Advancements in Geological Mapping Integrating Remote Sensing, GIS, and Digital Technologies

Abstract

Geological mapping has undergone a structural transformation over the past two decades, moving from an essentially analogue, observation-limited craft to a data-intensive, computationally mediated science. A mixed-methods design was adopted, combining a structured review of the literature with a comparative meta-analytic assessment of classification accuracy, data-acquisition efficiency and positional reliability across three generations of mapping workflow: conventional analogue survey, remote sensing supported by geographic information systems, and fully integrated digital mapping incorporating unmanned aerial vehicles, three-dimensional virtual outcrop models and machine learning. Results indicate that ensemble and deep-learning classifiers substantially outperform conventional parametric methods, with convolutional neural networks achieving a mean overall accuracy of 88.9 per cent compared with 71.4 per cent for maximum likelihood classification. Fusion of optical imagery with digital elevation and geophysical layers improves overall accuracy by a further six to eight percentage points across all sensor families, and hyperspectral data yield the highest single-sensor accuracies. Integrated digital workflows reduce field time per unit area by approximately seventy-four per cent and positional error by an order of magnitude relative to analogue practice, while increasing the volume of structural measurements collected per field day by a factor of nearly five. The paper concludes that remote sensing, geographic information systems and digital field technologies are most effective when treated as a single integrated system rather than as separate tools, and that the principal remaining constraints are the availability of reliable training data, the interpretability of complex models and the uneven distribution of technical capacity between geological surveys.