Potential and limitations of LiDAR altimetry in archaeological survey. Copper Age and Bronze Age settlements in southern Iberia

The archaeological use of images and data obtained through devices carried on mobile platforms (such as airplanes satellites) is already one century old. Today, aerial photography remote sensing are routinely used to capture, process analyse evidence present the surface earth, which reflected in a large body literature—see Bewley (1999), Corsi et al. (2013), Palmer Cox (1993), Piccarreta Ceraudo (2000), Riley (1987) Wilson (1982) for former Campana Forte (2001), Lasaponara Masini (2012), Limp (1989), Lyons (1981), Wiseman El-Baz (2007) Wheatley Gillings latter. In last two decades, there has been steady increase usage altimetric analysis based high-resolution techniques aimed at detection architectural elements both above ground underground difficult detect conventional air methods. Prominent among those airborne laser scanning (ALS), which, like terrestrial (TLS), allows measurement microtopographies with level precision not attainable surveying photogrammetric restitution (Challis al., 2008; Chase 2010; Doneus & Briese, 2006; Kühteiber, 2013; Fernandez-Diaz 2014; Fontana, 2022; Gallagher Josephs, Harmon Opitz, Opitz Cowley, Risbøl, Risbøl Gustavsen, 2018). As well-known, this technology uses active LiDAR (light ranging) sensors emit beam polarized infrared light discretized pulses order measure distance between sensor scanned object by time difference pulse emission reception its reflection (time flight, TOF). This offers value relative position regards sensor, turn must be converted absolute coordinates within geodesic system an accurate position, altitude, orientation speed means global navigation satellite (GNSS) differential correction inertial unit (IMU). When fixed airplanes, decimetric levels accuracy achieved, may centimetric helicopters or drones. final result three-dimensional scatter points treated digital 3D-modelling applications create precise models, using first returns produce model (DSM) (filtered) terrain (DTM) (Opitz, 2013). application ALS extensive reconnaissance fairly recent. Over decade, proven extremely useful, particularly densely forested regions northern Europe, American continent Southeast Asia, although Mediterranean environments still limited. After initial phase testing calibration, highly innovative even ground-breaking results have achieved—see, example, Barnes (2003), Briese (2006), Challis (2008), (2010), Crutchley Evans (2016), Canuto (2018), Historic England Guyot (2021), Prümers (2022). Spain, public, freely accessible updated issued periodically since 2014, fostered variety number disciplines, including Archaeology. Spanish survey appeared over 5 years, case studies centred Neolithic monuments, Iron Age cities Roman camps—see Cerrillo-Cuenca López (2020) synthesis. them examined area Portuguese Alentejo region Extremadura known fortified sites ditched enclosures, 1 m resolution DTM from datasets facilities National Geographic Institute (IGN acronym) (Cerrillo-Cuenca Bueno Ramírez, 2019). same IGN were map topography Age, Ancient Medieval Cordoba (Monterroso-Checa 2021), amphitheatre city Torreparedones, well suggest new location Phoenician temple Melkart (Hercules) San Fernando, Cádiz, combining altimetry sonar bathymetry produced Oceanography (Monterroso-Checa, 2017, 2019, 2021). Other very recent examples also include reconnaisance 135 ‘castros’ (hillforts) Galicia, 25 previously unknown ones, buried features, ditches, pathways, field boundaries levelled defensive (Parcero-Oubiña, fresh cartography pre-Roman ‘castro’ Irueña, Salamanca, surveys GIS (Berrocal-Rangel 2017) study military presence fringe Duero basin, where 66 discovered thanks combined different open access geospatial datasets, mainly photography, imagery (Menéndez Blanco 2020). Recent papers looked into possibilities megalithic mounds Galicia (NW Spain) (Carrero-Pazos Carrero-Pazos Vilas Estévez, 2016). A line work looks development algorithms automatic archaeologically relevant mining artificial intelligence, led successful great (Berganzo-Besga 2021; Cerrillo-Cuenca, Thus, while explore relatively wide range mounds, temples, towns camps, date, no attempt made examine most powerful ‘segments’ Iberian Late Prehistory, namely, Copper (c. 3200–2200 BCE) Bronze 2200–850 settlements. general aim paper test what extent (in case, public available existing allow identifications characterizations specific features found settlements dating (CA) (BA). will explained below, only wealth such awaiting examination but us that much can gained build scientific understanding them. prehistoric major stone-walled pervasive phenomenon across southern half Iberia. early stages, starting c. 3000–2900 BCE, was linked gradual sedentarisation communities coupled reduction residential mobility, demographic growth economic intensification, allowing greater accumulation staple surplus. At various locations, ‘fortified’ ‘walled’ interpreted resulting either needs monumentalisation practices (García Sanjuán Murillo-Barroso, Gonçalves 2013: 35; Jorge, 2003; Mataloto Boaventura, 2009: 59). advanced stage, beginning 2200 BCE (Early BA), resulted nucleated well-defended located inaccessible hilltops, demanded construction terraces capable providing horizontal space live on. CA BA presenting identified Iberia (see distribution shown Figure 1). Southwest, they common provinces Huelva, Cabezo de los Vientos, Juré, El Trastejón La Papúa excavated 2011; Hurtado Pérez, García Sanjuán, Hunt Ortiz, Nocete Calvo 2004; Pérez Macías 2019; Sánchez Díaz Badajoz, Blas, Pijotilla, Palacio Quemado, Los Cortinales Castillo Alange explored (Enríquez Navascues, 1990; 1999; Pavón Soldevila Duque Espino, 2014). On side border, along basins rivers Guadiana Tagus, Santa Justa, Vila Nova São Pedro, Leceia Zambujal several decades (Cardoso, 1997). Southeast, numerous showing stone architecture explorations Luis Henry Siret late 19th (Siret Siret, 1890). Of course, would too list here. Although some flourished CA, notably Millares, became more frequent impressive so-called ‘Argaric’ culture Early 2200–1550 BCE)—see Chapman, Aranda Jiménez 2015 good syntheses English. Jaén Granada, Argaric (not always well-published) Peñalosa, Castellón Alto, Terrera del Reloj, Cerro la Virgen Cuesta Negro Huerta Morales Hervás, 2004). province Almería, hilltops provided substantial civil excavated, eponymous site Argar others Fuente Álamo, Oficio, Vermeja, Lugarico Viejo, Picacho Gatas (Gilman Thornes, 1985). these sites, steep slopes often surrounded acropolis, sometimes cisterns other infrastructures (Molina González Cámara Serrano, Numerous neighbouring Murcia province, Placica Caravaca, Covaticas, Cuchillo, Morra Moro Ifre (Eiroa, 2004), as, recently, Bastida Almoloya. With than 4 ha stepped houses areas 10 70 m2 built, largest Southeast. perimeter, protected three sides cliffs, stands out wall could stretched 300 m, negotiating up 40%. sector 3 wide, bastions project 3.50 external face separated 2.80 4.70 m. Considering volume collapsed materials, original height lowest end wall, ‘covering gate’ flanked forts, make narrow 1.5 corridor (Lull Further north, Mancha (the Central Plateau), dozens described (Ruiz Taboada, 1997; Fernández-Posse 2007; etc.) grouped types referred literature ‘motillas’ Las Cañas, Romeros, Retamar, Acequión Azuer), ‘morras’ ‘castillejos’ (El Quintanar, Encantada, Acebuchal, Dornajos, Recuenco, Alberquillas Romeral). applies ‘Valencian BA’, central coast (Tarradell, 1961), seen Antón Orihuela, Callosa Segura, Lloma Betxí, Tabaiá, Muntanya Assolada Terlinques (Pedro Michó Martí Oliver, Altogether, share underlying locational preferences Iberia, considerably smaller southwest southeast, stone-made appears less frequently scale, sets apart patterns Sierra Morena region. Given that, throughout third second millennia show well-defined topographic caused large-scale intrinsic them, departs assumption contribute planimetry their morphology features. LIDAR examining potential limitations future research management. Strictly, assess ability identify locate ‘new’ sufficiently demonstrated 20 years (as briefly discussed above). Instead, it possible characterize matching superior quality achieved means, principally, fieldwalking (obviously, cannot match attained excavation hand drawing photogrammetry). mind, selected knowledge previous fieldwork, excavation, available. involves series highlands southwestern Spain (Figure 2). Between 1980s 1990s, University Sevilla developed (northern Huelva Sevilla). part said project, seasons undertaken discovery characterization multiple settlements, II Trastejón, 1988, 1990 1994 Mondéjar Fernández Quincoces, Romero Bomba, Rivera Jiménez, 2011). fieldwork capture executed team (University Sevilla) criteria methods, therefore, unity intensity sites. Up 36 documentation exists form plans, photographs written descriptions, improve expand records. Specifically, establish parameters size, perimeter associated stone-built walls gates, towers, terraces, dwellings streets roads, examined. if surpass mapped basis direct observation fieldwalking, then method safely analogous geographic settings principle, management purposes, lead creation inventories legally LiDAR-based alone (or without) ulterior truthing. technologies consolidation methodology well-tested phases, processes (Adamopoulos Rinaudo, 2020; Lozić Štular, usual workflow includes processing, analysis, interpretation representation data. turn, each phases incorporates processes, obligatory optional 3). For study, we captured 2014 density p/m2, reduced 0.32 p/m2 once ground-classified, rasterized DTMs pixel size open-source programme LASTools visualized relief visualization toolbox (RVT) 3D module QGIS named Qgis2threejs interpret structures. information normally winter, minimize ‘masking’ effect vegetation. Flight altitudes oscillate 200 drone-based projects (UAVs), 650 3000 frequencies 45 500 kHz. altitude frequency settings, point densities ranging 0.5 21 per square metre differences official run state agencies multipurpose coverage (between 8 Netherlands, example) tailor-made achieve 16 (Table hand, barely influenced flight root-mean-square error (RMSE) Z reaches cm, whereas account induced GNSS IMU systems, bigger altimeter independent altitude. view (FOV) usually limited maximum 50°, given higher angles, probability hitting low. Once captured, subjected control, geometric correction, transformation datum, georeferencing coordinate system, colour attribution orthophotos, differentiation classification returns, organization squares compression LAS LAZ formats (Lorite Martínez 2017). our 100 × 40 km defined Centre Information (CNIG acronym),1 supplies coverages framework Plan Aerial Orthophotography (PNOA), 6-year periodicity whole country. CNIG LAZ-format files scatters arranged 2 cells, automatically classified FWF coloured RGB cm orthophotos. being free, licence copyright cession reutilization any lawful purpose, condition recognizing acknowledging source data, citing authorship. product funded Cartographic System, recognition expressed © LiDAR-PNOA CC-BY 4.0 scne.es. technical specifications PNOA's set concerning sensors, cartographic satisfy (IGN, According specifications, orthometric EGM08 reference geoid, UTM projections zones, density—or series—and minimum RMSE precision, average. 50° angle, kHz scan frequency, ≤1.41 spacing, four discrimination vertical least 8-bit radiometric lower 30 X, Y These regarded standards, basically satisfied Spain. western Morena, provides better quality. fact cases derive raster resolution. addition, average 0.99 below 0.7 Table shows processed study. full-waveform (FWF) signal decomposition echoes it, assigning objects differentiating first, intermediate returns. according type (Doneus 2020), return gap used. differentiate buildings high vegetation, vegetation index (NDVI), calculated imagery, addition. automatized connecting nearest subsequent errors done manually Filtering obtain bare removed issue. Scarcity soil imprecision classifications become sensing. LiDAR, 6.3 million 20.7 corresponded Class (ground), 4.7 assigned rest, almost half, 12 (overlay), 7 (noise) (unclassified), causes lost. practice, 30% usable peaks 0.12 0.77 p/m2. It important note when (ground) fell 0.25 all cases. processed, downloading national agencies, divided (by comparison, England) typically ground-classified eliminate and/or buildings. sufficient depending classification, labelled 6 (building) contain (low vegetation) (Costa-García 2017; Costa-García Fonte, Depending level, (Opitz had allowed interpolation if, instead rasterizing values K-Nearest Neighbours Algorithm (KNN) algorithm, derived tridimensional triangulated irregular network (TIN) (Štular LASTools, integrated package QGIS, calculate GeoTIFF format. Starting DTM, entire algebra. variation details, automatizing demand visual (Verschoof-van der Vaart Microtopography-enhancing methods simple hillshading (HS), complex calculations slope aspect. Among latter sky factor (SVF), openness, local (LRM), principal components (PCA), dominance (LD), cumulative visibility (CV), multiscale integral invariants (MSII) Laplacian-of-Gaussian (LoG) (Bennett 2012). Especially, red image maps (RRIM) technique enhances subtle Slope, Hillshade Differential Openness (Daxer, Thanks needed now precise, azimuth, elevation, filtering radius, directions search (Kokalj Hesse, identification combination necessary, factors feature terrain, land alterations here, terraced relief, useful slope, SVF, PCA RRIM, software RVT lidar (LiVT) techniques, images, whether automatic, identification, inventory algorithm GRASS function r.param.scale, measures convexity direction (Arnau-Rosalén Normally, distinction polygonal burials; linear canals walls; (Mlekuž, photographs, morphology, distinguishing types: structure, ditch, bank slope. conversion shapes format vectorial entities, stored spatial databases systems (GIS) (Gillings many offered semiology. intuitive representations akin natural vision used, shading low illumination, 10° 35° 315° azimuth colours Somrak, virtual environments, recurrent ways display disseminate perspective views, interactive published web (Popovic mapping render 3.2.2 SAGA 2.3.2 7. There single computing handling execution fully environment, Open Toolbox integrates archaeology-specific unified interface, GRASS, libraries Therefore, pieces phases. treatment raw licenced Trimble MX Leica LSS, tool point-cloud generation cleaning free programmes, plug-ins applications. packages MCC-Lidar, Fusion commercial TerraSolid, VR Mesh MARS. visualization, filtering, cutting, union, exportation, ArcGIS, programmes MeshLab, Geomagic XOS FugroViewer. To convert file file, Surfer Whitebox Geospatial Analysis Tools available, exportation TIF, BIL, IMG, CSV ASCII. format, packages, valuable archaeology-oriented tools developed, LiVT. stage automated all, supported generic packages. Visualization perspective-viewing publication, (Geoweb3d) (Qgis2threejs) approach leads significant usability Pro