo & Pagnozzi, 2015;data from Celico, 1978;Celico et al., 2008;Fiorillo & Guadagno, 2012).located in the western sector of the Matese massif, where cold gas vents have been also observed (Ascione et al., 2018).In particular, CO 2 fluxes characterize the areas between Ciorlano and Ailano, and are well evident at Ciorlano spring.Near Boiano village, there are three main spring groups having a mean annual discharge of about 2.80 m 3 /s (Civita, 1969): the Maiella group (west of Boiano village), the Pietrecadute group (located in the village), and Riofreddo group (east of Boiano).These springs are fed by the karst system of the north-central sector of Matese massif and are located along the debris deposit covering a fault separating limestone and flysch sequences (Petrella & Celico, 2009).Most of the springs have been tapped by Biferno Aqueduct during the 1960s and supply the Molise region.A branch of this aqueduct is joined with Campano Aqueduct through a tunnel in the eastern sector of Matese massif.

Main endorheic areas and their hydroelectric exploitation
In many karst areas of the world, the origin of wide endorheic areas is connected to structural and tectonic features (faults and grabens), which control the form and dimension of the depressions.A large closed basin with a flat bottom, karstic drainage, and steep peripheral slopes can be defined as polje (Gams, 1978), even if there are different definitions of karst polje (Bonacci, 2013).A seasonal or permanent sinking lake is generally located at the bottom of the depression, with great environmental and ecological value (Pardo-Igúzquiza et al., 2022).
Previous studies highlighted the contribution of endorheic areas in the recharge processes of the Matese massif (Fiorillo & Pagnozzi, 2015) and the role of faults in controlling groundwater flow (Petrella et al., 2009).
The origin of the widest endorheic areas is connected to the upper Pliocene-Pleistocene tectonic activity, which caused a general uplift of the area by normal faults and the formation of grabens.These tectonic depressions evolved into endorheic areas during the following continental environment (Pleistocene-Holocene) thanks to karst processes.Currently, some of them have seasonal lakes and are drained by one or more ponors.
During the last century, dams have been built to exploit the potential energy of waters stored at high ground elevations to generate hydroelectric energy.In fact, dams avoid the runoff sinking into ponors and allow water storage in lakes.Currently, four hydropower plants exploit the water stored in endorheic areas of Matese, Letino and Gallo Lakes.Table 2 shows hydropower plants of the Matese massif, classified according to their type and power capacity.
The largest endorheic area is that of Matese Lake polje, located between 1000 and 2050 m a.s.l.It is the major karst morpho-structural feature of the massif, whose evolution has been controlled mainly by WNW-ESE normal faults.
The Matese Lake is a permanent water reservoir sustained by fine sediments and pyroclastic ash coming from volcanos located along the Campania coast (Vesuvius and Phlegraean Fields).In the past, the lake was naturally drained by several ponors located below the lacustrine sediments and by two other main ponors, named Brecce and Scennerato, located along the southern side of the polje.The hydraulic works of the 1920s have isolated the main ponors by earth dams (Ruggero, 1926), and have tried to waterproof the lake bottom.Before dam construction, the Brecce and Scennerato ponors limited the extension of the lake.Currently, the maximum water level reaches 1012 m a.s.l., which corresponds to a volume of about 15 Mm 3 (Fiorillo & Pagnozzi, 2015;Rasulo, 1976).
Both hydrological observations carried out before and during the hydraulic works of the 1920s (Ruggero, 1926) and tracer tests (Civita, 1969) have confirmed the connection between the Scennerato ponor and Torano spring.The hydrogeological cross-section in the main map (modified from Fiorillo & Pagnozzi, 2015) shows the connection between the endorheic area of the Matese Lake and the Torano spring.On the other hand, the connection between Matese Lake and other springs is controversial.Gauther (1910) described how temporary dams in the Matese Lake polje influenced the regime of the Grassano-Telese springs, whereas Ruggero (1926) argued that the lake had a limited or no role in the regime of these springs.However, tracer tests carried out in 1953 would prove the connection between the Matese Lake polje and the Grassano-Telese springs.
Based on the different elevations between the southern basal springs (Torano and Maretto springs) and the northern basal springs (Boiano springs) and speleological surveys, a fault system affects the geometry of the saturated zone, as illustrated by the hydrogeological cross-section in the main map.In particular, 4 km north-west from Mutria Mt., speleological surveys have ascertained the depth of 'Pozzo della Neve' shaft system in the vadose zone of the aquifer, which extends between 1363 and 318 m a.s.l.The role of faulting on the groundwater flow was also described by Aquino et al. (2015) in a north-western sector of the Matese massif.
The first hydropower plant of the Matese Lake started operating in the mid-1920s.Currently, the hydropower system is made of two powerplants.The lake water is first conveyed through a penstock towards the San Gregorio Matese hydropower plant, located at 526 m a.s.l.After the first hydraulic jump, the outflow is channelled into a floodway and is stored in a charge basin.The water is then conveyed towards the Piedimonte Matese hydropower plant, located at 178 m a.s.l.The amount of water adducted from the Matese Lake for hydropower purposes varies according to the energy demand during the year: the energy is mainly generated from December to June, and a long period of non-adduction occurs during summer.The hydropower system capacity is about 34 MW and can be classed as a large hydropower system.
The Letino and Gallo Lakes are also dam reservoirs.The lakes were created by the interruption of the Lete and Sava Rivers, respectively, but can be considered part of wide endorheic areas.Before dam construction, both rivers sunk into two ponors located downstream the dams.In particular, the Sava River ponor, namely 'Acqua Spuzzata', has been partially inactivated by the dam which has limited its functioning.A part of the runoff escapes from the endorheic areas via a man-made channel that flows in the Delizia Valley.
The Letino Lake (893 m a.s.l.) has powered a hydropower plant located in Prata Sannita Village up to 1966, which fell into disuse after the construction of the Gallo Lake dam.Since 1966 the Letino Lake water has been directed towards the Gallo reservoir and currently powers the Gallo powerplant (842 m a.s.l.), which has a capacity of 2.5 MW.Furthermore, the water of Gallo Lake (836 m a.s.l.) is conveyed toward the pumped-storage hydropower plant of Capriati (186 m a.s.l.), located in the Volturno plain.After going through the hydroelectric turbines, water is pumped up from the Capriati reservoir to Gallo Lake during periods of excess of electric energy production.The Capriati powerplant has a capacity of 113 MW and is the largest hydropower plant in the Matese area.
In the past, a hydropower plant existed in San Massimo village, which exploited the water of Capo d'Acqua spring, a perennial spring located in the Campitello Matese endorheic area.Nowadays, this powerplant is fallen in disuse.With an extension of 7.5 km 2 , the Campitello endorheic area is one of the main karst features of the Matese massif.The area gets flooded during rapid snow melting periods, and a temporary lake originates.A ponor located at 1417 m a.s.l. in the central part of the Campitello plain drains the lake water, which dries out in several days.
Another two hydropower plants are located in Gioia Sannitica and Telese Terme.The Gioia Sannitica powerplant, namely Auduni powerplant, adducts water from the Biferno aqueduct to produce electricity.The head of water available for power generation is 288 m, and the powerplant has a capacity of 11MW.
The Telese Terme powerplant has a capacity of 0.120 MW and exploits the hydraulic energy of the Grassano River, which is fed by the Grassano springs.

Data and mapping methods
The presented map was made using Esri ArcGIS® software to digitalize the hydrogeological and man-made hydrological features.
Hydrological features include endorheic areas, dolines, ponors, cave entrances, and karst springs.The man-made hydraulic features include dams, water reservoirs, lakes, and components of hydropower plants.Table 3 summarizes the sources of topographic and geo-thematic data used for mapping.
Dolines and endorheic areas were mapped through a semi-automated method from a 5-meter DEM, described in detail in section 3.1.Doline mapping was supported by a visual inspection of 1:25000 topographic maps of Istituto Geografico Militare (IGM) (n.d.), official cartography of Geological Survey and Soil Protection of Campania Region (2001), and Google Earth images.The dolines of Montepulgliano Mt. and Telese Terme plain, located in the southernmost sector of the Matese karst system, were manually mapped by visual analysis of a hillshade model and a contour map derived from a 1-meter LiDAR DEM as well as Google Earth images.Field surveys were made to verify the precise location of small and anthropized dolines of the area.
The location of ponors and cave entrances are provided by the Speleological Federation of Campania (2004).Further information can also be found in Russo and Capasso (2005), Russo et al. (2011).Subordinately, ponors of major endorheic areas were located through inspection of Google Earth images and additionally verified during field surveys.It must be underlined that the number of identified ponors is underestimated: many ponors are hidden among the vegetation or not appreciable by direct inspection, while others are hidden below cover deposits or located on the bottom of lakes.
The 1:25000 and 1:50000 IGM topographic maps, and the 1:5000 Technical Regional Chart of the Campania Region, were used to map karst springs and the surface and subsurface components of both historical and currently working hydropower plants.

Dolines and endorheic areas delineation
An original mapping of dolines and endorheic areas was made by the procedure described in the following: (1) The ArcGIS Fill tool was applied to the 5-meter DEM (Figure 1a) to fill all the depressions of the digital surface up to their spill elevation (i.e. the elevation at which water ideally flows out of the depression) to create a depression-less DEM (Figure 1b,c).The original DEM was then subtracted from the depression-less DEM to generate a difference raster (Figure 1d) representing depression location and depth (Doctor & Young, 2013;Wu et al., 2016).This procedure identifies both false and true closed depressions.(2) More precisely, only a part of the identified depressions represented actual dolines.The difference raster was therefore filtered with a pixel depth threshold of 0.55 m to reduce the number of false depressions.In this way, all pixels in the difference raster having a depth lower than the threshold value were removed (Mukherjee, 2012).
The threshold value of 0.55 m was chosen by analysing the statistical distributions of the pixel depths of depressions identified in two specific areas: one characterized by a high concentration of false depressions (flat area; Figure 2a) and one where true karst depressions are located (rough area; Figure 2b).Both sectors cover an area of 4 km 2 and were verified by field surveys and analysis of topographic maps (Table 3).
Figure 2c and d show the depth distribution of pixels of depressions identified in areas illustrated in Figure 2a and b, calculated by Matlab (v.R2019b) statistical tools.As 95% of pixels of false depressions are shallower than 0.55 m, a threshold of 0.55 m was chosen.Figure 2e and f show the results of filtering: in the flat area, most closed depressions are not actual depressions and have been deleted after filtering; in the rough area, depressions have been preserved and represent actual dolines.
(3) A visual check of the automatically identified depressions was carried out.Various informative layers (Figure 3) were visually inspected to drive doline mapping: (i) the filtered difference raster representing depression location, (ii) a contour map and a hillshade of the study area, both generated from the 5-meter DEM, (iii) Google Earth images, and (iv) IGM topographic maps.
The doline perimeter was then drawn according to the outermost closed contour method (Bondesan et al., 1992;Denizman, 2003;Šegina et al., 2018;Verbovšek & Gabor, 2019).Based on this method, the uppermost contour bounding a depression represents the upper rim of the doline (Figure 4).A vertical contour interval of 0.5 m was used.
(4) Finally, the ArcGIS Watershed hydrological tool was used to delineate endorheic areas from the DEM (Figure 4).The tool is designed to delineate watershed of a river (i.e. the upslope area that contributes to the river flow) and needs two input information: (i) the location of the watershed outlet, generally the lowest point along the stream channel, and (ii) the surface flow directions derived from a DEM.
Dolines represent the most depressed sites within endorheic areas.Therefore, the latter can be automatically delineated using the actual depressions identified in the difference raster (c.f., Figure 2f) as input for running the Watershed tool.The whole doline area, represented by a set of neighbouring pixels in a GIS environment (c.f., Figure 1d), should be considered to have a hydrologically correct delineation of the endorheic area drainage divide, instead of a single lowest point (single pixel) at the doline bottom (c.f., Bouear, 2015).Following this criterion, the endorheic area boundary should never intersect the doline upper rim when applying the Watershed tool.

Hydrological features of dolines and types
Numerous and different types of closed depressions characterize the Matese massif.Depressions affect both the recharge zone located at high altitudes and the discharge zone located at low altitudes along the boundary of the massif.
This study distinguishes the mapped dolines into solution and collapse dolines based on their hydrological characteristics and role.
Solution dolines characterize the recharge sectors of the massif and are associated mainly with an epigenic karst (Frumkin et al., 2015), being their genesis related to surface hydrological processes.Ponors are often associated with solution dolines and represent concentrated and direct infiltration points, even if they are often buried below the cover deposit.Solution dolines constitute the most depressed zones of endorheic areas, and the internal runoff which occurs in endorheic areas thus directly sinks into the vadose zone of the aquifer through dolines and ponors.Therefore, the wider the endorheic area, the larger the contribution of dolines and ponors to groundwater recharge.
Collapse dolines, also known as collapse sinkholes (Ford & Williams, 2007), originate from underground karst voids, whose widening leads to the collapse of the cavity's roof and the consequent formation of a doline on the surface (Gutiérrez et al., 2011;Nisio, 2008;Waltham et al., 2005).In turn, the genesis and evolution of these karst voids in depth are associated with hydrological processes occurring in the discharge sectors of wide karst systems, which are characterized by the rising CO 2 and H 2 S-rich waters from depth (Frumkin et al., 2015;Klimchouk, 2011).Collapse dolines have no relationship with surface hydrology, but rather are the results of the ascendent flows characterizing the discharge sectors (Fiorillo et al., 2019).The collapse dolines involving the karst bedrock are named bedrock collapse dolines, while those involving the thick cover deposits mantling the karst bedrock are named cover collapse dolines.

Morphometric parameters
The morphometric parameters of dolines and endorheic areas were measured in GIS environment and then statistically analysed.These parameters are: the area, elevation of the lowest point, maximum elevation difference, circularity index (CI), and longest axis orientation.Solution dolines only constituted a statistically significant sample and were therefore considered in the morphometric analysis.
The area, lowest point elevation, and maximum difference elevation were determined from the DEM by the ArcGIS Zonal Statistics tool and provide  3) used in doline identification and mapping processes.The layers were visually inspected using GIS to verify the results of doline automatic identification.The minus symbol '−' marks dolines in the official 1:25000 topographic maps.information on landform dimensions and elevation distribution.The maximum elevation difference is the difference between the highest point on the perimeter and the lowest point at the bottom of a doline or endorheic area.In particular, this parameter represents the depth of the closed depression in the case of a doline.
The CI was calculated as follows (De Carvalho Junior et al., 2013): Where A and P are the area and the perimeter of a doline, respectively.The index ideally ranges from 0 (elongated shape) to 1 (circular shape).
The ArcGIS Minimum Bounding Geometry tool was used to measure the orientation of dolines having CI < 0.9 and endorheic areas.The tool calculates the azimuth of the longest axis of the rectangle bounding the perimeter of the landform.

Results
Our study provides new insights into the morphometric and hydrological features of endorheic areas of the Matese massif.In addition, it gives an original description of the dolines and their relationship with hydrological processes.Based on this last aspect, dolines of the recharge zone differ from those of the discharge zones.Table 4 shows the doline types and their main features.
Solution dolines (N = 433) characterize the depressed sectors of endorheic areas and play a key role in the recharge of groundwater resources.
Cover collapse dolines (N = 18) affect the sedimentary layers covering the karst bedrock of the Telese Terme Plain (Fiorillo et al., 2019;Leone et al., 2019) and the slope debris located west of Raviscanina village.This type of dolines seems to be also located in the Lete River Plain (c.f., Del Prete et al., 2004).
Differently from other dolines occurring in the recharge area of the Matese massif, which have formed by dissolution during the infiltration processes, the genesis of Montepugliano and Telese dolines is associated with a local ascendent flow, connected to deep groundwater flows occurring into the saturated zone of the karst aquifer.In particular, the Montepugliano  relief can be considered the discharge zone of a wide portion of the Matese massif.The groundwater come out through the Grassano springs located along the southern side of the Montepugliano relief at 54 m a.s.l., in the lowest altitude zone of the Matese massif (Fiorillo et al., 2019).Therefore, the geomorphological processes that have led to the formation of these dolines are of a hypogenic type (Palmer, 2016) and are connected to ascendant flows transporting CO 2 and H 2 S in solution from the depth (Fiorillo et al., 2019).
A statistical analysis of the main morphometric parameters of the 433 solution dolines and the 321 associated endorheic areas was carried out.
Solution doline area ranges from 1.2 × 10 -4 to 0.21 km 2 , with 50% of doline smaller then 2.5 × 10 -3 km 2 (Figure 5a).The depth ranges from 0.42 to 56.0 m, with 50% of dolines deeper than 2.5 m (Figure 5b).Doline bottom elevation ranges from 430 to 1847 m a.s.l.(Figure 5c).The values appear normally distributed, with a mean 1160 m a.s.l.Solution dolines developed mainly within an altimetric zone located above the mean outcropping elevation of the karst terrains (1000-1100 m a.s.l.).The highest (solution) doline density zone is located in the north-western sector of Matese massif, north-east of the Gallo Lake, between about 1000 to 1500 m a.s.l.Almost 31% of the mapped solution dolines (N = 135) concentrate here and affect a weakly faulted Upper Cretaceous limestone block.
Solution dolines having CI < 0.9 (N = 321) represent 75% of the whole population.These dolines have a N140°-150° preferential orientation (Figure 6a).On the contrary, a clear preferential orientation cannot be observed for dolines of the north-western sector of the Matese massif (Figure 6b).
The endorheic areas occupy the high elevated zone of the aquifer, with more than 80% of the endorheic surface above 1000 m a.s.l., as shown by histograms in main map.Endorheic areas cover 167.4 km 2 (31% of the karst massif).Their area ranges from 3.1 × 10 - 3 to 43.4 km 2 , with 50% of endorheic areas smaller than 5.2 × 10 -2 km 2 .Most of them (N = 271) are smaller than 0.33 km 2 , while the largest is the Matese Lake polje, occupying an area of 43.6 km 2 .The maximum elevation difference varies from 8.1 m to 1111.0 m, with 50% of the values higher than 67 m.The majority of the endorheic areas are oriented N40°-50° and N90°-100° (Figure 7).Structural features have controlled the orientation of the largest endorheic areas of Matese massif, which show a general NW-SE trend.In addition, strata setting would have affected the development and evolution of small endorheic areas and dolines (Pagnozzi et al., 2019).

Conclusions
Karst features of the Matese massif were mapped in this study because of their importance in controlling water infiltration.An accurate GIS-based procedure was applied to make an original mapping of dolines and endorheic areas, while ponors were identified mainly based on published information, and subordinately on field surveys and Google Earth images inspection.We identified and mapped 489 dolines and 321 endorheic areas, with 89% of the dolines located in the recharge sector of the Matese massif.
In the recharge zone of the aquifer, the internal runoff occurring in endorheic areas infiltrates into the underground through dolines and ponors, which are the main points for concentrated infiltration and, therefore, the main entry points for pollutants.
The basal karst springs located along the boundaries of the massif drain the huge groundwater resource, and collapse dolines demonstrate the ascendant groundwater flow phenomenon occurring in the discharge zone.
The study also describes the Matese massif hydropower plants, which mainly exploit the potential energy of water reservoirs on the bottom of the largest endorheic areas to generate energy.
The presented map gives an overview of the major karst features associated with surface waters and groundwater circulation, as in the case of karst springs and collapse dolines.Thus, the map appears helpful in many practical applications, including delineating protection zones for wells and springs, making artificial tracer tests, and water balance, but also in environmental safeguard, and ecological development.The Matese massif hydrological features are common to other karst massifs of central-southern Italy and of the world.Therefore, the methodology used for doline identification and endorheic area delineation can also be applied in other karst areas.

Software
Digital Elevation Model analyses and map compilation were carried out using Esri ArcGIS® 10.3.1.Statistical analyses were performed using Matlab 2019b.

Figure 1 .
Figure 1.Automatic identification of the closed depressions in a digital surface model.Hillshade and contour map (contour interval equal to 2 m) derived from 5-meter DEM of the Matese massif (a); dashed red lines mark the uppermost closed contours bounding one or more closed depressions.Filled depression DEM derived by the ArcGIS Fill tool (b); all closed depressions are filled up to their spill elevation as shown by cross-section AA' (c).The difference raster (d), representing location and depth of depressions, was computed by differencing DEMs in figures a and b; colours represent depression depths measured from the spill level (as shown in figure c).

Figure 2 .
Figure 2. Location and depth of depressions automatically identified by GIS tools in a flat area (eastern sector of the alluvial plain of Rio River, a) and a rough area (north-western sector of the Matese karst massif, b).Depth distribution of pixels inside the depressions of the flat (c) and rough (d) areas; the blue dashed line indicates the threshold value.Closed depressions in the flat (e) and rough (f) areas after the filtering: white stars mark actual closed depressions (dolines).

Figure 3 .
Figure 3. Layers of GIS data (Table3) used in doline identification and mapping processes.The layers were visually inspected using GIS to verify the results of doline automatic identification.The minus symbol '−' marks dolines in the official 1:25000 topographic maps.

Figure 4 .
Figure 4.A 3D representation of the dolines (in red) and the boundary of endorheic areas (in blue) located in the north-western sector of the Matese karst massif (Figure 2b).

Figure 5 .
Figure 5. Frequency distribution of doline morphometric features: area (a), depth (b), and elevation of the lowest point (c).Dolines having area >0.1 km 2 (N = 7) are not shown in Figure a.

Figure 6 .
Figure 6. Rose diagrams showing orientation (azimuth) of the solution dolines for the whole Matese massif (a) and north-wester sector (b).

Table 3 .
Table of the main sources of topographic and geo-thematic data used for mapping.

Table 4 .
Classification of dolines of the Matese massif based on the formation process.