veroandreo / foss4g2022_grass4rs Goto Github PK

• Prepare S2 data for NC location, proposed computational region: same as landsat scenes
• Decide on exercises/steps to show
• Obtain new land use land cover data for NC (osm?)
• Create a merge between neteler's and vero's notebooks

• edit/adapt the maxlik classifcation section of diff scenes using signatures + band ref
• check section naming and numbering
• add section titles to README.md for a quick overview
• upload (reduced) Sentinel-2 data somewhere
• add section for importing the landuse map
• run everything from scratch
• complete instructions in the README file and send to workshop participants
  • software download
  • data download
  • Registration at Copernicus Open Access Hub, store username and password in two lines in a text file called esa_credentials.txt

Here is a list of things I suggest:

Section 4

For printing GRASS version I suggest:

import subprocess
print(subprocess.check_output([grassbin, "--config", "version"], text=True))

Section 5

For gisenv use:

gs.gisenv()

For g.list use:

gs.list_grouped(type="vector")

Creating new mapset - this is not completely solved API. Note that grass.grassdb package was developed for internal use only, it's not linked in the Python documentation, although it is there when you search for it. So it will likely change. You can create mapset and switch with (NOT recommended, unstable API and when you rerun the cell it will fail):

import grass.grassdb.create as gdb
gdb.create_mapset(grassdata, location, "sentinel2")
session.switch_mapset("sentinel2")

Alternatives: Call grass with bash

!grassbin -c -e ....
session.switch_mapset(...)

Or use an existing mapset user1:

session.switch_mapset("user1")

This is something we should address better for the next version.

Section 7

Avoid manual parsing:

import pandas as pd
from io import StringIO

pd.read_csv(StringIO(list_prod), delimiter=" ", usecols=[0, 1, 2, 4, 5, 6, 7],
            names=['uuid', 'scene', 'date', 'cloud', 'product', 'size', 'unit'])

Printing file content - use context manager (with open("") as f:) or just use !cat ...

Section 10

For getting center of region use:

gs.region(complete=True)

Section 13

Instead of:

# Set color table
colours = ["11 195:20:0", "13 255,255,168", "14 100:240:100", "31 77:105:0", "51 0:204:242"]
colourise = gs.feed_command("r.colors", map="sentinel_rf", rules="-", quiet=True)
colourise.stdin.write("\n".join(colours).encode())
colourise.stdin.close()

Use:

gs.write_command("r.colors", map=..., rules="-", stdin="\n".join(colours))

Instead of parse_command with r.report, use print(read_command(...)), there may be more places like that.

1. GRASS GIS basics (database, region, modules, show the GUI a bit)
2. GRASS GIS & Python
3. Why Jupyter Notebooks and how to use them?
4. Paths and variables
5. Download NC location and other data
6. Initialization of GRASS GIS in the Jupyter notebook session
7. Create a new mapset within NC
8. Setting the computational region
9. Importing geodata into GRASS GIS
10. Sentinel-2 processing overview
11. Search and download S2
12. Import S2 data and show different visualizations
13. Cloud masking - Skipped for now, no time
14. Computing NDVI and NDWI for one scene only
15. Create a time series of NDVI using semantic label syntax and show some other python utilities like plots
16. Creating an image stack (imagery group)
17. Image segmentation
18. Supervised Classification: RandomForest
19. Supervised Classification: Maximum Likelihood (show classification of different scenes based on signatures with semantic labels)

The explicit call of session.finish is not needed, remove it. Leaving it there means that after executing that cell, you can't get back to any other cell. Generally, you may want that behavior or you may want to align that with the standard non-Jupyter init use (grass.script[.setup].init), but for workshop focused on remote sensing, I recommend greater flexibility.

grass.jupyter.setup.init doc says:

The session is ended when session.finish is called or when the object is destroyed when kernel ends or restarts. This function returns a copy of an internally kept reference, so the return value can be safely ignored when not needed.

The jupyter_tutorial.ipynb notebook says:

When the kernel for the notebooks shuts down or is restarted, the session ends automatically. The session can be explicitly ended using session.finish(), but that's usually not needed in notebooks.

In other words, explicit use of session.finish() is neither required nor considered a best practice according to the documentation. Feel free to extend it as needed.

Not sure about it, so I wanted to ask in time: the ubuntugis-unstable doesn't offer GRASS 8.2 but 8.0.2:
was this version meant for the workshop (probably not, as 8.2 exists) and if not, how to add the 8.2 in Ubuntu?

Thanks.