This article is part of the SECREI project conducted by the UR 3795 - GEGENA and co-financed by the University of Reims Champagne-Ardenne and the Communauté urbaine du Grand Reims. The authors would like to thank the Association des Amis du Fort de Nogent-l’Abbesse (AAFNA) for their valuable support and Guillaume Elmerich for his kind assistance with the translation of this paper.

1. Introduction

1It is only in the last two decades that the geomorphological traces of ancient conflicts ­ also referred to as war landforms (De Matos Machado et al., 2019) or landforms of war (LOWs; Thestorf and Makki, 2022) ­ have increasingly attracted interest among geomorphologists, particularly within the framework of the archaeogeomorphological approach (Thornbush, 2012, 2013) applied to warfare (Hesse, 2014). In French literature, these traces are termed "polémoformes" (polemoforms; De Matos-Machado et al., 2016, 2019; Taborelli et al., 2017a, 2017b; De Matos Machado, 2018, 2024; Taborelli, 2018; Devos et al., 2021; Ancelin et al., 2023), a concept whose origins lie in the pioneering studies of Amat (1987, 2001) and of Boulanger (2001) on "polémo-paysages" (polemo-landscapes or conflict/war landscapes).

2In Western Europe, the most extensively studied landforms predominantly date back to WW1 (World War One), a conflict that took an industrial dimension (Archer et al., 2002) whose study received renewed attention on the occasion of its centenary (Weinrich and Patin, 2022). This conflict, among the most morphogenic in history due to the intensity of combat, the caliber of artillery and the amount of shells fired (approximately one billion; Prentiss, 1937), left deep marks on the Western Front landscape, stretched over 700 km from Belgium to Switzerland (Amat, 2001).

3In the late 2000s, Hupy and Schaetzl (2006; 2008), Bausinger et al., 2007 and Certini et al., 2013 have explored the geomorphological impacts (bombturbation), pedological disturbances (pedoturbation) and geochemical effects caused by explosive munitions, which are responsible for creating craters of varying diameters as well as for soil disruption. Mine warfare, practiced since antiquity to besiege fortified places (Triolet J. and Triolet L., 2011) and significantly employed during WW1, is undeniably the most morphogenic conflict-induced disturbance, producing huge craters, overlapping structures or crater chains. Such consequences can be found on 69 sites along the Western Front (Taborelli et al., 2017a, b; Taborelli, 2018).

4In 2014, Hesse's pioneering article highlighted the value of a landscape-based approach employing DTMs (Digital Terrain Models) to analyze the polemoforms shaped by past. This methodological framework has since been validated through various studies addressing their geomorphological impacts (e.g., De Matos-Machado, 2018; Taborelli, 2018), morphometric characteristics (e.g., De Matos-Machado et al., 2019), and spatial organization (Taborelli et al., 2017a), taking advantage of high-resolution topographic data acquired via ALS (Airborne Laser Scanning). These advanced techniques are also instrumental in the nuanced management of “dark heritage” (Hesse, 2014), which is often perceived negatively due to its association with recent violent conflicts (Biran et al., 2011). Detection and cartographic inventory represent an essential preliminary step in the patrimonialization process of polemoforms (Hesse, 2014), including the determination of their degree of erasure. For example, regarding WW1 trenches, the erasure rate is of approximately 40 % within the Forest of Verdun (De Matos-Machado, 2018) and of 30 % within the small massif of the Brimont butte, north of Reims (Ancelin et al., accepted a), while it tends to be absolutely complete in agricultural areas (Taborelli, 2018) due to post-war land reclamation efforts (Devos et al., 2021).

5Similar to conflict archaeology, research in conflict geomorphology has thus far focused on landforms associated with battlefields. However, the nature of polemoforms varies along a polemological gradient (De Matos-Machado, 2018) as well as their location within defense networks (Devos et al., 2015, 2019), sometimes including areas far from the front, such as the Hinterland (Bracke et al., 2018) or the so-called "Rear" and "Etapes" zones. Their formation process also differs (Hesse, 2014), encompassing features linked to "combat without war" such as training activities (e.g., instruction camps responsible for training soldiers in trench warfare tactics; Stichelbaut et al., 2009; Brenot et al., 2022), or to logistical support infrastructures for the war effort (e.g., military communication routes, military hospitals, ammunition depots, rest camps, supply centers, geomaterial extraction sites; Passmore et al., 2013; Caps Tunwell et al., 2016; Perarnau et al., 2022). Their origin can also be defensive, particularly in the context of permanent fortifications, which account for the majority of polemoforms from pre-industrial era due to the absence of high-power morphogenic weaponry (Hesse, 2014; fig. 1A-E). While these structures may not always result from violent conflicts, they often reflect a perceived threat or a desire to project power (Armit et al., 2007).

6The "Séré de Rivières" fortified system, erected by France as a response to the country’s defeat to Prussia and a coalition of German states (1870–1871), exemplifies this category of defensive polemoform well (fig. 2A-D). By 1886, 372 forts, organized into three defensive lines, are integrated into the eponymous system to ensure in-depth protection, including 234 first-generation forts (type 1874) built ex nihilo following standardized principles (Diest, 2022). Each fort comprises a plurihectometric masonry complex dedicated to artillery (e.g., firing platforms, traverse shelters, powder magazines) or ancillary functions such as barracks, supply depots, logistics and close defense (e.g., caponiers), all shielded by several meters of earth and surrounded by a dry ditch several meters deep (Ancelin et al., 2023). Between 1886 and 1914, more than 200 additional modernized structures with distinct architectural and morphological features are added to the defensive system (Frinjs et al., 2008).

7In archaeology, these fortifications are often merely a secondary subject, either in studies unrelated to war (e.g., Chailluz Forest, Doubs; Fruchart, 2014) or within explicitly military works (e.g., Château de Joux; Metral, 2020). They are also rarely discussed objects in WW1 research, be it archaeological, such as studies on the fortified camp of Paris (Dardignac and David, 2015), or geographical, such as battlefield analyses in the Pays de Reims, Champagne, Argonne (Taborelli et al., 2017a; Taborelli, 2018) and Verdun (De Matos-Machado, 2018). Covered with anthropogenic superficial deposits and vegetation, these forts now stand out as unique geomorphological features (Taborelli et al., 2017a; Ancelin et al., 2023) and distinct geomorphological case studies (Ancelin et al., 2023, 2024), raising significant archaeological challenges (Ancelin et al., accepted a). More than mere architectural remains, these structures are man-made landforms born of complex geomorphological dynamics, encompassing all morphostratigraphic types of artificial ground defined by the BGS (British Geological Survey; Mac Millan and Powell, 1999; Ford et al., 2010). Interest in "Séré de Rivières" polemoforms arises from two main factors. First, these fortifications reflect an exceptional rationalization of geomorphological parameters, as geomorphology – alongside related geographic and geological fields – played a major role in 19^th century French military strategy (Boulanger, 2019). This approach led to the multiscale adaptation of fortifications to terrain characteristics (Ancelin et al., 2023). Secondly, their construction has had a profound topographic impact, reshaping the landscape to a degree sometimes surpassing that of WW1 polemoforms. This reshaping resulted from the mobilization of significant volumes of material, ranging from tens to hundreds of thousands of cubic meters, both in situ (e.g., ditch excavation, earthworks, central mass) and ex situ (e.g., extraction sites; Ancelin et al., 2023, 2024, accepted a). Besides changes in the landscape, these forts host topoclimatic conditions that support unique plant communities and serve as biodiversity reservoirs linked to ecological corridors within the Green and Blue Infrastructure framework (fig. 2A-D; Decoq et al., 1996).

8Alongside a strictly geomorphological perspective, the post-conflict trajectories of these fortifications have been examined within the humanities and social sciences as part of military heritage (Truttmann, 1994). This has fostered studies on their historical significance (Diest, 2013, 2022), architectural features (Truttmann, 1992, 1994), and landscape dimensions within geography (Amat and Becker, 1994; Durup de Baleine, 1998, 2001; Mathis A. and Mathis D., 2014). Often geographically isolated (Durup de Baleine, 1998) and burdened by an absence of an engaging and valorizing image (Mathis A. and Mathis D., 2014), these structures have largely been abandoned since being decommissioned, resulting in limited patrimonialization efforts.

9Scientific studies have yet to comprehensively examine the geomorphological and environmental heritage value of the hybrid landform associated with the "Séré de Rivières" fortifications. To this end, this research undertakes a geomorphological and geohistorical analysis of the Reims belt of "Séré de Rivières" fortifications (12 forts, 1875-1885; fig. 3A-L), drawing inspiration from De Matos-Machado’s (2018) work on WW1 polemoforms on the Verdun battlefield. Reims was selected as the study area for several criteria ranging from territorial (funding from Grand Reims and the University of Reims Champagne-Ardenne), academic (continuity of research by GEGENA), scientific (a novel geographical approach to the Reims fortifications, with LiDAR coverage of half the belt in 2015 and HD LiDAR in 2023), and logistical reasons (geographic proximity).

10The aim is to evaluate the "Séré de Rivières" polemoforms as landforms worthy of conservation and transmission to future generations (André et al., 2013). These structures are positioned within an expanded concept of geoheritage, encompassing both their geomorphological and environmental values, assessed through scientific as well as complementary criteria (e.g., ecological, economic, cultural). This concept of what qualifies heritage aligns with contemporary challenges, where environmental and ecological considerations are increasingly central (Gray, 2004). These aforementioned qualities provide meaningful opportunities for the valorization of structures often painted in a negative light due to the legacy of military conflict and war (dark heritage; Biran et al., 2011).

11In this context, it is essential to assess the extent of morphological alteration over the past 150 years, because the integrity of these features influences both their heritage relevance and their environmental and ecological functions. This study therefore proposes a novel, preliminary methodology based on the processing of high-resolution ALS data to quantify erosion through numerical indices of slope and ruggedness. By comparing results across the forts, this analysis further investigates the role of natural and anthropogenic erosive factors in the erasure of the "Séré de Rivières" polemoforms.

2. Study area

12The Reims fortified belt (12 forts) is located in the eastern Paris Basin, at the junction of two tiered plateaus separated by the Île-de-France cuesta (fig. 4A-C). On the eastern side, the plateau is composed of Mesozoic chalks from the Champagne sèche, while to the west, it transitions into Cenozoic formations characteristic of the Laonnois, Soissonnais, Tardenois, and Montagne de Reims regions.

NGF
A : carte morphostructurale simplifiée du secteur de Reims ; B : Carte géologique de France au 1/1 000 000^e ; C : coupe géologique des plateaux étagés montrant le rôle d’assise du relief de cuesta dans l’organisation de la ceinture fortifiée de Reims (les tracés sont représentés sur la figure 4A).
1. Formations fluviatiles (Quaternaire) ; 2. Limons (Quaternaire) ; 3. Argiles à meulières (Rupélien) ; 4. Calcaires et marnes (Lutétien) ; 5. Sables (Yprésien supérieur, « Cuisien ») ; 6. Argiles, lignites et sables (Yprésien inférieur, « Sparnacien ») ; 7. Sables, grès et argiles (Thanétien) ; 8. Craie blanche (Crétacé) ; 9. Fronts de cuesta et principales buttes ; 10. Tracé A’B’ de la coupe géologique de la figure 4C ; 11. Points cotés (altitude en mètres NGF) ; 12. Principaux cours d’eau ; 13. Fortifications “Séré de Rivières” ; 14. Localités ; 15. Tranchées et boyaux français en 1918 ; 16. Tranchées et boyaux allemands en 1918 ; 17. Revers de côte ; 18. Buttes-témoins ; 19. Avant-buttes ; 20. Calcaires lutétiens, marnes et caillasses (Éocène moyen, Lutétien) ; 21. Calcaires grossiers (Lutétien inférieur à moyen) ; 22. Marnes et calcaires (Lutétien, Priabonien) ; 23. Sables (Yprésien supérieur) ; 24. Argiles (Yprésien inférieur) ; 25. Sables argiles et marnes (Thanétien supérieur) ; 26. Craie (Campanien) ; 27. Fortifications « Séré de Rivières » ; 28. Communes ; a. Réduit de Chenay ; b. Fort de Saint-Thierry ; c. Batterie de Loivre ; d. Fort de Brimont ; e. Batterie du Cran de Brimont ; f. Fort de Fresne ; g. Fort de Witry-lès-Reims ; h. Batterie de Berru ; i. Vigie de Berru ; j. Fort de Nogent-L’Abbesse ; k. Fort de la Pompelle ; l. Fort de Montbré ; MdBe : Mont de Berru ; MdBr : Mont de Brimont ; MdR : Montagne de Reims ; MdST. Massif de Saint-Thierry ; Ep : Épernay ; Fi : Fismes ; Re : Reims ; t1. Trigny ; t2. Pouillon ; t3. Brimont ; t4. Berru ; t5. Nogent-l’Abbesse ; t6. Rilly-la-Montagne.

13The fortifications are organized in a circular arc facing the funnel-shaped consequent valley of the Vesle, located at least 5 km apart from the urban area to remain out of range of enemy artillery, all the while maintaining proximity to ensure continuous artillery coverage. The Île-de-France cuesta provides the geomorphological foundation for this fortified belt, designed to protect both the Reims and the Vesle Valley, a strategic communication route to Paris (Lacroix, 1909).

14This geomorphological context gives rise to three categories of sites occupied by three initially planned fortified groups, supplemented by interval forts (fig. 4C). To the west, the Chenay redoubt (a) and the Saint-Thierry fort (b) form the Saint-Thierry group, located on the eponymous limestone massif, which represents the reverse slope of the Île-de-France cuesta (Lutetian limestone of the “Calcaire grossier”). In this area, the massif narrows into an interfluve, marking the easternmost extension of the Cenozoic formations of the eastern Paris Basin.

15To the northwest, the Brimont fort (d) sits atop lithified Thanetian sands, overlooking the Chalk depression from the summit of the eponymous residual hill (170 m; 2 km²). It is supported by two lower fortifications on either side, forming the Brimont group (Lacroix, 1909): the Loivre battery (c) on the chalky western flank (Campanian chalk) and the Cran de Brimont battery (e) on the sandy eastern flank of the hill.

16To the east of Reims, the residual hill of Berru (270 m; 10 km²) dominates the landscape, hosting multiple fortifications that comprise the Berru group. At its summit, the Berru watchtower (i) rests on Lutetian-Bartonian marls. Below, the main defensive line extends with the Witry-lès-Reims fort (g), built on Thanetian sandy marls partially stripped during its construction, the Berru battery (h), set on Lower Ypresian marls, clays, and lignitic sands, and the Nogent-l’Abbesse fort (j), resting on Thanetian continental clays and marls, violet marls, Lower Ypresian sands, and Campanian chalk.

17The final category of geomorphological sites includes forts positioned on isolated chalky outliers of low elevation, enabling control over the intervals flanking the Berru hill (fig. 4C). To the south, the Pompelle fort (k), overlooking the Vesle Valley, and the Montbré fort (l), likely located on remnants of former connecting glacis (Wiedemann, 1976; Ancelin et al., 2023), secure the interval with the Montagne de Reims. To the north, the Fresne fort (f), perched on a chalky rise, ensures control over the interval with the Brimont hill.

18These belt fortifications, part of the second defensive line of the “Séré de Rivières” system (Barros, 1999; Truttmann, 2017), saw minimal modernization following the shell-torpedo crisis (Ancelin et al., 2023) and were decommissioned on the eve of WW1. During the war, two-thirds of the city of Reims was destroyed, and the forts were integrated into the defense networks, including trenches and tunnels (Devos et al., 2015). They became key resistance centers, requiring morphogenetic adaptations (Devos et al., 2019; Ancelin et al., 2023). The Reims fortified belt is therefore among the few locations directly exposed to combat, particularly artillery fire, alongside the forts of Maubeuge (Nord department) and Verdun (Meuse department).

3. ALS data processing workflow

19LiDAR data, with its ability to provide high-resolution 3D views of micro- and macrotopography, is a crucial resource for studying polemoforms (Hesse, 2014). The methodology proposed here, structured into three phases, aligns closely with the documentation and standardization framework outlined by Lozic and Štular (2021). Their approach, comprising 18 processing steps organized into four phases, addresses the lack of a strict and consensual normative framework in the specialized literature, both in France (e.g., Ancelin, 2020; Berthe, 2024; Georges-Leroy et al., 2011; Nuninger et al., 2010; Ritz, 2020) and internationally (e.g., Cowley, 2011; Crutchley and Crow, 2010; Fernandez-Diaz et al., 2014).

3.1. Raw data acquisition and processing (phase 1)

20The lasergrammetric data used in this study originates from the national High-Density LiDAR acquisition campaign (2021–2026) led by the IGN (fig. 5A; available at https://geoservices.ign.fr/​lidarhd). This data is extracted from the "MD" tile, encompassing 2500 km² and covering the Champagne-Brie region, Laonnois, and the valleys of the Marne, Vesle, and Aisne, as well as the entirety of the Reims fortified belt (fig. 5B). The dataset comprises 2500 point clouds, each representing 1 km², published in standard LAS binary format (LAZ 1.4). The data follows the ASPRS (American Society for Photogrammetry and Remote Sensing) specifications and have been pre-classified by the IGN.

21Fig. 5 ­ Location of ALS data.
Fig. 5 ­ Localisation des données LiDAR aéroportées.

22This acquisition was conducted in March 2023 to ensure optimal phenological conditions for capturing a very high-resolution DTM (25 cm) beneath deciduous forests, suitable for geomorphological analysis (Höfle and Rutzinger, 2011). By contrast, many tiles in southern France were surveyed during the summer to support fire risk management efforts. Data acquisition and processing specifications, as well as related reports, remain unpublished and are not considered in this study. This study also builds on the 2015 LiDAR acquisition over the residual hills of Brimont and Berru, conducted as part of the IMPACT 14–18 program, which has already been used in the study of polemoforms (Ancelin et al., 2023; Devos et al., 2019; Taborelli et al., 2017b).

3.2. Point cloud processing and derivation of products (phase 2)

23The analysis focuses on 13 tiles of 1 km², each intersected by one of the forts in the Reims belt (fig. 5B). Ground point density varies according to land use (Ancelin et al., accepted b), ranging from approximately 70 pts/m² in tiles containing forts, where small forest patches interrupt the extensive openfield crops of Champagne, to lower densities in intra-forest areas where vegetation intercepts more laser rays.

24For instance, the tile covering the Saint-Thierry fort, located within the eponymous forest massif, contains 20,313,493 ground points, resulting in an average density of approximately 20.3 pts/m² (fig. 6A-D). The density map (with a cell resolution of 25 cm) demonstrate the strong isotropy of DTM cells interpolated across data gaps, which represent approximately 21.5 % of the surface area (1 m resolution = 0.3 %; 50 cm = 0.8 %). These gaps are primarily attributed to built-up areas (filtered out of the DTM), water surfaces (ray absorption), and forest cover types, such as the eastern forest band dominated by pines (fig. 6D).

25DTMs facilitate the visualization and interpretation of archaeological features and topography while reducing computational constraints related to data volume. Their use has become standard in ALS data processing specific to archaeology (Challis et al., 2011; Doneus et al., 2020) and geomorphology (Johnson et al., 2015; Dupuis et al., 2017; Öhrling et al., 2018), as well as, more specifically, for the subjects of war landscapes (Hesse, 2014 ; De Matos Machado et al., 2016, 2020 ; Taborelli et al., 2017a, 2017b ; De Matos Machado, 2018 ; Gheyle et al., 2018 ; Ancelin et al., 2023). Following this well-established use, DTMs are employed in this methodological approach, as they are essential for assessing the erasure of polemoforms through ruggedness and slope indices. To process DTMs at a 25 cm resolution ­ optimal for HD LiDAR feature detection in the Reims area ­ point cloud interpolation was performed using the "LAS to Raster" tool in ArcGIS 10.3. The IDW (Inverse Distance Weighting) and Nearest Neighbor interpolation methods were selected to ensure suitability for geomorphological analysis (Guo et al., 2010; Stular et al., 2023). IDW was chosen for its ability to preserve local topographic variability while minimizing interpolation artifacts, ensuring a more accurate representation of features. Nearest Neighbor was employed to retain the integrity of abrupt topographic transitions, crucial for analyzing the anthropogenic morphologies of the “Séré de Rivières” fortifications without introducing artificial smoothing effects. The footprint of each fort was then extracted from the DTMs using a polygon layer as a mask. Boundaries were defined either at the crest of the covered way (tab. 1) or based on the initial theoretical location reconstructed using historical planimetric and photographic documents (using the "extract by mask" tool on ArcGIS 10.3).

26The original and clipped DTMs were processed on QGIS 3.28.12 and ArcGIS 10.3 to generate three derived images optimizing the detection of topographical anomalies (Kokalj and Hesse, 2017), which form the basis of the statistical analysis (tab. 1):

• the local relief model (LRM), obtained by subtracting the original DTM from a smoothed DTM using a low-pass filter (Hesse, 2010);

• the terrain ruggedness index (TRI; Riley et al., 1999), which assesses the ruggedness of the terrain based on elevation variations;

• the slope index (Bennett et al., 2012), which measures the terrain's inclination.

27The ruggedness and slope indices are based on similar calculations, using the difference in elevation between a pixel and its 8 neighboring pixels. Figure 7A presents three theoretical case studies highlighting the relationship between ruggedness and elevation variations for the TRI: a peak and a depression, which show similar ruggedness indices despite different configurations, and a gently sloping terrain, associated with a low ruggedness value.

3.3 Archaeogeomorphological interpretation (phase 3)

28The terrain ruggedness and slope indices were primarily calculated to quantify and analyze the erasure of the "Séré de Rivières" polemoforms (tab. 2).

3.3.1. Terrain Ruggedness Indices

29To obtain a numerical value representing terrain ruggedness, expressed by the TRI (fig. 7A), zonal statistics were calculated on QGIS 3.28.12 using DTMs (tool "Zonal Statistics"), allowing for the determination of the arithmetic means. The analysis of ruggedness was refined further by the creation of a global index at the fort scale (RgI), proportional to the ruggedness of the relief, defined as the ratio between the draped surface and the flat surface of each fort's footprint (fig. 7B). To calculate this index, a TIN (Triangulated Irregular Network) layer was generated from the point cloud using the "Las Dataset to TIN" tool in ArcGIS 10.3. This vector layer, representing the terrain's ruggedness as triangles, includes in its attribute table the areas of the triangles, which were used to calculate the draped surface.

3.3.2. Statistical Distribution of Slope Values

30The analysis of the slope index is based on the hypothesis that standardization, being a characteristic feature of the "Séré de Rivières" fortifications (Ancelin et al., 2023), is highest at the level of the rampart and the ditch, as illustrated by Figures 8A-C. In contrast, morphological complexity appears to increase visually from the periphery towards the center (fig. 3A-L).

1. Crest of the covered way; 2. Outer crest; 3. Fire crest (inside crest); 4. Shooting terrace; 5. Natural soil; 6. Rampart; 7. Parapet; 8. Parapet slope; 9. Outside slope; 10. Glacis; 11. Counterscarp wall; 12. Scarp wall; 13. Ditch; 14. Position of the cunette; 15. Slope value; 16. Associated frequency.
A : Traitement de visualisation « indice de position topographique » (TPI) dérivé du MNT LiDAR ; B : Coupe topographique du rempart du front de tête du fort de Brimont réalisée à partir des données microtopographiques issues du LiDAR ; C : Profil théorique du rempart d’une fortification « Séré de Rivières de type » 1874 (d’après Guillot, 1949) ; D : Diagramme de distribution des valeurs de pente théorique du fossé (d’après Guillot, 1949).

1. Crête du chemin couvert ; 2. Crête extérieure ; 3. Crête de feu (crête intérieure) ; 4. Terre-plein ; 5. Sol naturel ; 6. Rempart ; 7. Parapet ; 8. Plongée ; 9. Talus extérieur ; 10. Glacis ; 11. Mur de contrescarpe ; 12. Mur d’escarpe ; 13. Fossé ; 14. Cunette ; 15. Valeur de pente ; 16. Fréquence associée.

31To test this hypothesis, the statistical distributions of slope values from the peripheral zone of each fort ­ situated between the crests of the covered way and the outer path ­ were compared with a theoretical distribution derived from a military fortification course (fig. 8B-D). The theoretical distribution was established by calculating the slopes of each developed section using the “ImageJ” image processing software and relating them to the plan length (. The parapet slope was excluded from the analysis due to the high frequency of traverse-shelters intersecting it (fig. 3A-L). This approach allowed for the identification of four slope values associated with specific elements (fig. 8D): (1) the bottom of the ditch (2.8°), (2) the outside slope (32.2°), (3) the embankment between the crest of the covered way and the top of the counterscarp wall (43.7°), and (4) the walls of the ditch (77°).

32To obtain such a distribution for each fort, the peripheral zone of each DTM was extracted using a polygonal vector layer covering the area between the crests of the covered way and the outer crest (fig. 9). After calculating the slope index, the values were exported to a spreadsheet for statistical analysis of three central measures: the arithmetic mean, the median, and modes. Focus was placed on the modes of the statistical distributions, represented as histograms. These modes were analyzed for their value and amplitude and compared to the relative frequencies and slopes of the theoretical distribution to identify potential anomalies.

3.3.3. Digitalization protocol

33Beyond the forts and their peripheral ditches, topographical anomalies located within the ditches, between the two crests, were digitalized to evaluate their potential role in the erosion of the "Séré de Rivières" polemoforms. Each anomaly type was defined following a digitalization protocol based on its morphological characteristics (fig. 10; Ancelin et al., accepted a). These definitions informed the cartographic standards, specifying the criteria feature identification as well as the mapping methods employed. To facilitate their identification, semantic information was supplemented with topographic profiles and plan views (fig. 10).

34Morphometric data (length, width, diameter) and archaeological information (field observations, dating) were recorded in an attribute table corresponding to the vector type used (point, line, or surface). The digitalized anomalies were organized into three thematic categories: (i) components of "Séré de Rivières" fortifications (Ancelin et al., accepted a); (ii) polemoforms inherited from WW1 (e.g., trenches, shell craters), drawing from the studies of Taborelli (2018) and De Matos-Machado (2018); (iii) landslides resulting from the leveling of certain ditch slopes.

35The digitalization of wall sections and shell craters specifically enabled for the calculation of preserved wall proportions and cratered surface areas.

36The protocol was validated through cross-checking with field observations, contemporary cartographic databases (e.g., IGN), as well as historical sources, including manuscripts, maps, and photographs (orthophotos, vertical, and oblique aerial imagery). Planimetric documents, such as the “plans directeurs des GCTA” (master firing plans), were georeferenced in ArcGIS using the "first-degree polynomial" transformation method, enabling their simultaneous projection with the LiDAR data and digitalized layers.

4. Results

37The ruggedness indices (calculated for entire forts) and slope indices (focusing on the ditch and peripheral rampart), combined with the proportion of preserved wall length, provide a statistically meaningful ranking of the forts (tab. 3). Generally, higher values correlate with better preservation of the fort’s initial shape, as suggested by visual interpretations of LiDAR-derived DTM visualizations (fig. 3A-L) as well as confirmed by field observations.

38The Montbré fort, the best-preserved structure in the Reims belt, exhibits the highest ruggedness and slope values. In contrast, the Berru watchtower, which retains only the general outline of the fort, displays the lowest values.

4.1. Fort-level analysis: terrain ruggedness study

39Bivariate statistical analysis of terrain ruggedness indices (TRI, RgI; fig. 11) reveals an exceptionally strong Pearson correlation (r = 0.961), enabling a consistent ranking of the fortified structures. The graphical representation of the two indices delineates two distinct groups, each characterized by closely clustered values and one outlier case (Berru watchtower and Montbré fort).

40The first group, associated with the lowest indices (TRI ≤ 0.26; RgI ≤ 1.12), includes the Berru watchtower (i), the forts of Fresne (f), Loivre (c), Witry-lès-Reims (g), as well as the Cran de Brimont (e) and the Berru batteries (h). In LiDAR-derived DTM visualizations, the internal organization of these structures is almost entirely erased or even entirely imperceptible. The Berru battery is a partial exception, with certain features, such as traverse shelters, the double caponier, and the peripheral ditch remaining identifiable.

41The second group comprises of forts with the highest terrain ruggedness indices (TRI ≥ 0.31; RgI ≥ 1.17). These structures generally retain, or even fully preserve, interpretable internal organizations, as revealed by LiDAR DTM visualizations. They include the fort of Nogent-l’Abbesse (j; fig. 3J; Ancelin et al., 2023), the redoubt of Chenay (a), and the forts of Saint-Thierry (b), Brimont (d; fig. 3D; Devos et al., 2019; Ancelin et al., accepted a), Pompelle (k) and Montbré (l).

4.2. Rampart and ditch: study of slope value distribution

42The use of descriptive statistics on the slope values of the peripheral zone of the forts corroborates the discriminate distribution into two groups obtained via the terrain ruggedness indices (tab. 3; fig. 12A-B).

43Group 1 (G1) is characterized by a median (Me) and mean (Mea) less than or equal to 16.5 and 19.6° respectively (forts c, e, f, g, h, i). The statistical distribution of slope values in the peripheral zone is bimodal, with a first peak (P1) between 2 and 7°, associated with a high relative frequency (F ≥ 4.16 %), a range (R) between 15 and 20, and a cumulative frequency distribution (CFD) between 55.9 and 84.4 %. The second peak (P2), established between 26 and 40°, has a more moderate F (≤ 2.2 %) and an R between 20 and 33 for a CFD between 11.8 and 32.6 %.

44On the other hand, Group 2 (G2) consists of polemoforms with a Me and Mea of the slope values greater than or equal to 23.5 and 22.7° respectively (forts a, b, d, j, k, l). The statistical distribution of the slope values is composed of three modes. P1 is located between 2 and 5°, with an F between 2.2 % and 4.2 %, an R between 11 and 14 and a CFD between 18.7 % and 36.4 %. P2 (30-33°) has an F of 2.7 % to 3.5 %, an R between 29 and 38 and a CFD of 42.5 % to 73.9 %. Additionally, there is a third peak (P3) at the end of the distribution (66-73°), sometimes statistically imperceptible, never exceeding 0.5 % in F, with an R between 14 and 22, resulting in a CFD of 0.5 % to 10.8 %.

A : diagramme de distribution des valeurs de pente ; B : statistiques associées.
1. Fréquence relative ; 2. Valeur de pente théorique (fig. 8C-D).

45The comparison with the distribution diagram established from the theoretical profile of a "Séré de Rivières" fortification (fig. 8C-D; Guillot, 1949) allows for matching of the peaks identified to the theoretical slope values associated with the ditch bottom (M1), the outer embankment (M2), and the retaining walls of the ditch (M4). It seems, however, that the values related to the reverse of the glacis (M3) are statistically imperceptible. The values of P1 are centered in the intervals 2-7° (G1) and 2-5° (G2), coinciding with M1 (2.8°). Apart from four forts, where the peaks are established between 2 and 3° (a, c, f, g), all other polemoforms show their peaks shifted towards higher values, between 3 and 7°. On the other hand, the two groups differ in their CFD, with G1 having a CFD greater than the theoretical CFD (G1 CFD > 38 %) and G2 having a CFD lower (G2 CFD < 38 %). The same logic applies to the values of P2, centered between 26 and 40° (G1), then 30 and 33° (G2), which intersect with M2 (32.2°; CFD = 44 %), but differ by having lower CFD for G1 and generally higher CFD for G2. One of the main discriminating factors is the absence of P3 in G1. In G2, the slope values of P3 (66-73°), although lower, correspond with the theoretical M4 (77°; CFD = 9.5 %), with CFD almost always well below, except for the Montbré fort (l).

4.3. Theoretical erosion scale of “Séré de Rivières” polemoforms

46The collection of statistical indicators of erasure refines the conceptual approach developed by Ancelin et al. (2023), which applies onto polemoforms the geomorphological evolution paradigms of reliefs described by Archambault et al. (1965). This approach shifts the focus from the geological timescale to the Anthropocene (protohistoric and historic) timescale—understood here not as a formally defined age, epoch or period (Gibbard and Walker, 2013), but as an ongoing event of planetary transformation, capturing the gradual, diachronic and intensified alterations of the Earth system resulting from human–environment interactions (Edgeworth et al., 2023). Just as a single term in stratigraphy can refer to both an age and a stage (e.g., the Thanetian, 59.2-56.0 Ma), the Anthropocene is likewise synonymous with a deposit, consisting either of a set of geological formations (Andreieff, 1997) or of a set of landforms and formations (Price et al., 2011). The second perspective, which is morphostratigraphic and more comprehensive, is the one being in this study (Mac Millan and Powell, 1999; Ford et al., 2010). The "Séré de Rivières" polemoforms can be classified into three evolutionary states: "primitive" relief, which remain well-preserved; "derived" relief, showing moderate erosion and infilling; and "attenuated" or "leveled" reliefs, which are heavily degraded or nearly erased (fig. 13A-B). These categories are reflected in terrain ruggedness and slope values, which systematically decrease as fort morphologies are altered. A less-erased fort shows a tri-modal slope distribution across the rampart and ditch, with the CFD of each peak (P1, P2, P3) closely aligning with the CFD of the dominant theoretical modal values (M1, M2, M4). This pattern is exemplified by Montbré fort, considered the best-preserved site in the Reims fortified belt. It exhibits the highest ruggedness indices (0.38; 1.27), consistent with its average slope (26°) and a multimodal statistical distribution aligned with the three dominant theoretical slope values recommended in fortification courses (tab. 3; fig. 8C-D). The distribution shows statistical peaks in CFD closely matching theoretical CFD values: 10.8 % compared to 9.5 % for P1 (M1); 42.5 % compared to 44 % for P2 (M2); and 36.4 % compared to 38 % for P3 (M4). In contrast, a heavily altered fort exhibits a bi-modal distribution, with two less balanced peaks and CFD values deviating further from the theoretical values, as observed at the Witry-lès-Reims fort.

47Beyond their statistical characterization, these evolutionary states also manifest distinct geomorphological trajectories, shaping their preservation potential as well as their morphological and historical legibility. "Primitive" landforms are characterized by remarkable preservation, reflecting simple trajectories and a strong potential for patrimonial valorization, as seen with the Montbré fort, which was only surrounded by a few trenches (fig. 3L) due to its position on the third line of defense during the war (fig. 4A). In contrast, "derived" landforms result from erosion and infilling processes, often associated with a multi-temporal evolution. These features sometimes exhibit deep morphological modifications, such as at Nogent-l’Abbesse fort (fig. 3J), where most of the ditch embankment has been regularized (cf.infra), or at the Malmaison fort, which was disfigured by military experiments (cf.infra). Finally, "attenuated" or "leveled" landforms gradually disappear from the landscape due to erosion, dismantling, or infilling. The Witry-lès-Reims fort (fig. 3G) provides such an example, with reliefs now entirely erased or significantly reduced (fig. 14A-D). While these classifications outline the progressive transformation of polemoforms and hint at the key processes involved, a more detailed examination is required for a deeper understanding of the factors driving their erasure.

5. Discussions

48The approaches at the fort scale as well as at the rampart and ditch scale reveal statistical disparities that raise questions about the factors and processes driving the erasure of the "Séré de Rivières" polemoforms within the Reims fortified belt. These mechanisms act either directly on the relief or on the masonry of the built structures, with subsequent impacts on the topography. However, these analyses have certain limitations, which will be addressed at a later stage.

5.1. Identification of Erasure Factors

49Anthropogenic factors – whether civilian, military, or polemological – appear to be the most decisive in explaining the erasure or lack thereof of "Séré de Rivières" polemoforms, while natural factors play a more marginal role. These elements impact either the terrain directly or the masonry of the structures, with subsequent effects on the topography.

5.1.1. Preservation factors

50The case of the Montbré fort (fig. 3L) exemplifies the key factors influencing the preservation of some "Séré de Rivières" polemoforms. Two primary factors, ranked by importance, account for this preservation: (i) late military use: the site was employed by the armed forces until 2013 (13^th Engineering Regiment of Épernay) and later repurposed by the owning municipality for recreational activities, protecting it from destructive interventions; (ii) strategic isolation: located on the third line of the French defense networks, the fort was spared from combat during trench warfare (fig. 4A).

51The role of vegetation cover in moderating erosion on topographic anomalies, including polemoforms, is likely limited for these large-scale features, due to their relatively short existence of 150 years (Durup de Baleine, 1998; Taborelli, 2018).

5.1.2. Land use reassignment

52In contrast, the highest rates of erasure, observed in the structures of G1, are directly linked to an early reassignment of the land, often at the expense of the fortified structures.

53Except for the Berru battery (fig. 3H), these sites display significant topographical alterations, including instances of leveling, resulting in a high CFD for P1 in the slope value distributions of ramparts and ditches (56.6-84.4 %), at the expense of CFDs for P2 and P3 (fig. 12A-B). Examples include destructive recreational activities, such as off-road circuits at the Loivre battery (fig. 3C) and Fresne fort (fig. 3F), or current military use, as seen at the Berru watchtower (fig. 3I), where the terrain has been leveled to accommodate a telecommunication center operated by the Army.

54At the Witry-lès-Reims fort (fig. 3G), a direct comparison of the 2015 and 2023 LiDAR acquisitions reveals the progressive erasure of the polemoform, characterized by the leveling of the rampart and, conversely, the infilling of concavities with construction debris (fig. 14A-D).

5.1.3. Geomaterial extraction

55Another significant process, closely tied to topographical reshaping and the masonry-based construction of these fortifications, involves the dismantling of dressed stone (medium and large masonry elements) for reuse. Beyond the loss of architectural components, this process has a direct impact on the topography, as the removal of retaining walls and structural elements alters slope stability and contributes to terrain modification. This phenomenon affects nearly all forts in the Reims fortified belt to varying degrees, with notable exceptions such as the Chenay redoubt (fig. 3A) and Montbré fort (fig. 3L, tab. 3).

56For G1 forts, stone extraction involves nearly all substructures, whether in sites repurposed and heavily altered (e.g., fig. 3C, F, I) or abandoned but still morphologically discernible (e.g., the Berru battery, fig. 3H). This dismantling contributes to the low ruggedness values observed in the RgI (tab. 3) and explains the absence of P3 in the statistical slope value distributions (fig. 12A-B), primarily due to the leveling of ditch slopes and the removal of retaining walls when not reshaped by human activity.

57In G2 forts, material extraction is more localized, often targeting specific facades (e.g., the barracks of Brimont fort, fig. 3D), ditch walls (e.g., Nogent-l’Abbesse fort, explaining the absence of P3, fig. 3J), or partial and sometimes complete substructures (e.g., half of the barracks and the double caponier at Saint-Thierry fort, fig. 3B, or the barracks at Witry-lès-Reims fort, fig. 3G).

58Although no direct documentation on the reuse of stones from these forts has been identified, it is plausible to associate this activity with the post–WW1 reconstruction period, recognized as the fourth major phase of material repurposing in the Reims region’s history (Fronteau et al., 2014). Additionally, a 1949 aerial photograph suggests that the dismantling of the substructures at Saint-Thierry fort occurred after World War II (fig. 15A).

5.1.4. War-related destruction

59Although less significant, war-related destruction also contributed to the deterioration of the fortifications. Shell impacts are particularly evident in the relief as craters or shell funnels (fig. 10), categorized into four morphological types: simple craters, with or without a peripheral ejecta ridge (fig. 16A-B), craters caused by the collapse of masonry structures (fig. 16C), and craters resulting from the subsurface collapse of casemate vaults (fig. 16D). These impacts reduce fort-scale ruggedness (RgI) by eroding convex micro-reliefs and filling concave features.

60Shell strikes on walls, in particular, contribute to the regularization of scarp and counterscarps by the collapsing masonry and colluviation, whose materials tend to fill the ditch, which directly alters the distribution of slope values (cf. infra). The cratered surfaces of Reims forts account for up to 7.2 % of the total area (tab. 3), a relatively moderate value compared to other forts, such as Malmaison (Laon stronghold), where craters cover at least one-fifth of the surface, or Douaumont Fort (Verdun stronghold), located in a much more exposed sector of the front, where impacts are four times as numerous (De Matos-Machado, 2018).

61This difference is partly due to extensive post-conflict reshaping, which can obscure the superimposition of polemoforms. For example, no craters are visible at the Loivre battery (fig. 3C), despite its position being on the front line for nearly two years. Similarly, the Pompelle fort barracks underwent significant modification during rehabilitation works in the 1990s (fig. 3K, 15B).

5.1.5. Bioerosion

62As with these anthropogenic processes (topographical reshaping, dismantling, war), natural deterioration affects both the masonry structures and the terrain of the fortifications. For the masonry structures, it results from the combined action of several erosive processes, such as bioerosion, frost weathering and hydraulic erosion especially after infiltration (Durup de Baleine, 1998). Bioerosion is induced as soon as the fortifications are built, through the implantation of defensive vegetation. This tree and shrub vegetation is primarily intended to conceal the fortification from enemy sight, prevent sapping work on the glacis (with tracing roots), increase the resistance of the outer slope to projectiles or facilitate the creation of abatis. Today, most of the forts in Reims are covered with multi-stratified forest stands in natural succession, inherited from the initial military plantings. These forest islands are either isolated in the middle of agricultural spaces (e.g., Montbré, La Pompelle, Nogent-l’Abbesse, Fresne forts) or completely integrated into adjacent forest masses (e.g., Berru, Brimont, Saint-Thierry, Chenay fortifications). This bioerosive activity is particularly evident in the presence of interstitial vegetation, whose root systems penetrate the cracks in the masonry. As an example, it can involve pine trees (Pinus sp.) rooted in the earthen massif (fig. 15C), that are capable of dislodging elements of the cornices, or ferns (Asplenium trichomanes subsp. quadrivalens) thriving on the walls of the ditches (Decoq et al., 1996).

5.1.6. Freezing

63The Reims area, although subject to less intense freezing than regions located further east (Vosges Massif or even the Lorraine cuestas), is also marked by the mechanical disintegration of materials due to freeze-thaw cycles. This phenomenon is amplified by the use of porous limestone rocks, such as Milioles and Campanile Giganteum limestones (Lutetian) from the tertiary plateaus to the west of Reims, or "Euville" type Entroquitic limestones (Upper Oxfordian) in the Meuse department (Ancelin et al., 2024). The deterioration of the structures ultimately facilitates water infiltration, contributing to hydraulic erosion that exacerbates the frost susceptibility of the construction stones. The destruction caused by the conflicts amplifies these effects by disrupting the internal water cycle in the fortified structures. This issue was anticipated from the design stage, notably through the installation of an impermeable cement layer between the substructure vaults and the overlying fill (fig. 15D), or by the inclusion of basal drainage outlets in the masonry walls. These devices, often damaged by conflicts, have lost their effectiveness, increasing the vulnerability of the structures. In all cases, these impacts on the masonry are only morphologically perceptible when they lead to collapse, which then affects the topography.

5.1.7. Gravitational hazard

64Natural deterioration impacts the relief, primarily through gravitational hazards. The embankments bordering the ditches undergo a regularization process similar to that observed on natural slopes (e.g., Guérémy and Vejux, 1987). This process is emphasized by the disappearance of lateral walls, leading to the loss of P3 in slope value distributions and the expansion of P2. This regularization contributes to a colluvial dynamic that progressively fills the ditches, increasing the slope value of P1.

65Additionally, pedogenesis processes, like those observed in shell craters (Hupy and Schaetzl, 2008; De Matos-Machado, 2018), likely occur in areas of water stagnation or accumulation within these ditches. The retention of surface water promotes the accumulation of organic deposits and leaching, thereby reducing terrain ruggedness.

66Depending on geological factors, this regularization can also lead to landslides of varying intensity. At the Nogent-l’Abbesse fort (fig. 3J), located on a hilltop capped by unstable Sparnacian (Ypresian) clayey-marly-sandy deposits (e.g., landslides in the Montagne de Reims; Bollot, 2014), this phenomenon is particularly pronounced. LiDAR data reveals that over 5000 m², mostly on the counterscarp, along a 500-meter section of ditch, are impacted by distinct embankment regularizations (fig. 17A-D). Alongside slower regularization via flow, numerous landslides were reported during the fort's active period (e.g., 1882-1887, 1904, 1907), within "[...] whitish clay [...] under the action of thawing and rain" (ministère de la Guerre, 1909, 1904).

67The influence of geology on the erasure of "Séré de Rivières" polemoforms remains an underexplored subject. Until now, bedrock (fig. 4A-C; tab. 3) has only been studied in its role as a multiscale conditioning factor for the fortification system (Ancelin et al., 2023). However, geological characteristics, such as porosity or resistance, play a crucial role in the preservation of fortified terrain. For instance, unconsolidated or highly porous bedrock promotes erosion and the gradual filling of concave forms, while more indurated bedrock with increased resistance contributes to the preservation of convex microreliefs and limits sliding or regressive erosion. This is illustrated by the Saint-Thierry fort (fig. 3B), where P3 persists, despite the dismantling of over 9/10^th of the ditch walls, in part due to the limestone rock outcrops backing them (fig. 12A-B; tab. 3).

68This issue also raises the question of determining the initial ground level, which helps distinguish the natural bedrock from the altered soil used to cover substructures and form earthen embankments, as these materials have vastly different behaviors.

69At the scale of the Reims fortified belt, preliminary results suggest that the theoretical initial ground level corresponds to a line passing between the crests of the scarp and counterscarp walls, as suggested in fortification treatises (fig. 8C). This configuration could explain the lower susceptibility of Nogent-l’Abbesse’s fort outer embankments to landslides (fig. 17A-D). A comparison of the topographic profiles of ditch embankments reveals a clear disparity between the clayey-marly embankments of Nogent-l’Abbesse, which are wider and less steep, subjected to more pronounced regressive erosion, compared to the other main forts, which feature ditch sections without walls, such as Pompelle fort.

5.2. Method limitations

70The study of the erasure of "Séré de Rivières" polemoforms reveals several methodological challenges, largely due to the complexity of these structures.

5.2.1 Morphological complexity

71A primary challenge arises from the vast hectometric areas occupied by these "mega-polemoforms" (Taborelli, 2018; Taborelli et al., 2017a) at the geofacies scale (Mercier, 2004) and the morphological complexity of their internal organization (Ancelin et al., 2023, accepted a).

72The fortifications indeed exhibit a serial architecture (Diest, 2022) and a high degree of standardization, both in their general form and infrastructure layout (Ancelin et al., 2023, accepted a), which also motivated the choice to focus on the rampart and ditch for slope value analysis (fig. 8A-D, 9). However, the level of modulation and adaptation remains significant, shaped not only by human decisions but also by the geographic, geological, and geomorphological conditions of the construction sites (Ancelin et al., 2023). A notable example is the Saint-Thierry fort, whose subtriangular shape follows the contours of the underlying interfluve.

73Regarding the forts with intact structures (G2), it is likely that the specific characteristics of each fort influence terrain ruggedness more significantly than the factors of erasure. This is exemplified by the high ruggedness indices of Brimont fort (RgI = 1.23; TRI = 0.37; fig. 3D, tab. 3) even though the fort’s surface being heavily impacted by shell craters within the Reims fortified belt (fig. 3A-L).

74The comparison of the results from the forts of Reims (fig. 12A-B) with the Malmaison fort (Chavignon municipality, Aisne department) reinforces this hypothesis regarding the superimposition of WW1 polemoforms (fig. 18A-C). This fort, covering 5.7 ha, located about 40 km west of Reims on the Soissonnais plateau in the Aisne department, displays a high density of craters (67/ha) covering more than 20 % of its surface (fig. 18A). These impacts resulted from the Malmaison experiments of 1886 (i.e., testing the power of newly created siege weapons; ministère de la Guerre, 1886) and artillery preparation phase during the Battle of the Malmaison (1917; fig. 18B). However, the ruggedness indices do not explicitly reflect this crater density: the fort has an RgI (1.18) higher than Nogent-l’Abbesse fort and a TRI (0.34) equivalent to that of Saint-Thierry fort, despite only 5 % and 1 % of their respective surfaces being affected by shell craters. Similarly, the slope value distribution of the rampart and ditch is equivalent to that of Nogent-l’Abbesse fort (fig. 18C), with a P1 at 4.5° (CFD = 30.3) and a P2 at 31.5° (CFD = 58 %).

75These observations suggest that analyzing terrain ruggedness could be enhanced by adopting a multiscale approach, involving the calculation of indices on smaller, predefined spatial grids of the polemoforms. This method would allow the creation of a weighted composite ruggedness index, combining both local and global measurements, while accounting for the internal spatial variability or the shape of the forts, thereby enabling a more accurate assessment.

76Another methodological challenge lies within forts constructed or modernized after 1885 (fig. 19A-E), whose features, adapted to new armor requirements, differ significantly from the 1874-type forts (fig. 8D-E; Guillot, 1949; Simoutre, 1889). These modifications reflect a focus on deeper burying and hardening (concreting, armoring), coupled with volumetric simplification (fig. 19B-D; Durup de Baleine, 1998). The Méroux fort (1906-1913; Belfort area; fig. 19A-E) serves as an example, showing higher ruggedness values than the Reims fortified belt, with slope values matching those of G2, warranting further investigation (fig. 19A). The method for analyzing slope values of the rampart and ditch seems effective, revealing three peaks comparable to the four theoretical modes.

5.2.2. Vertical complexity and invisbility of fort substructures

77A second limitation stems from the subsurface, or "subterranological," nature of the "Séré de Rivières" fortifications, precursors to the Maginot Line in the 20^th century (Truttmann, 2009). LiDAR provides topographic information on the morphology and internal organization of the forts, and even on their volumetry, allowing an assessment of how much a site has been topographically altered. However, it does not directly capture the masonry substructures located beneath a plurimetric layer of fill. Beyond the horizontal complexity of the surface, there is also a more intricate vertical dimension, distinguishing these fortifications from the polemoforms typically analyzed with ALS data. This highlights the essential role of field studies, which complement LiDAR data by characterizing the military elements, shining light on biogeographical aspects, assessing the functional system (Durup de Baleine, 1998), examining the construction stones (Ancelin et al., 2024) and their degradation state, and cross-referencing these observations with archival documents, such as construction plans. In addition, this vertical dimension enhances the relevance of geophysical prospection – already applied to 19th-century fortified sites (Nickel and Hunt, 2002; Joyce and Van Beckum, 2015) – which, while not particularly suited for studying erasure processes, proves valuable for earthen embankments (e.g., for early medieval fortifications; Milo et al., 2022).

6. Conclusion

78The archaeogeomorphological and geoheritage approach of the "Séré de Rivières" fortifications underscores the historical and scientific significance of these structures, which remain underexplored in academic research. This study aligns with three emerging scientific trends in conflict geomorphology: (i) the analysis of war landscapes using GIS and high-resolution ALS data (e.g., Hesse, 2014; Taborelli et al., 2017b; De Matos-Machado, 2018; Gheyle et al., 2018; Taborelli, 2018; De Matos-Machado et al., 2019; Stichelbaut et al., 2023); (ii) the exploration of polemoforms as patrimonial entities valued beyond cultural and memorial dimensions (De Matos Machado, 2018, 2024) despite their controversial associations with "dark heritage" (Biran et al., 2011); (iii) the study of "Séré de Rivières" fortifications through the lens of geomorphological factors influencing their design as well as the geomorphological impacts of their construction and use (Ancelin et al., 2023, 2024, accepted a).

79This research introduces an innovative method to assess the erasure rate of first-generation "Séré de Rivières" polemoforms (type 1874) using high-resolution ALS data. The method addresses the pressing need for standardized practices in the context of widespread national and regional LiDAR coverage (Kokalj and Somrak, 2019; Georges-Leroy, 2022). By establishing morphological indices – ruggedness indices for all forms and a slope index targeting ramparts and ditches – this method enables evaluation of the preservation state, temporal alteration, and contributing factors to the erasure of these structures. In future developments, artificial intelligence may offer promising opportunities to further standardize the classification and interpretation of LiDAR-derived morphometric patterns across fortification sites.

80This analysis identified two distinct fortification groups, revealing a correlation between high ruggedness and slope values with a good state of preservation among Reims’ forts, or conversely, heavily altered polemoforms exhibiting significantly reduced indices. Comparison between the statistical distribution of current slope values (LiDAR data) and theoretical values (fortification manuals; Simoutre, 1889; Guillot, 1949) reinforces this distinction, based on statistical mode analysis. Future research should explore the applicability of this method to post-1885 fortifications, assuming similar levels of standardization. These findings suggest the potential for developing automated detection methods to evaluate erasure levels in other complex polemoforms, such as fortifications. While most current approaches focus on simpler or more repetitive polemoforms (e.g., craters; De Matos Machado et al., 2019), recent advances in object-based and machine learning techniques (Lozic and Štular, 2021) open promising perspectives for the automated identification and morphometric assessment of more complex, multi-scalar, and heterogeneous military landforms – such as the standardized “Séré de Rivières” forts – provided that adequate training datasets and formal geometric criteria are available.

81The results highlight the multifaceted nature of erasure processes, driven primarily by anthropogenic factors – such as integration into WW1 defense networks, biogeographical characteristics, and post-conflict trajectories (e.g., reassignment, abandonment) – and often tied to dismantling for recovery of construction stone. Anthropogenic influences, whether civil, military, or polemological, dominate; with natural factors playing a marginal role. Just as war generates short, intense morphogenetic sequences, human activities are uniquely effective in erasing polemoforms in less than 150 years. Concurrently, human interventions exacerbate natural erosion processes affecting both the masonry substructures (bioerosion, infiltration, frost weathering) and the earthen mass (colluvium, landslides, pedogenesis).

82Beyond their status as military remnants, this study advocates for reevaluating polemoforms as geoheritage assets. It contributes to the discourse on recognition of polemo-geomorphosites (Bétard et al., 2017; De Matos-Machado, 2024), interrogating the integration of these artificial forms into the science of reliefs, tied to societal events and short morphogenetic sequences of centennary or even decadal scales. Incorporating them into conservation and heritage valorization strategies offers an opportunity to deepen understanding and ensure preservation for future generations. These elements are not only historical markers but also geomorphological indicators, reflecting rare, globally significant morphogenetic processes, both rapid and disruptive.

83While the geomorphological study of polemo-landscapes examines the forms, formations, and processes contributing to their creation and destruction (Ancelin et al., 2023), it simultaneously suggests a geoheritage perspective. This approach recognizes these landforms as components worthy of protection and transmission, enriching the geomorphological dimension of cultural heritage.

*Auteur correspondant : +33 (0)3 26 91 36 83 pierre-yves.ancelin@univ-reims.fr (Pierre-Yves Ancelin)