Optimization of Statistical Modeling Parameters for Geophysical Fields in Permafrost Conditions

This study examines the geocryological environment of a site near Norilsk’s Nickel Plant slag dump. The rectangular area spans 600 by 1000 meters. The goal is to assess permafrost properties within the geological section. The section is analyzed using geophysical methods to depths of 15 meters, with data validation through boreholes averaging 15 meters, one reaching 20 meters. Sparse and heterogeneous data necessitate interpolation for continuous models. Interpolation algorithms, including a three-dimensional Bayesian approach, were used with parameter tuning for search radius, neighbors, and covariance function type. This approach accounts for soil property variability and improves spatial model accuracy. The study adapts methods for reliable geocryological modeling. Analysis uses ArcGIS Pro, employing the empirical Bayesian method with validation through borehole and geomorphological data. Key conclusions include a methodology integrating geophysical investigations and statistical processing for permafrost modeling. The three-dimensional approach better captures environmental variability and enhances accuracy, confirmed by borehole data. For instance, the seasonally thawed layer’s thickness identified through geophysics aligns with geomorphological and lithological features. A three-dimensional method, Bayesian Kriging 3D, was adapted for permafrost conditions. Parameters like covariance function type, partitioning scale, and neighbors were studied. This is the first evaluation of empirical Kriging’s effectiveness in this area. The results support infrastructure planning and resource management, demonstrating advanced geostatistical techniques’ applicability for Arctic permafrost modeling.

Geocryological environment, Permafrost, Spatial modeling, Bayesian Kriging, Geophysical investigations, Interpolation, Data validation, Covariance function, Interpolation algorithms, Norilsk