lsdtopytools

Repository for lsdtopytools, a python wrapper on LSDTopoTools. It binds c++ code from LSDTopoTools with lsdtt-xtensor and exposes it to python via numpy. It allows us to take advantage of both c++ speed and python flexibility. The lsdtopytools suite can be installed using conda, built in native linux or within a Docker container.

This repository hosts the static version of the code available for installation; it is updated at each release (usually when we are publishing a new manuscript utilising new features). Tutorials and example notebooks can be found in this dedicated repository.

Features

• Raster i/o and basic operations
• Preprocessing depressions (carving, filling, nodata, ...)
• River extraction (drainage area threshold)
• ksn from segmented profiles (Mudd et al. 2014)
• concavity index (Mudd et al., 2018, Gailleton et al., 2021 (preprint))
• polyfit metrics
• ...

Installation using docker

See https://github.com/LSDtopotools/lsdtt_pytools_docker for intructions on how to install the Docker container.

Installation using conda

Installation through conda and pip

You need to first have a valid installation of anaconda on your computer. We suggest miniforge see readme here which guarantee a full open-source and open-license installation.

Installation from conda has been tested for windows10, MacOS and different linuxes (ubuntu, Debian, Redhat, WSL). Other OSes will need to build from source (explanation bellow).

First create a conda environment with python <= 3.8 (3.9 will come shortly) and activate it:

conda create -n lsdtopytools python=3.8

conda activate lsdtopytools

RECOMMENDED STEP IF YOU HAVE ANOTHER DISTRIBUTION THAN MINIFORGE: you need to fix an annoying tendency of gdal to rely on messed up dependencies:

conda config --prepend channels defaults
conda config --prepend channels conda-forge
conda config --set channel_priority strict

Then, you can install by simply running:

conda install lsdtopytools

Something went wrong

If something went wrong, you need to check several things:

• You are on a previously created conda environment and trying to install lsdtt-xtensor-python in it, check that you DO have python=3.7.X
• You have a 32 bit OS/processor, You need to install from source (also it has been over 15 years since all the computers are 64 bit, this is the reason we do not guarentee compatibility)
• You are on an antiquated computer unable to get recent libc, libc++ or libstdc++, there is nothing we can do unfortunately although I'd be surprised if it happens. We need c++14 for lsdtt-xtensor and that is a rigid dependency unfortunately.

Installation from source

You need to get the code (so far from the development repository under src/driver_analysis_BG/), and you need a c++14-ready compiler (tested with gcc 5+ and MSVS so far, should work with clang, theoretically icc as well). Make sure all the dependencies (see above) are installed, especially xtensor and xtensor-python. There are different ways to install a python package from source. From far the safest is to generate a wheel, i.e. a python binary.

For lsdtt-xtensor-python:

cd lsdtt-xtensor-python
python setup.py bdist_wheel
pip install dist/XXX.whl

where XXX is the name of the created wheel file. Then for lsdtopytools:

cd ../lsdtopytools
pip install .

Done.

Troubleshooting

• [LINUX] GLIBC error: I am working on it, but basically your linux is older than the one with which I installed it and does not have as recent glibc. The good news is that there is a solution, bad news is that It will take me (BG) a bit of time to implement and you need to install from source. I need to build manylinux2010 wheels which involves using docker, which is not compatible with any of my hardware without accessing the BIOS and I cannot.

So far in early development stage, all of that code will evolve very rapidly and probably drastically. Contact boris.gailleton@gfz-potsdam.de for questions.