visualize optical time series from sentinel-2 lesson

[rbavery]

this could follow the example of https://datacarpentry.org/r-raster-vector-geospatial/12-time-series-raster/index.html but instead use a Sentinel-2 optical time series from a previous stackstac episode. This could involve the following learning objectives:

• assembling band arrays into a single multiband array with xarray
• compute multispectral indices, either by hand or using xarray-spatial as an example https://github.com/makepath/xarray-spatial
• plot change over time at a pixel, or other summary plots. we can use hvplot + xarray for this: https://github.com/makepath/xarray-spatial

[rbavery]

other idea: netherlands crop boundaries, plotting change in spectral indices for a field over time

[rbavery]

this comment applies to this issue: #85 (comment)

@rogerkuou

[rbavery]

@rogerkuou @fnattino I've selected an initial area for this episode, the goal is to compute focal stats with xarray-spatial on a time series of sentinel-2 derived ndvi and see which farms have the highest average greeeness over the growing season (or something like that).

What do you two think of this initial area? I selected a spot that bordered the north of Amsterdam since globally that might be the most recognizable, but if there is an area with a more interesting story to tell through the episode that either of you know of, I'm down to change it.

[rogerkuou]

Thanks Ryan! Looks good! I am working on a Notebook for #88 and selected almost the same area like this. I will finish the notebook and push today, we can always align in the end to make the AoI the same.

[rbavery]

closing in favor of the stackstac issue #82 , these episodes could probably one and the same