Reproducible Drone Photogrammetry DSM and DTM Workflow with Public LiDAR Terrain Reference for a Heterogeneous Site

Accurate elevation surfaces supported planning and drainage screening, and uncrewed aerial vehicle imagery provided rapid collection where airborne Light Detection and Ranging, LiDAR, products did not match project timing. Prior comparisons often omitted key acquisition and processing details and did not control vertical reference selection, which limited reproducibility and weakened cross sensor interpretation. We aimed to document a reproducible workflow that generated drone based digital surface model, DSM, and digital terrain model, DTM, products and compared them with a LiDAR derived reference surface. We collected imagery at Ag Heritage Park, Alabama, on February 26, 2025, over 27.8 acres, using a DJI Mavic 3 platform, and we flew one nadir and one oblique mission at 70 m height with 80 percent front and side overlap and 1.89 cm per pixel ground sampling distance. We processed 810 images in Pix4D to export merged GeoTIFF DSM and DTM rasters, and we used ArcGIS Pro to convert a public LiDAR point cloud to a terrain raster, align extents, generate hillshades, and compute raster difference layers. The comparison artifacts showed concentrated disagreement near the pond and along flight coverage margins, and they supported interpretation of texture driven photogrammetry artifacts distinct from broad area elevation shifts. We also observed systematic offsets attributable to vertical reference mismatch between ellipsoidal heights and mean sea level heights. This workflow supports transparent drone to LiDAR surface comparison for small sites in applied terrain studies and guides selection of DSM for above ground feature mapping and DTM for bare earth analysis.