Introduction to Geospatial Data

This repository serves as a brief introduction into the field of geo-spatial data. The objective is to play around with various solutions proposed by the geo-spatial and GIS communities.

Specifically, this repository covers

1. Spatio-Temporal Asset Catalog (STAC) creation and querying using pystac and pystac_client.
2. Creating monthly mosaics of satellite data using Dask.
3. Cloud Optimized GeoTIFFs (COGs) and the use of tilers using titiler.

Get Started

git clone git@github.com:alexberndt/geospatial-data
cd geospatial-data
poetry install

Get started with a jupyter notebook by running

poetry run jupyter-lab

Notebooks

This repository consists of

1. STAC Registry and Querying

   1. src/ml_notebooks/stac/stac.ipynb
   2. src/ml_notebooks/stac/read_stac_api.ipynb

   with helper functions written in a separate file src/ml_notebooks/stac/helper.py to aide with code readability.

   Although catalog.validate() passed all checks, I was struggling to query data using the pystac_client tool. Any idea as to what I was doing wrong?

   To avoid being blocked by this, I used a publicly available STAC catalog to test STAC queries with (see read_stac_api.ipynb notebook).

2. Exploration of Provided Satellite Data

   1. src/ml_notebooks/stac/explore_data.ipynb

3. Monthly Mosaic

   1. src/ml_notebooks/conda/monthly_mosaic.ipynb

   As mentioned above, I was struggling getting the STAC queries to work without error, so ended up testing the coiled-based cloud cluster using the example STAC provided by Planetary Computer. The results are documented in this notebook:

   1. src/ml_notebooks/conda/median_mosaic.ipynb

4. COG Visualization

   1. src/ml_notebooks/cog/cog_visualizations.ipynb

Assets Folder

The assets are saved as follows

Technologies

Here follows a list of technologies commonly used to solve ML-related problems in the geo-spatial industry:

1. Dask
2. Xarray
3. Coiled
4. Dagster
5. PySTAC