A global spectral library, and earth engine tools for satellite land cover mapping..

Documentation: earth-chris.github.io/earthlib

Source code: earth-chris/earthlib

earthlib is a python package featuring a global spectral library of reference spectra, and a set of software tools for satellite-based land cover mapping in Earth Engine.

The library contains several thousand unique spectral endmembers representing green vegetation, soil, non-photosynthetic vegetation, urban materials, and burned materials. The reflectance data cover the visible to the shortwave infrared wavelengths (400-2450 nm) at 10 nm band widths.

The software tools (1) resample these data to match the wavelengths of popular satellite and airborne earth observing sensors and (2) run spectral mixture analysis in Google Earth Engine via the earthengine python package.

The goal is to quantify spatial and temporal patterns of change in global vegetation cover, as well as patterns of soil cover, burned area, non-photosynthetic vegetation, and impervious surfaces. With earthlib, you can do this using a wide variety of public satellite data sources.

This library can be installed via pip.

pip install earthlib

You can also clone the source repository and install it locally.

git clone https://github.com/earth-chris/earthlib.git
cd earthlib
pip install -e .

Christopher Anderson^{1 2}

This work was primarily developed at the Stanford Center for Conservation Biology with the Natural Capital Project.