GreenIT coding guidelines tutorial

[tromain]

This tutorial's goal is to show good practices to develop an image processing chain in python, how to optimize the code, the size of the data, and monitor your application to find the bottlenecks and speed up your processing

[ytanguy]

Some comments

• cell "work directories" : set a data_dir for inputs and a separate output_dir. Users should set the first as their favorite tile in datalake, and the second directory as a temp directory in their scratch for instance
• the two cells about OTB should also declare the output_file as f"{output_dir}/img_ndvi_otb_py.tif" (for instance)
• there are a lot of warnings about "RuntimeWarning: invalid value encountered in divide" : I think we can disable these warnings
• cell "simple LZW compression" and "create a cog" : 'raster' is not defined => you should add a cell to define a path to a PHR image.
• this cell could also measure elapsed time with %%time
• be careful because this cell only writes 1 band : if the input image has multiple bands, you have to loop on the different bands (I did not find an other way to do it..)

[ytanguy]

In compute RIOXarray cells, you don't define phr_product_lzw / _cog...

[tromain]

In compute RIOXarray cells, you don't define phr_product_lzw / _cog...

They are defined in the work directories as "input_greenit/phr_product_lzw/cog.tif" , I generated them by hand using the code in the cell "compression", but maybe I could modify the code in this cell to take phr_product.tif as input and produce phr_product_lzw and cog instead of generating them outside the notebook