This is a MATLAB script demonstrating a deep learning-based geodetic bathymetry prediction using gravity gradient tensors as input signals.
The MATLAB wrapper for the Generic Mapping Tools (GMT) is extensively used in this code.
You only need to install GMT 6.3.0 or above, available at https://github.com/GenericMappingTools/gmt/releases.
This demonstration was done on a computer with 64 GB RAM, i7-12700H CPU and RTX 3050Ti GPU. It has been tested on MATLAB R2020b and above.
The datasets are grids of gravity gradient tensors (Txx.nc, Tyy.nc, Tzz,nc, Txy.nc, Txz.nc and Tyz.nc), and ship-borne depths (Sounding.mat).
It takes roughly 3 days to fully run this script.
There is a possibility of running out of memory as the demonstration area (lon -1 ~ 60, lat -31 ~ 30) is very large.
Access to a more powerful GPU is highly recommended. It is advisable to cut out a study area smaller than, or preferably half of, this one if your computer has a 32 GB RAM.
There are further instructions as you run the code.
It is advisable to run it section-by-section in a livescript (.mlx) file.
This code supports a research article currently under review. I will provide the reference for the article immediately it is published.
By: Richard Fiifi ANNAN (richannan@outlook.com)
School of Land Science and Technology
China University of Geosciences (Beijing)
No. 29 Xueyuan Road, Haidian District, Beijing, China