To achieve the stated goal, the following objectives will be accomplished:
Automation of processing and analysis of large volumes of geological data is a key condition for increasing the efficiency of geological exploration. AI algorithms, such as deep learning and convolutional neural networks, make it possible to automate the extraction of meaningful features from large datasets and filter noise. This enables the creation of mineragenic models that allow reconstruction of geological processes and, as a result, identification of territories prospective for mineral deposits. This approach significantly reduces analysis time and improves the accuracy of identifying promising areas for further exploration activities.
This objective addresses the problem of accessibility and automation of data processing for a wide range of users, from government agencies to private companies. Integrating machine learning and AI methods into an expert system within the “Minerals.gov.kz” platform is an important step toward digitalizing geological exploration and creating a system that provides more accurate and timely geological information. The application of AI methods allows the identification of hidden patterns and anomalies in data that cannot be detected using traditional analysis methods. Pilot operation of the prototype will enable testing on real data, improving functionality, and adapting algorithms to user requirements.
ArcGIS provides powerful tools for analyzing remote sensing data, including satellite and aerospace imagery. Using the ArcGIS Image Analyst module, spectral analysis can be applied to identify mineralogical indicators. Remote sensing data, including hyperspectral and multispectral imagery, allow the detection of typomorphic minerals and ore-bearing associations based on their spectral characteristics. The use of ArcGIS for testing and mapping on real test sites enables integration of all collected data into a unified GIS system. This allows the construction of mineragenic models of the area, which is critical for accurately identifying geological structures and mineralogical indicators. Testing the methodology on real sites will help evaluate its effectiveness under various geological conditions and refine it for specific areas.
Regional structural elements, such as geological belts and zones, are important indicators of mineral resource potential, as they influence the spatial distribution of ore bodies. Local structures, including tectonic faults, fracture zones, and metasomatically altered rocks, also play a significant role in ore localization. Metasomatic alterations are often associated with mineralization, and accurate mapping of such zones can significantly improve the accuracy of prospectivity assessment. Clustering methods, such as k-means, can group data based on rock characteristics and identify lithologically heterogeneous zones that may host mineral resources. Identification of metasomatic alterations is a complex task that requires analysis of multiple factors. Deep learning algorithms can be trained to analyze remote sensing and geochemical data to detect such changes. Traditional structural analysis methods are labor-intensive and subject to human bias; machine learning algorithms enable automation of this process. The technological readiness level (TRL) of the developments is assessed according to the scale from 0 to 9. At the application submission stage, the TRL is 2, as the concept has been formulated and scientifically substantiated. Upon completion of the program, an expert distributed system integrated with the Unified Platform will be created, increasing the TRL to 3.