The aim of this study was to confirm if predictive regression algorithms can provide reliable results in missing geophysical logging data in the western and eastern parts of the Drava Super Basin, especially Gola Field, and to apply unsupervised machine learning methods for a better understanding of lithological subsurface relations. Numerous regression models have been used for the estimation of prediction accuracy, along with some clustering algorithms to support the estimation of lithology distribution estimations in well log datasets, consisting of 20 wells in total. Tree-based algorithms and the boosting algorithm have been optimized and proven valuable in predicting well log data when they are not measured or are unavailable at all depth intervals. For blind datasets, predictions become much less reliable. For this purpose, neural networks with at least one Long Short-Term Memory (LSTM) layer have significantly improved the accuracy and reliability of predictions, not in terms of absolute values but in the aspect of the trends in values that change with the depth and other well features, as well as in terms of the magnitudes. Trendlines can further be used for pattern recognition or as a newly engineered feature. Unsupervised learning has confirmed reliability in lithology recognition on validation sets and has proven to be a great asset in distinguishing variabilities in the petrophysical properties of sediments.

In subsurface geological mapping, it is more than advisable to compare different solutions obtained with neural and other algorithms. Here, for such comparison, we used the previously published and well-prepared dataset of subsurface data collected from the Bjelovar Subdepression, a 2900 km² large regional macrounit in the Croatian part of the Pannonian Basin System. Data on depth were obtained for the youngest (the shallowest) Lonja Formation (Pliocene, Quaternary) and mapped using neural network (NN), inverse distance weighting (IDW), and ordinary kriging (OK) algorithms. The obtained maps were compared based on square error (using k-fold cross-validation) and the visual interpretation of isopaches. Two other algorithms were also tested, namely, random forest (RF) and extreme gradient boosting (XGB) algorithms, but they were rejected as inappropriate for this purpose solely based on the visuals of the obtained maps, which did not follow any interpretable geological structures. The results showed that NN is a highly adjustable method for interpolation, with adjustment for numerous hyperparameters. IDW showed its strength as one of the classical interpolators, and its results are always located close to the top if several methods are compared. OK is the relative winner, showing the flexibility of variogram analysis regarding the number of data points and possible clustering. The presented variogram model, even with a relatively high sill and occasional nugget effect, can be well fitted into OK, giving better results than other methods when applied to the presented area and datasets. This was not surprising because kriging is a well-established method used exclusively for interpolation. In contrast, NN and machine learning algorithms are used in many fields, and these algorithms, particularly the fitting of hyperparameters in NN, simply cannot be the best solution for all.

Electrical resistivity tomography (ERT) in an aquatic environment is usually measured using two basic systems. One system uses floating electrodes, which is usually applied in shallow water, and the other uses underwater cables with integrated electrodes, which is often carried out as a mobile system. The aim of this research was to test an underwater ERT measurement with floating cables on the water surface, but with electrodes submerged at the water bottom. This type of cable spread makes it possible to lay all the electrodes vertically to the bottom without the distance between them being reduced by the uneven topography and various underwater obstacles. Prior to the field survey that was conducted in a 40 m – deep lake, the response of common electrode arrays was tested using synthetic models. Two models were used that correspond to the geological condition in the field, higher resistivity bodies in a lower resistivity environment and a model with inverse relationship of resistivity to the first one. The Wenner Schlumberger and dipole-dipole arrays resolved the resistivity range, size and shape of the bodies very well and were therefore used in the field. The field data quality was very good and it was shown that ERT measurements in freshwater depths of more than 40 metres can provide very good results. As expected from the modelling, the dipole-dipole array led to a high-resolution resistivity model that enabled the characterisation of the lakebed sediments.

Understanding elastic properties of reservoir rocks is essential for seismic modeling under different saturation conditions as well as lithology discrimination. Experiments on elastic properties of limestones are significantly less published compared to siliciclastic sedimentary rocks. The current study presents the results of laboratory measurements on Pannonian Basin limestone cores. The research was carried out for the first time for a hydrocarbon reservoir in the Bjelovar Depression, located in the southern part of the Pannonian Basin. Ultrasonic velocity measurements and determination of dynamic elastic properties were performed on limestone plugs, in dry and saturated condition under different confining pressure steps. Based on the results obtained in laboratory conditions, an empirical relationship between shear wave velocity (Vs) and compressional wave velocity (Vp) has been defined. The saturated samples show an effect of shear modulus weakening, while three samples have a shear modulus strengthening effect. Two models were used in the interpretation of the measured data, the Kuster and Toksöz and the Xu-Payne model. The results show that the Xu-Payne model describes the data well and the dominant pore type system in the limestone samples can been identified. The interpretation revealed an interparticle pore system with a fraction of microcracks from 20% to 35%. The results have helped to understand the elastic properties of limestones from the southern part of the Pannonian Basin, which are necessary for any process of reservoir characterization, such as porosity distribution and permeability variation.

The Žune barite-fluorite ore body in northwestern Bosnia and Herzegovina has been explored in order to determine spatial distribution of the ore body. At the Žune site, barite mineralization occurs in the form of veins and is hosted by dolostone. A geophysical survey was carried out with the aim to detect anomalies associated with the barite-fluorite deposit and to characterize the abandoned old deposit. Therefore, 2D electrical resistivity tomography (ERT) was measured along four profiles. Prior to surface geophysical measurements, detailed geological field mapping of the ore body and host rocks was undertaken. The geometry of the abandoned mining cut with steep walls and complex structural setting with barite that incorporates coarse fragments of host rock and fluorite makes interpretation of the results challenging. The collected geological data were used to calibrate the geophysical models. Forward modelling in addition to inversion of the ERT data helped to reduce the ambiguity of geophysical data interpretation. The results revealed that the barite vein in the area of mining cut is limited to a depth of about 10 m and length is about 40 m. There is no indication for barite mineralization in the area south of the mining cut, while anomalies that may be associated with ore bodies are present in the northern and northeastern part.