Automatic Digitization of Jma Strong-motion Seismograms Recorded on Smoked Paper Using Deep Learning
Year: 2025
Type: article
Abstract: Furumura et al. (2023) trained convolutional neural network (CNN) models to automatically digitize the waveforms scanned from the Japan Meteorological Agency (JMA) analog strong-motion seismographs recorded on smoked paper. We validated the CNN models using automatically digitized seismograms of the 1940 Kamui-Misaki-Oki earthquake by applying CNN models to the scanned images of the seismograms that were not used for CNN training. We compared the resulting data with manually traced data. The automatically digitized data agreed well with the manually traced data in most cases, although some data required correction. Using the CNN models substantially reduced the effort required to digitize analog records.
DOI: 10.5610/jaee.25.8_83
Authors: Mitsuko FURUMURA, Ritsuko S. MATSU'URA
Language: English
Subfield: Artificial Intelligence
Field: Computer Science
Domain: Physical Sciences
Open Access status: diamond
OpenAlex ID: W4411768481