Multi-temporal satellite images embedded in GoogleEarth Pro were used to determine changes in the vegetation pattern across a time span of almost 25 years (Fig. S1 in the Supplement). Morphometric investigations were supported by the digital elevation model DGM1 of the Federal State of Baden-Württemberg, established from 2000 to 2005 (LUBW, 2024). All GIS-based investigations were carried out using QGIS software v. 3.34.8.

Geophysical prospection was based on 2-D ERT using a GeoTom MK1E100 device with Schlumberger and Dipol-Dipol configurations, with each configuration being measured with 100 electrodes and 0.5 m spacing, as described in Kneisel (2003). The ERT line from NNW to SSE runs perpendicular to the investigated linear anomaly. ERT data were post-processed by the calculation of standard inversions without filtering using Res2Dinv software. Erroneous data points, e.g., due to electrodes disconnecting during the measurement, were deleted from the raw dataset prior to data modeling.

Two pits were dug down to a depth of 170 cm on plot (Flur) 57 (“Rindlache”), parcel (Flurstück) 153. Pit 1 in the southwestern part of the land parcel and pit 2 in the northeastern part of the land parcel (Figs. 2A, C and S5) both cut the anomalies and allow for a three-dimensional study of the stratigraphies, which was described in the field according to Ad-hoc AG Boden (2005) and the Munsell Soil Color Charts (Munsell Color Company, 2000). Samples were taken from both stratigraphic sections.

Samples taken at pit 2 were dried, carefully pestled by hand, and dry-sieved (2 mm). Grain size distributions of the <2 mm fraction were measured using a laser particle sizer (Fritsch Analysette 22 NeXT Nano) at the Laboratory for Geomorphology and Geoecology, Institute of Geography, Heidelberg University. Samples were pre-treated with H₂O₂ (30 %) to dissolve organic matter and Na₄P₂O₇ (55.7 g L⁻¹) to dissolve soil aggregates. Univariate statistical measures were calculated using GRADISTAT v9.1 (Blott and Pye, 2001). Organic matter (LOI550) and carbonate content (LOI950) were determined by loss-on-ignition (LOI) following a slightly modified protocol from Heiri et al. (2001). Samples of 3–5 g were combusted in a muffle furnace, first at 550 °C for 4 h and then at 950 °C for 4 h. C, N, and S were simultaneously measured using high-temperature combustion (Elementar VarioMax organic elemental analyzer) on aliquots homogenized using a mixer mill (Retsch MM 301). X-ray fluorescence (XRF) spectroscopy analysis was carried out using a Thermo Scientific Niton XL3t handheld device inside a radiation-proof measurement chamber. For XRF analysis, dried, homogenized aliquots were pressed into pellets with a perfectly smooth surface to guarantee a high reproducibility of the measurements. Several NIST (National Institute of Standards and Technology) soil standards were measured together with the samples for data quality control. After centered log ratio transformation following Bertrand et al. (2024), the XRF data were subjected to principal component analysis (PCA) using Past software v4.16 (Hammer et al., 2001) to support the facies model.

To supplement facies interpretation and identify the source area of the sand found in the pits, particle shape, grain size, and mineral composition were qualitatively studied under a Keyence VHX-2000 digital microscope with reflected light for five samples. Morphoscopic interpretation was mainly based on Liang and Yang (2023).

From the sand unit creating the linear anomalies, six samples in total were taken for OSL dating in lightproof steel cylinders (samples HDS-1818 to HDS-1820 from pit 2; samples HDS-1821 to HDS-1823 from pit 1; Table S2). As experiments on quartz and feldspar minerals showed that quartz bleaches significantly faster than feldspar (Godfrey-Smith et al., 1988), we extracted quartz coarse grains (125–212 µm) from the OSL sediment samples for equivalent-dose (De) determination. Sample preparation occurred under subdued red light in the Heidelberg Luminescence Laboratory following standard procedures (Fig. S13). For the potentially poorly bleached samples, the analysis of single grains or subsamples (aliquots) with only a few grains would be preferential to identify the well-bleached grains in a population of variably bleached grains. However, the relatively dim luminescence signal of the samples only allowed us to investigate “small aliquots” (∼ 10² grains each; ø 4 mm; grains strewn on aluminum cups, ø 10 mm, and fixed with silicon oil). The measurements were performed on an automated luminescence reader, model TL/OSL DA15 (upgraded to DA20; Bøtter-Jensen et al., 2000; DTU Physics, 2021), applying a blue-light-stimulated luminescence (BLSL) single-aliquot regeneration (SAR) protocol (Murray and Wintle, 2000) (Fig. S14) and detecting the UV quartz emission around 340 nm through a set of glass filters (3× U340, Hoya, 2.5 mm each). BLSL stimulation occurred for 40 s at 125 °C (400 data channels; 0.1 s per data channel). Potential feldspar contamination was tested in a final cycle of the SAR protocol by IR stimulation prior to BLSL (OSL IR depletion ratio) (Duller, 2003). De determination was performed with the Luminescence Analyst software, v4.31.9 (Duller, 2015). De's were calculated for the integrals 0–0.1 and 0–0.4 s, using 30.1–40 s for late-light subtraction. Central doses and minimum doses (Galbraith et al., 1999) for central age model (CAM) and minimum age model (MAM) age determination were calculated using the functions of Burow (2023a, b) with the R package “Luminescence” (Kreutzer et al., 2012). Radionuclide determination was done using the μ dose system (Tudyka et al., 2018; Kolb et al., 2022). The cosmic dose rate was assessed according to Prescott and Hutton (1988) and Hutton and Prescott (1992) with a function of the R package Luminescence (Burow, 2024). For effective dose rate calculation, the ratio of a sample's wet weight to dry weight (Δ) was assumed to be 1.05±0.05. CAM and MAM ages (Galbraith et al., 1999) were calculated based on data in Table S13. In addition, for graphical representation, apparent OSL ages of the individual aliquots, derived by dividing the De's of the individual aliquots by a sample's mean dose rate, were plotted in ascending order of OSL ages (Fig. 5). All methodological details are provided in Supplement Sect. S3.

Information on the historical land use of the study site was acquired from historical archives of the state of Hesse (Hessisches Landesarchiv, Hessisches Staatsarchiv Darmstadt). We searched digital copies (Digitalisate; available at Arcinsys Hessen; https://arcinsys.hessen.de, last access: 9 January 2026) of files on land use in the area of Viernheim (G 15 Heppenheim V 598) which include official permits; directories of names of owners of land parcels (Namensverzeichnis zum topographischen Güterverzeichnis von Viernheim; H 23 No. 3491); and, for the geographical allocation and matching of the historical plans with modern land parcel numbers, an atlas of the land parcels (Parzellenatlas; P 4 No. 2737).

Noticeable changes in vegetation vitality based on multi-temporal GoogleEarth imagery are visible during different seasons on the following dates: 1 June 2000, 19 April 2015, 8 July 2018, and 30 June 2019 (Figs. 2A and S1). They all show a linear pattern of pale-green and yellowish colors versus fully green colors running from SSW to NNE. The linear pattern covers an area of approximately 100 m ×50 m. While stripes in the eastern part are less pronounced, the western part exhibits incomplete and shorter stripes.

The maximum width of pale-green stripes (up to 3.5 m) occurs in the central part. The DEM shows that the surface of the linear pattern (“upper-level surface”) lies ca. 0.4 m above the present-day surface of the neighboring paleo-channel (“lower-level surface”).

The transect ERT-19 reaches down to ca. 7 m below ground level (b.g.l.). For the Schlumberger configuration (Fig. 2D), the base between 7–4 m b.g.l. consists of low resistivity values between 20 and 75 Ωm with a horizontal layering. Between 4–2 m b.g.l., the lowest resistivities are in the range of 10–20 Ωm, also reflecting horizontal layers. The only area with higher values at this depth is at 9–10 m horizontal distance, where resistivity reaches values of 200–300 Ωm. The uppermost 2 m are characterized by alternating resistivity values varying between sections of 10–30 and 300–600 Ωm. Overall, there are nine circular high-resistivity anomalies from NNW to SSE, coinciding with the individual stripes. Between 38 and 49.5 m horizontal distance, sections of high resistivity are less pronounced. The Dipol-Dipol configuration shows a similar pattern, with less pronounced horizontal layering for the lower units (Fig. S2).

The sampling pits (Fig. 2C) cut the linear anomaly. Thus, the profiles show disturbed (pit 1 and eastern profile of pit 2) and undisturbed (western profile of pit 1) stratigraphic sections. The general stratigraphic pattern is very similar in both pits; here, we describe the undisturbed part based on the western profile of pit 2 (Fig. 3). The stratified, light-yellowish-brown base unit S10 (170–140 cm b.g.l.) consists of carbonate-rich (LOI950 >20%), moderately to poorly sorted sandy silt. Iron oxidation occurs along root casts, whereas organic content is moderate (LOI550 ∼ 12 %). S10 represents a feF ° Gor horizon according to the updated German classification of pedological horizons in Ad-hoc AG Boden (2024) (see Table S1 for further explanations of pedological horizons). It is overlain by a black peat (S9 [fnH], 140–104 cm b.g.l.), void of carbonate but very high in organic matter (LOI550 ∼ 70 %), S (0.28 %), and N (2.55 %), whereas the Fe/Mn ratio is low. The overlying black unit S8 (eMm, 104–98 cm b.g.l.) consists of an organic-rich (LOI550 ∼ 24 %), moderately sorted clayey silt, high in S (0.23 %) and Fe/Mn but low in carbonate (LOI950 ∼ 3.8 %). It is overlain by poorly sorted black to dark-gray fine sandy silt (S7 [eMm], 98–61 cm b.g.l.) with similar carbonate (LOI950 ∼ 3.5 %) and lower organic (LOI550 ∼ 6.7 %) contents, whereas the Fe/Mn ratio is high. It contains white fine roots and some gastropod shell fragments. Unit S6 (61–49 cm b.g.l.) is a poorly sorted, carbonate-rich (LOI950 ∼ 20 %) loam, showing a mottled pattern resulting from iron oxides and carbonate precipitation, especially around root casts (Fig. S8). S6 (Gor – eMcm) has the highest number of gastropod shells and shell fragments out of the entire stratigraphy. Unit S5 (Gor – eMm, 49–40 cm b.g.l.) consists of yellowish dark-gray, poorly sorted sandy loam with a gritty texture, iron stains, and abundant gastropod shell fragments. The carbonate content is relatively high (LOI950 ∼ 11 %), and the organic content is also increased (LOI550 ∼ 15 %). Unit S4 (Dj, ∼ 40 cm b.g.l.; up to 3 cm thick) is a yellowish-brown, well-sorted thin and discontinuous medium sand with small dark-brown mud clasts and very sharp boundaries. Organic (LOI550 ∼ 1.6 %) and carbonate (LOI950 ∼ 1 %) contents, as well as S (0.01 %) and N (0.06 %) values, are relatively low. It is overlain by unit S3 (Dj, 40–29 cm b.g.l.), a very dark-gray, poorly sorted loamy sand with abundant iron stains, root penetration, and gastropod shell fragments (plus one entirely preserved shell). S2 (Dj, 29–25 cm b.g.l.) resembles the medium sand of S4, including the discontinuous nature, very sharp undulating boundaries, and fine mud clasts (Fig. S7). The top layer (S1 [Dj], 25–0 cm b.g.l.) consists of poorly sorted, very dark-gray silty loam with abundant roots, few angular to subrounded gravel components, a brick fragment, and a few fine gastropod shell fragments.

At the opposite wall, the eastern profile of pit 2, a massive, unstratified unit of medium sand cuts into the stratigraphic pattern down to unit S9 (Figs. 2C, 4b). Underneath the sand deposits, Unit S10 consists of the same carbonate-rich sandy silt as in the undisturbed section of pit 2. The grain size distribution, sorting, and geochemistry of the unstratified sand strongly overlaps with those of S2; it furthermore contains mud clasts and drapes of mud (Figs. S3, S4). It is only covered by S1, with some grayish humic material incorporated into the upper part of the infill (Figs. 2C and S5).

The northern profile of pit 1, ca. 50 m south of pit 2, shows the same disturbed stratigraphy as the eastern profile of pit 2, starting with unit S10 at the base, followed by a massive, unstratified unit of medium sand and covered by unit S1. In pit 1, the sand infill also contains some centimeter- to decimeter-scale mud clasts and mud lenses. Locally, an up to 3 cm thick layer of whitish, powdery carbonate was observed at a depth of around 35 cm b.g.l (Fig. S5). Furthermore, a hollow brick (Hohlziegel) was observed in the uppermost part of the light-colored sand infill (Fig. S6), identified as a survey point of the land consolidation campaign Viernheim DF 459 from the 1970–1980s (personal communication, Dieter Hogen, Amt für Bodenmanagement, 23 September 2020).

Microscopic investigations of sand grain characteristics show great similarities with the samples from the Bettenberg dune. Both the massive sand unit of the linear anomalies and the discontinuous sand layer S2 in pit 2 show well-rounded to very well-rounded and frosted quartz grains, slightly soiled, with only a few angular and shiny grains. Grain sizes are exclusively in the fine and medium sand fractions (Fig. 4a, b, Table 1). In contrast, samples from the active BSN channel (samples RL09/16 and RL09/23 from Engel et al., 2022) reveal rounded and angular, predominantly shiny and dirty or coated quartz grains in the fine and medium sand fractions, as well as calcite grains in the silt and fine sand fractions (Fig. 4c).

In total, 236 aliquots of the six OSL samples from the two pits gave acceptable results. OSL ages for the six different samples calculated following the CAM range from approximately 2.1 ka (HDS-1821) to approximately 6.5 ka (HDS-1823), while OSL ages calculated following the MAM range from approximately 0.6.–0.7 ka (HDS-1822) to approximately 3 ka (HDS-1820). The results are based on 20 aliquots (HDS-1818, HDS-1920) to 113 aliquots (HDS-1822) per sample passing the acceptance criteria for the De determination (Sect. S3). The apparent ages of the individual aliquots of all six samples are given in ascending order of OSL ages in Fig. 5A. The sandy material used for the infilling of the trenches may have been of pre-Holocene age, possibly up to LGM age. Aliquots pointing to material handling within the last 1000 years are given in Fig. 5B.

A file on brickyards and brick manufactures (Akte zu Ziegeleien und Backsteinbrennereien in Viernheim; G 15 Heppenheim V 598, 1875) lists several brickworks (licensed to the names of Kühner, Bläß, Samstag, and Martin); however, these are located outside the study site (former cadastral sections VII and I of Martin). A license for the brickworks site of Peter Minnig from 1865 CE (digital version, image 88) does not mention the land parcel studied here. However, attached to it is a plan with land parcels adjacent to the border between Hesse and Baden-Württemberg (formerly Großherzogthum Baden) on the Hessian side (digital version, image 90). Land parcel no. 2 is marked as a field of the mayor Michael Keller, and no. 3 (Klafter 1473) is marked as a field of P. Minnig. Therefore, in the directory of the topographical list of land parcels (Namensverzeichnis zum topographischen Güterverzeichnis von Viernheim; H 23 No. 3491) we searched for land parcels of Michael Keller and P. Minnig, which are located in the same cadastral section and carry similar plot numbers. These criteria are fulfilled by plot no. 34 (M. Keller) and 33 and 41 (P. Minnig) in the cadastral section XXI. The position of the former cadastral section XXI was retrieved from the plot atlas (Parzellenatlas; P 4 No. 2737). The respective land parcels are found in section A, image 21 (Abteilung A, Bild 21) (Figs. S9, S10). The former land parcel no. 33 (owner P. Minnig; see reference above) covers the linear anomalies of the present-day land parcel 153. In addition to the plan, the conditions under which a license for clay mining was granted are detailed:

“Field no. 1 is the plot on which the brickworks site shall be established. The plot is situated 1/2 hour south of Viernheim, and it is 450 feet wide and 360 feet long. The substrate for the brick manufacture will be excavated in the center of the plot and the pits will be refilled immediately. From time to time the bricks are put together approximately in the center of the plot and burned with stone coal. Viernheim 1^st June 1865” (for the original wording in German, see Sect. S2).

The anthropogenic excavations and sand infilling inferred from the historical sources (Sect. 4.5) reach the depth of the black clay (S8) or peat (S9), respectively. This observation suggests that the reason for the trenching was the mining of the clay-rich units S6 to S8. The immediately underlying peat uncovered during clay mining may have been used as an energy source as peat still used to be a relevant burning material in the northern URG at that time (Mone, 1826; Tasche, 1858).

The trench walls from pit 2 are slightly sloped or sometimes stepped-sloped towards the interior of the trench, and they were most likely dug by hand with a spade or a similar tool. Excavators or other equipment were introduced for peat mining only after 1865 CE (Günther, 2020). Not all trenches run continuously from end to end. Several of the outer trenches on the northwestern side run parallel to the other trenches but represent shorter and disconnected trench sections in a line. This observation supports the interpretation that the trenches were dug manually, perhaps with different people starting at different sections along a predefined line and not always excavating until the next dugout was reached. It also suggests that the trenching and the immediate refilling of the excavated trenches were done one after another.

Activities of clay mining are known of in the area from at least the Middle Ages. Approximately 850 m to the northwest of land parcel 153, a former clay pit of approximately 100 m ×150 m areal size and 3 m depth is still visible as a rectangular depression in the present-day landscape. The site name, “Egelsee”, refers to a clay pit once drowned by groundwater and populated by leeches, as documented for 1482 CE. Along with the brickwork site of 1865 CE, other former brickwork sites are mentioned for Viernheim, less than 2 km away from Rindlache, for the 19th century. Licenses for clay manufactures in the study area are documented for the periods 1827–1828, 1837, 1864–1873, and 1898–1905.

The general use of river clays for mud-brick production, however, dates back to PPNB (Pre-Pottery Neolithic B) contexts ca. 9500 years ago in the southwestern Fertile Crescent (sun-dried brick) and the early 5th millennium by the Indus civilization (baked bricks) (Stordeur and Khawam, 2007; Van Beek and Van Beek, 2008; Gibling, 2018). In western Europe, baked mud bricks were introduced during the time of the Roman Empire (Van Beek and Van Beek, 2008; Lehouck, 2019), with traces of systematic clay mining in Holocene floodplains both in the Roman Heartland (Scalenghe et al., 2015; Vanzani et al., 2025) and also north of the Alps, with examples in Germany (Obrocki et al., 2020), Switzerland (Maggetti and Galetti, 1993), or Britain (Claughton, 2016). Mud bricks as building material disappeared after the Romans left, mostly until the 12th century CE (Lehouck, 2019), when clays of the Holocene floodplain were used again for mud-brick production throughout western Europe (Van Dinter, 2013; Claughton, 2016; Goemaere et al., 2019). The demand for local clay sources must have been high since then; the present study shows that, still in the 19th century, these resources were exploited at a local scale as soon as they were discovered or became accessible – e.g., through human-induced lowering of the groundwater table, as in the present case.