Quantifying the influence of DEM elevation errors on regional gravity forward modeling through DEM correction and fusion

Le résumé fourni par la source

High-precision characterization of the Earth’s gravity field is essential for geological structure analysis, volcanic activity monitoring, and resource exploration. As direct measurement of a high-resolution global gravity field remains impractical, forward modeling based on Digital Elevation Models (DEMs) is a key method for recovering its high-frequency components and is essential for regional high-resolution gravity field modeling. However, the accuracy of this method is severely limited by substantial errors in existing global DEMs, particularly those induced by vegetation cover. Currently, a systematic research quantifying the full pathway from DEM error correction to its impact on gravity field accuracy is lacking. To address this gap, this study proposes an integrated framework for correcting DEM errors and evaluating their propagation in gravity field modeling. Specifically, we develop a hybrid Random Forest-Inverse Distance Weighting (RF-IDW) method to fuse multi-source remote sensing data and generate a regionally optimized, high-precision DEM. The proposed framework is evaluated in the Wudalianchi volcanic region using in-situ gravity observations. Comparative forward modeling based on the original TanDEM-X DEM and the optimized DEM demonstrates a 42.35% improvement in elevation accuracy, resulting in a 17.89% increase in the accuracy of the modeled terrain gravity field. Furthermore, DEM elevation errors exhibit a moderately strong correlation with gravity anomaly error (R = 0.648), accounting for approximately 42.0% of their variance. These findings provide both a quantitative assessment of elevation errors in gravity field modeling and a transferable methodological framework for enhancing the accuracy of DEM-based gravity field solutions in diverse topographic and vegetation environments.

Ce résumé expose les affirmations des auteurs. BNTIC ne l’interprète pas comme une validation indépendante des résultats.