1:45 PM - 2:00 PM
[HTT17-01] Estimation of Paleotopography in the Okayama Airport Area Using Machine Learning Methods
★Invited Papers
Keywords:Digital Elevation Model, Paleotopography, Machine Learning, Modeling
Okayama Airport (OKJ) is a regional airport located in Kita Ward, Okayama City, Okayama Prefecture, approximately 15 km northwest of the city center. The area currently occupied by the airport once featured undulating terrain composed of the middle layer of the Tomiyoshi Formation within the Kibi Group. This formation is characterized by alternating layers of sandy deposits from meandering river channels and muddy floodplain deposits. However, this original terrain was lost due to the construction of the airport. This study aimed to estimate and examine the pre-construction topography of the Okayama Airport area using machine learning applied to a Digital Elevation Model (DEM).
The DEM used in this study was derived from LiDAR-based topographic data provided by the Geospatial Information Authority of Japan (GSI), specifically the Fundamental Geospatial Information DEM05A (5-meter mesh). The machine learning processes were conducted using Google Colaboratory, a Python IDE (Google, 2025). Pre-construction aerial photographs were obtained from GSI tiles (GSI, 2025), and geospatial analyses were performed using QGIS (QGIS Development Team, 2025). The Contour plugin in QGIS was used to generate contour lines.
Two machine learning algorithms were developed to estimate the paleotopography of the Okayama Airport area. The results of both algorithms indicated the presence of continuous topographical features extending from outside the airport area into its boundaries. Additionally, three depressions were identified within the airport area. However, the southwestern depression was inconsistent with the features observed in pre-construction aerial photographs, suggesting the need for further refinement of the machine learning algorithms for topographical estimation.
The DEM used in this study was derived from LiDAR-based topographic data provided by the Geospatial Information Authority of Japan (GSI), specifically the Fundamental Geospatial Information DEM05A (5-meter mesh). The machine learning processes were conducted using Google Colaboratory, a Python IDE (Google, 2025). Pre-construction aerial photographs were obtained from GSI tiles (GSI, 2025), and geospatial analyses were performed using QGIS (QGIS Development Team, 2025). The Contour plugin in QGIS was used to generate contour lines.
Two machine learning algorithms were developed to estimate the paleotopography of the Okayama Airport area. The results of both algorithms indicated the presence of continuous topographical features extending from outside the airport area into its boundaries. Additionally, three depressions were identified within the airport area. However, the southwestern depression was inconsistent with the features observed in pre-construction aerial photographs, suggesting the need for further refinement of the machine learning algorithms for topographical estimation.