search

earthlab / firedpy

A Python Command Line Interface for classifying fire events from the Collection 6 MODIS Burned Area Product.
MIT License
36 stars 20 forks source link

Edge case process is, actually, completely broken #64

Open admahood opened 2 years ago

admahood commented 2 years ago

file:///home/a/Desktop/edge_case_process_is_completely_broken.png

Check out this figure... this is supposed to be spatial parameter of 1, and a temporal parameter of 2. Whatever that edge case process is doing is not working.

admahood commented 2 years ago

edge_case_process_is_completely_broken

admahood commented 1 year ago

@Ckster are you considering working on this issue at all, once the low-hanging fruit bugs are tackled? I'm also starting to look into seeing if I can find someone here to work on it, since we might be able to use work study to get someone from CSU. @EllieLindrooth got started on it and I think she was close to figuring it out, so there might be a branch that has it half done somewhere on this repo.

Ckster commented 1 year ago

Hey @admahood sorry did not see this until now. Yes I will be trying to solve this as I work on the 2.0.0 branch

maxwellCcook commented 1 year ago

@admahood do you remember what you ran for this? I.e., can we try to reproduce it to work on it? At least which tile that is?

CibeleAmaral commented 1 year ago

Firedpy 2 0 - FIREDpy_2 0 1

The team is working on this now. Updates will come in the next few months.

admahood commented 1 year ago

It's Queensland, Australia, FID == 59228 (That info is actually also in the screenshot). It was in 2004, so it might be easier to process only that year to reproduce the error and do checks. Australia has a LOT of fires so it takes a long time. It might be worth it to explore other, smaller countries that fall along a lot of tile intersections and see if you can find edge-related errors there, so you don’t have to do such a resource intensive operation to reproduce errors.