BUILDING AND ANALYZING STRUCTURE-FROM-MOTION POINT CLOUD DATA FOR VISIBLE AND BURIED TRACKSITES AT DINOSAUR STATE PARK, ROCKY HILL, CT

New Structure-From-Motion (SfM) photogrammetry is increasingly used to develop and analyze geologic base maps and detailed topographic models, in some cases at resolutions greater than that available from LiDAR. Here we describe techniques and associated outcomes arising from our efforts to develop, analyze, and manipulate high resolution digital models of two theropod tracksites at Dinosaur State Park (DSP) in Rocky Hill CT. The well-known Interpretive Center (IC) tracksite, consisting of >600 predominantly Eubrontes tracks was re-imaged in 2017 and modelled using 602 overlapping full-frame images and 26 laser-surveyed ground control points (GCP). As well, we have modelled the original and much larger buried tracksite (BT) adjacent to the IC that has been covered by ≈2 m of sandy soil for more than 40 years . BT modelling is based on 54 historical (ca. 1970) photographs that include ≈50 painted crosses at ≈3 m spacing that we use as GCP’s. Point cloud models for both tracksites are large (IC≈239.5 M points; BT≈87.3 M points), high resolution (IC≈0.5 mm/pixel; BT ≈1.1 mm/pixel) and are reasonably precise (linear error from independent scale bars: IC≈1.6mm, BT≈1.8 cm). However, imperfect overlap for historical images causes holes in the BT dataset limiting analyses. That said, we demonstrate that SfM data may be used to rectify historical sketch-maps that identify individual trackmakers for both sites. As well, we use IC data to construct detailed height maps and inverted 3D prints for ≈30 tracks to evaluate the reliability of digital measurements as compared with conventional (caliper) measures of track morphometry. Also hypsometric analysis of selected tracks is used to quantify changing track morphology made by the same individual but preserved in different stratigraphic horizons. We find that the quality of models and derived products (like rectified maps) are strongly influenced by the manner and order in which GCM data are incorporated. Moreover, digital modeling at DSP illustrates the value of re-framed coordinate systems and inverting models as means to normalize topography, quantify form, and improve visualization.