(1) Overview Context
The Kokentau mountain range, located in East Kazakhstan, preserves one of the most extensive concentrations of pigment-based rock art in the country. Although a comprehensive quantitative inventory of Kazakhstan’s rock art has not yet been undertaken, published national surveys consistently indicate that the overwhelming majority of recorded sites consist of petroglyph complexes executed by pecking or engraving with tools [1, 2]. In contrast, pigment-based rock art sites are known only from isolated grottos and individual localities. In this context, the Kokentau assemblage represents the largest systematically documented concentration of painted rock art currently known in Kazakhstan, comprising dozens of shelters and hundreds of figures within a single landscape setting. The Kokentau shelters are situated in an archaeologically rich micro-region, surrounded by Bronze Age settlements and burial sites, suggesting long-term ritual use of the landscape [3].
Until recently, knowledge of these paintings was limited to short mentions in regional surveys and unpublished archives. Before our work began, the literature mentioned only eleven shelters with paintings, without clear geographic coordinates, and illustrated them only with fragmentary photographs of the panels [4].
Systematic documentation of Kokentau began in 2024 within the framework of the project “Interdisciplinary Study of the Oldest Rock Paintings of East Kazakhstan” (Ministry of Science and Higher Education of Kazakhstan). As a result, 75 rock shelters with paintings were documented in 2024–2025.
The Kokentau assemblage expands the known distribution of pigment-based rock art in Kazakhstan and provides new material for understanding early symbolic practices associated with Bronze Age pastoralist communities. The Kokentau assemblage currently represents the most extensively documented corpus of ochre pictographs in the country recorded using modern digital documentation methods. While other sites with ochre pictographs are known elsewhere in Central Asia, none have been surveyed at a comparable spatial scale or with equivalent methodological detail.
Spatial coverage
Location: Kokentau Mountain Range, East Kazakhstan, Republic of Kazakhstan
Northern boundary: 49.86974° N
Southern boundary: 49.84226° N
Eastern boundary: 79.61602° E
Western boundary: 79.58288° E (WGS84)
Description
The Kokentau mountain range is an isolated granite massif located in East Kazakhstan (Figure 1). The study area represents a semi-arid steppe environment with granite outcrops and shelters naturally suited for pigment rock art preservation. The coordinates and boundaries provided here define the systematically surveyed area within the Kokentau massif that contains the highest known concentration of painted shelters. While additional isolated panels may occur beyond the surveyed zone, no comparable concentrations of pigment-based rock art have been identified elsewhere within the massif. The documented area covers approximately 5 km2.
Figure 1
Location of the Kokentau study area in Eastern Kazakhstan. Coordinate grid in WGS 84. Made with Natural Earth [5].
Temporal coverage
Most researchers attribute the Kokentau paintings, along with similar pigment rock art of East Kazakhstan, to the Early Bronze Age (late 3rd–early 2nd millennium BCE) [6, 7, 8, 9]. This attribution is primarily based on stylistic comparisons with other Early Bronze Age rock art assemblages in the region and on the archaeological context of the surrounding landscape, which includes Bronze Age settlements and burial sites. However, the chronological attribution remains debated due to the absence of associated archaeological materials suitable for direct or contextual dating.
(2) Methods
Compilation of sources and preliminary research
All available published and archival materials on rock art in the Kokentau Mountains were reviewed to identify previously recorded localities and to establish a base map of known outcrops.
Field documentation and spatial mapping
The most recent fieldwork conducted during 2024–2025 documented 34 outcrops containing 75 rock shelters, within which a total of 702 pictorial figures and signs were identified (Figure 2).
Figure 2
Distribution of the 34 outcrops documented within the Kokentau massif. Coordinate grid in WGS 84. Made with Copernicus GLO-30 Digital Elevation Model [11].
Geographic coordinates were recorded in the WGS 84 using smartphone GNSS positioning with expected horizontal accuracy approximately ±5 m [10]. The boundaries of rock art clusters were delineated according to geological features and the distribution of pictorial surfaces (Figure 2).
Photographic documentation
Each panel was photographed under natural light using DSLR cameras. For surfaces with faint pigment traces, image enhancement was applied using the DStretch plugin for ImageJ, a Java-based open-source image-processing platform widely adopted in rock art research [12, 13]. Both original and enhanced images were archived in high resolution.
Photogrammetric processing and orthophoto generation
Orthophotos of selected panels were produced using Structure-from-Motion (SfM) photogrammetry implemented in Agisoft Metashape Professional (v. 2.1.0). Orthorectified images were saved in GeoTIFF format to preserve the scale within the image file.
Image enhancement and pigment analysis
DStretch was applied systematically to all recorded panels to enhance the visibility of ochre pigments. Portable X-ray fluorescence (pXRF), a non-invasive analytical technique, was applied to selected shelters and confirmed the presence of iron oxides consistent with natural ochre sources.
Sampling strategy
A systematic survey was conducted by four field teams within the defined study area of the Kokentau granite massif, corresponding to the sector of the mountain range with the highest known concentration of painted shelters and bounded by the coordinates specified in the Spatial coverage section (Figure 2). Preliminary reconnaissance in a radius of 1–2 km around the surveyed area revealed no additional painted shelters with comparable concentrations of pictorial figures, although the presence of isolated pigment traces or non-figurative markings beyond the surveyed zone cannot be ruled out and requires further systematic investigation. After the initial documentation phase, an additional control expedition by a single team re-examined the entire survey area on foot to verify and reconcile all recorded data.
All visibly discernible traces of ochre were documented, including non-figurative elements such as isolated lines, dots, smears, and ambiguous marks that could not be confidently assigned to specific motifs. When ochre residues or pictorial panels were identified on an outcrop, every shelter on that formation was examined using the mobile version of DStretch, enabling in-field enhancement of pigment visibility and precise delineation of the spatial extent of pictographs.
Around 15% of the recorded panels were discovered exclusively through DStretch enhancement and remained invisible to the naked eye under natural light. This dual approach—direct visual observation and real-time digital enhancement—ensured comprehensive detection of pigment traces across the entire study area.
Metadata Embedding
Each photograph retains the original technical capture metadata recorded by the camera (EXIF; e.g., camera/lens information, exposure settings, and acquisition date/time). We additionally used ExifTool to embed descriptive metadata, including the image title, country, geographic coordinates, creators, a description of the fieldwork, and keywords. These fields were written to XMP and duplicated in the IPTC metadata of each image.
Quality Control
Quality control was implemented at all stages of data collection and processing. Four independent field teams followed standardized recording protocols, and a subsequent control expedition re-examined the entire study area to verify coordinates, panel counts, and the completeness of photographic documentation. All photographs were checked for focus, exposure, and scale accuracy; reshooting was carried out when necessary.
Each DStretch-enhanced image was visually inspected by two independent researchers to identify and eliminate any misleading colour artifacts, and all filters (YRE, LDS, CRGB, etc.) were applied according to a unified standardized protocol. The classification of motifs (anthropomorph, zoomorph, geometric and undetermined motifs) was likewise independently reviewed by two team members, with any discrepancies resolved through collective discussion.
Metadata were entered using a standardized set of predefined categories, and all files underwent systematic checks for correct naming, versioning, and linkage between panels, images, and metadata fields.
(3) Dataset description
Object name
The Kokentau Ochre Paintings: An Open Archaeological Dataset
Data type
Primary data, secondary data, processed data and interpretation of data.
Dataset Structure and Contents
The Kokentau dataset is organised according to a three-level spatial hierarchy:
Outcrop – a distinct granite formation visible at the landscape scale.
Shelter – a smaller geomorphological unit within an outcrop (overhang, niche, or shallow cavity) where painted surfaces occur.
Panel – a single painted surface within a shelter, functioning as the primary analytical unit for recording motif counts, pigment traces, preservation, and spatial arrangement.
This hierarchical structure ensures consistent spatial referencing and facilitates integration of the dataset into GIS and comparative studies.
All Kokentau Ochre Painting files are provided under the Creative Commons Attribution 4.0 International License (CC BY 4.0) and may be reused without restriction, provided appropriate credit is given.
The database includes both the original, unprocessed photographs and orthophotos, as well as versions enhanced using the DStretch plugin (decorrelation stretch method). Provision of unprocessed images was intentional as the processing configuration applied in the project is not finite. By providing the originals, we enable other researchers to reprocess the images using different DStretch parameters or alternative software and digital enhancement techniques. We deliberately chose not to include traced drawings of the painted panels in the database. Any such tracings would inevitably be subjective and merely offer an interpretative reconstruction of the imagery and composition. Since the database provides both the original photographs and their DStretch-enhanced versions, other researchers can perform their own visual analyses and produce independent tracings based on the openly accessible primary data.
In total, the database includes:
34 photographs of outcrops (JPEG)
119 original photographs of pictographs (JPEG)
119 DStretch-enhanced photographs (JPEG)
23 orthophotos of pictorial panels (TIFF, orthorectified)
23 orthophotos (JPEG)
23 DStretch-enhanced orthophotos (JPEG)
The dataset is organized following the three-level spatial hierarchy described above and is implemented through a nested folder structure. At the top level, the dataset contains 34 folders corresponding to individual outcrops (named out01, out02, etc.).
Each outcrop folder includes overview photographs of the rock formation and subfolders corresponding to individual rock shelters within that outcrop. Rock shelter folders are named using a compound identifier (e.g., out01_sh01, out02_sh03), where the first number refers to the outcrop and the second to the shelter number within that outcrop.
Within each rock shelter folder, image files corresponding to individual painted panels are stored. For each panel, both the original photograph or orthophoto and a single DStretch-enhanced version are provided. In addition, orthophotos are provided as GeoTIFF files with preserved metric scale. These GeoTIFFs are referenced to the origin of a local coordinate system (units: meters), allowing measurements to be performed directly in GIS software while maintaining the original scale. The enhanced image represents the most informative enhancement selected for that panel during data processing. The naming of processed image files records the enhancement parameters applied; for example, out01_sh01_p02_crgb20 denotes an image from outcrop 1, rock shelter 1, panel 2, processed using the CRGB DStretch decorrelation stretch with a scale value of 20.
All orthophoto-related files include the “_ortho” suffix (e.g. out06_sh02_p04_ortho.tif, out06_sh02_p04_ortho.jpg, out06_sh02_p04_ortho_lds20.jpg). This standardized naming convention ensures transparency, traceability, and reproducibility of image processing and facilitates automated reuse of the dataset.
The dataset also includes an Excel spreadsheet and a GeoJSON file containing an attribute table for all recorded figures, which was used to generate descriptive statistics.
The Kokentau assemblage comprises four main motif categories: anthropomorphic stick figures (273; 38.9%), panel-bodied anthropomorphs (120; 17.1%), zoomorphs (31; 4.4%), and geometric and undetermined motifs (278; 39.6%) (Figure 3). These values include all identifiable motifs as well as non-figurative pigment traces documented during fieldwork, providing a quantitative overview of the composition of the painted corpus.
Figure 3
Motif categories in the Kokentau Ochre Painting corpus and their proportional distribution.
In this study, we employ the descriptive term “panel-bodied anthropomorphs” to denote figures in which a rectangular element is incorporated into the human form in one of two ways: either positioned perpendicular to the torso (overlying it) or placed in the location of the head. What exactly this rectangular element represents remains unclear. It may depict an enclosure, an item of clothing, a structure, or a symbolic element that is not yet fully interpretable. The term is used in a broad, descriptive sense and does not imply that the rectangular shape necessarily represents the body itself. Thus, panel-bodied anthropomorphs serves as a convenient graphic label for a recurring formal combination—a schematic stick figure combined with a rectangular motif—without presupposing its functional or semantic meaning.
Anthropomorphic imagery constitutes the largest component of the Kokentau corpus. Together, stick figures and panel-bodied anthropomorphs comprise 56% of all documented figures. Against the background of the Eastern Kazakhstani petroglyphic tradition, typically executed in pecking or engraving techniques, the painted panels of Kokentau appear highly distinctive. The imagery does not include explicit scenes of conflict or warfare that are common in many pecked or engraved Central Asian rock art complexes. Instead, the compositions are dominated by anthropomorphic figures and geometric elements arranged in extended linear or grouped formations. Particularly noteworthy is the frequent depiction of both anthropomorphic types—stick figures and panel-bodied anthropomorphs—shown in close spatial association, often forming long chains or continuous sequences, which may reflect collective or coordinated activities.
Format names and versions
TIFF, JPEG, GeoJSON, XLSX
Creation dates
May 2024 – October 2025
License
The dataset is released under the Creative Commons Attribution 4.0 International License (CC BY 4.0), which permits unrestricted use, distribution, and reproduction in any medium, provided appropriate credit is given.
Repository location
Publication date
23.12.2025
(4) Reuse potential
The Kokentau Ochre Paintings dataset offers extensive potential for reuse across multiple research domains. Because the collection provides both raw and processed imagery (including DStretch-enhanced versions) as well as structured metadata and orthorectified photogrammetric outputs, it can serve as a methodological and analytical resource in several ways.
First, the dataset enables comparative studies of pigment-based rock art in other regions of the world, allowing researchers to examine motif distributions, anthropomorphic variability, and stylistic parallels such as with other Bronze Age painted traditions. Second, its documented spatial referencing supports GIS-based landscape and visibility analyses, including assessments of site clustering, shelter orientation, proximity to Bronze Age settlements, burial sites, and natural resources. Third, the availability of both raw and enhanced images makes the collection suitable for archaeometric and pigment provenance research, including reprocessing with alternative DStretch parameters or machine-learning–based classification of pigment traces.
Beyond archaeological interpretation, the dataset provides high-quality material for digital heritage applications, such as the development of virtual reconstructions, interactive museum displays, and educational resources. It is also suitable for training machine learning and computer vision models on pigment detection, surface segmentation, and feature recognition—applications that are increasingly important in rock art research.
By adhering to FAIR principles—ensuring findability, accessibility, interoperability, and reusability—the dataset is designed for long-term preservation and seamless integration into broader archaeological repositories and digital research infrastructures.
Competing Interests
The authors have no competing interests to declare.
Author Contributions
Zaytseva O. V.: writing – original draft, funding acquisition, project administration, conceptualization, methodology, data curation; Vodyasov E. V.: conceptualization, data curation, writing – review & editing, supervision; Doumani Dupuy P. N.: writing—review and editing, conceptualization; Zhuniskhanov A. S.: data curation, resources; Vavulin M. V.: methodology, data curation, visualization, figure preparation; Rakhmankulov Ye. Zh.: data curation, resources; Pushkarev A. A.: data curation; Polovtsev M. Yu.: data curation. All authors have read and agreed to the published version of the manuscript.