Approximately 2,000 km of 2D and 4,800 km² of 3D seismic reflection data across the northwestern and central CLB were utilized. We performed sequence stratigraphic analysis by identifying and mapping key sequence boundaries (specifically SH-5 and SH-7, which bound the Lower Miocene). Seismic facies distributions were mapped based on reflection characteristics (amplitude, continuity, configuration) and calibrated by tying seismic data to 32 well locations. Well log interpretation (Rider, 1996) and conventional core data provided direct lithological and environmental information, allowing us to ground-truth the interpretation of seismic patterns (e.g. chaotic, high-amplitude reflections corresponding to channelized sandstones).

Two conventional cores from the Lower Miocene section in wells WT10 (sample BK03) and WT19 (sample BK04) were strategically chosen for U-Pb dating to represent different positions within the basin. An additional 200 thin sections from various wells were analyzed for petrography. This involved point counting (minimum 300 points per slide) to determine mineralogical composition. To aid in provenance interpretation, modal compositions from the petrographic dataset (N=200) were plotted on a Quartz-Feldspar-Lithic fragment (QFL) ternary diagram following the methodology of Dickinson (1985).

Zircon grains were separated from the two core samples (BK03, BK04) using standard heavy liquid and magnetic techniques. Cathodoluminescence (CL) imaging was performed to reveal internal structures and guide laser spot placement. U-Pb isotopic analyses were conducted in-situ using Laser Ablation Inductively Coupled Plasma Mass Spectrometry (LA-ICP-MS), following procedures described by Hieu et al. (2013). Raw data were processed using GLITTER software (Griffin et al., 2008). Analytical results exhibiting discordance greater than 10% were excluded from the final geological interpretation to ensure data robustness. For zircons <1000 Ma, the ²⁰⁶Pb/²³⁸U age is used; for grains >1000 Ma, the ²⁰⁷Pb/²⁰⁶Pb age is preferred. Age distributions were plotted as probability density plots (PDPs) with kernel density estimates (KDE) using IsoplotR (Vermeesch, 2018). Age peaks were statistically identified from the maximum on the generated KDE curves.

The Lower Miocene interval is clearly bounded by regional seismic horizons SH-5 (base) and SH-7 (top) and comprises two main sequences, BI.1 and BI.2 (see Figure 3). An illustrative seismic line shows the typical basin architecture, with fault-bounded horst and graben structures inherited from earlier rifting phases. Seismic facies analysis reveals significant lateral variations. In the northwestern part of the study area, the BI.1 and BI.2 intervals are often characterized by discontinuous, chaotic to sub-parallel, moderate-to-high amplitude reflections (see Figure 4). Tying these facies to wells reveals they correspond to sections with blocky, sand-prone log signatures, interpreted as channelized fluvial and distributary channel systems. Towards the east and southeast, these facies transition into more continuous, parallel to sub-parallel, lower-amplitude reflections, which correspond to more heterolithic log signatures, interpreted as lower-energy coastal plain, floodplain, or distal deltaic deposits as illustrated on the seismic profile (see Figure 3) and paleo-environmental maps (see Figure 4).

Well log correlation across the study area shows that the Lower Miocene interval generally thickens from west to east (see Figure 5). The BI.2 reservoir interval ranges from 250 m to over 600 m thick. In the SD field, petrophysical analysis indicates good reservoir quality with porosity (φ) from 17% to 27% and net-to-gross (NTG) from 28% to 61.5%. The underlying BI.1 reservoir varies from 110 m to over 650 m thick. In the HS area, it exhibits porosities of 19-22% and NTG of 55-70% (see Table 1).

Core observations from the BI.1 and BI.2 intervals show light- to dark-gray, fine- to medium-grained sandstones with features including cross-bedding, laminations, and fining-upward trends (see Figure 6). Petrographic analysis reveals the sandstones are predominantly Arkose and Lithic Arkose. They are compositionally immature, containing abundant feldspar (32-34%, mostly K-feldspar) and lithic fragments (2-8%, mostly volcanic and granitoid), in addition to quartz (30-33%) (see Table 2). The modal composition data, when plotted on a Dickinson (1985) QFL diagram, falls overwhelmingly in the "Dissected Arc" and "Transitional Arc" provenance fields (see Figure 7). This strongly indicates that the sediment was derived from the erosion of a magmatic arc terrane.

The U-Pb dating results from both samples (BK03 and BK04) are strikingly similar (see Figure 8). Sample BK03-WT10 yielded a dominant Cretaceous age population (63.3% of grains) with a prominent peak at ca. 108 Ma. Sample BK04-WT19 also yielded a dominant Cretaceous age population (67.4% of grains) with a prominent peak at ca. 105 Ma. Both samples also contain subordinate populations of older, likely recycled zircons, including Permian-Triassic (ca. 210-280 Ma), Ordovician-Silurian (ca. 440-460 Ma), and Precambrian grains. The Cretaceous zircons are typically euhedral to subhedral with clear oscillatory zoning, indicative of a magmatic origin and relatively short transport (see Figure 9).

The detrital zircon U-Pb age data provide robust, quantitative evidence for the primary sediment source. The overwhelming predominance of Cretaceous zircons (ca. 105–108 Ma) in both samples points unequivocally to the Dalat Zone on the adjacent Vietnamese mainland as the principal source terrane (see Figure 8). This age range correlates perfectly with the well-documented Cretaceous granitoid intrusions and volcanic rocks of the Dalat Zone (Shellnutt et al., 2013; Hieu et al., 2017; Tod et al., 2021). This interpretation is strongly supported by the petrographic data, as the arkosic composition classifies the sandstones within the "Dissected Arc" field on the QFL diagram (see Figure 7), a classic signature of sediment derived from an eroded granitic-volcanic arc. The euhedral to subhedral morphology of the Cretaceous zircons (see Figure 6) suggests relatively short transport distances, consistent with derivation from the adjacent mainland.

Our results provide the first direct geochronological confirmation for the provenance of these specific reservoir units. While previous studies (e.g. Breitfeld et al., 2022) inferred a Dalat Zone source, our data quantify its dominance (~65%). Furthermore, the lack of a significant Permian-Triassic zircon peak, which is characteristic of the Mekong River system, suggests that a "Proto-Mekong" source had minimal to no influence on deposition in the northwestern CLB during the Early Miocene. The subordinate older zircon populations were likely recycled from Paleozoic-Mesozoic sedimentary cover rocks or basement terranes exposed within the Dalat Zone's eroding hinterland.

Integrating core, well log, and seismic data allows for a refined interpretation of depositional environments. The Lower Miocene succession records a transition from terrestrial to marine-influenced settings.

Formation BI.1 is interpreted as being deposited primarily in alluvial and upper delta plain environments. The blocky, sand-prone seismic and well-log facies represent fluvial channel belts.

Formation BI.2 represents a landward shift of environments, with deposition occurring in lower delta plain to coastal plain and marginal marine settings. The interbedded nature of sands and shales reflects deposition in distributary channels, mouth bars, and adjacent lower-energy interdistributary bays or floodplains.

The overall sediment transport pathway was from the uplifted Dalat Zone in the northwest towards the subsiding basin center in the southeast. Reservoir quality is highest in the sandstone facies associated with higher-energy channel axes (see Table 1).

Early Miocene sedimentation occurred during the post-rift thermal sag phase of the CLB's evolution. The primary control on sediment supply was the significant uplift and erosion of the Dalat Zone magmatic arc. This uplift was likely driven by a combination of regional tectonic forces, including post-rift thermal doming and far-field stresses related to the ongoing India-Asia collision (Hieu, 2015; Breitfeld et al., 2022). The creation of accommodation space was controlled by regional thermal subsidence, while the distribution of depositional systems was heavily influenced by the underlying basement topography of horsts and grabens, which focused sediment fairways into structural lows. The overall landward-to-seaward transition from BI.1 to BI.2 reflects a relative sea-level rise, consistent with a long-term transgressive-to-highstand sequence set. The stacking patterns observed on seismic data, with fluvial systems (BI.1) overlain by more widespread deltaic/coastal systems (BI.2), are consistent with a long-term transgressive-to-highstand sequence set (see Figure 3).

This integrated study provides a robust source-to-sink model for the Lower Miocene reservoirs in the CLB. The confirmation of a proximal, stable Dalat Zone source and the delineation of depositional fairways controlled by basement structure are critical for predicting reservoir distribution and quality in less explored parts of the basin. The provenance conclusions are based on two sample locations, and while their consistency provides confidence, further spatial variability could exist. Similarly, the density of well and core data limits the resolution of paleo-environmental mapping. Future research could expand the detrital zircon dataset to other parts of the basin to test for spatial or temporal changes in provenance and further refine sequence stratigraphic models.