Using Convolutional Neural Networks for Detection and Morphometric Analysis of Carolina Bays from Publicly Available Digital Elevation Modelss

Carolina Bays are oriented and sandy-rimmed depressions that are ubiquitous throughout the Atlantic Coastal Plain (ACP). Their origin has been a highly debated topic since the 1800s and remains unsolved. Past population estimates of Carolina Bays have varied vastly, ranging between as few as 10,000 to as many as 500,000. With such a large uncertainty around the actual population size, mapping these enigmatic features is a problem that requires an automated detection scheme. Using publicly available LiDAR-derived digital elevation models (DEMs) of the ACP as training images, various types of convolutional neural networks (CNNs) were trained to detect Carolina bays. The detection results were assessed for accuracy and scalability, as well as analyzed for various morphologic, land-use and land cover, and hydrologic characteristics. Overall, the detector found over 23,000 Carolina Bays from southern New Jersey to northern Florida, with highest densities along interfluves. Carolina Bays in Delmarva were found to be smaller and shallower than Bays in the southeastern ACP. At least a third of Carolina Bays have been converted to agricultural lands and almost half of all Carolina Bays are forested. Few Carolina Bays are classified as open water basins, yet almost all of the detected Bays were within 2 km of a water body. In addition, field investigations based upon detection results were performed to describe the sedimentology of Carolina Bays. Sedimentological investigations showed that Bays typically have 1.5 m to 2.5 m thick sand rims that show a gradient in texture, with coarser sand at the bottom and finer sand and silt towards the top. Their basins were found to be 0.5 m to 2 m thick and showed a mix of clayey, silty, and sandy deposits. Last, the results compiled during this study were compared to similar depressional features (i.e., playa-lunette systems) to pinpoint any similarities in origin processes. Altogether, this study shows that CNNs are valuable tools for automated geomorphic feature detection and can lead to new insights when coupled with various forms of remotely sensed and field-based datasets.

[1]  S. Stanley,et al.  Evidence for an extraterrestrial impact 12,900 years ago that contributed to the megafaunal extinctions and the Younger Dryas cooling , 2007, Proceedings of the National Academy of Sciences.

[2]  A. Goudie,et al.  Pan morphology, Distribution and formation in Kazakhstan and Neighbouring areas of the Russian federation , 2016 .

[3]  L. C. Glenn.  SOME NOTES ON DARLINGTON (S. C.), 'BAYS.'. , 1895, Science.

[4]  Spatial Distribution of Playa Basins on the Texas High Plains , 2016 .

[5]  C. W. Cooke,et al.  Discussion of the Origin of the Supposed Meteorite Scars of South Carolina , 1934, The Journal of Geology.

[6]  M. Breuß,et al.  A modified Mask region‐based convolutional neural network approach for the automated detection of archaeological sites on high‐resolution light detection and ranging‐derived digital elevation models in the North German Lowland , 2021, Archaeological Prospection.

[7]  Antonio P. Carandang,et al.  Land use characterization and change detection of a small mangrove area in Banacon Island, Bohol, Philippines using a maximum likelihood classification method , 2015 .

[8]  G. Hugelius,et al.  GIS‐based Maps and Area Estimates of Northern Hemisphere Permafrost Extent during the Last Glacial Maximum , 2016 .

[9]  Aaron E. Maxwell,et al.  Mapping the Topographic Features of Mining-Related Valley Fills Using Mask R-CNN Deep Learning and Digital Elevation Data , 2020, Remote. Sens..

[10]  K. Di,et al.  Investigation of the depth and diameter relationship of subkilometer-diameter lunar craters , 2018, Icarus.

[11]  H. French,et al.  Permafrost at the time of the Last Glacial Maximum (LGM) in North America , 2014 .

[12]  D. A. King,et al.  Spatial distribution and geomorphology of lunette dunes on the High Plains of Western Kansas: implications for geoarchaeological and paleoenvironmental research , 2018 .

[13]  W. Watts.  Late-Quaternary Vegetation History at White Pond on the Inner Coastal Plain of South Carolina , 1980, Quaternary Research.

[14]  M. Piehler,et al.  Burning peat and reworking loess contribute to the formation and evolution of a large Carolina-bay basin , 2012, Quaternary Research.

[15]  Hong-Yuan Mark Liao,et al.  YOLOv4: Optimal Speed and Accuracy of Object Detection , 2020, ArXiv.

[16]  M. Depledge,et al.  Spending at least 120 minutes a week in nature is associated with good health and wellbeing , 2019, Scientific Reports.

[17]  Antonio Zamora.  A model for the geomorphology of the Carolina Bays , 2017 .

[18]  W. Schriever,et al.  The Carolina "Bays": Are They Meteorite Scars? , 1933, The Journal of Geology.

[19]  D. E. Fenstermacher,et al.  Distribution, Morphometry, and Land Use of Delmarva Bays , 2014, Wetlands.

[20]  J. Orestes Cerdeira,et al.  Contextual classification of remotely sensed images with integer linear programming , 2006, CompIMAGE.

[21]  Thomas Blaschke,et al.  Land cover change assessment using decision trees, support vector machines and maximum likelihood classification algorithms , 2010, Int. J. Appl. Earth Obs. Geoinformation.

[22]  Weixing Zhang,et al.  Deep Convolutional Neural Networks for Automated Characterization of Arctic Ice-Wedge Polygons in Very High Spatial Resolution Aerial Imagery , 2018, Remote. Sens..

[23]  Francesco Mancini,et al.  Image classification methods applied to shoreline extraction on very high-resolution multispectral imagery , 2014 .

[24]  G. Brook,et al.  Late pleistocene paleowind directions, Atlantic Coastal Plain, U.S.A , 1989 .

[25]  A. McEwen,et al.  Relative depths of simple craters and the nature of the lunar regolith , 2017 .

[26]  M. Hodgson,et al.  How many Carolina bays? An analysis of Carolina bays from USGS topographic maps at different scales , 2017 .

[27]  Robert C. Weih,et al.  OBJECT-BASED CLASSIFICATION VS . PIXEL-BASED CLASSIFICATION : COMPARITIVE IMPORTANCE OF MULTI-RESOLUTION IMAGERY , 2010 .

[28]  Stephen L. Egbert,et al.  A GIS-based Approach to Identify and Map Playa Wetlands on the High Plains, Kansas, USA , 2010, Wetlands.

[29]  C. Swezey.  Quaternary Eolian Dunes and Sand Sheets in Inland Locations of the Atlantic Coastal Plain Province, USA , 2020 .

[30]  M. J. Brooks,et al.  CAROLINA BAYS: TIME CAPSULES OF CULTURE AND CLIMATE CHANGE , 2010 .

[31]  D. Wysocki,et al.  Synthesis on Quaternary aeolian research in the unglaciated eastern United States , 2015 .

[32]  Niall O' Mahony,et al.  Deep Learning vs. Traditional Computer Vision , 2019, CVC.

[33]  Ross B. Girshick,et al.  Mask R-CNN , 2017, 1703.06870.

[34]  Wei Guo,et al.  Geospatial Object Detection in High Resolution Satellite Images Based on Multi-Scale Convolutional Neural Network , 2018, Remote. Sens..

[35]  Shuai Li,et al.  Improved Mask R-CNN with distance guided intersection over union for GPR signature detection and segmentation , 2021 .

[36]  Laurence Smith.  Solution Depressions in Sandy Sediments of the Coastal Plain in South Carolina , 1931, The Journal of Geology.

[37]  K. Staver,et al.  Macroinvertebrate community convergence between natural, rehabilitated, and created wetlands , 2016 .

[38]  Nathan Howell,et al.  Guide to the littoral zone vascular flora of Carolina bay lakes (U.S.A.) , 2016, Biodiversity data journal.

[39]  Y. Gao,et al.  A COMPARISON OF THE PERFORMANCE OF PIXEL-BASED AND OBJECT-BASED CLASSIFICATIONS OVER IMAGES WITH VARIOUS SPATIAL RESOLUTIONS , 2008 .