Geomorphological Mapping and Erosion of Abandoned Tailings in the Hiendelaencina Mining District (Spain) from Aerial Imagery and LiDAR Data

: The Hiendelaencina district in Spain was the most important silver producer in Europe during 1844–1925. At the end of the 20th century, with mines having closed, some waste rock dumps were reprocessed, and the sludge from the ﬂotation process was stored in two tailings ponds. When this activity ceased, the residues began to be eroded and disperse. In this study, the state of degradation of both deposits was evaluated using historical mapping and light detection and ranging (LiDAR) data, incorporated into a Geographic Information System. In the aerial images (1946–2018), mine tailings and their main erosive and sedimentary forms were mapped. Geoforms linked to hydrological (channels, gullies, alluvial cones), wind (eolian mantles), hydric–gravitational (colluvium) and anthropic (motorbike tracks) processes which move sludge into the surrounding areas were identiﬁed. A net loss of 8849 m 3 of sludge, a release of 10.3 t of potentially polluting substances and a high erosion rate of 346 t/ha*year were calculated based on LiDAR data from 2009 and 2014. The ponds show a current high degree of erosion that could increase due to both human activity and the growing frequency of drought and torrential rain periods if stabilization measures are not undertaken.


Introduction
Spain was one of the main producers of metals during the 19th century, and the ceasing of mining activity led to the abandonment of numerous slagheaps, such as waste rock dumps and mining ponds, throughout the 20th century [1,2]. Mining sludge is a mixture of crushed rock, solid residues from the extraction and treatment of the material, washing liquid, and heavy metals resulting from metallurgical flotation processing, and has a medium-to-fine grain size. When the ponds are abandoned, the sludge dries up and is weathered and eroded by water and wind. The mobilization of metals causes a serious environmental impact, especially in the surrounding areas [3][4][5][6][7][8].
Weather conditions are a determinant in the dispersion of pollutants both by water and by wind [3,5,7,9]. Regarding water erosion, torrential rains have great erosive force and can create steep escarpments and deep gullies in the ponds. Wind erosion is significant in regions with a Mediterranean climate characterized by low humidity, which translates into low cohesion of the sludge and the scarce development of vegetation cover. In this context, the airborne transport of the finer fractions of sludge can affect soil, water, and wildlife.  [30,31]). ZEC-ZEPA, Zone of Special Conservation-Zone of Special Bird Protection. The red circle in the regional map locates the study area. CS, Central System. IR, Iberian Range.
Over recent decades, several studies have been carried out in the mining district of Hiendelaencina on mineralization, mining, mining heritage, and tourist use [26][27][28][32][33][34], but none on physicochemical state, stability, or the extent to which the abandoned mine ponds are affecting the area. Reclamation techniques have not been implemented in the  [30,31]). ZEC-ZEPA, Zone of Special Conservation-Zone of Special Bird Protection. The red circle in the regional map locates the study area. CS, Central System. IR, Iberian Range. Over recent decades, several studies have been carried out in the mining district of Hiendelaencina on mineralization, mining, mining heritage, and tourist use [26][27][28][32][33][34], but none on physicochemical state, stability, or the extent to which the abandoned mine ponds are affecting the area. Reclamation techniques have not been implemented in the area to reduce erosion and, in order to evaluate the possible effects on the environment, it is essential to characterize these deposits as a first step. Therefore, the aim of this study is to estimate the state of erosion of the Hiendelaencina ponds by mapping the main geomorphological features and their evolution over time, and estimate the rate of erosion and the release of potentially hazardous elements into the surrounding areas. Although the use of unmanned aerial vehicles (UAVs) and terrestrial images is providing highspatial-resolution 3D models [14,15,17,20], abundant aerial imagery and aerial LiDAR data are available, free of charge, for the study area. Thus, geomorphological mapping was conducted from the 2018 orthophoto and field observations. The areal evolution of the ponds was carried out based on aerial images from 1946 to 2018. The erosion rate was calculated from the volume of sludge obtained from the LiDAR data from 2009 and 2014. The amounts of potentially hazardous elements released were calculated from various physicochemical parameters of the sludge from Hiendelaencina and other mining areas. The Geographic Information System (GIS) software ArcGIS was used to manage and analyze aerial imagery and aerial LiDAR data.

Study Area
The Hiendelaencina mining district was the most important European silver-producing area between 1844 and 1925 [25][26][27][28]. In 1844, thirteen silver veins were discovered and, in the most productive one of them, twelve mines started to operate. Although the origin of the ore was the result of various orogenic processes, silver was deposited mainly in quartz-rich vertical fractures that crossed a Paleozoic gneiss and quartzite basement, which were reactivated during the Alpine orogeny. In the gneisses, acanthite (Ag 2 S), stephanite (Ag 5 SbS 4 ), and kerargyrite (ClAg) were intergrown with barite (BaSO 4 ) and galena (PbS). The American or European amalgamation method was used to extract the silver from the ore, and the wastes were deposited in dumps near the mines. After a succession of prosperous and declining periods, mining came to a standstill in 1925.
Between 1980 and 1990, mining was resumed with the processing of materials from waste rock dumps [26][27][28]. The flotation process was carried out in the vicinity of the old San Carlos mine, 1 km north-east of the town of Hiendelaencina. The resulting sludge was deposited in two ponds, with upstream recharge, on a bedrock of tertiary gravel and clay discordant over the gneisses. Although the ponds are currently abandoned, they have an active drainage system that allows them to achieve high compaction and mechanical stability. The estimated volume of sludge is 300,000 m 3 in the north pond (San Carlos 2) and 330,000 m 3 in the south pond (San Carlos 1) (Figure 1) [29].
Regarding the main geographical features, this region is in the area where the mountain belts of the Central System and the Iberian Range meet. The ponds are specifically located on a foothill with scrubland vegetation at an average altitude of 1085 m, with uniform, less than 15% gradient slopes. The foothills are bordered by the Diógenes and Cal creeks, which are tributaries of the Bornova river which is dammed in this area for water supply and irrigation ( Figure 1). To the west of the ponds, and within the municipality of Hiendelaencina, the Sierra Norte de Guadalajara Natural Park and the ZEC-ZEPA (Zone of Special Conservation-Zone of Special Bird Protection) zone of the Sierra de Ayllón can be found, owing to the significant geodiversity, landscape and botanical value of the mountains [35]. The climate is a warm-summer Mediterranean type according to the Köppen classification (Csb) [36]. The average annual rainfall is around 600 mm, with average annual maximum daily rainfall in May and October (46 mm/day on average) and lows in August. The average annual temperature is 12.5 • C, with maximum temperatures in July and August and minimum temperatures in December and January. The average wind speed is 14.2 km/h, with maximum gusts of 127 km/h [37]. The windiest periods are between October and May. Winds from the north and west prevail, although in dry periods they come from the west, east and south [38].

Time Evolution and Geomorphological Mapping of Mine Tailings
The time evolution and geomorphological mapping of the ponds were carried out by analyzing the aerial images available from the National Geographic Information Center of the Spanish National Geographic Institute (IGN) [30]. Data were collected from 1946 to 2018, with flight scales from 1:43,200 to 1:20,000 and pixel sizes from 0.27 to 1 m ( Table 1). The 1973The -1986The , 1980The -1986The and 2002 (quinquennial flight of IGN) flights were georeferenced with mean residual errors of 0.32, 0.57 and 0.48 m, respectively. The rest of the data were already georeferenced by IGN with RMSE x,y ≤ 1.00 m [30]. The mapping consisted of identifying the contours of the ponds and the landscape geoforms that had been generated on them by the action of geological agents (water, wind, gravity) and by human activity since they were abandoned. Prints of the aerial photographs at a scale of approximately 1:2000 were viewed under a tabletop stereoscope, except for the 1997-1998 flight where the orthophotos were used. All the information was integrated and processed in the ArcGIS Geographic Information System, version 10.7.1 (Esri, Redlands, CA, USA), considering the different format of the data source. Landscape geoforms and mine pond boundaries, identified in the stereo pair images, were digitized and stored in shapefile format and from them sludge mine pond areas were obtained. For geomorphological mapping, the orthophoto of the 2018 flight was used and this interpretation was contrasted with a field survey on 26 July 2021. The perimeter of the ponds was walked to observe the slopes, and east-west parallel transects were also made to observe features from both the north and south ponds. In addition, biological activity data were collected, and a photographic inventory was made.

Volume, Loss-Gain and Erosion Rate of Sludge and Toxic Element Mobilization
To map and quantify the sludge volume variations between 1954-2009 and 2009-2014, altitude data from the 1954 topographic map (1:25,000 scale and 10 m contour interval) [39] and from the 2009 and 2014 LiDAR flights were used (point density 0.5 points/m 2 ). The horizontal (RMSE x,y) accuracy of the LiDAR data is ≤0.30 m, and the accuracy for the vertical (RMSE z) component is ≤0.15 m [30]. The ArcGIS software allowed us to import the x,y,z coordinates of the LiDAR point cloud and convert them into three-dimensional models, from which DTMs were derived. LASTools were used to convert the laz LiDAR data to las format. After that, the LAS dataset vegetation was removed and las format was converted to raster.
In order to calculate the sludge volume in 2009, a reference surface was estimated from the visual analysis of the stereoscopic images of the 1946 and 1956 flights, and it was found that the sludge was deposited on a summit area with flat topography. This area is surrounded by a 1110 m contour line in the georeferenced 1954 topographic map [39]. Therefore, a constant height raster of 1110 m was generated and the 2009 LiDAR data were superimposed onto it. This constant elevation DTM was used as a reference surface to calculate the volume of sludge generated until 2009.
The variation in sludge volume between 1954-2009 and 2009-2014 was estimated from the raster data using the ArcGIS cut and fill tool. Analysis of volume variation between the mine pond border and the surrounding area and between the north and south pond boundary was carried out using polygonal layers of the specific areas. The border between both mine ponds was based on the change in slope of the 2014 LiDAR elevation map. The layers were applied as a mask by using the cut-and-fill tool ( Figure 2). The rest of the volumes, for which no mask was generated, were calculated by mathematical relation (e.g., surrounding area contour = 2009 contour − 1983 contour). The uncertainty of the volumes measured was calculated using the arithmetic mean and standard deviation of the data with the ArcGIS statistical tool. This provided the standard error presented here. The error of the volumes estimated by mathematical relations was the sum of the errors of these values. was estimated from the raster data using the ArcGIS cut and fill tool. Analysis of volume variation between the mine pond border and the surrounding area and between the north and south pond boundary was carried out using polygonal layers of the specific areas. The border between both mine ponds was based on the change in slope of the 2014 LiDAR elevation map. The layers were applied as a mask by using the cut-and-fill tool ( Figure 2). The rest of the volumes, for which no mask was generated, were calculated by mathematical relation (e.g., surrounding area contour = 2009 contour − 1983 contour). The uncertainty of the volumes measured was calculated using the arithmetic mean and standard deviation of the data with the ArcGIS statistical tool. This provided the standard error presented here. The error of the volumes estimated by mathematical relations was the sum of the errors of these values. Finally, the erosion rate of the ponds and the amount of potentially released toxic elements were calculated through the volumetric balance between 2009 and 2014. To obtain the sludge mass, the dry density was estimated from the metallic content and mineralogical composition of the sludge from Hiendelaencina. In addition, published values of porosity and grain density in sludge from similar mining areas were also used.

Geomorphological Mapping of Mine Tailings
Geomorphologies and deposit characteristics of hydrological, gravitational, wind, biological and anthropic processes were identified in both ponds ( Figure 3). Water processes have formed four abrupt gullies in each of the slopes of the north pond which connect to the surface with small river valleys of ephemeral streams. In the south pond, there is an emerging gully on the south-east slope. The eroded sludge transported by water is de- Finally, the erosion rate of the ponds and the amount of potentially released toxic elements were calculated through the volumetric balance between 2009 and 2014. To obtain the sludge mass, the dry density was estimated from the metallic content and mineralogical composition of the sludge from Hiendelaencina. In addition, published values of porosity and grain density in sludge from similar mining areas were also used.

Geomorphological Mapping of Mine Tailings
Geomorphologies and deposit characteristics of hydrological, gravitational, wind, biological and anthropic processes were identified in both ponds ( Figure 3). Water processes have formed four abrupt gullies in each of the slopes of the north pond which connect to the surface with small river valleys of ephemeral streams. In the south pond, there is an emerging gully on the south-east slope. The eroded sludge transported by water is deposited at the end of the gullies forming alluvial cones (1-2 in Figures 3 and 4). On the slopes of the ponds, the streams flow in rills (4 in Figures 3 and 4). Due to the way in which the ponds were enlarged, two depressions were formed in their central areas where temporary flooding occurs. In the dry season, saline efflorescence precipitates on the sludge (3 in Figures 3 and 4).     The joint action of hydrological and gravitational processes on the slopes of the ponds has led to their retreat, which is evident by the presence of trees with exposed roots (2-3 in Figures 3 and 4). The south-west-oriented slopes of the south pond are stepped due to the differential erosion of alternating coarser-and finer-grained sludge levels. The eroded sludge is deposited around the perimeter of the ponds, forming a colluvium, the extent of which is greater towards the south (5 in Figures 3 and 4).
The parts of the pond that lack vegetation cover are exposed to the wind during dry periods. This has led to the formation of 2-3 cm thick eolian mantles on which ripples develop (5 in Figures 3 and 4). The grain size of the sludge, based on the USDA textural classification, varies from loamy sand (predominantly medium sand 0.5-0.2 mm or fine sand 0.2-0.05 mm,~40% by weight) to sandy loam (~48% by weight of silt fraction 0.05-0.002 mm). However, the texture of the eolian mantle deposits is sandy and the majority of the fraction is medium sand (~59% by weight on average). The ripples in the western deposit of the north pond have a N132 • E average orientation, while in the eastern deposit of the south pond the ridge direction varies from N112 • E to N198 • E. These sands are in turn eroded away during the rainy season and transported by ephemeral currents to the alluvial cones, colluvium or depressed areas.
In addition to these geological processes, human activity has modified the ponds, altering the stream processes. Motorbike tracks on the slopes generate preferential paths, which in some cases are across the steepest slope (2-3 in Figures 3 and 4). The surface of both ponds is used as a dirt track, compacting the surface and creating depressions under the tracks, as well as mounds due to the accumulation of material raised by motorcycles and quad bikes from riding (6 in Figures 3 and 4). On the other hand, based on the evolution of the mine tailings over time ( Biological activity also moves a small amount of sludge. There are sand martin nests on the slopes of the gullies, bee-eater nests on the more vegetated slopes, and small mammal burrows (3 in Figures 3 and 4). These animals break up and weaken the slopes. On the other hand, in the areas of temporary flooding, as well as in the loose materials of the colluvium, flora species typical of wetlands have developed (7 in Figures 3 and 4). Vegetation cover in the ponds occupies 29.5% of the surface, with herbaceous and shrub plants predominating in the south pond and trees in the north pond. Biological activity also moves a small amount of sludge. There are sand martin nests on the slopes of the gullies, bee-eater nests on the more vegetated slopes, and small mammal burrows (3 in Figures 3 and 4). These animals break up and weaken the slopes. On the other hand, in the areas of temporary flooding, as well as in the loose materials of the colluvium, flora species typical of wetlands have developed (7 in Figures 3 and 4). Vegetation cover in the ponds occupies 29.5% of the surface, with herbaceous and shrub plants predominating in the south pond and trees in the north pond.  Table 2). From 1983 to 2018, the area occupied by sludge increased by 31.5% to 76,161 m 2 . In all cases, the estimated error is below 1 m (RMSE x,y ≤ 1.00 m [30]). After the closure of mines at the beginning of the 1990s, changes are recognizable in the new terrain created by the ponds, which are associated with geological, biological and anthropogenic processes, and which progress over time. In the 1997-1998 flight, rills on the southern slope of the south pond, two gullies on the north-east slope of the north pond, isolated trees, and the extraction of sludge in the southwest corner of the north pond are identified (quarry in Figure 3). The area covered by sludge in 1997-1998 increased by 11.5% compared to the perimeter of the mine ponds when they were operational (1983). This is the largest increase detected between two consecutive flights (Table 3). In 2002, the formation of the north-west slope gully stands out, connected with a diffuse stream towards the surface of the deposit and with the alluvial cone on the perimeter. In the south pond, colluvium/eolian mantle deposits are observed towards the south-east, and the more depressed central part begins to be reclaimed by herbaceous-type vegetation. In comparison with the previous flight, the surface occupied by sludge has increased by 5.4%. The landforms described above were still in place in 2006, and on the surface of the ponds there are pale patches that could be related to the precipitation of saline efflorescence during summer. However, the most significant changes from previous years are the motorbike tracks along the perimeter of the south pond surface. In the 2009 image, traces of tracks can be seen on the north pond and on the slopes, and a greater number of streams, the formation of the gully on the slope of the quarry, and the greater extension of the colluvium/eolian mantle in the south-east are also recognized. There is almost no change in the area occupied by sludge in the period 2002-2009. In the 2012 image, the only relevant issue is the change in the hillside in the north pond quarry due to a new sludge extraction. The south-east colluvium/eolian mantle reaches its maximum extent in 2015, and some runoff is identified across it. An average 2.1% increase in sludge area is obtained between 2009-2012 and 2012-2015. Lastly, in 2018 there was a strong erosion in the quarry slope gully, and the rest of the elements already described in the previous subsection can also be recognized. However, sludge surface decreases by 3.3% compared to the previous flight.

Formation and Evolution of Mine Tailings
Overlapping of the sludge contours of all images shows that the residue movement has not been homogeneous over time ( Figure 6). The west, north-west and north-east hillsides have maintained their position as roughly constant, except for the occasional deposition beyond the initial perimeter of the ponds that is associated with stream processes and human activity. However, on the south-east and south hillsides, there has been a greater displacement of sludge due to the formation of colluvium/eolian mantle. Some areas show a higher settlement of sludge removed in past, e.g., the alluvial cone at the north-west hillside of the north pond. This is due to the growth of vegetation on the sludge after deposition. While the area occupied by the sludge was changing, the position of scarp hillsides receded by an average retreat of 3 m from 1983 to 2018, reaching up to 10 m in the gully areas.
Remote Sens. 2022, 14,4617 north-west hillside of the north pond. This is due to the growth of vegetation sludge after deposition. While the area occupied by the sludge was changing, the p of scarp hillsides receded by an average retreat of 3 m from 1983 to 2018, reachin 10 m in the gully areas. The mixed action of natural (surface erosion) and anthropic processes not on ifies the extent occupied by the sludge, but also determines the value of the pond (Figure 7). The hillsides of the north pond had slopes greater than 60% in 2009 t increased by 2014. In contrast, the hillsides of the south pond had slopes in the r 30-60%, locally steepening to more than 60%, which tended to decrease. These diffe together with the greater height of the north pond compared to the south pond, the reason why the north pond has steep gullies on its slopes, which are conne small valleys on the surface, while the south pond has only rills. In that period, the were excavated by the ephemeral streams, the inland depressions with temporary ing deepened and even slightly increased in size, and the tracks caused by mot also became more pronounced. The mixed action of natural (surface erosion) and anthropic processes not only modifies the extent occupied by the sludge, but also determines the value of the pond slopes ( Figure 7). The hillsides of the north pond had slopes greater than 60% in 2009 that had increased by 2014. In contrast, the hillsides of the south pond had slopes in the range of 30-60%, locally steepening to more than 60%, which tended to decrease. These differences, together with the greater height of the north pond compared to the south pond, may be the reason why the north pond has steep gullies on its slopes, which are connected to small valleys on the surface, while the south pond has only rills. In that period, the valleys were excavated by the ephemeral streams, the inland depressions with temporary flooding deepened and even slightly increased in size, and the tracks caused by motorbikes also became more pronounced.

Volume, Loss-Gain and Erosion Rate of Sludge
To quantify the amount of sludge movement since mining activity ceased, and since there is no detailed altitudinal information prior to 2009, it has been assumed that the eroded material was deposited in the surrounding area closest to the ponds. In 2009, there was 632,582 ± 754 m 3 of sludge distributed over an area of 68,068 m 2 . This volume and those of each pond (Table 2) are equivalent to the ones listed in the IGME catalogue file [29]. Considering the initial boundary of the ponds in 1983 (57,932 m 2 ), the amounts of sludge inside the 1983 contour and in the surrounding area in 2009 were 598,941 m 3 and 33,642 m 3 , respectively. Therefore, between 1983 and 2009, 5.3% of the total volume of sludge was eroded from the ponds and deposited into the surrounding 10,136 m 2 , increasing the occupied area by 17.5% ( Figure 6). The volume of sludge movement from the south pond to its surrounding zone (5.8%) is higher than that from the north pond (4.7%). Similarly, the increase in the area occupied by sludge around the south pond (19.0%) also exceeds that of the north pond (15.1%).
Based on the altitude data collected by LiDAR flights, the distribution of sludge loss and gain between 2009 and 2014 has been quantified and mapped (Table 4 and Figure 8

Volume, Loss-Gain and Erosion Rate of Sludge
To quantify the amount of sludge movement since mining activity ceased, and since there is no detailed altitudinal information prior to 2009, it has been assumed that the eroded material was deposited in the surrounding area closest to the ponds. In 2009, there was 632,582 ± 754 m 3 of sludge distributed over an area of 68,068 m 2 . This volume and those of each pond (Table 2) are equivalent to the ones listed in the IGME catalogue file [29]. Considering the initial boundary of the ponds in 1983 (57,932 m 2 ), the amounts of sludge inside the 1983 contour and in the surrounding area in 2009 were 598,941 m 3 and 33,642 m 3 , respectively. Therefore, between 1983 and 2009, 5.3% of the total volume of sludge was eroded from the ponds and deposited into the surrounding 10,136 m 2 , increasing the occupied area by 17.5% ( Figure 6). The volume of sludge movement from the south pond to its surrounding zone (5.8%) is higher than that from the north pond (4.7%). Similarly, the increase in the area occupied by sludge around the south pond (19.0%) also exceeds that of the north pond (15.1%).
Based on the altitude data collected by LiDAR flights, the distribution of sludge loss and gain between 2009 and 2014 has been quantified and mapped (Table 4 and Figure 8). Within the original perimeter of the ponds (1983 contour), the volumes of lost and gained sludge are similar, but the areas affected are larger in the south pond. The surrounding zone of the south pond has the highest values for calculated volumes and areas with loss and gain, but this area increased in size from 2009 to 2014. The combined action of the erosive agents over six years has produced a net loss of 8849 m 3 of sludge, with a retention of 282 m 3 in the surrounding area of the ponds.

Toxic Elements Mobilized into the Environment
Once the net loss of sludge is known, it is possible to estimate the tonnes of potentially hazardous metals that have been eroded and transported to the surrounding areas of the ponds during this 6-year period. With this aim in mind, we used the metal contents and mineralogical composition of the ponds obtained by X-ray diffraction and published by [40]. In this work [40], tailings were collected with an Eijkelkamp soil core manual sampler for undisturbed samples, with a known volume and diameter. Twenty samples were collected with a vertical constant spacing of 50 cm from the pond surface to a depth of 9.5 m along the gully vertical hillsides at the SW corner of the north mine pond. This

Toxic Elements Mobilized into the Environment
Once the net loss of sludge is known, it is possible to estimate the tonnes of potentially hazardous metals that have been eroded and transported to the surrounding areas of the ponds during this 6-year period. With this aim in mind, we used the metal contents and mineralogical composition of the ponds obtained by X-ray diffraction and published by [40]. In this work [40], tailings were collected with an Eijkelkamp soil core manual sampler for undisturbed samples, with a known volume and diameter. Twenty samples were collected with a vertical constant spacing of 50 cm from the pond surface to a depth of 9.5 m along the gully vertical hillsides at the SW corner of the north mine pond. This procedure, also used in the La Naya mine pond [23], allows us to determine variations in element contents from the surface to the bottom and, therefore, work with representative and average values.
From the mineralogical composition of the sludge [40], the density of the solid phase was calculated (Table 5). Porosity and grain density data from sludge from similar ponds in Sweden [41], Ireland [42] and South Africa [43] reveal the existence of a linear correlation between both parameters (Figure 9). Non-linear models were also tested but they did not provide significantly better results. From the linear equation and our calculated grain density values, we obtained the porosity values that were needed for the subsequent calculation of dry density values (Table 5). density values, we obtained the porosity values that were needed for the subsequent calculation of dry density values (Table 5). Figure 9. Grain density vs. porosity plot and parameters of the linear fit. The fit has been computed with a 95% confidence (blue zone). Data from Mansson [41], Quille and O'Kelly [42] and Mhlongo [43]. Finally, with this dry density value and the volume of net sludge loss (Table 4), the total mass eroded from the ponds was calculated, followed by the tonnes of potentially hazardous metals eroded. A value of 4.2 ± 0.2 t of metals was calculated for the north pond, and for the south pond 6.1 ± 0.4 t of metals were eroded and deposited in the surrounding areas. Particularly noteworthy are the values of As (2.7 ± 0.16 t), Pb (3 ± 0.17 t) and Zn (2.9 ± 0.16 t).
Considering the estimated density for these sludge deposits of 1,705 ± 0.046 t/m 3 and a pond surface of 7.27 ha, the erosion rate during this period was 346 ± 9 t/ha*year. The loss of sludge on the surface of the ponds is related to the erosive action of the surface streams and the wind, increased by the breakdown of material by motorbike riders. On the surface of the south pond there are areas of sludge gain build-up ( Figure 8). Those areas that are close to the north pond coincide with mounds created by motorbikes, while in the areas bordering the central zone of temporary flooding, particles are retained by the higher sludge moisture content and vegetation cover. Sludge loss on the slopes is caused by surface runoff processes (in sheets, rills, gullies) and gravitational processes. Sludge gain in the surrounding zone is associated with sedimentation in the alluvial cones and colluvium. Figure 9. Grain density vs. porosity plot and parameters of the linear fit. The fit has been computed with a 95% confidence (blue zone). Data from Mansson [41], Quille and O'Kelly [42] and Mhlongo [43].
Finally, with this dry density value and the volume of net sludge loss (Table 4), the total mass eroded from the ponds was calculated, followed by the tonnes of potentially hazardous metals eroded. A value of 4.2 ± 0.2 t of metals was calculated for the north pond, and for the south pond 6.1 ± 0.4 t of metals were eroded and deposited in the surrounding areas. Particularly noteworthy are the values of As (2.7 ± 0.16 t), Pb (3 ± 0.17 t) and Zn (2.9 ± 0.16 t).
Considering the estimated density for these sludge deposits of 1,705 ± 0.046 t/m 3 and a pond surface of 7.27 ha, the erosion rate during this period was 346 ± 9 t/ha*year. The loss of sludge on the surface of the ponds is related to the erosive action of the surface streams and the wind, increased by the breakdown of material by motorbike riders. On the surface of the south pond there are areas of sludge gain build-up ( Figure 8). Those areas that are close to the north pond coincide with mounds created by motorbikes, while in the areas bordering the central zone of temporary flooding, particles are retained by the higher sludge moisture content and vegetation cover. Sludge loss on the slopes is caused by surface runoff processes (in sheets, rills, gullies) and gravitational processes. Sludge gain in the surrounding zone is associated with sedimentation in the alluvial cones and colluvium.

Discussion
The combination of UAVs and terrestrial photogrammetric methods over recent years is revealing as an accurate procedure to obtain 3D MDTs that are the closest approximations of surface morphology [14,15,17,20]. With this approach, more than one z-coordinate per xy set can be obtained and, thus, complex topographies, such as the ones in gullies with undercut walls, are detected and modelled with detail. On the other hand, the spatial and temporal coverage of satellite data is increasingly available in some countries, and 2.5D models can be obtained over large territories [18,44]. In this study, we have estimated the erosion dynamics of the Hiendelaencina abandoned mine tailings from a methodological approach, combining the use of aerial photographs, orthophotos and satellite data [30]. Although there are a scarcity of LiDAR data to estimate sludge erosion rate accurately, the two DTMs obtained from satellite imagery are helpful for a preliminary estimation. Furthermore, erosion rates are usually higher in mining deposits than in natural landscapes [12,21]. Thus, large variations in elevation can be detected in the terrain over a period of a few years. Field work was essential to recognize rills from motorcycle tracks on some slopes or mapping eolian mantles, and check that there were no significant differences in landscape forms from those already observed in the photointerpretation. Therefore, surface processes affecting a mine site can be better understood with the integration of different remote sensing imagery and data. However, erosion dynamics cause the dispersion of pollutants in metallic mineral districts, and tailings characteristics can also be used to identify environmental pollution problems [1][2][3][4][5][6][7][8][9]. We obtained the amount of metals dispersed to the surrounding areas of the mine ponds by using their physicochemical features.
Several studies demonstrate the environmental impact of the erosion of mining deposits and the displacement of tailings outside different mining areas through surface runoff [4,5,10]. The amount of material transported depends on variables such as grain size and amount of precipitation [45]. In the Cartagena-La Unión mining district (SE Spain), 2.14 kg/m 2 of mining sludge were lost after a 60.5 mm rainfall event [8]. However, in the Santa Engracia mine (Guadalajara, Spain), water erosion processes have been identified with lower precipitation values (8.2 mm in a rain event on 25 August 2007) [46]. The highest rainfall episodes in Hiendelaencina are not very severe (46 mm/day on average) [38], but they have generated very pronounced erosive geoforms (gullies, valleys, rills) in the mining ponds since they were abandoned, as well as hillside scarp retreating and the formation of colluviums (Figures 3-5). High amounts of toxic and mobile elements, such as arsenic, have been found in the sediments of the Cal and Diógenes streams, confirming the effective transport of tailings by surface runoff from mines and slagheaps [47].
Although no erosional landforms associated with gravitational processes on the hillsides have been identified in the Hiendelaencina ponds, the colluvium may also include material accumulated by gravity. These processes should be more noticeable on the steep slopes of the gullies and hillsides of the north pond ( Figure 4). Catastrophic gravitational processes are responsible for the breakage of pond dams with steep slopes and large volumes of sludge, as seen in the case of the Aznalcóllar mining pond [48]. However, the smaller size of the two Hiendelaencina deposits and their good mechanical stability reduce the probability of occurrence of this type of phenomenon.
Much of the finer mining waste in semi-arid areas is transported over long distances by high winds [3,4,7]. The collisions that occur between particles due to wind saltation modify the cohesion of the sludge, allowing the separation of the finer fractions that include heavy metals and are displaced by winds of 14.4-25.2 km/h [49]. The effective wind that can move sludge from the San Carlos ponds to the urban center of Hiendelaencina occurs from mid-June to September, with a north-west to south-east orientation and average speeds of 12.9 km/h, which can reach 25 km/h in June [38]. The ridges of eolian mantle ripples are parallel to the slope of the hillside from which they are formed (Figure 4). This arrangement is consistent with small shifts in the prevailing wind directions around the ponds. On the other hand, the dominantly sandy grain size of the sludge with the small presence of finer fractions suggests that the wind has blown the latter away. If this released fraction contains metallic particles, this could pose a health risk to the inhabitants in the surrounding area.
Although some plants can extract metals from the soil and bioaccumulate them, native species colonizing the studied mining ponds are not effective in fixing these elements [50]. Vegetation improves water retention and soil structure, and helps reduce erosion processes [51], but in semi-arid climates a minimum cover of 50% is needed, especially of the herbaceous type [52]. The lower vegetation cover in the Hiendelaencina ponds (29.5%) would, therefore, be insufficient to prevent sludge erosion. However, the distribution of vegetation is not consistent and its contribution could be varied. There is a predominance of zones with isolated growing trees and shrubs, but on steep slopes (>30 • ) their roots usually become exposed by erosion. On the other hand, there is a higher density of vegetation, including herbaceous species, in the central depressed area of the south pond, which is subject to temporary flooding and, therefore, better retains the mobilized tailings. Steep hillsides in abandoned mines register maximum erosion rates due to the higher surface runoff along rills and gullies [21,53]. Recent studies suggest geomorphic mining reclamation to mitigate the loss of mine waste [19]. The slope of linear hillsides should be lessened to 32% or transformed to a concave profile, and the surface should be covered with topsoil. This integrated management of topography and soil allows spontaneous vegetation colonization and the stabilization of hillsides.
The grain density, porosity and dry density values obtained for the sludge from the Hiendelaencina mining ponds are similar to those published by other authors for this type of mining sludge [23,54]. Although the content of potentially hazardous metals in these ponds is not significantly high when compared to similar deposits in other mining districts in Spain [6,23,55,56], significantly high net amounts of mining sludge lost through erosion have been found (more than 10 t of metals). This is due to the high rate of erosion calculated from these deposits (346 ± 9 t/ha*year). The estimated erosion rate in abandoned deposits in semi-arid areas ranges between 100 and 150 t/ha*year (NW Mexico [57], SE Spain [10]). The higher value calculated in Hiendelaencina may be linked to the coarser granulometry of the sludge and to the specific climatic, geomorphological, and anthropic characteristics of the region. Higher annual rainfall can lead to a larger amount of sludge movement. As the ponds are in a flat, elevated and completely open uphill area, the entire sludge perimeter is exposed to the combined effects of water, wind and gravity. In addition, the recreational use of these deposits for motorbike racing strongly alters the cohesion of the sludge, increasing erosion. Therefore, considering climate change, where periods of drought and torrential rains are likely to become increasingly intense, measures to restore and/or stabilize the ponds should be undertaken to reduce the high rate of erosion and the dispersion of potentially toxic metals.

Conclusions
The study of geomorphological processes in abandoned mining sludge ponds and their time evolution, together with the estimation of erosion rates from aerial imagery and LiDAR data, has revealed the state of degradation of these deposits, as well as the movement of potentially toxic materials into the environment, and provides information for planning future restoration. In the tailings of the Hiendelaencina mining district there are hydrological, wind, gravitational and anthropic processes that erode the sludge and deposit it at the foot of the hillsides. Although vegetation cover has been developing on the surface of the ponds, its low density and irregular distribution are not enough to prevent particle movement. The climatic, geomorphological and anthropic characteristics of the region determine the high rate of erosion recorded, and measures should be undertaken to prevent potential damage to population living nearby.