Integrated multispectral remote sensing and soil-gas measurements for exploring natural hydrogen near the Kempirsai ultramafic massif, western Kazakhstan

Abstract

Natural hydrogen (H₂) has recently attracted increasing interest as a potential low-carbon energy resource. In this context, the surface expressions and geological controls governing hydrogen generation and migration require more in-depth investigations, particularly in ultramafic terranes. This study focuses on the north–central Aktobe region of western Kazakhstan, along the margin of the Kempirsai ultramafic massif, where extensive serpentinized ultramafic rocks favor hydrogen-related redox processes. A multi-proxy remote-sensing approach is integrated with soil-gas measurements, and petrographic and geochemical data to identify surface anomalies and propose hydrogen sources and migration pathways. At the regional scale, spectral indices consistently align with lithological boundaries, suggesting strong lithological and structural control over the distribution of surface anomalies. Focused analysis of the Bogestay–Kenoy area reveals clustered moisture, redox, salinity, vegetation, and thermal anomalies associated with sub-circular surface depressions where soil-gas surveys record hydrogen concentrations exceeding background levels and locally surpassing 50 ppm. Petrographic and geochemical results show that the serpentinized harzburgite in the study region, containing olivine and chromite, provides a reactive mineralogical framework for Fe^(II)-driven redox reactions and sustained hydrogen generation during water–rock interaction. Moreover, the area also hosts granitic intrusions which may provide a secondary source of hydrogen via water radiolysis. These results support a coupled hydrogen system in which ultramafic rocks act as the main generation domain. The study demonstrates that integrated remote sensing, when constrained by geological reasoning and field validation, provides an effective and transferable approach for screening and prioritizing natural hydrogen prospects.

Supplementary Information

The online version contains supplementary material available at 10.1038/s41598-026-55055-1.

Introduction

Natural hydrogen (H₂) has recently attracted increasing scientific and industrial interest as a clean and continuously generated energy resource in specific geological contexts^1–5. Natural hydrogen forms within the Earth through a range of geological reactions, including the serpentinization of ultramafic rocks, radiolysis of water, and deep water–rock interactions^3,6. These processes can generate hydrogen over geological timescales and sustain its migration toward the surface through fractures and permeable zones. Despite being considered a carbon-free energy source, exploration of natural hydrogen is still at an early stage, perhaps due to a poor understanding of hydrogen sources, migration and accumulation mechanisms (hydrogen systems), as well as a lack of clear exploration strategies, including the standardized detection methods. Most existing studies have focused on geochemical sampling, borehole gas measurements, or subsurface modeling, whereas the identification of surface indicators that relate to subsurface processes has received far less attention^7–14. Remote sensing provides an effective and non-invasive tool to address these challenges by detecting indirect surface manifestations associated with subsurface hydrogen migration^15,16. Variations in vegetation vigor, soil moisture, oxidation state, and surface temperature can serve as proxies for gas seepage and fluid migration zones^8,17,18. Multispectral and thermal satellite data enable the mapping of such anomalies with high spatial consistency, offering a cost-effective alternative to field-intensive exploration. While remote sensing has been widely applied in hydrocarbon and mineral exploration^19–24, its application in natural hydrogen exploration requires further research and development. Establishing a remote sensing framework for natural hydrogen exploration may thus provide an essential first step toward identifying potential seepage zones and guiding exploration field surveys.

Over the past decade, natural hydrogen seeps were correlated with rounded sub-circular depressions (SCDs), commonly named “fairy circles”. The earliest documented observations were reported in Russia, where elevated H₂ concentrations accompanied by minor methane traces were interpreted as surface manifestations of subsurface hydrogen-generating processes²⁵. Subsequent studies in North Carolina, USA, and Amadeus Basin in Australia confirmed similar hydrogen anomalies in comparable geomorphic contexts^5,26. More recently, fairy circles associated with hydrogen concentration anomalies have been described across several continents, including Brazil²⁷, Namibia^6,28, and Australia^8,29. Although hydrogen concentrations in the soil can exhibit spatial and temporal variability^25,27 when present, SCDs are widely regarded as surface indicators of hydrogen migration originating from the generating rocks but more likely from subsurface reservoirs or gas-bearing aquifers^29,30.

Notably, most SCDs occur in regions underlain by iron-rich Precambrian metamorphic and igneous basement rocks, where diverse geochemical and radiolytic processes act as primary drivers of molecular hydrogen generation. Among these processes, Fe²⁺ oxidation, serpentinization reactions, and water radiolysis are widely recognized as the most efficient mechanisms responsible for continuous H₂ release from the crystalline basement^5,29,31. Consequently, such surface expressions have emerged as key first-order indicators in the search for natural hydrogen systems. To enhance the practical exploration significance of SCDs, it is imperative to establish robust and standardized methodologies for their detection, mapping, and monitoring. The integration of these geomorphological features with multispectral and remote sensing proxies offers a promising pathway to refine hydrogen prospectivity assessments. This approach not only facilitates accurate spatial identification of hydrogen-emitting zones, but also provides insights into their temporal evolution, thereby distinguishing actively degassing structures from relic or inactive ones.

Kazakhstan is one of the few countries worldwide that hosts a wide spectrum of natural hydrogen–generating rock types (H₂-GRTs) (Fig. 1). These include extensive ultramafic massifs, uranium-rich rocks, coal-bearing basins, and mature hydrocarbon provinces, within which natural hydrogen seepages have previously been documented³. SCD’s are also widespread across these geological domains. However, the potential genetic relationship between these depressions and active hydrogen seepage has not yet been systematically investigated. The present study focuses on the Aktobe region of western Kazakhstan, where the Kempirsai ultramafic massif forms part of the Southern Ural orogenic belt, which is in Kazakhstan commonly referred to as the Mugozhdar Range (Fig. 2). The area is characterized by extensive ultramafic and mafic complexes, major fault systems, and abundant surface depressions, collectively providing a favorable geological framework for hydrogen generation, migration, and possible accumulation. Despite the promising setting, no previous research has applied remote sensing techniques to identify or evaluate hydrogen-related surface indicators in the region, underscoring a significant knowledge gap that this study aims to address.

Fig. 1.

A geologic map illustrating the distribution of different hydrogen-generating rock types H₂-GRTs in Kazakhstan and the already reported occurrences of H₂³.

Fig. 2.

A schematic geologic map of the study area including the Kempirsai ultramafic massif and the surrounding sedimentary and igneous terrains of the Mugozhdar zone and Mugozhdar microcontinent. The Bogestay–Kenoy region (on the right) in the southeastern part of the study area is characterized by the extensive occurrence of SCD’s. A-B shows the location of the geological cross-section displayed in Fig. 14.

In this study, we apply an integrated remote sensing framework to identify potential hydrogen seepage zones in the study area. The approach combines multispectral, thermal, and topographic datasets processed within the Google Earth Engine (GEE) environment. Rather than focusing on individual spectral indices, the analysis emphasizes the spatial relationships among vegetation, moisture, oxidation, and thermal variations as surface expressions of underlying geochemical and structural processes. Topographic data derived from the Shuttle Radar Topography Mission (SRTM) were incorporated to provide geomorphological context and assess potential pathways for fluid migration. This integrated approach demonstrates how remote sensing can serve as an effective indicator of subsurface hydrogen-related processes, offering a new screening tool for early-stage exploration.

Geologic setting

The Kempirsai represents one of the most extensive ultramafic massifs within the Ural orogenic belt. The massif is an elongate, north–south–trending ophiolitic body covering an area of approximately 200 km² and extending for nearly 90 km along its principal axis (Figs. 1 and 2). The Kempirsai massif hosts globally significant chromium ore deposits. These orebodies, among the world’s largest known occurrences within ophiolitic complexes, constitute the second-most abundant chromite reserves after those of the Bushveld Complex (South Africa)^32,33. Chromite deposits in ophiolitic settings are commonly classified as podiform due to their highly irregular geometries, lenticular to pod-like morphologies, and their typical enclosure within dunite bodies. Such characteristics sharply contrast with the tabular and laterally extensive chromitite reefs found in stable, layered mafic–ultramafic intrusions. These chromite deposits may play a crucial role in the generation of abiotic gases via water-rock interactions^34,35. The Kempirsai ultramafic massif lies within the Sakmara Allochthon of the Central Ural Uplift. It overlies the West Uralian zone, which comprises shelfal sedimentary successions and is bounded to the east by the Main Ural Deep Fault³². The Uralian orogen (Mugozhdar Range), to the east of the Sakmara’s ultramafics, includes the thrusted west Mugozhdar Zone and Mugoshdar microcontinent, characterized by complex tectonic and magmatic evolution associated with the development and closure of the Uralian palaeo-ocean^36,37.

The Sakmara Allochthon comprises a heterogeneous assemblage of sedimentary, metavolcanic, and metamorphic units forming part of the broader Uraltau anticlinorium. Between the major ophiolite massifs and the Main Ural Fault, a narrow belt of metasedimentary rocks and meta-ophiolitic sequences is exposed, reflecting the polyphase deformation, accretion, and metamorphism characteristic of the Uralian orogen^36,38,39. The magmatic stratigraphy of the Kempirsai massif and surrounding region represents a fully developed ophiolitic succession, encompassing the entire lithospheric sequence from upper-mantle to supra-crustal levels (Fig. 2). It includes fertile mantle peridotites transitioning upward into depleted harzburgitic tectonites, overlain by ultramafic–mafic cumulate assemblages, a well-preserved sheeted dyke complex, and laterally extensive basaltic lava flows capped by deep-marine to pelagic ocean-floor sedimentary units^38,40.

Methodology

This study employed an integrated workflow combining satellite data acquisition, spectral index computation, and geospatial analysis to identify surface indicators potentially associated with natural hydrogen systems (Fig. 3). Moreover, soil gas measurements are integrated with the remote sensing data to verify the existence of natural hydrogen seeps inferred from remote sensing analysis. Mineralogical and petrographic analysis of ultramafic lithotypes exposed in outcrops near the Aktobe region (Figs. 1 and 2) was conducted to confirm their capability to generate natural hydrogen.

Fig. 3.

A schematic flowchart illustrates the integrated workflow used in the current study. The integrated approach encompasses data collection, processing, spectral index derivation, and final map generation for the study region.

Remote sensing analysis

Data sources and preprocessing

Multispectral Sentinel-2 Level-2 A data (bands B2, B3, B4, B8, B11, B12) were used to derive optical and shortwave infrared indices (supplementary materials, Appendix 1). Thermal data from Landsat-8/9 Collection 2 Level-2 (band B10) were used to support surface temperature (TIR) analysis, while Shuttle Radar Topography Mission (SRTM) DEM data provided topographic context for geomorphological interpretation. All datasets were accessed and processed directly within GEE, which enables cloud-based data handling and automated large-scale analysis⁴¹. Images were filtered by the Area of Interest (AOI), acquisition period (June–September 2024), and cloud coverage (< 10%) to minimize atmospheric and seasonal effects. Median composites were generated to produce cloud-free images representative of the dry season. Cloud masking was applied to remove residual artifacts and ensure clean surface reflectance before index computation.

Spectral indices and image processing

Several spectral indices were computed to highlight surface characteristics potentially associated with hydrogen seepage zones. The selected indices are sensitive to variations in vegetation vigor, surface moisture, oxidation state, and mineral alteration, all of which may indicate near-surface geochemical or thermal activity caused by hydrogen seepage. The Normalized Difference Vegetation Index (NDVI), calculated as (NIR − Red)/(NIR + Red), was used to assess vegetation health and density^42–45. Healthy vegetation exhibits high NDVI values due to strong near-infrared (NIR) reflectance and red-light absorption, whereas low NDVI values indicate stress or sparse vegetation, serving as an indirect indicator of soil chemistry alteration and possible gas-induced stress.

The Normalized Difference Moisture Index (NDMI), calculated as (NIR − SWIR)/(NIR + SWIR), was applied to detect variations in surface and vegetation moisture^46–48. Water strongly absorbs shortwave infrared (SWIR) radiation while reflecting near-infrared (NIR), making NDMI effective for mapping moisture content. High NDMI values indicate wetter soils or vegetation, potentially marking areas influenced by fluid discharge or subsurface moisture accumulation, whereas lower values suggest dryness or moisture stress. The Normalized Difference Soil Index (NDSI) was used to highlight soil brightness and alteration features related to salinity, hydrothermal activity, and possible fluid interaction^49–52. It was calculated using two formulations: (Red − NIR)/(Red + NIR) and (Green − NIR)/(Green + NIR), where Red, Green, and NIR represent the respective spectral bands. Saline or altered soils typically reflect more red or green light and absorb NIR, resulting in higher NDSI values. Using both formulations enhances detection accuracy across varying surface conditions and vegetation covers by capturing complementary spectral responses of altered or saline materials.

The Iron Oxide Ratio (IOR), calculated as Red/Blue, was employed to detect and map iron oxide minerals and surface oxidation zones^21,53–56. Iron oxides such as hematite and limonite strongly absorb blue light while reflecting red wavelengths, resulting in higher IOR values over oxidized or altered surfaces. This index is effective for identifying zones of chemical alteration and oxidation potentially linked to fluid migration or gas-related processes.

The Ferrous Mineral Index (FMI), calculated as SWIR/NIR, was used to detect and map ferrous iron–bearing minerals^54,55. These minerals, such as olivine, pyroxene, and chlorite, exhibit strong absorption in the shortwave infrared (SWIR) region relative to near-infrared (NIR) reflectance. High FMI values indicate enrichment in ferrous minerals, typically associated with ultramafic or altered rocks where hydrogen may be generated through serpentinization or reductive processes. Lower FMI values are commonly observed in oxidized or moist areas, complementing IOR and NDMI patterns.

The Thermal Infrared (TIR) data, derived from Landsat 8–9 Band 10 (ST_B10), were utilized to map surface temperature variations. This band measures emitted thermal radiation within the 10.6–11.2 μm range and was converted to land surface temperature (°C) using USGS-provided scaling factors and offsets. High-resolution TIR data enable detection of localized thermal anomalies that may indicate zones of fluid migration or exothermic serpentinization reactions, providing complementary insight into subsurface thermal and geochemical activity^57,58. The Digital Elevation Model (DEM) derived from the Shuttle Radar Topography Mission (SRTM) dataset was employed to analyze topographic variation and structural control across the study area^59,60, while Principal Component 1 (PC1), derived through eigen decomposition of Sentinel-2 bands (B2, B3, B4, B8, B11, and B12), enhanced lithological contrasts and captured the majority of the spectral variance (78.84%)^61,62. The indices, their equations, and corresponding spectral bands are summarized in Table 1.

Table 1.

Summary of the spectral indices used in this study, including their full names, computation methods, and corresponding Sentinel-2 or Landsat bands.

Index	Name	Equation/Method	Data source/Bands
NDVI	Normalized Difference Vegetation Index	(NIR − Red)/(NIR + Red)	Sentinel-2: B8, B4
NDMI	Normalized Difference Moisture Index	(NIR − SWIR1)/(NIR + SWIR1)	Sentinel-2: B8, B11
NDSI	Normalized Difference Soil Index	Red − NIR)/(Red + NIR) or (Green − NIR)/(Green + NIR)	Sentinel-2: B4, B8/B3, B8
IOR	Iron Oxide Ratio	Red/Blue	Sentinel-2: B4, B2
FMI	Ferrous Mineral Index	SWIR/NIR	Sentinel-2: B11, B8
TIR	Thermal Infrared	Derived from thermal band emissivity (ST_B10)	Landsat-8/9: B10
DEM	Digital Elevation Model	Derived directly from SRTM DEM dataset	SRTM (30 m)
PC1	Principal Component 1	Derived from covariance matrix via eigen decomposition	Sentinel-2: B2, B3, B4, B8, B11, B12

All these computed indices were clipped to the Area of Interest (AOI). Cloud masking and median compositing were applied to obtain clean, cloud-free surface reflectance data. A consistent color palette (Turbo) was applied across all spectral index maps to enhance visual comparability and maintain perceptual uniformity, providing a balanced representation of low- to high-value anomalies that facilitated interpretation of spatial patterns. The resulting layers were visually inspected and subsequently exported from Google Earth Engine (GEE) in GeoTIFF format for integration, overlay analysis, and final thematic map generation in QGIS.

Soil-gas measurements

The Biogas GA 5000 instrument was used in the field to measure hydrogen concentration in the soil. The measurements were performed in the SCD’s associated with spectral anomalies, at a sampling depth of approximately 50–80 cm. GA 5000 measures the concentrations of H₂ (ppm), CO₂ (%), CH₄ (%), O₂ (%) and N₂ (%) in the bulk gas. The corresponding sampling coordinates, measurement dates, and gas concentrations, along with summary statistics (minimum, maximum, mean, standard deviation, and coefficient of variation), are provided in Supplementary Appendix 1.

Petrography, mineralogy and geochemistry

Polished thin sections were prepared from four ultramafic rock samples collected from outcrops within the study area to characterize their mineralogical and textural attributes. Petrographic examination was conducted under plane-polarized light (PPL) using conventional optical microscopy to identify constituent mineral phases and assess their microtextural relationships. To complement the petrographic observations, scanning electron microscopy (SEM) analyses were performed on carbon-coated thin sections at the Core Facilities Department, Nazarbayev University (Kazakhstan), using a JSM-IT200 instrument equipped with an energy-dispersive spectrometer (EDS) for semi-quantitative elemental characterization.

Bulk mineralogical and geochemical compositions were determined through X-ray diffraction (XRD) and X-ray fluorescence (XRF) analyses conducted both before and after the experimental procedures. XRD analyses were carried out on powdered samples using a Rigaku SmartLab diffractometer following the analytical workflow described by Zhou et al.⁶³ Mineral identification was undertaken using the PDF-4 + database in conjunction with the Bruker DIFFRAC.EVA software, and relative mineral abundances were estimated using the reference intensity ratio (RIR) method⁶⁴. Major-element geochemistry was obtained via XRF using an Axios Max PANalytical spectrometer, with analytical uncertainties ranging between ± 0.5 wt% and ± 2 wt%.

Results and interpretations

Regional remote-sensing screening for hydrogen-related surface anomalies

A regional-scale analysis of remote-sensing data was conducted across the northern central Aktobe area (Figs. 1 and 2) to evaluate whether surface features could reflect subsurface processes associated with potential natural hydrogen seepages. Specifically, multiple spectral indices FMI, IOR, NDMI, NDSI, NDVI, and TIR were integrated with geomorphological and geological information to assess whether their spatial patterns align with drainage networks, valley systems, topographic gradients, and geological boundaries (Figs. 4A). Each index represents an environmental variable relevant to hydrogen-related processes: oxidation–reduction conditions (IOR, FMI; Figs. 4B), moisture distribution (NDMI; Fig. 4A), vegetation response (NDVI; Fig. 4A), salinity variation (NDSI; Fig. 4D), surface temperature (TIR; Fig. 4E), and elevation transition (DEM; Fig. 4F). These proxies are sensitive to surface–subsurface interactions such as fluid movement, mineral alteration, and water–rock interaction, key components of hydrogen generation and migration.

Fig. 4.

Regional multi-index remote sensing screening of the study area (see Fig. 2). Moisture–vegetation (A), oxidation–reduction (B, C), salinity (D), thermal response (E), and elevation (F) patterns indicate geomorphological and structural zones where multiple indicators converge. These integrated signals help screen for hydrogen-related surface expressions and shallow geochemical activity. The indices were derived from Sentinel-2 and Landsat 8/9 data acquired during June–September 2024.

Collectively, the indices reveal spatial trends that consistently follow major geomorphological and structural features. Moisture and vegetation proxies (NDMI, NDVI; Fig. 4A) outline the regional-scale drainage networks, suggesting zones of potential fluid concentration. Oxidation-sensitive indices (IOR, FMI; Figs. 4B and C) respond to slope transitions and lithological contacts, hinting at possible redox alteration or long-term weathering. NDSI identifies localized salinity signatures in selected depressions and valley margins, potentially linked to shallow fluid accumulation or water–rock interaction (Fig. 4D). TIR variability distinguishes moisture-rich zones from drier oxidized surfaces (Fig. 4E), while DEM highlights elevation contrasts that influence hydrological flow direction (Fig. 4F). When interpreted collectively, the spatial convergence of multiple indices within geomorphologically constrained areas demonstrates the value of spectral overlap as a screening tool for detecting surface anomalies that may reflect underlying geochemical processes relevant to hydrogen systems.

Detailed analysis of the Bogestay–Kenoy area

The Bogestay–Kenoy area was selected as one of several promising zones identified during the remote-sensing assessment in the north-central part of the Aktobe region, western Kazakhstan (Figs. 4A), and was chosen for detailed examination due to its manageable size and suitability for high-resolution analysis and field validation. Within this area, the combined remote-sensing dataset highlights clusters of moisture, oxidation, ferrous, and salinity anomalies that enable a focused evaluation of surface expressions potentially linked to near-surface geological processes. The DEM and spectral indices together reveal relationships between valley morphology, slope transitions, shallow depressions, and the distribution of key geochemical indicators, forming the basis for the following detailed interpretation.

Geological and geomorphological framework for interpreting surface anomalies

In the Bogestay–Kenoy area, the major north–south valley, broad elevation gradients, and mapped lithological contacts define the structural framework controlling shallow fluid behavior (Fig. 5). The main valley and its tributaries act as pathways for focused groundwater movement and enhanced near-surface water–rock interaction, while valley intersections may represent zones where fluids accumulate or shift laterally. Lithological contacts contribute to structural heterogeneity, and where these boundaries intersect valley trends, they may create permeability contrasts favorable for upward or lateral fluid movement. Because the 30 m DEM cannot resolve fine-scale tributaries or small depressions, NDMI, NDVI, and NDSI were used to delineate smaller drainage features and moisture-related surface patterns that are not captured by the DEM (Fig. 5). These indices respond to variations in surface moisture, vegetation, and soil conditions, enabling the detection of subtle surface expressions associated with water pathways. Their finer spatial resolution (10–20 m) further supports the identification of localized features. Integrated with the regional geological framework, these features provide additional constraints for interpreting surface anomalies and help refine the identification of localized surface processes and potential near-surface pathways.

Fig. 5.

Integrated geomorphological and lithological map used to delineate valley pathways, tributary networks, and lithological contacts that guide the interpretation of spectral anomalies in Bogestay–Kenoy. Major drainage patterns and lithological boundaries were manually extracted using NDMI, NDVI, NDSI, and the geological map.

Topographic controls on fluid pathways

The DEM provides a broad representation of elevation across the Bogestay–Kenoy area, showing a dominant north–south valley aligned with wide upland–lowland transitions between roughly 255 and 300 m (Fig. 6). Although this resolution cannot resolve small depressions, narrow tributaries, or detailed slope variations, it still defines the major elevation gradients that influence regional-scale surface and subsurface flow. Areas with noticeable elevation contrasts, such as the main valley floor and adjacent slopes, represent zones where groundwater flow and surface runoff naturally converge, enhancing water infiltration and water–rock interaction^65–67. The main valley serves as the primary topographic pathway, guiding shallow fluids toward lower elevations, while low-elevation convergence zones at valley intersections provide favorable conditions for fluid accumulation or movement. Where valley trends intersect lithological contacts, structural heterogeneity (e.g., faults, fractures, or permeability contrasts) may further facilitate vertical or lateral fluid migration. In this context, zones combining low elevation, broad valley orientation, and proximity to lithological boundaries represent priority targets for natural hydrogen generation.

Fig. 6.

DEM-derived elevation model (30 m resolution) of the Bogestay–Kenoy region (see black rectangle in Fig. 2), highlighting the main north–south valley and regional elevation gradients controlling surface drainage, derived from the SRTM DEM dataset (30 m resolution).

Ferrous minerals index (FMI): distribution of reduced iron

FMI values in Bogestay–Kenoy range from 1.0 to 1.65, with the strongest responses occurring along the valley margins and several slope breaks (Fig. 7A). High FMI values indicate zones enriched in reduced Fe²⁺-bearing minerals. These areas are particularly significant for natural hydrogen studies, as Fe²⁺ plays a key role in water–rock reactions that can generate hydrogen within ultramafic or iron-rich substrates at depth^68–70. In the Bogestay–Kenoy, high FMI zones align with elevated slopes and lithological contacts, suggesting the presence or influence of reduced mineral material and possibly indicating pathways where reduced iron persists or is brought near the surface. Lower FMI values dominate the valley floor and moist areas, where oxidation or surface coatings (e.g., thin weathering layers, clay films, or moisture films) can mask the spectral signature of reduced minerals.

Fig. 7.

Key spectral indices revealing geochemical and thermal variations across Bogestay–Kenoy: (A) FMI highlights reduced Fe²⁺-bearing minerals; (B) IOR marks oxidation halos and redox transitions along valley sides and depressions; (C) TIR shows cooler moisture-rich areas and warmer oxidation-related surfaces; (D) PC1 separates moisture-dominated zones from oxidized surfaces. The indices were derived from Sentinel-2 and Landsat 8/9 data acquired during June–September 2024.

Iron oxide ratio (IOR): oxidation halos and redox transitions

High Fe³⁺ zones, which result from strong oxidation of Fe²⁺-bearing minerals, are closely linked to the natural hydrogen (H₂) potential. The oxidation of Fe²⁺ to Fe³⁺ is a key indicator of past or ongoing hydrogen-generating reactions^16,71–73. IOR values capture the oxidation of Fe²⁺ to Fe³⁺ at the surface, forming a complementary signal to the FMI distribution. In the study area, IOR values range from 1.0 to 1.4, with the highest responses occurring along valley sides, tributary intersections, and around the sub-circular surface depressions (fairy circles, Fig. 7B). These high IOR zones reflect strong surface oxidation associated with long-term weathering or through oxidation fronts produced when reduced fluids rise along fractures and mix with oxygen-rich shallow environments. Moderate IOR values form transitional belts that often border high-FMI zones, marking areas where oxidized and reduced iron coexist. Low IOR values occur in moist or low-relief areas where water suppresses spectral iron features or where relatively unoxidized surfaces are exposed. The spatial match between zones with elevated IOR values and major geomorphological boundaries, especially along valley sides, slope transitions, and around the circular surface depressions, suggests that oxidation patterns are shaped by structural or permeability contrasts that may help guide shallow fluid migration.

Normalized difference moisture index (NDMI): moisture accumulation and alteration

Moisture is a fundamental driver of natural hydrogen generation, as it enables water–rock interactions that promote redox water-rock interactions and natural hydrogen generation^69,74. It also influences the physical and chemical conditions that control hydrogen migration and accumulation. Therefore, greater moisture availability enhances the potential for H₂ generation and supports fluid mixing within structurally controlled zones^{69,70,74–77}. NDMI values in Bogestay–Kenoy range from − 0.2 to 0.2, indicating moderate moisture levels that are sufficient to differentiate wetter surface conditions from well-drained or exposed areas (Fig. 8). Positive NDMI values indicate increased surface or near-surface moisture, which is more suitable for hydrogen generation. These moisture anomalies occur along the master valley, at tributary and sub-tributary junctions, and near lithological contacts where permeability contrasts and fault-controlled pathways can promote fluid-rock interaction. Elevated NDMI values are also detected within several sub-circular depressions: deeper basins (3–6 m), which coincide with saline discharge features, and shallow depressions (≤ 1 m), whose alignment with tributary pathways and lithological boundaries suggests they may act as minor seepage or discharge points rather than simple erosional features. In contrast, low NDMI values are found along elevated ridges and well-drained slopes, reflecting dry or unaltered surfaces that are typically distal to structural conduits. The overall clustering of high NDMI at valley junctions and surface depressions is consistent with the main drainage geometry and supports the interpretation of Bogestay–Kenoy as an area where shallow hydrological processes interact with structural controls.

Fig. 8.

Normalized Difference Moisture Index (NDMI) map showing moisture accumulation along the master valley, tributary intersections, lithological boundaries and sub-circular surface depressions. High NDMI zones are associated with structural controls and potential shallow fluid migration, derived from Sentinel-2 data acquired during June–September 2024.

Normalized difference salinity index (NDSI): saline discharge and soil chemistry

Soil salinity can act as a shallow tracer of water–rock interaction and fluid movement, particularly when influenced by structural or lithological boundaries⁷⁸. Variations in salinity may develop through enhanced groundwater flow, mineral alteration, or mixing with mineralized fluid at depth, especially where valleys intersect lithological contacts or fault-controlled pathways^78–80. NDSI values in Bogestay–Kenoy range from − 0.30 to 0.05 (NDSI_RedNIR), indicating generally low salinity but allowing clear separation between saline and non-saline soils (Fig. 9). The highest NDSI values occur along the sides of the master valley, several tributary and sub-tributary channels, and near lithological boundaries, marking zones where mineralized water or chemically altered soils accumulate. These saline anomalies frequently coincide with high NDMI values, suggesting moist, saline depressions that may serve as shallow discharge points for near-surface fluids. Two types of depressions are evident: (1) deeper depressions (3–6 m), located near lithological contacts, which show elevated salinity and are likely influenced by groundwater discharge or prolonged water–rock interaction; and (2) shallow depressions (≤ 1 m), which generally exhibit low NDSI values and reflect surface-controlled runoff and evaporation, though their alignment along valley margins or lithological contacts may still indicate weak lateral diffusion or subtle structural influence. Low NDSI values also appear in well-drained or vegetated areas (Fig. 9) where moisture and vegetation likely suppress the salinity signal, and soil-free areas where ultramafic or intrusive exposures lack soil development. These salinity patterns collectively highlight the interplay between hydrological conditions and structural controls on shallow fluid movement within the Bogestay–Kenoy area.

Fig. 9.

Normalized difference salinity index (NDSI) maps for the Bogestay-Kenoy region derived from (A) Red–NIR and (B) Green–NIR highlight saline discharge and chemically altered zones, with elevated values along valley margins and lithological contacts indicating shallow fluid activity. On the right, close-up views and topo-profiles of two 6 m anomalies identified by Green- and RedNIR are shown on the left derived from Sentinel-2 data acquired during June–September 2024.

Normalized difference vegetation index (NDVI): vegetation response to moisture and soil chemistry

Vegetation can reflect shallow environmental variations linked to moisture availability, soil chemistry, or subtle geochemical alteration. NDVI is therefore useful for identifying areas where vegetation is either supported by favorable hydrological conditions or stressed by salinity, oxidation, or chemical alteration. When interpreted in a geological context, anomalous vegetation patterns may act as indirect surface expressions of fluid movement, especially along structural boundaries, valley pathways, or zones of water–rock interaction. NDVI values in Bogestay–Kenoy range from approximately 0.14 to 0.60, capturing clear variations in vegetation vigor across the area (Fig. 10). High NDVI values closely match with high NDMI zones, particularly along valley sides and low-slope junctions where moisture accumulates and supports healthier vegetation. These moisture-supported areas may represent shallow groundwater or trap-like zones where fluids tend to collect or discharge at the surface. In contrast, low NDVI values occur in locations where vegetation would normally be expected, especially near oxidation anomalies (high IOR) and within saline depressions highlighted by NDSI. These vegetation stress zones likely reflect the influence of elevated salinity, Fe³⁺-rich oxidation, or chemically altered soils that inhibit plant growth. In natural hydrogen–prone environments, such vegetation anomalies may serve as indirect indicators of shallow geochemical changes or fluid-related alteration⁸. In Bogestay–Kenoy, the spatial correspondence between NDVI, NDMI, IOR, and NDSI suggests that vegetation patterns are strongly influenced by moisture availability and soil chemistry shaped by near-surface processes.

Fig. 10.

Normalized difference vegetation index (NDVI) highlights vegetation vigor and stress patterns in the Bogestay-Kenoy region. High NDVI corresponds to moisture-supported vegetation, while low NDVI correlates with oxidation (IOR) and salinity (NDSI) zones, indicating possible soil alteration or seepage influence. On the right are Google Earth images of two depressions located within the area indicated by red arrows on the left, derived from Sentinel-2 data acquired during June–September 2024.

Thermal infrared (TIR): temperature variability and surface material differences

TIR values in Bogestay–Kenoy range from approximately 30 to 41 °C and show subtle but meaningful temperature contrasts that help distinguish variations in moisture, vegetation cover, and surface material properties (Fig. 7C). Relatively warmer zones appear along several valley margins, where lower moisture content or the presence of oxidized surface coatings may increase surface temperature. In multiple locations, these warm anomalies coincide with high IOR values, suggesting that oxidation-related alteration may contribute to localized heating. Conversely, cooler TIR responses occur along the valley floor and within the SCD’s, matching areas of high NDMI and high NDVI where moisture retention and vegetation cover suppress surface temperatures. Although TIR data alone cannot confirm deeper processes, the alignment of warm TIR zones with oxidation anomalies and the cooling effect observed in moisture-rich depressions supports the interpretation of contrasting surface conditions and possible shallow alteration zones. These relationships likely indirectly indicate areas where fluid movement, surface geochemical alteration, or near-surface redox activity is influencing the thermal response.

H₂-seepage zones

Results of the remote sensing analysis, guided by the multi-proxy target selection workflow (Fig. 11), enabled the identification of a series of prospective H₂-bearing depressions selected for soil-gas measurements. These depressions correspond to zones where geomorphological, moisture, redox, and surface indicators spatially converge, making them favorable locations for potential gas seepage. Soil-gas measurements acquired within these targets yielded H₂ concentrations ranging from background levels (0 ppm) to values exceeding 50 ppm (Fig. 12), confirming the utility of remote-sensing tools and the utilized workflow as a powerful exploration screening tool. Notably, the highest H₂ concentrations were consistently recorded along the margins of the sub-circular depressions rather than their centers (Fig. 12). This distribution pattern aligns with previous observations from documented “fairy circles” associated with active hydrogen seepage systems^18,27,29, suggesting that leakage pathways are preferentially developed along the structural or lithological discontinuities that define the boundaries of these depressions. In addition to H₂ measurements, the GA5000 instrument also recorded CH₄, CO₂, O₂, and N₂ concentrations. The results show negligible CH₄ (0%) and very low CO₂ (mean ~ 0.05%), while O₂ (~ 21%) and N₂ (~ 78%) are consistent with atmospheric composition (Supplementary Appendix 1), indicating that the measurements were obtained within a predominantly aerobic soil environment at shallow depth (50–80 cm). Under such conditions, the absence of methane is expected and does not provide a basis for excluding microbial processes. Although gas composition alone is insufficient to constrain the origin of H₂, the detection of elevated H₂ concentrations (up to ~ 96 ppm) within this non-anaerobic setting, when considered together with the spatial distribution of anomalies and the structural–geological framework, is well-matched with a deep-seated hydrogen leakage model.

Fig. 11.

Stepwise workflow for identifying priority soil-gas sampling locations based on the spatial convergence of geomorphological, moisture, geochemical, and surface indicators. Maps on the right show the filtered outputs used in target selection.

Fig. 12.

Spatial distribution of soil-gas H₂ measurements across selected sub-circular depressions in the Bogestay–Kenoy region. The central map shows the locations of sampling points, while enlarged views illustrate H₂ concentrations (ppm) within individual depressions. Elevated H₂ values are predominantly observed along the peripheral zones of the depressions.

Petrography and mineralogy

The investigated ultramafic rocks are predominantly composed of serpentinized harzburgite. The primary mineral assemblage consists of serpentine-group minerals (lizardite and chrysotile), olivine, pyroxene, and chromite (Figs. 13A and B). Olivine occurs as both fresh and partially altered grains, typically exhibiting serpentinization along grain boundaries. Serpentine minerals occupy microfractures and cleavage planes and occur as cross-cutting vein networks, reflecting pervasive fluid–rock interaction. Serpentine is the volumetrically dominant phase in all examined samples, constituting more than 70% of the modal composition (Fig. 13C).

Fig. 13.

(A, B) Thin-section microphotographs illustrating the mineralogical and textural features of the ultramafic samples, including the distribution and alteration patterns of serpentine, olivine, pyroxene, and chromite. (C) X-ray diffractogram highlighting the dominant mineral phases and confirming the pervasive serpentinization within the studied rocks. (D) SEM–EDS analyses documenting the microtextural attributes and chemical compositions of the principal ultramafic mineral phases, providing insights into their alteration pathways and element redistribution during serpentinization.

SEM–EDS analyses indicate that olivine corresponds to Mg-rich forsterite, characterized by an average composition of ~ 47 wt% Mg, ~ 39 wt% Si, and ~ 10 wt% Fe (Fig. 13D). The vein-forming serpentine displays slightly higher Si content and lower Fe concentration relative to olivine, with a representative composition of ~ 44 wt% Mg, ~ 46 wt% Si, and ~ 8 wt% Fe (Fig. 13D). These compositional variations reflect mineralogical transformation during serpentinization and fluid-mediated element redistribution.

Major element compositions

The major-element geochemical composition of the investigated ultramafic samples is dominated by Mg, Si, and Fe (Table 2), reflecting their olivine- and serpentine-rich mineralogy as established through petrographic and mineralogical analyses. In all samples, MgO, SiO₂ and Fe₂O₃ collectively account for more than 95 wt%, except for one sample that contains exceptionally high Cr₂O₃ (25.1 wt%), indicative of significant chromite enrichment. Aluminum concentrations are uniformly low (Al₂O₃ < 1 wt%) in most samples except for the chromite-rich one, consistent with Al incorporation into chrome-spinel. The very low concentrations of Na₂O, K₂O, and CaO (< 1 wt%) across all samples further suggest the absence of feldspar- and mica-bearing phases, and reaffirm the highly depleted, ultramafic character of the examined lithologies.

Table 2.

Major element composition (wt%) of the investigated ultramafic samples.

Sample	MgO	Al2O3	SiO2	CaO	Cr2O3	MnO	Fe2O3
1	46.5	0.67	43.9	0.08	0.31	0.12	7.6
2	50	0.33	42.1	0.5	0.41	0.1	5.81
3	51.3	0.27	40.3	0.12	0.31	0.1	6.25
4	38.3	3.15	25.6	0.4	25.1	0.82	5.83

Discussions

Principal component analysis (PC1) of spectral proxies

The north–central Aktobe region lies adjacent to the Kempirsai ultramafic massif, a major ophiolitic body of the Ural orogenic belt hosting globally significant podiform chromite deposits^32,81–83. Surrounding sedimentary, metavolcanic, and intrusive successions introduce additional contrasts in permeability and lithological interfaces, which may influence fluid flow and shallow geochemical alteration. The multispectral proxies applied in the Bogestay–Kenoy region at the boundary of the Kempirsai massif highlights the major surface trends associated with potential hydrogen seeps. Principal component analysis (PC1) integrates the different spectral proxies into a single layer that captures the dominant surficial patterns linked to redox reactions and hydrogen seepage. Low PC1 values consistently align with high IOR zones, marking oxidation-dominated areas that may represent shallow alteration or redox transitions influenced by fluid seepage or surface geochemical alteration (Fig. 7D). In contrast, high PC1 values correspond closely with high NDMI responses, reflecting moisture-rich, brighter surfaces within valleys, shallow depressions, and low-slope junctions. This distribution reinforces the two key surface processes captured by the individual indices: (1) oxidation and surface alteration concentrated along valley sides and structural boundaries, and (2) moisture accumulation and related chemical alteration in topographic depressions. PC1 therefore provides a concise summary of the spatial relationships observed across the dataset and helps distinguish oxidation-driven zones from moisture-dominated areas within the Bogestay–Kenoy area.

The combined interpretation of the DEM, FMI, IOR, NDMI, NDSI, NDVI, TIR, and PC1 highlights several zones where multiple indicators converge, particularly at master–tributary junctions, along valley-side slope breaks, and within the sub-circular depressions. These depressions, especially where they align with valley margins or lithological contacts, may reflect weak lateral diffusion or subtle structural influence, consistent with the patterns observed in the salinity, vegetation and moisture indices⁸⁴. Across these convergence zones, moisture accumulation, oxidation–reduction transitions, salinity accumulation, and vegetation responses occur together, indicating active interactions between geomorphology, surface conditions, and near-surface geochemical processes. Although remote sensing alone cannot resolve the precise subsurface mechanisms, the clustering of anomalies along structural and topographic pathways suggests that Bogestay–Kenoy contains favorable conditions for shallow fluid movement or chemical alteration. The integrated dataset, therefore, provides a coherent framework for identifying priority targets and guiding field validation and near-surface investigation within the study area.

Soil-gas measurements demonstrate that spectral anomalies coincident with sub-circular surface depressions systematically correspond to zones of enhanced H₂ seepage. The spatial distribution of H₂ is highly non-uniform, with consistently elevated H₂ concentrations recorded along the margins of these depressions rather than at their centers (Fig. 12). This peripheral enrichment pattern closely mirrors the boundary-controlled hydrogen anomalies documented in fairy circles reported from other geological provinces^25,27,29. Such a recurrent spatial association strongly suggests a structural or permeability-controlled mechanism governing hydrogen migration and near-surface expression.

Hydrogen generation system

The study area is located along the margin of the Kempirsai Massif, which represents one of the largest and best-exposed ultramafic complexes on Earth^85,86. In addition, the study region includes the Precambrian to Paleozoic granitoid terranes of the Mugozhdar microcontinent, which may represent an additional hydrogen-generating system through water radiolysis. Petrographic investigations indicate that the ultramafic assemblage is predominantly composed of harzburgite and dunite that have undergone variable degrees of serpentinization (Fig. 13). The examined samples contain both fresh and partially altered olivine grains. In contrast, serpentine-group minerals constitute the most abundant secondary mineral phases.

Textural relationships observed in thin section reveal extensive serpentinization of primary olivine by circulating fluids. This interpretation is supported by the preferential development of serpentine along vein networks and at the margins of olivine grains, indicating fluid-mediated alteration pathways. Whole-rock major oxide geochemistry further corroborates the petrographic observations, revealing a dominance of Mg-rich silicate phases characteristic of olivine and serpentine. Iron represents the third most abundant major oxide after MgO and SiO₂ (Table 2), and the three oxides are likely incorporated in the olivine and serpentine minerals. However, iron could also be incorporated in chromite (Fig. 13).

Recent reports linked the presence of chromite in ultramafic-host rocks with active generation of hydrogen^35,87. Fe^(II)-bearing chromite provides highly reactive surfaces that can promote H₂ generation through surface-mediated redox reactions. In this process, electrons derived from both structural Fe^(II) within the spinel lattice and adsorbed Fe^(II) species are transferred to interfacial water molecules or surface-bound protons⁶⁹. Moreover, H₂ production may be sustained over time through the continuous oxidation of other Fe^(II)-bearing phases such as olivine, serpentine and Fe-brucite^88–90, which supplies Fe^(II) to the solution. The dissolved Fe^(II) subsequently adsorbs onto chromite surfaces and undergoes repeated oxidation, effectively maintaining the redox cycle required for ongoing H₂ generation⁶⁹. Accordingly, the presence of chromite-bearing ultramafic rocks in the immediate vicinity of the study area strongly suggests that these ultramafic lithologies act as a source for natural hydrogen documented at the surface locations (Figs. 7, 8, 9, 10, 11, 12 and 14). This allows for proposing a hydrogen system model, in which the ultramafic massif acts as an “H₂ generation system” (H₂ source) by maintaining reactive mineral surfaces and Fe^(II) supply pathways, while local structural permeability may facilitate upward migration of generated gas toward shallower levels, including documented seepages. In this context, the Bogestay–Kenoy area is structurally linked to the regional tectonic framework of the Main Ural Deep Fault (MUDF), which likely exerts a first-order control on the orientation, depth, and connectivity of fluid-migration pathways (Fig. 14). Fluid and gas migration are further controlled by fault zones, fracture networks, and associated permeability anisotropy developed during regional deformation. These structural features provide vertically connected conduits linking ultramafic source rocks at depth to shallow levels and surface seepage zones.

Fig. 14.

Simplified geological cross-section of the study region (see location in Fig. 2) illustrating the principal tectonic domains hosting potential hydrogen-generating lithologies (modified after⁹²). The distribution of the fairy circles (blue stars) spatially coincides with zones where three major geological hydrogen-generation mechanisms may operate: (1) shallow to intermediate-depth, low-temperature serpentinization, where the ~ 60 °C isotherm is estimated to occur at depths of approximately 2.5 km; (2) deep-seated, high-temperature serpentinization processes occurring at depths generally exceeding 10 km; and (3) radiolytic hydrogen generation associated with granitoid bodies within the western Mugodzhar zone.

In association with the ultramafic source is a complex assemblage of thrust-faulted Upper Proterozoic and Paleozoic volcaniclastic successions and exotic terranes of the Mugozhdar microcontinent, which are comprised of granites, gabbros, and plagiogranite (Figs. 2 and 14). These rocks provide not only a plausible migration and accumulation domain in which hydrogen could be trapped and accumulated (Fig. 14), but also an additional, if not exclusive, source of hydrogen can be generated in shallower intervals by granite-driven radiolysis. However, the widespread occurrence of active surface seeps and fairy circles yielding > 50 ppm H₂ suggests that the seal may be locally breached, fractured, or regionally discontinuous rather than perfectly effective⁹¹. Accordingly, the system may involve episodic accumulation followed by leakage along tectonic discontinuities, rather than long-term, regionally continuous storage beneath a uniformly intact caprock. If sufficiently sealed in restricted structural compartments, the Proterozoic and Paleozoic volcaniclastics may still provide accumulation zones for migrated H₂, but the surface anomalies indicate that preservation potential is likely heterogeneous and leakage pathways remain active. The juxtaposition of a reactive ultramafic source beneath a sedimentary package capable of hosting reservoir–seal pairs is therefore consistent with a coupled “generation–migration–storage” petroleum-system analog adapted to hydrogen. However, storage efficiency is likely limited by deformation-related breaching and ongoing or localized degassing in parts of the basin.

Conclusions

• The integration of multiple spectral indices with TIR, DEM, and geological maps successfully identifies surface anomalies that consistently align with geomorphological and structural features. Spectral maps overlap is proven to be a powerful first-order screening tool for detecting surface expressions potentially linked to hydrogen seeps, indicative of the underlying subsurface hydrogen system.

• In the Bogestay–Kenoy area, valley margins, tributary junctions, slope breaks, and lithological contacts exert primary control on moisture distribution, redox patterns, salinity accumulation, vegetation response, and thermal variability, highlighting their role as preferential pathways for shallow fluid movement.

• Sub-circular surface depressions, referred to as fairy circles, represent surface expressions of hydrogen seeps. Soil-gas measurements demonstrate that elevated H₂ concentrations are preferentially developed along the margins of these features, which is consistent with structurally or permeability-controlled seepage rather than purely surface-driven processes.

• Petrographic and geochemical data confirm that the Kempirsai massif is composed of serpentinized harzburgite, dominated by olivine and chromite. These minerals provide reactive Fe^(II) sources and catalytic surfaces that are conducive to sustained hydrogen generation via water–rock interaction.

• The study area is interpreted as a coupled generation–migration–accumulation framework: ultramafic rocks of the Kempirsai Massif and possibly the granitoids of Mugozhdar micro-continent act as H₂ generation engine, fractures and shear zones may facilitate upward migration, and overlying Precambrian and Paleozoic volcaniclastic units represent a plausible storage domain, while studied surficial fairy circles are natural seeps.

• The workflow presented here demonstrates how remote sensing, when attached to geological reasoning and validated by field and laboratory data, can effectively guide hydrogen exploration. This approach is transferable to other ultramafic provinces worldwide where direct subsurface data are limited.

Supplementary Information

Below is the link to the electronic supplementary material.

Author contributions

M.L. and A.Y. conducted rhe remote sensing analysis and wrote the main manuscript text. M.L., I.M. and N. B. conducted the soil-gas measurements and interpreted the results. M.F. and R.H. interpreted the mineralogical and geochemical results. All authors reviewed the manuscript.

Funding

This work has been funded by the Ministry of Science and Higher Education of the Republic of Kazakhstan under grant No. AP26104405 as well as Nazarbayev University Faculty Development Competitive Research Grants Program No. 040225FD4717.

Data availability

All datasets utilizied in this study were listed in the supplementary material.

Declarations

Competing interests

The authors declare no competing interests.