Bridging the Dimensional and Temporal Gaps in Large-Scale Gully Erosion Modeling: Challenges and Solutions

Significant progress has been made in gully erosion research since the early 2000s. However, quantifying the 3-dimensional evolution of gullies at regional and global scales remains a major challenge in soil erosion science.
The first challenge is the multidimensional gap. While global assessments have successfully mapped gully head distributions (0D) and the characterization of  linear gully densities (1D) has been achieved in specific countries and regions, accurate gully sediment budgeting requires a shift towards three-dimensional (3D) volumetric quantification of gully structure . The primary technical requirement is the large-scale inversion of gully depth, which can be addressed through the integration of modern geodetic techniques—including high-precision field surveys, Unmanned Aerial Vehicle (UAV) photogrammetry, and satellite stereoscopic mapping—with AI-driven predictive algorithms. Parallel to this is the temporal gap. Most regional datasets remain static. Resolving these spatial and temporal challenges is the prerequisite for transitioning from purely morphological descriptions to robust, process-based gully erosion models at large scales. The convergence of multi-source remote sensing now offers opportunities to reconstruct gully development (in 3-D) over recent decades. Coupled with the advancement of AI, which facilitates a transition from labor-intensive manual digitizing to automated, high-throughput workflows, it is now feasible to achieve rapid, dynamic, and intelligent extraction of gully information.
This report systematically examines these methodological challenges and evaluates potential solutions through the integration of multi-modal remote sensing, field measurements, and advanced analytical frameworks. To demonstrate these solutions, we present case studies from the Chinese Loess Plateau and the Northeast Black Soil Region based on small-watershed units, including 1,300 remote-sensing survey units and 65 high-precision units integrating field measurements with UAV surveys. These data provide the foundation for validating AI-driven depth inversion and automated extraction methodologies at a regional scale. Finally, we propose an international cooperation initiative to harmonize data collection and standardize validation protocols. Such a collaborative framework is essential to effectively integrate gully erosion into the next generation of global soil loss and Earth system models.