Image of Where are the outcrops? Automatic delineation of bedrock from sediments using Deep-Learning techniques

Text

Where are the outcrops? Automatic delineation of bedrock from sediments using Deep-Learning techniques

The delineating of bedrock from sediment is one of the most important phases in the fundamental process of regional bedrock identification and mapping, and it is usually manually performed using high-resolution optical remote-sensing images or Light Detection and Ranging (LiDAR) data. This task, although straightforward, is time consuming and requires extensive and specialized labor. We contribute to this line of research by proposing an automated approach that uses cloud computing, deep learning, fully convolutional neural networks, and a U-Net model applied in Google Collaboratory (Colab). Specifically, we tested this method on a site in southwestern Norway using both a set of explanatory variables generated from a 10 m resolution digital elevation model (DEM) and, for comparison, cloud-based Landsat 8 data. Results show an automatic delineation performance measured by an F1 score between 77% and 84% for DEM terrain derivatives against a manually-mapped ground truth. Overall, our automated bedrock identification model reveals very promising results within its constraints.

Ketersediaan
151551.136Perpustakaan BIG (Eksternal Harddisk)Tersedia
Informasi Detail
Judul Seri
Applied Computing and Geoscience - Open Access
No. Panggil
551.136
Penerbit
Amsterdam : Elsevier.,
Deskripsi Fisik
10 hlm PDF, 10.364 KB
Bahasa
Inggris
ISBN/ISSN
2590-1974
Klasifikasi
551.136
Tipe Isi
text
Tipe Media
-
Tipe Pembawa
-
Edisi
Vol.18 June 2023
Subjek
Cloud Computing
Deep learning
U-net
Bedrock
Sediment
Google Earth Engine (GEE)
Info Detail Spesifik
-
Pernyataan Tanggungjawab
Versi lain/terkait

Tidak tersedia versi lain

Lampiran Berkas
  • Where are the outcrops? Automatic delineation of bedrock from sediments using Deep-Learning techniques
    The delineating of bedrock from sediment is one of the most important phases in the fundamental process of regional bedrock identification and mapping, and it is usually manually performed using high-resolution optical remote-sensing images or Light Detection and Ranging (LiDAR) data. This task, although straightforward, is time consuming and requires extensive and specialized labor. We contribute to this line of research by proposing an automated approach that uses cloud computing, deep learning, fully convolutional neural networks, and a U-Net model applied in Google Collaboratory (Colab). Specifically, we tested this method on a site in southwestern Norway using both a set of explanatory variables generated from a 10 m resolution digital elevation model (DEM) and, for comparison, cloud-based Landsat 8 data. Results show an automatic delineation performance measured by an F1 score between 77% and 84% for DEM terrain derivatives against a manually-mapped ground truth. Overall, our automated bedrock identification model reveals very promising results within its constraints.
Komentar
Anda harus masuk sebelum memberikan komentar