Abstract: Vehicle-borne ground penetrating radar system can efficiently detect road cavities and play an important role in urban safety management. The stability of integrated GNSS / IMU positioning for the vehicle-borneground penetrating radar system is significantly degraded in urban canyons due to the accuracy decreases of GNSS. The robustness of GNSS / IMU integrated navigation can be improved by robust filter based on the normalized innovation squared. However, clear variations arise in outlier suppression and useful information utilization, due to the distinct attenuation characteristics and truncation mechanisms of different robust kernel functions. To investigate the suitability and robustness differences of various kernels in complex observation environments, the GNSS / IMU integrated navigation performance of four robust kernel functions, namely Cauchy, Student-t, Tukey, and Huber, is compared within the error-state Kalman filter (ESKF) framework using measured data collected on the urban expressway in Changchun. The results indicate that the Cauchy and Student-t kernels achieve superior robustness performance through their heavy-tailed distributions with asymptotic attenuation characteristics, with the latter being slightly more robust than the former. Compared with the conventional ESKF, the position RMSE, velocity RMSE, and the 95th percentile of position error ( P95) are improved by 11. 1% and 12. 1% , 3. 0% and 4. 9% , and 22. 3% and 21. 0% for the Cauchy and Student-t kernels, respectively. The Huber kernel exhibits relatively weak robustness due to its linear attenuation and lack of truncation. Compared with the conventional ESKF, the improvements are limited to 4. 5% , 0. 6% , and 15. 3% in position RMSE, velocity RMSE, and P95, respectively. The Tukey kernel exhibits the lowest robustness among the four candidates, as its strong truncation results in insufficient utilization of useful information and severe accuracy degradation. This study provides a reference for robust kernel function selection and algorithm design in GNSS / IMU integrated navigation of vehicle-borne ground penetrating radar system in complex urban environments. 

Abstract: Diabase has been increasingly verified as an important reservoir in hydrocarbon exploration and development, however, its strong heterogeneity makes predicting favorable reservoirs difficult. Current studies have not yet established a quantitative response relationship between crystallization degree and reservoir properties, and the lack of a unified classification evaluation standard severely restricts the fine-scale evaluation and exploration deployment of such reservoirs. This study investigates diabase with different crystallization degrees from the third member of Shahejie Formation in Linnan Subsag. A crystal size distribution (CSD) method with manual intervention was used to quantify crystallization degree, and in combination with the cooling and crystallization characteristics of shallow intrusive rocks, a three-tier classification scheme-fine-grained, medium-grained, and coarse-grained diabase-was established for Linnan Subsag using crystal size thresholds of 200 μm and 400 μm for the first time. Porosity-permeability measurements, cast thin sections, scanning electron microscopy, and high-pressure mercury intrusion analyses were integrated to examine the relationship between crystallization degree and reservoir properties, revealing a positive correlation between them. Mechanistically, during the magmatic cooling and crystallization process, coarse-grained diabase experienced longer mineral crystal growth times and larger crystal sizes, leading to well-developed intergranular pores. Furthermore, the connected pores between feldspar crystals in the coarsegrained structure facilitate the influx of later diagenetic fluids and promote dissolution, resulting in the formation of dissolution pores. Consequently, coarse-grained diabase exhibits significantly superior pore structures and storagepermeability properties compared to medium- and fine-grained diabase, identifying it as the most promising lithology for diabase reservoirs.

Abstract: Accurate detection of fire zones is an important prerequisite for effective coal mine remediation. Coalfield fire zones occur in steep terrain and are accompanied by severe surface smoke cover, which restricts ground geophysical surveys and makes it impossible to guarantee detection accuracy. Based on the characteristics of high magnetic susceptibility and low resistivity of fire zones, the synergistic detection technique combining semiairborne electromagnetic and airborne magnetic methods is suitable for complex terrain conditions, compensates for the limitations of individual methods, and enhances the resolution of the spatial distribution of fire zones. However, existing 3D inversion methods for electromagnetic and magnetic data do not adequately address remanent magnetization from combustion or complex terrain, which leads to poor delineation of subsurface fire zone distributions. The authors developed a 3D magnetization intensity vector and resistivity inversion technique for complex terrain, which effectively obtains the subsurface 3D resistivity and magnetization intensity distributions and provides a precise and efficient technical means for fire zone remediation. The Haizhou open-pit coal mine, once the largest open-pit mine in Asia, has undergone severe ground landslides, subsidence, and fissuring after closure due to subsurface fire zones formed by spontaneous combustion of autochthonous residual coal and historical mining activities. The terrain is complex, and conventional 3D electromagnetic and magnetic inversion methods do not meet detection requirements. Therefore, the synergistic detection technique combining semi-airborne electromagnetic and airborne magnetic methods was applied in the Haizhou open-pit coal mine area. After systematic data processing, the developed complex terrain inversion technique was used to perform 3D magnetization intensity vector and resistivity inversion. Based on the distribution patterns of high magnetization intensity and low resistivity, a comprehensive interpretation identified 17 fire zones. Field investigations validated the accuracy of the inferred results, and the identified fire zones were classified into 3 types: type Ⅰ corresponding to subsurface spontaneous combustion of residual coal (4 sites), type Ⅱ corresponding to shallow coal gangue combustion or accumulation (12 sites), and type Ⅲ corresponding to suspected fire zones (1 site). 

Abstract: Downhole heaters are critical equipment for the in-situ conversion of oil shale, and their heat transferperformance directly influences heating efficiency and energy consumption. Based on conventional heat exchangers, the authors propose a downhole convective shell-and-tube electric heater adapted to the wellbore diameter constraints of oil shale in-situ conversion through dimensional optimization. A helical baffle design is introduced to enhance the shell-and-tube heat exchange structure. According to this, two types of heater prototypes with different structures have been developed respectively. To analyze the heat transfer performance and irreversible losses of the downhole heating system for oil shale in-situ conversion, a high-temperature and high-pressure experimental platform was established. Experimental investigations were conducted on both shell-and-tube and helical structures under various heating powers and air volumetric flow rates, focusing on key heat injection parameters. The performance of the two configurations in terms of heat exchange and energy utilization was analyzed by comparing parameters such as wall temperature, heat transfer coefficient, thermal efficiency, comprehensive performance index, dimensionless entropy production, and thermal resistance. The experimental results demonstrate that, compared to the shell-andtube structure, the helical structure effectively reduces the peak wall temperature of the electric heating rods. Within the tested power and flow rate ranges, the heat transfer coefficient increased by 51. 4% to 66. 5% , and thermal efficiency improved by 2. 7% to 3. 7% . Regarding the comprehensive performance indicator K / Δp 1 / 3 , the helical structure showed an increase of 17. 1% to 37. 5% over the shell-and-tube structure, indicating superior heat transfer and flow resistance synergy. Furthermore, it exhibited lower irreversible losses in terms of entropy production and thermal resistance, demonstrating a clear advantage in enhancing energy utilization. 

The increasing demand for target detection accuracy in applications such as land resource surveys, urban planning, ecological monitoring, and disaster early warning, using high-resolution remote sensing images from drones and satellites, has exposed limitations in traditional detection methods. These limitations are particularly noticeable when there are significant scale differences, complex backgrounds, low contrast, and sparse samples. In particular, detecting small targets in multi-scale and complex background scenes has been challenging. To address these issues, this paper proposes an improved YOLO small target remote sensing image recognition algorithm called PSC --YOLO. It is based on the YOLO11n framework and incorporates multiple feature enhancement and optimization mechanisms. These additions improve the models ability to handle small targets and complex backgrounds. First, the PMKCA module is introduced into the backbone network. It uses parallel multi-core convolution and channel attention mechanisms to enhance multi-scale feature extraction and improve global feature modeling. Second, the SPPCA module is embedded into the SPPF feature fusion layer. It employs multi-scale pooling and channel attention mechanisms to boost small target perception. Next, the C3k2 HLCA module is integrated into the C3k2 structure. It enhances local feature representation and optimizes feature expression using channel attention mechanisms, which further improves small target detection accuracy. Finally, an adaptive confidence threshold adjustment function is designed to optimize the NMS process. This reduces redundant detection boxes and improves performance. Experimental results show that PSC--YOLO outperforms YOLOv5, YOLOv10, YOLO11n, and YOLOv12 on the NWPU VHR--10 dataset. For the mAP50 and mAP50--95 metrics, PSC--YOLO achieves a 0. 8% and 2. 0% improvement over YOLOv5, a 1. 5% and 2. 0% improvement over YOLOv10, a 0. 8% and 2. 0% improvement over YOLO11n, and a 1. 5% and 2. 0% improvement over YOLOv12. The results demonstrate that PSC --YOLO effectively reduces missed detections in complex background and multi-scale target scenarios. It also improves small target detection accuracy and provides better stability. This method offers more precise information support for the automated interpretation and intelligent decision-making of remote sensing images.

To finely characterize the reservoir structure of unconfined slope channels in gravity flow deposits of
Baiyun Sag, Pearl River Mouth Basin, the authors took the X Gas Field in Baiyun Sag as the research object. Com-
prehensive use was made of core, logging, seismic inversion and other data to study the internal structure and quan-
titative characteristics of unconfined channel reservoirs at different levels. A fine characterization method for the in-
ternal structure of unconfined channels was established, and the contacted and connected relationship of channel
sandbodies were revealed. The research results show that: ① The study area mainly develops medium-fine-grained
feldspathic quartz sandstone. Core grain size data points are basically parallel to the C = M baseline, with massive
mud gravels and intense bioturbation without obvious incised bedforms, indicating an unconfined channel system.
② The channel system in the study area develops four composite channel sequences, multiple composite channels
and isolated channels. The width of composite channel sequences is on the kilometer scale, with an aspect ratio of
133--136. The width of composite channels is approximately 1 000 m, with an aspect ratio of 10--55. The scale of
solated channels is difficult to determine, with a width of about 50--100 m and an aspect ratio of 4--8. ③ There are
four superposition modes of sandbodies inside the unconfined channels, including vertical isolated, channel-channel
lateral superimposed, channel-channel lateral juxtaposed, and isolated channels. Among them, the incised-super-
imposed reservoir structure has the best connectivity, the juxtaposed reservoir structure has good connectivity, and
the isolated reservoir structure is disconnected. This research achievement can be used to guide the precise develop-
ment and development adjustment of oil and gas fields in unconfined gravity flow channels.

Abstract: The magmatic rocks exposed in Xiyinkeng area (Jurong City, Jiangsu Province) belong to the Anjishancomplex. The rock type is mainly granodiorite porphyry, which is intruded by a small amount of intermediate-basicdykes such as quartz diorite porphyrite, diorite porphyrite, and diabase porphyrite. To determine the timing of mag-matism (crystallization / emplacement age) and geodynamic background, this paper conducted petrographic, zirconU--Pb chronology, and geochemical characteristics research on Xiyinkeng magmatic rocks. Zircon U--Pb dating in-dicated that granodiorite porphyry crystallized at (104. 45 ± 0. 29) Ma, followed by the emplacement of quartz dio-rite porphyrite at (104. 66 ± 0. 52) Ma, and subsequently diabase porphyrite intrusion at (101. 37 ± 0. 34) Ma, all occurring during the Early Cretaceous. Geochemical data show that the magmatic rocks in Xiyinkeng area are characterized by high potassium [w (K2O) 1. 96%--4. 05% ] and high alkalinity [w (Na2O) 2. 29%--4. 68% ], with A/ NK values ranging from 1. 38--2. 15, A/ CNK values ranging from 0. 62--0. 96, and Littmann indexes ranging from 1. 76--2. 59, indicating that they are metaluminous calc-alkaline series I-type granites. The diabase porphyrite exhibits higher rare earth elements enrichment ( ΣREE = 144. 61 × 10 - 6 - 182. 84 × 10 - 6, avg. 163. 73 × 10 - 6), with LREE/ HREE, (La/ Yb)N, and (Gd/ Yb)N ratios of 10. 81--12. 76, 13. 60--17. 62, and 2. 16--2. 33, respectively. In contrast, the rare earth element contents of granodiorite porphyry, quartz diorite porphyrite and diorite porphyrite are lower than those of diabase porphyrite (ΣREE = 94. 43 × 10 - 6 --143. 47 × 10 - 6, avg.
118. 79 × 10 - 6). The light and heavy rare earth element ratios (LREE / HREE) range from 14. 43--24. 48, the (La / Yb) N and (Gd / Yb) N are 22. 16--43. 70 and 2. 57--2. 87 respectively. The rocks are relatively enriched in large ion lithophile elements (Rb and La) and relatively depleted in high field strength elements (Nb, Ti and Ta), showing obvious characteristics of island arc magmatic rock. The Nb / Ta ratio is 11. 57--24. 68, which is close to that of continental crust, indicating the presence of a large amount of crustal material. Combined with previous research results, it is believed that the magmatic rocks in the Xiyinkeng area was formed in a back-arc extensional environment during the subduction and retreat of the Pacific Plate. It is the product of interaction between the lower
crust and mantle materials, classified as crust-mantle granite.

Abstract: Banded iron formation (BIF) is an important type of iron deposit formed during the early evolutionary history of the Earth. The Naxiguole BIF deposit is situated in the western segment of the East Kunlun Orogenic Belt. It is hosted within a medium-to high-grade metamorphic plagioclase amphibolite assemblage of the Neoprotero-zoic Jinshuikou Group and occurs as stratiform-lenticular bodies. With estimated resources of 75 Mt and iron grades ranging from 15% to 28. 83% , the ore primarily consists of magnetite quartzite exhibiting typical banded struc-tures. To gain a deeper understanding of the deposit genesis and provide a basis for regional mineral exploration, based on systematic field geological investigations combined with electron probe microanalysis (EPMA) of magne-
tite and whole-rock major and trace element geochemistry, this study systematically elucidates the geological charac-teristics, ore-forming material sources, and genetic type of the deposit. EPMA results reveal that magnetite exhibits high TFeO content (90. 01% --92. 98% ) and very low concentrations of TiO2, MgO, MnO, CaO, and Al2O3 (all < 0. 4% ), consistent with the composition of magnetite from typical BIFs. Whole-rock geochemical data indicate variable w (TFeO) (15. 45%--74. 74% ) and w (Al2O3) (1. 84%--17. 04% ) in the ore. The ore displays signifi-cant positive Eu anomalies (Eu / Eu∗PAAS = 1. 38--2. 12), positive Y anomalies (Y/ Y∗
PAAS = 0. 98--1. 42), low Eu /Sm ratio (0. 30--0. 51), and high Sm/ Yb (0. 69--2. 17) and Y/ Ho (26. 55--40. 73) ratios. Collectively, these geochemical characteristics indicate that the ore-forming materials were primarily derived from seawater mixed with minor proportions (0. 1% ) of high-temperature hydrothermal fluids. The absence of significant negative Ce anoma-lies in the ore, combined with the attributes of the host rocks, suggests formation in a reducing marginal sea envi-ronment. Integrating evidence from its strictly stratabound geological occurrence, typical magnetite quartzite mineral assemblage, and the characteristic geochemical signatures outlined above, this study identifies the Naxiguole iron deposit as a classic Superior-type BIF deposit.

Abstract: To address the limitation of existing suitability evaluation methods for underground space develop-
ment and utilization that predominantly rely on two-dimensional (2D) analyses and fail to accurately characterize
three-dimensional (3D) underground space, the authors propose a multi-elements 3D suitability evaluation method
based on a regular voxel model. By employing a voxel model as a unified evaluation framework, this study utilizes
multiple 3D spatial analysis techniques to construct geological models and multi-factor 3D voxel attribute models of
underground space. Combined with the analytic hierarchy process (AHP), an evaluation system incorporating con-
straint and graded indexes is constructed to evaluate the suitability of underground space development and utilization
in the Zhengzhou High-Tech Zone. The results indicate that over 65% of the underground space in the study area is
classified as highly suitable for development, with only a small proportion classified as restricted areas. Overall, the
engineering geological and hydrogeological conditions are favorable for development and utilization. In terms of spa-
tial distribution, the shallow layer exhibits the highest suitability, with a slight decrease in suitable areas as depth
increases. Horizontally, the western part of the study area shows greater development potential and suitability than
the eastern part. Compared with traditional 2D evaluation methods, the multi-factors 3D suitability evaluation meth-
od for underground space development and utilization based on 3D voxel model enables organization, management,
computation, and visualization of evaluation results at any spatial location. It effectively addresses the loss of verti-
cal information inherent in traditional 2D layered evaluation, enhances the evaluation resolution and result visibility
in the vertical dimension, and provides more comprehensive information and services for underground space plan-
ning and construction.

Abstract: To determine the stratigraphic position of the Naitoushan marble in the eastern margin of the
Changbai Mountains basalt-covered area, the authors conducted a study on its paleosedimentary environment using
petrology and petrochemistry. The results show that the main rock types of the Naitoushan marble are medium-thick
bedded marble and medium-thin bedded dolomitic marble, indicating formation in a coastal-shallow marine evapora-
tive environment slightly closer to the continent. The w ( SiO2 ) of the major elements ranges from 2. 04% to
6. 21% , with a correlation coefficient of 0. 38 between SiO2and Zr, suggesting that the SiO2 content is not signifi-
cantly related to terrigenous detritus and is mainly of chemical sedimentary origin. The w (MgO + CaO + LOI)
ranges from 91. 37% to 96. 77% , indicating that the rock is predominantly composed of carbonate minerals with
minimal detrital components. The CaO/ MgO ratio ranges from 1. 41 to 24. 34, suggesting deposition in a relatively
enclosed coastal-bay marine environment. The Fe3 + / Fe2 + ratio is between 0. 05 and 0. 80, possibly reflecting for-
mation in a relatively dynamic and weakly neutral environment. Trace and rare earth elements analysis show that the
correlation coefficient between Ce anomaly (Ce/ Ce∗) and Eu anomaly (Eu/ Eu∗) is -0. 46, and -0. 25 with total
rare earth elements (REE). The Mn / Sr ratio ranges from 0. 01 to 0. 05, indicating that the rare earth elements in
the marble samples from the study area were hardly affected by diagenesis. The correlation coefficient between Zr
and Th is 0. 75, 0. 14 with ∑REE, 0. 47 with Ce / Ce∗, and - 0. 48 with Eu / Eu∗, indicating that the REE distri-
bution in the marble samples was barely affected by external materials. The Ce / Ce∗ ratio of 0. 82--1. 00 indicates a
weakly oxidizing environment during marble formation. The (Tb / Pr) N ratio of 1. 02--3. 13, and the (Tb / Yb) Nratio
of 0. 96 --1. 56 indicate a slight enrichment of middle rare earth elements, suggesting possible hydrothermal input.
The average Eu / Eu∗ value is 1. 17, with correlation coefficient of 0. 12 with Fe2 + and 0. 56 with Mn2 + , and 0. 20
with (Fe2 + + Mn2 + ), also implying hydrothermal influence. The average Gd / Gd∗ is 1. 05, indicating that human
factors in sample testing have been basically excluded. The Y/ Ho ratio ranges from 31. 53 to 44. 00, with an
average of 35. 17, suggesting formation in a inshore or restricted marine environment. Combined with regional
geological data, it is believed that the Naitoushan marble was formed in a relatively closed coastal-shallow marine
environment under oxidized to weakly oxidized seawater, accompanied by hydrothermal input. The rock assemblage
is correlated with the lower part of the Zhenzhumen Formation, representing its eastern extension.

Abstract: This study focuses on two fine-flake graphite metallogenic belts in the Shanxi Fault Uplift and South
Qinling regions of Henan Province. Through field geological surveys, sample testing, and comprehensive analysis, the
authors systematically investigated the distribution characteristics, enrichment regularity, and genetic mechanism of
fine-flake graphite. Research indicates that fine-flake graphite ore bodies primarily occur in the Chishangou Formation of
the Yinyugou Group within the Shanxi Fault Uplift and the Dagou Formation of the Douling Group in South Qinling,
demonstrating significant stratigraphic specificity. Spatiotemporal distribution analysis reveals that the ore-hosting strata
are mainly Paleoproterozoic, with deposits exhibiting a northwest-southeast trend, consistent with regional structural
trends. The ore bodies are strictly strata-bound and conformable with the host strata. The ore-forming materials originate
from a combination of organic and inorganic sources, with the deposit type classified as regional metamorphic.
Further research demonstrates that the fine-flake characteristics of graphite are primarily influenced by tectonic
stress, metamorphism and migmatization. These findings provide a scientific basis for a deeper understanding of the
metallogenic patterns of fine-flake graphite deposits in Henan Province.

Abstract: Water source quality safety is a major livelihood issue. With numerous water quality indicators and
complex influencing factors, timely evaluation of water source quality is particularly important. Most current evaluation
methods are relatively macro-level assessments based on existing quality standards, lacking relatively micro-level scien-
tific evaluation of water bodies in water source areas across different periods. Therefore, establishing a comprehensive
evaluation system can provide a theoretical basis and technical support for pollution risk early warning in drinking
water sources and for improving water environment quality. Based on the single-factor index evaluation method and
the Nemero index method, the authors introduce the concept of an environmental capacity index to construct a
multi-factor comprehensive environmental capacity index evaluation model and a Nemero method environmental
capacity index model. These two models were used to conduct a comprehensive evaluation of a drinking water source
in Dafeng District, Yancheng City, Jiangsu Province. Furthermore, based on the water quality evaluation, a health risk
assessment was introduced to analyze the human health risks associated with the water source, comprehensively ensuring
drinking water safety. The evaluation results indicate that the overall environmental quality of this water body shows
a trend of improvement, while the chemical oxygen demand (COD) indicator shows a deteriorating trend; all other
indicators show improving trends. Environmental quality in August was relatively poor within the year. The established
multi-factor comprehensive environmental capacity index model and the Nemero method environmental capacity index
model both can be used to evaluate water bodies in water source areas with stable categories. Among them, the improved
Nemero method environmental capacity calculates the weight of each pollution factor, avoiding the neglect of certain
pollutants with low concentration but high hazard. The total health risk of the water source area was around 1. 81 ×
10 - 4 a - 1-- 2. 30 × 10 - 4 a - 1, classified as high risk. Carcinogenic arsenic contributed the most to this risk,
followed by carcinogenic hexavalent chromium. Compared to non-carcinogenic risks, carcinogenic substances are
the main source of potential risk through the drinking water pathway.

To investigate the rare earth element (REE) fractionation patterns, grades, and their influencing factors in different ore blocks of the Bayan Obo REE--Nb--Fe symbiotic deposit, and to provide guidance for research on REE patterns and directional mineral processing at Bayan Obo, elemental mass fraction was conducted using inductively coupled plasma mass spectrometry on samples from different ore blocks of TK13--04 core in the main orebody. Results show that two distinct REE fractionation patterns exist in the main orebody. The REE fractionation pattern of the middle dolomite Fe--REE ores differ from that of the upper and lower dolomite REE ores. The former exhibits an “initial upward inclination followed by rightward inclination” characteristic with increasing from La to Nd and then decreasing from Nd to Lu. In contrast, the latter shows a “rightward inclination” pattern with gradually decreasing from La to Lu. The middle dolomite Fe--REE ores have a lower mass fraction of total REEs compared to the upper and lower dolomite REE ores. The total REE mass fraction is closely related to the degree of light-to-heavy REE (LREE--HREE) fractionation, the REE fractionation pattern, and the total iron (TFe) mass fraction. Higher degrees of LREE --HREE fractionation and higher La/ Nd ratios correlate with higher total REE mass fraction. When the TFe is below 20%, there is no definitive correlation between TFe and the REE mass fraction. However, when the TFe mass fraction exceeds 20%, the total REE mass fraction decreases significantly. Compared to the average proportions of individual light REEs (LREEs) in the mining area, the proportion of Ce in the middle dolomite Fe--REE ore block of TK13--04 is significantly lower ( ~37%), while the proportion of Nd is significantly higher ( ~40. 7%), exceeding the proportion of Ce that is typically the dominant REE in the deposit. The proportion of La in the middle dolomite Fe--REE ore block is also reduced ( ~9%) and the proportions of Pr and Sm are increased. Therefore, the middle dolomite Fe --REE ore blocks hold greater potential for the utilization of Nd, Pr, and Sm elements. For the utilization of La and Ce, development of the upper and lower dolomite REE ore block would likely yield more substantial benefits.

This study is the first to analyze the geothermal resource occurrence conditions and the formationevolution mechanisms in the southern part of Taiyuan City, Shanxi Province, from both geothermal geophysical and geothermal water isotope geochemical perspectives. Based on geophysical exploration data and geothermal water isotope chemical analysis, and combined with regional geological data, a comprehensive study of the geothermal field in southern Taiyuan is conducted. The geophysical methods primarily include gravity, magnetic, and electrical surveys, which are used to analyze the structural characteristics and reservoir distribution of the geothermal field. The geochemical methods mainly include hydrochemical and isotopic analyses, aimed at revealing the origin, evolution, and water-rock interaction processes of thermal groundwater. The geophysical exploration results indicate that the Taiyuan Basin as a whole exhibits a low gravity field and negative magnetic anomalies, reflecting a deeply buried basement and strong tectonic activity, which provide favorable heat sources and reservoir spaces for the formation of geothermal fields. The southern geothermal field of Taiyuan is located at the southeastern margin of the basin, where the gravity field shows a distinct gradient zone, and the magnetic anomalies are characterized by negative values, suggesting the presence of concealed fault structures that serve as channels for deep thermal flow upwelling. Moreover, an apparent arcuate low-resistivity zone exists within the geothermal field, inferred to represent the geothermal reservoir, further demonstrating the favorable conditions for geothermal resource accumulation in this area. Geothermal water isotope geochemical analysis reveals that the thermal groundwater in the southern Taiyuan geothermal field originates from meteoric water, undergoing processes of infiltration, leaching, and transformation into sedimentary (semi-confined) water. During this process, water interacts with surrounding rocks through leaching, dissolution, and cation exchange. Stable components accumulate in the groundwater, while unstable components precipitate, thus forming the hydrogeochemical characteristics of modern geothermal water. The formation of the geothermal field in southern Taiyuan is mainly controlled by the favorable regional geological structure, the development of concealed faults, and the good storage capacity and permeability of the geothermal reservoir.

novel integrated learning-based method for igneous reservoir fluid identification is proposed to address the limitations of traditional approaches in handling complex lithological variations and heterogeneous reservoir spaces, which are crucial for global oil and gas resource development. In this paper, the Adaptive Multi-- Objective Swarm Crossover Optimization (AMSCO) innovatively combined with an engineered extreme gradient boosting (XGBoost) based on deep forest method for fluid identification in complex lithologic igneous reservoirs using conventional logging data sets. Methodologically, firstly, the AMSCO algorithm is used to optimize the imbalanced conventional logging data set, effectively solving the problem of class imbalance in the data set, providing a more balanced data basis for subsequent model training. Secondly, a cross-adaptive XGBoost and deep forest (CXDF) is constructed by fully utilizing XGBoost􀆳s advantages in processing large-scale and high-dimensional data, as well as the excellent performance of deep forest in feature extraction and classification tasks. Thus, the
accurate identification of reservoir fluids in complex lithologic igneous rocks is achieved. Then, to verify the effectiveness of this method, the model was applied to the simulated well together with support vector machine (SVM), XGBoost and XGBoost-based deep forest for comparison. Finally, the model is applied to the actual stratum. The results show that the evaluation index of the proposed method in the simulated well is superior to other methods, especially in the identification of non-water-producing reservoir fluids. In the application to actual formations, this method maintains high identification performance in different reservoirs with different fluid structures, and shows good generalization ability and stability.

Multiple wells in the carbonate gas reservoirs of the Lower Permian Qixia Formation in the central Sichuan region of the Sichuan Basin have encountered industrial gas flows, but the anisotropy of the single well reservoir is strong and the physical parameters are obviously different, so it is difficult to guide the well location deployment of this type of gas reservoir by conventional geophysical means. Therefore, clarifying the main controlling factors of the Qixia Formation gas reservoir, accurately predicting the distribution of thin dolomite reservoirs, and further improving the development efficiency of the gas reservoir are the key and difficult points of geophysical work. This article combines core, logging, seismic and other data to first determine the main controlling factors for the enrichment and high yield of the Qixia Formation gas reservoir in the study area, and summarizes the rock physical characteristics and seismic response patterns of the dolomite reservoir. Innovation has formed key technical processes for thin reservoir prediction, such as high-resolution processing technology, qualitative reservoir prediction based on wavelet reconstruction, fine characterization of ancient landforms constrained by layers, inversion technology constrained by reconstruction curves, and fracture and cave prediction technology constrained by porosity curves. This has improved the ability to vertically resolve thin reservoirs of 1--8 m in the research area, and ultimately characterized a favorable area for the development of dolomite reservoirs as 75 km2. The research results indicate that: ① There is an obvious positive correlation between the thickness of fracture-vuggy dolomite thin reservoir, fracture-cave development degree and productivity in the Qixia Formation gas reservoir in the study area. ② Before the sedimentation of the Qixia Formation (paleogeomorphology), reservoirs were developed in the lower part, and the reservoir properties in the slope area during the karst period were better. ③ The reservoirs of Qixia Formation in the study area are mainly developed in the middle and upper parts, and the seismic profile shows the reflection characteristics of “weak peaks” in the middle and upper parts, and the more the reservoirs are developed, the more obvious the response characteristics of “weak peaks” are. This article combines seismic facies, paleogeomorphology, reservoir thickness, and fracture distribution parameters for the first time to establish a series of key technologies for predicting dolomite reservoirs suitable for the study area, achieving effective prediction of the spatial distribution of thin reservoirs in the Qixia Formation and laying the foundation for increasing reserves and production in oil fields.

Abstract: Geochemical characteristics and potential evaluation of source rocks is a core component of thepetroleum system and plays a crucial guiding role in the exploration and development of hydrocarbon resources. To systematically assess the hydrocarbon generation potential of source rocks in the Shahezi Formation within the LishuFault Depression of the Songliao Basin and to clarify their geochemical characteristics and distribution patterns, the authors utilized core observations and well log interpretations to identify the macro sedimentary facies and vertical distribution of organic-rich mudstone intervals. A series of geochemical analyses (total organic carbon content meas-urement, rock pyrolysis, kerogen carbon isotope analysis, reflectance in oil testing, and gas chromatography-mass spectrometry of saturated hydrocarbon biomarkers) were employed to quantitatively characterize key parameters of the source rocks. The research results indicate that the lower part of the Sha1 Member and the middle part of the Sha2 Member of the Shahezi Formation are developed with mudstone deposits belonging to coastal-shallow lacustrine and semi-deep to deep lacustrine subfacies, which constitute the main source rock intervals of the Lishu Fault Depression. The depositional environment is characterized by weak redox conditions, and the migration of the depositional center controls the distribution of high-quality source rocks. In terms of organic geochemical character-istics, the source rocks of the Shahezi Formation exhibit high organic matter abundance, with TOC values ranging mainly from 1. 0% to 6. 8% and an average of 2. 5% . The vast majority of samples exceed the quality source rock threshold (w(TOC)≥ 2. 0% ). The kerogen is predominantly Type Ⅲ, followed by Type Ⅱ2, with a few samples exhibiting characteristics of TypeⅡ1 and TypeⅠ. Thermal maturity assessment indicates that the source rocks have reached the mature to over-mature gas generation stage (Ro > 1. 3% , average Tmax = 455. 1 ℃). Comprehensive evaluation indicates that the Shahezi Formation in the Lishu Fault Depression of the Songliao Basin is characterized by the rich main source rock development, high organic matter abundance dominated by gas-prone types, high thermal maturity having entered the gas generation stage, and distinct spatial differentiation, demonstrating excellent hydrocarbon generation potential. 

Abstract: Outang landslide is a typical giant creeping landslide in Three Gorges Reservoir area, with a volume of up to 90 million m3. To date, its maximum surface horizontal displacement has reached nearly 1 600 mm, and the maximum deep displacement exceeds 500 mm. The surface deformation exhibits extensiveness, persistence, and accelerating rate, whereas the deep deformation is marked by bedding movement, multi-layered activity, and sustained creep. The mechanism of deep reactivation-deformation and long-term stability of landslide are hot topics in the field of geological hazard research. Regarding Outang landslide, key issues such as it exhibits deep-seated integral sliding, the evolutionary mechanisms of reactivation-deformation, major potential risk sources, and effective
mitigation strategies were investigated. Based on 14 years of high-resolution monitoring data and multi-source exploration
results, the authors conducted comprehensive research integrating geological conditions, landslide characteristics, and reactivation-deformation models. This included long-term continuous geological tracking, retrospective analysis of deformation stages, and correlation studies between deformation rates and rainfall. The findings confirm the existence of deep-seated integral sliding in Outang landslide. Its integral deformation pattern follows a chain-reaction mechanism characterized by “Tier--3 sliding body pushing Tier--2 sliding body, which in turn drives Tier--1 sliding body. ” The intrinsic triggers of reactivation-deformation mechanism originate from underlying multi-layered weak intercalated structure prone to sliding. Factors such as high-frequency heavy rainfall, intense reservoir water inundation effects, and insufficient stability of Tier--3 sliding body have collectively exerted significant influence on integral reactivation-deformation of landslide. Among these, heavy rainfall has emerged as the dominant driving force of persistent deformation. Under heavy rainfall conditions, the progressive deformation characteristics are as follows: Tier--1 sliding body deformation rate (0. 3 mm/ d) < Tier--2 sliding body deformation rate (1. 0--3. 2 mm/ d) < Tier--3 sliding body deformation rate (1. 6--6. 0 mm/ d). Based on these observations, the criterion of identifying severe deterioration trends in deep slip zone soil of creeping landslides is established as: a sustained surface displacement rate exceeding 1. 0 mm/ d for more than 90 consecutive days, with significantly higher rates during flood seasons or rainy periods. The key to risk control lies in enhanced monitoring and numerical early warning for Tier--3 sliding body and the toe uplift zone.

Abstract: In order to study the effects of dry density and matric suction on the shear strength characteristics of unsaturated volcanic soils, this paper takes volcanic soils in Antu County, Jilin Province as the research object, and prepares soil samples with three dry densities and six water content levels. The soil-water characteristic curvesof volcanic soils with different densities were determined by the pressure plate instrument method, and the soil-water characteristic curves were fitted using the Van Genuchten model. Direct shear test was used to obtain theshear strength indexes of the samples under different water content conditions. The results indicated that the SWCC of volcanic soils were influenced by dry density, and the matric suction of volcanic soil increased with the rising of dry density under the same volumetric water content condition. The dry density of the soil has a good relationship with the VG model parameters (the reciprocal of intake value a, parameters n and m, and residual water content θr). Both the internal friction angle φ and cohesion c increased with dry density, while they increased initially and decreased with matric suction. The shear strength of volcanic soils is affected by the combination of dry density and matric suction. An rising in dry density not only enhances the occlusion between particles, but also reduces the internal pores of the soil and increases matric suction, which leads to an rising in the shear strength of volcanic soils. This study established the relationships between the internal friction angle φ, cohesion c, dry density and matric suction in volcanic soils, providing a reference for selecting strength parameters in geotechnical engineering projects in volcanic soil regions.

To investigate the conductivity mechanism and fluid distribution characteristics of tight sandstone reservoirs in the Ordos Basin and to construct an accurate saturation model, this study conducted a series of petrophysical experiments, including porosity-permeability tests, resistivity measurements using the centrifuge method,and NMR (nuclear magnetic resonance) response analysis, on 11 rock samples from the study area. Nitrogen gas was used in the porosity-permeability experiments, revealing porosity values ranging from 5. 21% to 11. 26%, and permeability values ranging from (0. 132 8--1. 261 1) ×10 -3 μm2. The resistivity tests under varying centrifugation durations provided key electrical parameters: the cementation exponent m (ranging from 2. 053 to 3. 293) and the saturation exponent n (ranging from 0. 714 to 2. 131). The NMR technique, which leverages the relaxation time of hydrogen nuclei within the rock, was used to characterize reservoir pore structures and fluid mobility. The T2 spectrum obtained through NMR enabled the distinction of pore sizes and their proportions. Results indicated that the tight sandstones in the study area exhibit similar properties and pore structures. The T2 spectrum showed a wide distribution (0. 1--1 000. 0 ms), typically presenting a bimodal pattern with a boundary at 10 ms. Based on empirical methods, mercury intrusion data, and NMR T2 spectrum characteristics, the T2 range for small pores was defined as 0. 01--5 ms, and for large pores as 5--10 000 ms. The findings show that fluid mobility is significantly
higher in large pores than in small ones, and water content in small pores increases during centrifugation. After converting T2 values into corresponding pore diameters, it was found that movable water is mainly distributed in the 1--10 μm range, followed by 0. 01--0. 1 μm and 0. 1--1 μm ranges. Using the fractal dimension method, a positive correlation was established between fluid distribution and electrical conductivity: greater fluid presence in large pores corresponds to smaller fractal dimensions and higher conductivity.

Abstract: Tiejiashan granite in the Anshan area, Liaoning Province, represents the oldest known K-rich granite
in the North China Craton. This study integrates petrography, zircon U--Pb geochronology, whole-rock geochemistry,
Lu --Hf isotopic and deformation structural analysis to elucidate the petrogenesis, tectonic setting, and dynamic mech-
anism of Mesoarchean K-rich granite. Zircon U--Pb geochronological data indicate that the Tiejiashan K-rich granite
formed during the Mesoarchean (3. 0--2. 9 Ga). The samples exhibit high w(SiO2)(68. 03%--71. 24% ), w(Fe2O3
T)(3. 31% --4. 90% ), w ( K2O + Na2O) (7. 73% --7. 99% ), and high Ga/ Al ratio, coupled with low w ( MgO)
(0. 53%--0. 80% ). Rare earth elements exhibit distinct negative Eu anomalies (δEu = 0. 21--0. 31). The rocks are
obviously enriched in large ion lithophile elements (LILE, e. g. Rb, Th, U, and K), while depleted in high field
strength elements (HFSE, e. g. Nb, Ta, P, and Ti). Additionally, they display high zircon saturation temperatures
ranging from 822 ℃ to 926 ℃. Based on these geochemical characteristics, the Tiejiashan granite is classified as an
A-type granite. No Mesoarchean mafic rocks have been identified in the Anshan area. Petrographic and geochemical
analyses reveal that the Tiejiashan granite demonstrates metaluminous to peraluminous affinities. The Tiejiashan
granite has low εHf (t) values ( - 8. 23 to - 14. 75), with TDM2 model ages between 4. 0 and 3. 7 Ga, suggesting
derivation from the partial melting of ancient crustal material. Variations in components such as TiO2, CaO, MgO,
P2O5, Sr, and Zr show linear correlations with w (SiO2) content on Harker diagrams, indicating experienced continuous
magmatic evolution and differentiation. Combining the geochemical data and tectonic deformation characteristics, it is
inferred that lithospheric mantle underplating induced regional extension in the Anshan area during the Mesoarchean.
Heat from the lithospheric mantle likely triggered partial melting of the lower crust and facilitated rapid diapiric
ascent of magma, lead to the formation of Tiejiashan granite.

Abstract: In recent years, moderate to strong earthquakes have frequently occurred in the Songliao Basin. To investigate the seismicity in the region, the authors calculate the minimum completeness magnitude (Mc) of the Songliao Basin based on the earthquake catalog from the Jilin Earthquake Agency (2009--2024) by using the maxi-mum curvature method and apply the Gutenberg--Richter law to fit the corresponding b-value. The results show that the Mc for the Songliao Basin is 1. 4, indicating either a strong monitoring capability of the local network or a higher density of small earthquakes. However, industrial noise in some areas may result in under-detection of low-magni-tude events. The fitted b-value for the Songliao Basin is 0. 59, lower than the b-value for the entire Northeast China
region (0. 74), suggesting that moderate to strong earthquakes will be likely to occur in the Songliao Basin region in the future. Further analysis of the relationship between the earthquake distribution and the geological structure of the underlying basin in the Songliao Basin is conducted, and statistical analysis is employed to explore the correspondence between earthquake distribution and hypocentral depth. The results show that the vertical distribution of the earthquakes is closely related to the key deep geological interfaces within the basin. Earthquakes mainly occurred from 5 km to 10 km. The International Continental Scientific Drilling Program (ICDP) borehole SK--2 in the Songliao Basin revealed that at the depth of 6 km, a regional unconformity interface of the seismic reflector named T5 separates the Cretaceous sedimentary cover from the underlying Triassic basement, and at the depth of
7 km, a basement detachment fault occurs within the dual-layer Triassic--Paleozoic basement system.

This study investigates members 2--3 of the Late Cretaceous Qingshankou Formation in the Gulong area of the Songliao Basin. By integrating core data, well-logging information, magnetic susceptibility and chromaticity measurements, wavelet transform analysis of logging curves, and cluster analysis, the coupling mechanism between high-frequency climatic oscillations and sequence development in continental lacustrine basins was systematically revealed. The principal findings are summarized as follows: ① Two depositional environments lacustrine and deltaic facies were identified in the studied interval, which were further subdivided into six microfacies: deep-lake mud, semi-deep lake mud, shell sand, turbidite, prodelta mud, and sheet sand. A delta-lacustrine sedimentary evolution model was established to characterize their vertical stacking patterns. ②Two complete third-order sequences were recognized based on lithologic associations, well-logging responses, and wavelet transform time-frequency analysis. These sequences were divided into four system tracts, lowstand (LST), transgressive (TST), highstand (HST), and regressive (RST). Sequence boundaries were characterized by GR curve inflection points, lithologic abrupt changes, and energy cluster transitions in time-frequency spectrograms. ③Paleoclimate
evolution was quantitatively reconstructed using magnetic susceptibility and chromaticity indices through SPSS cluster analysis. The results revealed a climatic cyclicity of “cooling-warming-cooling-warming-cooling” with an anomalously
warm-humid phase during 86. 895--86. 364 Ma closely linked to lacustrine anoxic events (LAEs). This phase was attributed to enhanced hydrological cycling-triggered terrestrial coarse clastic input and intensified reducing conditions. Overall, paleoclimate governed sequence differentiation by regulating hydrological cycling and sediment supply. During warm-humid phases, strengthened water circulation promoted fine-grained mudstone deposition, forming TST and HST system tracts, whereas under relatively dry-cold conditions, weakened water circulation favored coarse-grained gravity flow sedimentation, developing RST and LST system tracts. 

Abstract: The Huoshiling Formation in the Longfengshan gas field, southern Songliao Basin, has achieved
cumulative proved gas reserves exceeding 10 billion cubic meters, demonstrating favorable reservoir properties and
exploration potential in deep volcaniclastic rock sequences. Clarifying the characteristics and controlling factors of
high-quality volcaniclastic reservoirs aids in expanding favorable zones for deep-buried gas reservoir exploration and
facilitating the deployment of development wells. Based on the drill core description, blue-dyed epoxy-resin impreg-
nated thin section observation, scanning electron microscopy analysis, the porosity-permeability measurements and
high-pressure mercury injection tests, a systematic study was conducted to characterize the lithology, lithofacies,
and reservoir development patterns of volcaniclastic rocks, in order to clarify the controlling factors and spatial
distribution of high-quality reservoirs. The results reveal that: the Huoshiling Formation in the Longfengshan gas
field is composed of thirteen lithological types in four categories of volcaniclastic rocks, formed in both subaerial
and subaqueous environments, and dominated by volcaniclastic deposits of explosive facies. Four distinct pore types
were identified, the intergranular pores, intragranular pores, intracrystalline pores, and microfractures.
The intergranular and intragranular porosity constitute the predominant reservoir space, especially for the volcanic breccia
which exhibits the highest porosity, 37% and 53% of intergranular and intragranular pores, respectively. The
porosity of volcaniclastic reservoirs ranges from 1. 6% to 13. 8% , and the permeability ranges from 0. 002 ×10 - 3 μm2
to 0. 751 × 10 - 3 μm2. They are classified as medium-porosity / low-permeability reservoirs, developing micron-scale
and nano-scale pore throats, the maximum throat ranges from 0. 020 μm to 2. 830 μm. Eogenetic chlorite cementa-
tion and albitization coating of the particles effectively inhibited compactional porosity loss, thereby preserving inter-
granular pores; Devitrification of volcanic glass/ ash matrix in shallow burial stage, and the organic acid dissolution
in deep burial diagenesis significantly enhances secondary porosity development, composing a key factor that controls
the development of high-quality volcaniclastic reservoirs. Three subfacies exhibit the highest proportions of high-quality
reservoirs (porosity > 6%), including diatreme subfacies of volcanic conduit facies, subaqueous volcaniclastic flow
deposits and fallout subfacies of explosive facies, which can be regarded as priority targets for deep gas exploration.

To deeply reveal the paleoclimate characteristics of the Sifangtai Formation in the Songliao Basin, infer the controlling factors of climate change, and improve the information implied by climatic evolution, magnetic susceptibility and chromaticity tests were conducted on the core (140. 0--380. 5 m) from the ZKQA1--1 in the Qian an area. Through Pearson correlation analysis, it was found that magnetic susceptibility and chromaticity data were weakly correlated with stratigraphic depth. Brightness (L∗) was negatively correlated with redness (a∗) and yellowness (b∗), while redness (a∗ ) was positively correlated with yellowness (b∗ ), indicating that the data were minimally affected byagenesis and had reliable paleoclimate significance. Through SPSS cluster analysis, high-frequency magnetic susceptibility, low-frequency magnetic susceptibility, frequency-dependent magnetic susceptibility, brightness (L∗), redness (a∗), and yellowness (b∗) were filtered once, and four filtered mean values were obtained for each dataset. The filtered values were classified into four climatic types (cold-dry, relatively cold-dry, relatively warm-wet, and warm-wet) based on paleoclimate proxy indicators. To avoid errors caused by
semi-quantitative analysis of climatic types using magnetic susceptibility and chromaticity data, the six datasets representing climatic types were assigned numerical values (cold-dry: 0. 5, relatively cold-dry: 1. 5, relatively warm-wet: 2.5, warm-wet: 3.5), averages were calculated based on depth summation, and comprehensive temporalscale analysis was performed in combination with lithological and well-log data. The results showed that the magnetic susceptibility and chromaticity data of the Sifangtai Formation core from the Well ZKQA1--1 indicated a relatively cold-dry climate during the mid-Campanian (76. 08--75. 65 Ma), and the paleotemperature decrease in this stage was associated with the migration of the Intertropical Convergence Zone (ITCZ). A relatively warm-wet to warmwet climate was identified during the mid-late Campanian (75. 65--74. 32 Ma), with an extreme paleotemperature peak observed at 75.55 Ma, while the slight paleotemperature decline from 75.55 to 74.32 Ma was linked to the Campanian--Maastrichtian Boundary Event (CMBE). During the late Campanian (74. 32--73. 19 Ma), a cold-dry to relatively cold-dry climate was recorded, with an extreme paleotemperature minimum at 74. 13 Ma, followed by a relatively warm-wet climate from 73. 19 to 72. 86 Ma, and the paleotemperature rise from 74. 13 to 72. 86 Ma might be related to intermittent volcanic activity prior to the late Campanian Deccan Traps eruption event. Overall, the
paleoclimate of the Sifangtai Formation was demonstrated to correspond well with global climate trends, representing a relatively warm and dry climatic type.

When using machine learning techniques to identify lithology based on the feature vectors of each sample in the logging dataset, it is generally imperative to first clean and preprocess the dataset to detect missing data and outliers in it. Affected by the distortion of logging curves or the disparities in the quantity of curves between wells, the logging dataset used for lithology identification is usually incomplete, with a large amount of missing data and outliers. This renders it challenging for the majority of machine learning methods to be directly applied. To tackle this problem, a K--Nearest Neighbors algorithm based on the Partial Distance Strategy (PDSKNN) is proposed. This algorithm improves the traditional K--Nearest Neighbors algorithm (KNN) based on a method (PDS) that capable of computing the distance between feature vectors containing missing values, enabling the direct application to incomplete dataset. The PDSKNN algorithm was experimentally implemented and tested in the lithology identification task of complex igneous rock reservoirs in a certain region. When the data missing rate was 2. 7%, the lithology identification accuracy of the PDSKNN algorithm was as high as 91. 90%. This result indicates that the PDSKNN algorithm effectively solves the problem of incomplete dataset at the algorithm level. For the purpose of further validating the effectiveness of the PDSKNN algorithm, certain data in the dataset was randomly deleted to elevate the missing rate, and the changes in the identification accuracy were observed. The experimental results reveal that as the missing rate increases, the identification accuracy gradually decreases. However, even when the missing rate attains 20%, the identification accuracy of the PDSKNN algorithm still remains above 80%. This result proves that the PDSKNN algorithm can maintain a relatively high identification accuracy in the case of severe data missing. Finally, by simulating the missing conditions of curves in different wells, a comparative analysis of the changes in the identification effect of the PDSKNN algorithm was carried out. As the number of missing curves increases, the lithology identification performance in the wells decreases to some extent, but the identification effect of most well sections is still guaranteed. It shows that the PDSKNN algorithm has strong robustness to the missing curves in the wells.

The exploration focus of the Junggar Basin has gradually shifted from the hydrocarbon generation center of the sag to the peripheral uplift area, and it is particularly important to understand the distribution characteristics of oil and gas reservoirs in the Dabasong Uplift, the authors have conducted research on the main controlling factors and development mode of high-quality reservoirs in the Dabasong Uplift by means of core observation, thin section identification, well logging interpretation and various experimental analyses. It is believed that the Dabasong Uplift was in the underwater reducing environment during the eruption period, and the carboniferous igneous rocks were mainly basic rocks, including diabase of volcanic channel facies, crystalline tuff of the underwater sedimentary subfacies, breccia clastic tuffs of the underwater volcaniclastic flow subfacies, basalt and andesite of underwater overflow facies, and sedimentary tuffs of the volcanic sedimentary facies. Among them, breccia clastic tuffs of the underwater volcaniclastic flow subfacies and the diabase of volcanic channel facies are the dominant lithologic lithofacies in the study area. The breccia clastic tuffs has more concentrated pore throat distribution, good connectivity and greater contribution of pore throat, while the diabase has more dispersed pore throat distribution and good connectivity. The reservoir space of igneous rocks in the study area is dominated by intrachrystalline solution pores, matrix solution pores, structural fractures and solution fractures. The porosity of carboniferous igneous rocks is less than 5% (51. 9%), between 5% and 12% (29. 6%), more than 12% (18. 5%), the permeability is less than 1 ×10 -3μm2 (81. 5%), between (1 and 5) ×10 -3μm2 (16. 6%), and more than 5 ×10 -3μm2 (1. 9%), the whole belongs to the medium and low porosity-low permeability reservoir. The formation of effective reservoirs is controlled by lithologic lithofacies, dissolution and fractures, and is a weathering and dissolution modified reservoir based on lithologic lithofacies and dominated by fractures. It is a favorable zone for oil and gas exploration in areas with dominant lithological lithofacies, weathering and dissolution, active fluid dissolution and fracture development.

Abstract: Aming to the unique temperate forest vegetation within the Northeast Tiger and Leopard National
Park, a study on multi-source remote sensing image coverage classification has been conducted, which aims to
utilize modern remote sensing techniques and advanced machine learning methods to perform fine classification of
complex vegetation types, thereby better understanding the ecological conditions of the region and providing scientific
basis for the conservation efforts in the Northeast Tiger and Leopard National Park. Firstly, the authors employed
multi-scale image segmentation technology to decompose the original remote sensing images into a series of small
regions with similar texture and color features, which helps reduce data redundancy and highlight different characteristics
of vegetation. Subsequently, these segmented images has been optimized by extracting feature parameters that effec-
tively distinguish different types of vegetation. Then, the authors input these optimized feature parameters into three
different machine learning classifiers for training and testing, including support vector machine (SVM), C5. 0 deci-
sion tree and random forest. The experimental results show that, in terms of classification accuracy, random forest
achieving the highest accuracy of 91. 33% , and C5. 0 decision tree reaching 89. 59% , with the Kappa coefficients
of 0. 90 and 0. 88, respectively. In comparison, the classification accuracy of support vector machine was lower, at
only 58. 96% , and the Kappa coefficient is 0. 54. These results indicate that, for distinguishing vegetation cover
types and identifying other land features, the random forest algorithm has significant advantages. It not only accu-
rately distinguishes different vegetation types in complex backgrounds, but also effectively identifies other features
such as roads and buildings.