Workaround for computing NDVI stats on all census tracts

[lauren-alexandra]

It was taking hours while looping over the census tracts to process even a quarter of the data in my local repository. I switched to using the free Google Colab Jupyter Notebook service on a standard runtime (default). I calculated NDVI statistics for each tract in just 40 minutes.

For the setup, I copied over what I had in my repository notebook and included the following package installs at the top of the Colab notebook. To get the package versions, I ran conda list in my local conda environment.

!pip install geopandas==1.0.1
!pip install geoviews==1.14.0 
!pip install holoviews==1.20.0
!pip install hvplot==0.11.1
!pip install rioxarray==0.17.0
!pip install shapely==2.0.6 
!pip install xarray==2024.11.0
!pip install cartopy==0.24.0
!pip install numpy==1.26.4
!pip install pandas==2.2.3
!pip install scipy==1.14.1 
!pip install scikit-learn==1.5.2
!pip install tqdm==4.66.5 
!pip install pystac==1.11.0 
!pip install pystac-client==0.8.4

If you want to download the stats CSV to work with it locally, open the folder in the Files panel storing your file and download. Example file path: /root/Projects/earth-analytics/data/chicago-greenspace/chicago-ndvi-stats.csv

[BioNCphilic]

Thanks. Wish I had seen this earlier.

[katherinesiegel]

Great find!