1. Introduction
The end-Cretaceous Deccan volcanism on the western continental margin of India constitutes one of the largest continental flood basalt provinces in the world. The volcanism is mostly linked to the Reunion mantle plume and straddles the Cretaceous-Palaeogene (K-Pg) boundary coinciding with Chicxulub impact. Hence, the province has been the subject of numerous studies in the past five decades. High precision Ar/Ar and U-Pb geochronology have constrained the Deccan eruption within a narrow age band (<1 Ma) across K-Pg boundary (Duncan and Pyle, Reference Duncan and Pyle1988; Courtillot et al. Reference Courtillot, Feraud, Maluski, Vandamme, Moreau and Besse1988; Venkatesan et al. Reference Venkatesan, Pande and Gopalan1993; Hofmann et al. Reference Hofmann, Feraud and Courtillot2000; Chenet et al. Reference Chenet, Quidelleur, Fluteau and Courtillot2007; Hooper et al. Reference Hooper, Widdowson and Kelley2010; Renne et al. Reference Renne, Sprain, Richards, Self, Vanderkluysen and Pande2015; Schoene et al. Reference Schoene, Samperton, Eddy, Keller, Adatte, Bowring and Gertsch2015, Reference Schoene, Eddy, Samperton, Keller, Keller, Adatte and Khadri2019; Sprain et al. Reference Sprain, Renne, Vanderkluysen, Pande, Self and Tushar2019; Tholt et al. Reference Tholt, Renne, Marzoli, Vanderkluysen, Mohabey, Samant, Dhobale, Pande and Self2023). Extensive studies of geochemistry (Krishnamurthy and Cox, Reference Krishnamurthy and Cox1977; Cox and Hawkesworth, Reference Cox and Hawkesworth1985; Beane et al. Reference Beane, Turner, Hooper, Subbarao and Walsh1986; Lightfoot and Hawkesworth, Reference Lightfoot and Hawkesworth1988; Mahoney Reference Mahoney and Macdougall1988; Peng et al. Reference Peng, Mahoney, Hooper, Macdougall and Krishnamurthy1998; Basu et al. Reference Basu, Yannopoulos and Chakrabarty2020;Tholt et al. Reference Tholt, Renne, Marzoli, Vanderkluysen, Mohabey, Samant, Dhobale, Pande and Self2023; Marzoli et al. Reference Marzoli, Reisberg, Capriolo, Callegaro, Renne, Chiaradia, Meyzen, Self, Vanderkluysen and Boscaini2026 and references therein) and palaeomagnetism (Courtillot et al. Reference Courtillot, Besse, Vandamme, Montigny, Jaeger and Cappetta1986; Vandamme et al.Reference Vandamme, Courtillot, Besse and Montigny1991; Chenet et al. Reference Chenet, Flutean, Courtillot, Gerard and Subbarao2008, Reference Chenet, Courtillot, Fluteau, Gerard, Quidelleur, Khadri, Subbarao and Thordarson2009; Jay et al. Reference Jay, Mac Niocaill, Widdowson, Self and Turner2009; Schöbel et al. Reference Schöbel, de Wall, Ganerød, Pandit and Rolf2014; Radhakrishna et al. Reference Radhakrishna, Mohamed, Venkateshwarlu, Soumya and Prachiti2019 and references therein) brought out petrogenetic aspects, polarity sequences and key end-Cretaceous palaeomagnetic pole for the Indian subcontinent. Field mapping and structural observations were used to suggest that the Deccan succession thins southward and displays overstep of older formations by younger formations (Mitchell and Widdowson, Reference Mitchell and Widdowson1991). However, the southward extent of the individual formations beneath the surface is not known.
Combined geochemical and geological studies were used to determine volcano-stratigraphy in many earlier studies of the Deccan, but mostly the work was done on less continuous volcanic stratigraphy (Basu et al. Reference Basu, Yannopoulos and Chakrabarty2020). Recently, Basu et al. (Reference Basu, Yannopoulos and Chakrabarty2020) provided results from contiguous set of samples in progressive succession from bottom upwards in all the lava formations of the Deccan based on their new geochemical data and by incorporating most of the previous results. Geochemical and palaeomagnetic data on same set of samples representing the continuous volcanic sequence vertically are expected to provide new insights into the evolution of Deccan volcanism in the southern Deccan Province.
The recent Continental Scientific Deep Drilling (CSDD) at Koyna (Figure 1) within the Western Ghats is valuable as it gives a unique opportunity to sample unbiased continuous vertical stratigraphic section of the Deccan down to the basement. We have carried out an integrated study of petrology based on major and trace elements, mineral magnetism and palaeomagnetism on one of the drill holes (KBH-7; KBH-8 also for palaeomagnetism) that recovered the Deccan pile of maximum thickness. Using the data, our previous work was focussed to assess the magmatic underplating model for explaining the rift-flank uplift and its mechanism along the volcanic continental margin (Radhakrishna et al. Reference Radhakrishna, Mohamed, Venkateshwarlu, Soumya and Prachiti2019) and comprehensive palaeointensity study to determine end-Cretaceous geomagnetic field strength and to infer linkage between geomagnetic behaviour, polarity reversals and deep mantle processes (Radhakrishna et al. Reference Radhakrishna, Mohamed, Venkateshwarlu and Soumya2020). Khanna et al. (Reference Khanna, Kanakdande, Bizimis and Arora2023) attempted demarcation of eruptive phases and bringing out petrogenetic process in the magma evolution. The current study was undertaken to demarcate the hitherto unknown subsurface stratigraphic formations down the drill hole to basement. Furthermore, palaeomagnetic results are used to constrain the duration of Deccan volcanism independent of geochronology.
Geological map showing areal spread of the Deccan flood basalt eruptions and location of the Koyna Continental Scientific Deep Drilling project drill hole (KBH-7) (redrawn from Radhakrishna et al. Reference Radhakrishna, Mohamed, Venkateshwarlu and Soumya2020).
2. Geology
Detailed geological descriptions of Deccan volcanism are available in the literature, and recent reviews (Kale et al. Reference Kale, Dole, Shandilya and Pande2020; Basu et al. Reference Basu, Yannopoulos and Chakrabarty2020; Krishnamurthy, Reference Krishnamurthy2020; Dole et al. Reference Dole, Das and Kale2022; Kale and Pande, Reference Kale and Pande2022; Tholt et al. Reference Tholt, Renne, Marzoli, Vanderkluysen, Mohabey, Samant, Dhobale, Pande and Self2023) provide comprehensive syntheses of its stratigraphy, volcanology and geochemistry. The Deccan lava succession comprises nearly horizontal basaltic flows with gentle regional dips. Based on volcanostratigraphic and chemostratigraphic characteristics, the Deccan Traps are subdivided into twelve formations. In the Western Ghats stratigraphy, the Wai Subgroup represents the youngest volcanic sequence and comprises, in descending order, the Desur, Panhala, Mahabaleshwar, Ambenali and Poladpur formations. These formations constitute the principal phase of Deccan volcanism in the Western Ghats. The Wai Subgroup is underlain successively by the Lonavala Subgroup (comprising the Bushe and Khandala formations) and the Kalsubai Subgroup.
The drill hole investigated in the present study is located near Koyna in the southwestern part of the Deccan Volcanic Province within the Western Ghats at regional elevation of 949 m above present-day mean sea level (MSL). It penetrates the entire Deccan volcanic succession in the region and extends into the underlying Archaean granite–greenstone basement. The drill hole recovered a 1251 m thick basaltic sequence, representing the greatest basalt thickness recovered from any drill hole in the Deccan Volcanic Province. Consequently, it provides a complete volcano-stratigraphic record of the Western Ghats region, extending from the surface to the basement.
A detailed lithological log of the Deccan succession recovered from the KBH-7 drill hole was presented by Misra et al. (Reference Misra, Bartakke, Athavale, Akkiraju, Goswami and Sukanta2017), and the principal observations are summarized here. The drill core contains a total of 46 lava flows, comprising 40 simple flows and 6 compound flows. A typical simple flow consists of a massive basaltic core overlain by vesicular and amygdaloidal basalt and is commonly capped by a brecciated flow-top horizon, which serves as a useful stratigraphic marker. Intervals of hiatus are recorded throughout the succession by centimetre-to metre-thick red bole horizons and zones of red and green clay alteration. These features indicate that the Deccan lava pile was emplaced during multiple discrete eruptive episodes. However, some bole horizons within flood-basalt sequences may have formed by weathering of tachylitic/glassy layers within the flows after the emplacement of the overlying lava flows or even long after the cessation of volcanic activity (Sheth et al. Reference Sheth, Naik, Janisar and Kumar2026).
Giant plagioclase basalt (GPB), characterized by conspicuous plagioclase laths measuring 1–4 cm in length and 0.5–4 mm in width, occurs within the basaltic succession at MSL elevations of 716–663 and 288–268 m. The uppermost part of the volcanic sequence has undergone intense weathering, resulting in the development of a lateritic cap. Recent high-precision U–Pb and 40Ar/39Ar geochronological data from the Western Ghats (Renne et al. Reference Renne, Sprain, Richards, Self, Vanderkluysen and Pande2015; Schoene et al. Reference Schoene, Samperton, Eddy, Keller, Adatte, Bowring and Gertsch2015, Reference Schoene, Eddy, Samperton, Keller, Keller, Adatte and Khadri2019; Sprain et al. Reference Sprain, Renne, Vanderkluysen, Pande, Self and Tushar2019) constrain the age of the Koyna Deccan drill-hole succession to approximately 65–66 Ma.
The studied samples are generally fresh and fine-grained, displaying characteristic basaltic mineral assemblages and textures. Clinopyroxene and plagioclase occur in roughly equal proportions and commonly form ophitic to subophitic intergrowths. Accessory Fe–Ti oxides are typically present as microphenocrysts, while olivine occurs only sporadically as an accessory phase. Both clinopyroxene and plagioclase are also present as phenocrysts in several flows.
3. Methods
Forty-four samples, covering the depth range down to basement within the KBH-7 hole, were analyzed for major oxides and trace elements (Cr, Ni and Zr) by X-ray fluorescence method at National Centre for Earth Science Studies, Trivandrum, India. Other trace elements were analyzed for 39 samples by Inductively Coupled Plasma-Mass Spectrometry method at National Geophysical Research Institute, Hyderabad, India. Analytical errors are within ±2% for SiO2 and better than ±8% for other oxides and generally better than 5–10% for trace elements. The full set of analyses along with reference standards used is given in Table S1 (Supplementary information).
Samples from chosen depth intervals of the drill hole were analyzed for different mineral magnetic characters. Complete set of mineral magnetic characters and their detailed descriptions are given in a recent publication (Asanulla et al. Reference Asanulla, Radhakrishna, Bansal and Ramakrishna2024). Characteristic palaeomagnetic directional measurements were undertaken on 126 samples across the KBH-7 drill hole and on 95 samples from KBH-8 (∼1198 m thick) drill hole through detailed alternating field and thermal demagnetization experiments and principal component analysis (see for details: Radhakrishna et al. Reference Radhakrishna, Mohamed, Venkateshwarlu, Soumya and Prachiti2019). Only inclination data are obtained as the samples are azimuthally un-oriented. Fisher statistics such as α95 and k values are not considered due to absence of declination data. However, maximum angular deviation calculated from principal component analysis to be less than 6° has been considered for reliable palaeoinclination determination. The inclination data are listed in Table S2 (Supplementary information).
4. Results
The samples are compositionally very monotonous in major element chemistry and basaltic in composition and intermediate, felsic or alkaline basalt compositions are totally absent. Their SiO2 content ranges within 49.63 ± 2.25 wt% with very restricted low K2O + Na2O (2.83 ± 0.47 wt%) and an iron enrichment (Fe2O3T = 14.05 ± 1.97 wt%), suggesting that the KBH-7 lava pile is Fe-rich sub-alkaline tholeiitic basalt. The samples are low in MgO (5.91 ± 1.41 wt %) and, correspondingly, their Mg numbers (Mg/(Mg + Fe+2)) are also lower (43.9 ± 7.8). Two samples at 131 m and 268 m below MSL have distinctively higher MgO (7.44 and 9.18 wt%, respectively) and higher Mg number (61.8 ± 0.3). Cr and Ni show compatible behaviour in basaltic melts and decrease with MgO. Abundances of both these elements are low in KBH-7 samples (Cr < 350 ppm; Ni < 1 12 ppm) and batch melting and Rayleigh fractionation models using Zr, Cr and Ni in KBH-7 samples are in agreement with olivine-rich picrite fractionation from primitive mantle melts (Radhakrishna et al. Reference Radhakrishna, Mohamed, Venkateshwarlu, Soumya and Prachiti2019).
Samples across KBH-7 section show generally variable enrichment in light rare earth elements (LREE) relative to heavy rare earth elements (HREE) in chondrite-normalized diagrams and variable levels of large ion lithophile element (LILE) enrichment in primordial mantle-normalized diagrams (Figure 2). Samples between ∼600 and 260 m MSL have lowest LREE and LILE concentrations. More conspicuous are the samples downward from ∼120 m below MSL, which have relatively more fractionated REE patterns and higher levels of more incompatible trace element concentrations. All samples show clear negative Sr spikes and mild negative spikes for Ti in Figure 2. Samples above ∼−120 m MSL characteristically have no or very mild negative Nb anomaly, whereas samples below this elevation have higher to mild negative Nb anomaly (Figure 2). These geochemical distinctions across the drill hole in relation to subsurface stratigraphy are described in the following section.
Chondrite normalized rare earth element patterns (a) and primordial mantle normalized incompatible trace element patterns (b) representative of Deccan basalt along KBH-7 drill hole near Koyna. Chondrite and primordial mantle values for normalisation are after Sun and McDonough (Reference Sun, McDonough, Saunders and Norry1989). The drill hole section depth-wise is classified into different stratigraphic formation as shown in red (Mahabaleshwar), green (Ambenali) and blue (Poladpur) of the Wai Subgroup on left and in brown (Bushe) and green (Kandala) of the Lonavala Subgroup on right. The boundary between the two subgroups is marked at ∼120 m below MSL. Details of stratigraphic formations are given in Section 4.2. Altitude of samples is given as MSL in parentheses in legend.
Mineral magnetic properties relevant to the current study are summarized here from the detailed descriptions (Asanulla et al. Reference Asanulla, Radhakrishna, Bansal and Ramakrishna2024). Most of the samples have very low magnetic viscosity index values (v; <5%) and good magnetic stability, which is evident in only a single component of magnetization. Positive magnetic susceptibility values of all samples (137 to 2020 × 10−5 SI) (mean: 465 × 10−5 SI) reveal ferrimagnetic mineral character. Both isothermal remanent magnetisation (IRM) acquisition and backfield curves of all samples recorded saturation magnetization between 300 and 400 mT applied field, indicating absence of canted antiferromagnetic mineral (e.g. haematite) character. All other magnetic parameters and/or inter-magnetic/inter-parametric ratios, such as anhysteretic remanent magnetisation (ARM) and saturation isothermal remanent magnetization, are calculated from IRM acquisition curves, while coercivity of remanence is calculated from IRM backfield curves. Soft and Hard IRM values are calculated from IRM forward and back field values; these values, together with hysteresis data from representative/selected samples, also reveal that all the samples are single domain (SD) and pseudo-SD (PSD) or PSD and multi domain grained ferrimagnetic mineral dominance.
The inclination data plotted against elevation (Figure 3) indicate a single polarity change from a normal-polarity sequence at the top to a reversed-polarity sequence, with the polarity flip occurring between 636 and 632 m MSL and the transition initiating below 632 m MSL. The normal/reverse polarities recorded from Deccan lava sequence are linked to 29N/29R (Vandamme et al. Reference Vandamme, Courtillot, Besse and Montigny1991; Chenet et al. Reference Chenet, Courtillot, Fluteau, Gerard, Quidelleur, Khadri, Subbarao and Thordarson2009; Jay et al. Reference Jay, Mac Niocaill, Widdowson, Self and Turner2009). The consecutive inclination values show significant variation at some elevation ranges, while at some elevations the values are statistically consistent. These variations are discussed further in the following section.
Plot of palaeomagnetic inclination vs altitude across the Deccan section of KBH-7 drill hole near Koyna. Mean values of each DG from KBH-7 (solid red) and nearby (∼18 km) KBH-8 (solid yellow) are also plotted. The elevation ranges and standard deviation of inclinations in each DG are marked in error bars.
Figure 3. Long description
Panel: A scatter plot representing the relationship between inclination and elevation. The horizontal axis shows inclination in degrees ranging from -90 to 90. The vertical axis shows elevation in meters ranging from -300 to 1000. The plot includes data points for KBH-7 all inclinations data, DGs in KBH-7, and DGs in KBH-8. Several clusters of data points are visible, with some marked as DG 1 to DG 8. The plot also includes error bars indicating the standard deviation of inclinations in each DG. The data points show a general trend of varying inclination with elevation.
5. Discussion
5.a. Demarcation of discrete eruptive phases of volcanism
The Deccan lava flows erupted in discrete eruptive phases is evident from the occurrence of red boles and alteration zones within the lava pile. Khanna et al. (Reference Khanna, Kanakdande, Bizimis and Arora2023) demarcated 15 discrete eruptive phases in KBH-7 and nearby (18 km) KBH-8 drill holes based on geochemical criteria. A closer examination reveals some difficulties in such demarcation. For example, their study demarcated ∼300 m thick lava pile (∼630–300 m MSL) into a single eruptive phase. Considering the thicknesses of other eruptive phases, it is hard to conceive such a thick lava pile to be a single eruptive phase, particularly when alteration zones or red boles are encountered within the interval that indicate a period of quiescence. That is, the lava pile in this elevation range comprises discrete eruptive phases but may have a similar petrogenetic history and recognizing different eruptive phases in such a situation is not easy based on geochemistry alone. However, the lava within ∼590–390 m MSL range appears to have subtle differences (relatively lower La/Smn, Ba/Nb and Ba/Y and higher Sr/Ba) compared to its immediate top (∼625–590 m MSL) and bottom (590–390 m MSL) sections, suggesting that the lava in this elevation may constitute geochemically discrete eruptive phases rather than one eruptive phase. Furthermore, geochemistry can change within a single eruptive cycle if a different magma chamber or part of the chamber is accessed, in which case it may be argued that defining eruptive phases by geochemistry alone is not perfect. Thus, it is necessary to be cautious in determining and interpreting eruptive phases based on geochemistry alone.
5.b. Demarcation of subsurface stratigraphy
The subsurface stratigraphy in the CSDD Koyna drill holes is not yet known. Here, an attempt is made to demarcate the stratigraphic formations across KBH-7 drill hole down to the basement to define subsurface stratigraphy of the southern Deccan Province. The recent results of Basu et al. (Reference Basu, Yannopoulos and Chakrabarty2020) are used to testify the demarcated subsurface stratigraphy down the drill hole, as their study fully characterized all the major lava formations of this province with respect to elemental and isotope geochemistry and is regarded as a reasonably complete volcano-stratigraphy of the Deccan.
The topmost section of the KBH-7 drill hole shows normal geomagnetic polarity and is designated as Mahabaleshwar Formation following the magnetostratigraphy of Chenet et al. (Reference Chenet, Flutean, Courtillot, Gerard and Subbarao2008) and Radhakrishna et al. (Reference Radhakrishna, Mohamed, Venkateshwarlu, Soumya and Prachiti2019). The geochemical data of KBH-7 samples (Table S1) and the elemental patterns in Figure 2 suggest the following. Samples in the top section display moderately fractionated light to heavy REE patterns and higher levels of incompatible element patterns. The section geochemically shows a change at about 612 ± 12 m MSL. Below the depth down to 260 ± 10 m MSL, the samples have somewhat lower Sr, Ba and Zr contents and have relatively lower LREE and incompatible trace elements compared to samples in the top section. Thus, the lava pile in this section is demarcated to constitute the Ambenali Formation. The underlying lava pile represents the Poladpur Formation downwards to a depth of −118 ± 12 m MSL. The change from Ambenali to Poladpur is marked by a slightly lower TiO2, high Ba and relatively enriched LREE and more incompatible trace element patterns. The geochemically demarcated boundaries between the Mahabaleshwar, Ambenali and Poladpur formations closely coincide with the occurrence of GPBs, which are considered marker horizons in the stratigraphy (Shandilya et al. Reference Shandilya, Chatterjee, Pattabhiram, Bodas, Pande and Kale2020).
The altitude level at 118 ± 12 m below MSL in KBH-7 is considered to represent the stratigraphic boundary between Wai and Lonavala Subgroups. The top Bushe Formation of Lonavala Subgroup extends down to 172 ± 10 m below MSL, and then the Khandala Formation down to basement. The Bushe characteristically contains highest Mg no (54–62), Cr (198–341 ppm) and Ni (109–125 ppm) and lowest Zr (79–104) compared to rest of the drill hole section. This formation contains lowest ∑REE (79 ± 5) and lowest (Tb/Yb)n (1.41 ± 0.04) in the entire section. Some of these geochemical distinctions of the Bushe from other formations are noted by Mahoney (Reference Mahoney and Macdougall1988). The change from Bushe to Khandala Formation is marked by lowering of Mg#, Cr, Ni and increase in TiO2, P2O5, Fe2O3, Sr, Ba, Nb, Zr and Hf. The change is also marked by sharp increase in ∑REE (151 ± 58) and HREE fractionation with (Tb/Yb)n (1.82 ± 0.03).
The geochemical distinction between the Wai and Lonavala Subgroups in KBH-7 is clearly evident in some of the incompatible trace element ratios. For example, the lower Lonavala have higher (La/Sm)n (1.98 ± 0.26), Ba/Nb (20.5 ± 9.6) and Ba/Y (6.5 ± 1.5), while samples of the overlying Wai Subgroup have lower (La/Sm)n = 1.45 ± 0.22, Ba/Nb = 9.7 ± 3.6 and Ba/Y = 3.6 ± 1.4. The elemental ratios as constrained in each of the stratigraphic formations across the KBH-7 drill hole are plotted along with those reported by Basu et al. (Reference Basu, Yannopoulos and Chakrabarty2020) in Figure 4. A reasonably good agreement between both the results testifies to the demarcation of the subsurface stratigraphic formations at depth in the KBH-7 drill hole. Figure 4 reveals that the Bushe Formation is characterized by high Ba/Y, Ba/Nb, (La/Sm)n, Sr/Nb and lowest Nb/Th. Compared to Bushe, Poladpur Formation shows diminished Ba/Y, Ba/Nb, (La/Sm)n and Sr/Nb and somewhat increased Nb/Th. Both Poladpur and the overlaying Ambenali formations have nearly similar compositions, but Ambenali appears to show slightly increased Nb/Th and a little lower Zr/Nb, Ba/Y and Ba/Nb than Poladpur. The Mahabaleshwar Formation, compared to the underlying Ambenali, shows relatively higher Ba/Y and (La/Sm)n and lower Sr/Nb and Zr/Nb. Similar trends for selected trace element ratios (Zr/Nb, Ba/Y) across the Wai Subgroup formations are discernible in many earlier studies (see Table 2 from Widdowson et al. Reference Widdowson, Pringle and Fernandez2000).
Selected incompatible trace element ratios within the demarcated stratigraphic formations along KBH-7 drill hole in comparison with the ratios of Deccan stratigraphic formations reported by Basu et al. (Reference Basu, Yannopoulos and Chakrabarty2020). Y values are not available in Basu et al. (Reference Basu, Yannopoulos and Chakrabarty2020) and therefore Ba/Y values are calculated from the compilation of Kale et al. (Reference Kale, Dole, Shandilya and Pande2020). A value much beyond error limits is treated as outlier and excluded. Blue and Gray circles indicate Basu et al. (Reference Basu, Yannopoulos and Chakrabarty2020) and KBH-7 ratios, respectively.
Whereas element plots and mantle-normalized trace-element patterns are useful for classifying basalt samples into stratigraphic formations, a statistical approach is more practical for treating large numbers of samples systematically. Discriminant Function Analysis (DFA) is a multivariate statistical technique that uses the chemical characteristics of samples from known groups to classify samples of unknown group membership. In simple terms, DFA first establishes the chemical characteristics that distinguish the known stratigraphic formations and then determines which formation an unknown sample most closely resembles based on its overall chemical composition. DFA has been successfully applied to large datasets for regional lava chemostratigraphy in flood basalt provinces.
In the present study, DFA was performed using IBM SPSS Statistics to classify the KBH-7 basalt samples into the established Deccan volcanic formations. A training dataset comprising 389 reference samples from the Mahabaleshwar, Ambenali, Poladpur, Bushe,and Khandala formations, compiled from Vanderkluysen et al. (Reference Vanderkluysen, Mahoney, Hooper, Sheth and Ray2011), was used for model development. Major- and trace-element concentrations (SiO₂, Al₂O₃, MgO, CaO, TiO₂, P₂O₅, K₂O, Ni, Ba, Sr, Zr, Y and Nb) were selected as predictor variables. The effectiveness of group discrimination was evaluated using Wilks’ Lambda and canonical correlation statistics. Posterior probabilities were calculated for each sample, and the unknown KBH-7 samples were assigned to the formation for which the posterior probability was highest.
The DFA results indicate that approximately 90% of the KBH-7 basalt samples are correctly classified into their corresponding formations recognized in the southwestern Deccan Province. These assignments are in close agreement with the stratigraphic subdivisions identified independently in this study. Furthermore, the Mahalanobis distances of the classified samples are comparable to those observed within the training dataset, supporting the robustness of the classification.
Figure 5 illustrates the distribution of KBH-7 samples in the space defined by the first two discriminant functions (F1 and F2) relative to the fields occupied by the corresponding formations in the southwestern Deccan Province. The F1 discriminant function clearly separates the Mahabaleshwar formation from the underlying Ambenali formation. In contrast, samples from the Ambenali and underlying Poladpur formations show partial overlap near their mutual boundary, consistent with the overlap observed in the elemental-ratio plots (Figure 4). This apparent geochemical overlap likely reflects the transitional chemical character of lavas within the Poladpur formation, as noted by Peng et al. (Reference Peng, Mahoney, Hooper, Macdougall and Krishnamurthy1998). The F1 discriminant function effectively distinguishes Bushe formation samples from those of the overlying Poladpur formation and Wai Subgroup, while F2 further resolves the separation between the Khandala and Bushe formations in the KBH-7 succession. Overall, the DFA results provide strong statistical support for the stratigraphic correlations established from geochemical data and reinforce the assignment of KBH-7 lava pile to the established formations of the southwestern Deccan Province.
Discriminant Function Analysis function F1 Vs Function F2 plot of KBH-7 drill hole samples for classification into stratigraphic formations of the Deccan lavas in Western Ghats of India. The fields representing the Deccan formations are defined using the Deccan training dataset. Samples from 172 ± 10 m below MSL downward show a distribution distinct from the overlying Bushe Formation and similar to that of the Khandala Formation, although they do not fall within the Khandala field. Given their stratigraphic position beneath the Bushe Formation and their geochemical affinity with the Khandala Formation, we classify them as Khandala Formation.
Figure 5. Long description
A scatter plot representing Discriminant Function Analysis with Function 1 on the horizontal axis and Function 2 on the vertical axis. The plot includes several clusters of data points, each representing different stratigraphic formations: Mahabaleshwar, Ambenali, Poladpur, Bushe, and Khandala. Each cluster is enclosed within an elliptical boundary, indicating the defined fields for each formation. The data points are color-coded: blue for Mahabaleshwar, orange for Ambenali, gray for Poladpur, yellow for Bushe, and green for Khandala. The plot shows that samples from 172 ± 10 meters below mean sea level (MSL) have a distribution distinct from the overlying Bushe Formation and similar to the Khandala Formation, although they do not fall within the Khandala field. These samples are classified as Khandala Formation based on their stratigraphic position and geochemical affinity.
The distinctions between different volcano-stratigraphic formations appear to be visible in some of the mineral magnetic properties also (Figure 6). The change from Bushe to Khandala is marked by fall in natural remanent magnetisation (NRM), susceptibility, ARM and soft IRM values and rise in Hcr. The change from Poladpur to Bushe is marked by increase in NRM and minor increase in susceptibility and soft IRM. The Poladpur and Ambenali, like in elemental ratios in Figure 4, share similar characteristics and are not distinguished. The Mahabaleshwar, compared to the underlying Ambenali, appears to be low in susceptibility, SIRM, Hcr and IRM (Soft and Hard IRM). Some of the properties exhibit overlaps in error bars across formations and therefore, further work would better constrain the observations of the present study as such results across the stratigraphic formations are not yet available.
Selected mineral magnetic properties within the stratigraphic formations demarcated along the KBH-7 drill hole near Koyna. NRM (×10−3 A/m); volume susceptibility (×10−5 SI); SIRM, saturation isothermal remanent magnetization (×10−3 Am2/kg); ARM, anhysteretic remanent magnetization (×10−5 Am2/kg); Hcr, coercivity of remanence calculated from IRM backfield curves (mT); Soft IRM (×10−3 Am2/kg); Hard IRM (×10−3 Am2/kg).
The subsurface stratigraphy proposed along Koyna drill hole broadly represents the stratigraphic sequence in the southern Deccan province. Thus, this work provides direct evidence for southward overstepping of older formations speculated from geological and structural observations (Mitchel and Widdowson, Reference Mitchell and Widdowson1991). Our results indicate that, in addition to the Wai Subgroup, the Lonavala Subgroup extends at least as far south as Koyna (17°18′N), whereas the older Kalsubai Subgroup is pinched out and/or overstepped to the north of Koyna.
5.c. Palaeomagnetic directional groups, secular variation and duration of volcanism
Palaeomagnetically recognizing directional groups is an efficient approach to constrain duration of flood basalt volcanism. Such directional groups represent short, quasi-instantaneous eruptions that do not meet the criteria for secular variation and represent single eruptive events. In this respect, palaeomagnetic data from KBH-7 and KBH-8 are more rewarding, as these drill holes provide thickest continuous succession in a vertical section down to the basement. In the absence of declination data, the inclination data are used to recognize directional groups. The current study considers more than three consecutive tightly constrained inclination values that are present in both drill holes at nearly comparable levels of elevation and may also cover >30 m thick lava section in either of the drill holes to constitute a directional group (DG). As an exception to this criteria, ∼20 m thick lava pile ∼750 m MSL towards the end of normal polarity is considered to represent DG (DG 2) because it is observed in both the drill holes and its inclination (−69–70°) compares well with that reported from a normal polarity section above the N/R boundary from nearby Deccan traverse (I = ∼70°; Chenet et al. Reference Chenet, Flutean, Courtillot, Gerard and Subbarao2008). Including this, KBH-7 and KBH-8 display 8 DGs (Figure 3). Seven DGs are at similar elevation levels in both drill holes; one DG (DG 6) displays some difference (∼100 m), but is preferred to be related to the same group because of good agreement in other DGs between the two drill holes. With respect to the stratigraphic formations demarcated in the previous section, 6 DGs are located within the Wai Subgroup. DG 1 and 2 are in Mahabaleshwar with normal polarity (DG 2 is regarded as transitional polarity). Other 4 are within a single reverse polarity located in Ambenali (DG 3 and 4) and Poladpur (DG 5 and 6). DG 7 and 8 are located in Lonavala Subgroup (one in Bushe and the other in Khandala formations). The only palaeomagnetic study identifying directional groups within the upper Deccan is along Mahabaleshwar–Poladpur (MP) section, ∼50 km north of Koyna (Chenet et al. Reference Chenet, Flutean, Courtillot, Gerard and Subbarao2008); their study also identified 6 DGs (2 each in Mahabaleshwar, Ambenali and Poladpur).
Successive lava sequences with statistically coherent palaeomagnetic directions were likely emplaced within a short time span that is not sufficient to record secular variation. The statistically tight within-group direction, considering the Holocene secular variation, may indicate eruption within a shorter time relative to the speed of palaeosecular variation that is estimated to be about 100 years (Chenet et al. Reference Chenet, Flutean, Courtillot, Gerard and Subbarao2008; Musgrave and Fussell, Reference Musgrave and Fussell2011; Usui and Tian, Reference Usui and Tian2017). The Wai Subgroup, which is widespread in the southern Deccan province, is the massive main Deccan volcanic phase and accounts for approximately 70% of the total Deccan basalt volume (Hoyer et al. Reference Hoyer, Haase, Regelous and Fluteau2023). The Wai Subgroup covers ∼1000 m in the drill hole, comprising 6 DGs put together ∼250 m thick lava pile and ∼750 m intermittent non-DG lava sequence. The 6 DGs may have erupted in not more than 600 years.
It may be possible that the ∼750 m thick intermittent lava sequence could have erupted on longer time scale. However, the rates of magma eruption during Wai subgroup are similar with only the top section witnessing increased rates of eruption (Renne et al. Reference Renne, Sprain, Richards, Self, Vanderkluysen and Pande2015), but recently Schone et al. (Reference Schoene, Eddy, Keller and Samperton2021) and Kale and Pande (Reference Kale and Pande2022) argue that there was no cognizable variance in the eruptive rates in the western sub-province. Thus, the eruption time scales in non-DG section cannot differ drastically from those of DG section. Even if the non-DG pile is considered to have taken double the time scales of DG eruption (i.e., 1200 years instead of 600 years for each 250 m lava pile), the 750 m non-DG sequence may have erupted in ∼3600 years and the entire Wai sequence thus could have erupted in ∼4200 years. A similar time frame was estimated for the Wai Subgroup by a robust consideration based on flow count in MP section (Chenet et al. Reference Chenet, Flutean, Courtillot, Gerard and Subbarao2008).
The time elapsed between the flows also needs to be accounted for to estimate total duration of the main Deccan eruption, and it can be constrained by the intermittent red boles and alteration zones developed between flow units. Palaeosoles and climate data estimated a cumulative time period of ∼200 ka for the formation of red boles and alteration zones within the Wai Subgroup (Chenet et al. Reference Chenet, Flutean, Courtillot, Gerard and Subbarao2008). Consequently, the 65 Ma main Deccan volcanism represented by Wai Subgroup, palaeomagnetically, lasted for a very short duration of a few thousand years (<4000–5000 years) within a total duration of about 200–300ka. Taken together, the palaeomagnetic directional groups provide evidence for rapid emplacement of the DG-bearing lava sequences, whereas the red-bole and alteration estimates indicate a substantially longer cumulative timescale for the development of interflow weathering horizons. These estimates are shorter than the duration inferred from high-precision geochronological studies of the Wai Subgroup (e.g., ∼600 ka; Schoene et al. Reference Schoene, Eddy, Samperton, Keller, Keller, Adatte and Khadri2019). As emphasized by Schoene et al. (Reference Schoene, Eddy, Keller and Samperton2021), achieving greater precision and accuracy in geochronological constraints is essential for resolving the eruptive history of the Deccan Traps, but remains challenging and requires integration with independent geological and geophysical observations.
Finally, the current study differed the discussion of the lowermost drill hole sections (Bushe and Khandala of Lonavala Subgroup) with a view that these may belong to the lower Deccan stratigraphic subgroups that are prevalent in northern Deccan Province, which are considered to belong to an earlier phase (e.g., Chenet et al. Reference Chenet, Flutean, Courtillot, Gerard and Subbarao2008).
5d. Assumptions and limitations in estimation of duration of volcanism
The estimates of the duration of volcanic activity are subject to the following assumptions and limitations. The ∼100-year timescale used here as a reference for palaeosecular variation is based on the rate of directional change inferred from the Holocene field. In the absence of a direct measure of geomagnetic field behaviour at 65 Ma, the six directional groups identified within the Wai Subgroup are assumed to represent relatively short eruptive episodes, using the Holocene reference timescale as an analogue.
The duration of the intervening non-DG lava sequences cannot be directly determined from the palaeomagnetic data and is instead estimated from the reported relatively uniform eruption rates during emplacement of the Wai Subgroup, together with the relative thicknesses of the DG and non-DG lava piles. These assumptions mean that the few-thousand-year duration estimated for the Wai Subgroup should be regarded as an approximate timescale rather than a precise determination of the duration of active volcanism.
The ∼200–300 ka estimate represents the cumulative timescale associated with palaeosol formation, alteration and weathering between lava flows and relies on red-bole thickness, climate-based estimates of weathering duration and alteration. This estimate is therefore subject to the associated uncertainties and should not be considered a direct measure of the duration of active volcanism.
Taken together, the palaeomagnetic directional groups provide evidence for rapid emplacement of the DG-bearing lava sequences, whereas the red-bole and alteration estimates indicate a substantially longer cumulative timescale associated with interflow weathering horizons. The resulting few-thousand-year timescale for active lava emplacement, in contrast to the ∼200–300 ka cumulative timescale associated with interflow weathering and alteration, should therefore be regarded as an approximate estimate, subject to the assumptions outlined above.
6. Conclusions
Detailed analysis of elemental geochemistry, mineral magnetism and palaeomagnetism of a continuous lava succession in southern Deccan province along one of the drill hole of maximum (1251 m) thickness reveals the following. An undisputed demarcation of discrete eruptive phases in the section does not seem to be feasible based on elemental geochemistry alone. However, incompatible trace element concentrations, their ratios and multivariant statistical analysis constructed hitherto an unknown depth-wise subsurface stratigraphy down to basement. The stratigraphy comprises thick (∼1000 m) Wai Subgroup (Mahabaleshwar, Ambenali and Poladpur formations) and relatively thin (∼250 m) Lonavala Subgroup (Bushe and Khandala formations), but the older Kalsubai Subgroup is totally absent. Mineral magnetic properties also show some distinctions across the formations but show some overlaps in error bars; hence, in the absence of similar studies, further studies would better constrain the distinctions.
The present results provide subsurface evidence to the southward overstepping of older formations by younger formations as speculated by geological mapping and structural observations (Mitchell and Widdowson, Reference Mitchell and Widdowson1991).
Palaeomagnetic inclination data from two thick and nearby drill holes (KBH-7 and KBH-8) identify eight directional groups, six within the Wai Subgroup and two within the Lonavala Subgroup. Based on the inferred geomagnetic secular-variation timescale and assumption-based estimates of eruption rates and interflow alteration, the directional groups suggest that the main 65 Ma Deccan volcanism represented by the Wai Subgroup was emplaced within a few thousand years (<4000–5000 years), whereas the cumulative timescale associated with interflow weathering and alteration may have extended to ∼200–300 ka. This few-thousand-year estimate is approximate and subject to the assumptions outlined above.
Supplementary material
The supplementary material for this article can be found at https://doi.org/10.1017/S0016756826101095
Data availability statement
All data used are available in electronic supplementary tables.
Acknowledgements
The samples for this work are provided by the Borehole Geophysical Research Laboratory, Karad, Maharashtra, to Drs. T Radhakrishna and M. Venkateshwarlu (principal investigators) of the MoES (Ministry of Earth Sciences, Government of India) project. The work was carried out under the financial grants of the MoES project no. MoES/P.O. (Seismo)/(195)/2013. This work was also carried out under CSIR Emeritus scientist programme (21/1041/17/EMR-II). We thank Dr P.K. Prachiti for her help during the sampling National Centre for Earth Science Studies (NCESS) and Council of Scientific and Industrial Research (CSIR)-National Geophysical Research Institute (NGRI) have provided all support and facilities. Dr. Harsh K Gupta motivated and inspired us to carry out this work. We appreciate Andrea Marzoli and an anonymous reviewer for their constructive comments, and thanks to Prof. Peter Clift and Eleanor Jennings for editorial handling.
Author contributions
T.R. Conceptualization; T.R. and M.V.: Funding acquisition; T.R., R.M.A. and M.V.: Equally in Data curation, Formal analysis, Investigation; Methodology, Validation, review & editing; Ch.R: Review & editing
Financial support
Funding information is not applicable/no funding was received.
Competing interests
The authors declare no competing interests.
Open access
