A BIBLIOMETRIC ANALYSIS OF MACHINE LEARNING TECHNIQUES FOR IMPROVING DIGITAL ELEVATION MODEL ACCURACY: TRENDS, GAPS, AND FUTURE DIRECTIONS

This study presents a bibliometric and thematic analysis of research focused on improving the accuracy of digital elevation models (DEMs) using machine learning (ML) techniques between 2005 and 2025. Drawing from Scopus and Web of Science databases, complemented by manual reference chaining, approximately 250 publications were analyzed. Results show a notable increase in scholarly activity after 2018, linked to the release of enhanced DEM products such as CoastalDEM and FABDEM. Keyword co-occurrence and thematic coding revealed four conceptual pillars: models, methods, applications, and data sources. Ensemble algorithms like Random Forest and LightGBM dominate the methodological landscape, while deep learning methods such as Convolutional Neural Network (CNNs) and Generative Adversarial Network (GANs) are emerging. Despite advancements, methodological homogeneity, reliance on Root Mean Square Error (RMSE), and underutilization of data fusion and semi-supervised learning strategies remain significant limitations. Silent themes and regional gaps emphasize the need for methodological diversification and broader global integration. Future research should prioritize algorithmic diversity, standardized multi-metric validation frameworks, open science practices, and regional model applications. This study offers a structural mapping of DEM–ML research and proposes strategic directions for advancing the field through interdisciplinary collaboration and innovation.