Global and local processes influencing altimetric error patterns in digital elevation models (DEM): An approach on vertical accuracy assessment and spatial aspects of DEM error

The field of geospatial data quality assessment is critical for ensuring the reliability and utility of Digital Elevation Models (DEM). DEM provide detailed elevation information, impacting various Earth sciences applications, including hydrology, geomorphology, environmental monitoring, land-use planning, and disaster management. However, uncertainties in DEM can propagate to derived products, which may lead to inaccurate predictions and decisions. This research addresses a significant knowledge gap in the field, particularly in understanding how terrain characteristics influence DEM vertical accuracy and how this impact varies across different spatial scales. The main objectives of this research are to investigate the vertical uncertainty of four open-source DEM, classify them according to cartographic standards, explore the correlation between DEM vertical error and terrain characteristics, provide a better understanding of error factors, identify local factors affecting DEM vertical accuracy, and investigate how terrain characteristics relate to altimetric error at different spatial scales. To achieve these objectives, we employed advanced geospatial techniques, including Geographically Weighted Regression (GWR) and Multiscale Geographically Weighted Regression (MGWR) to analyse local relationships and spatial variability in DEM altimetric errors. Our research reveals that elevation and slope impact DEM vertical accuracy, with higher altitudes and steeper terrains corresponding to increased altimetric errors. Furthermore, Land Use and Land Cover (LULC) also influence altimetric errors, particularly in areas with artificial structures and forest vegetation. The major contributions of this work include a nuanced understanding of DEM vertical accuracy and the role of terrain characteristics, emphasizing the importance of addressing spatial non-stationarity in DEM vertical accuracy assessments. Our research highlights the significance of terrain characteristics on DEM vertical error at different spatial scales and offers valuable guidance for researchers and practitioners working with these data. By enhancing the understanding of these influences, this research advances the field of geospatial data quality assessment, leading to better-informed decisions in several applications relying on these products.