Translate all remaining R functionality by brettedw · Pull Request #13 · cffdrs/cffdrs_py

brettedw · 2026-01-27T21:50:06Z

This PR is a comprehensive conversion of the CFFDRS (Canadian Forest Fire Danger Rating System) R package to Python.

closes implement rest of package #2

Key Changes

Implemented as scalar operations to match existing python code, without batch modes or vectorization

New Modules

Core Models cffdrs/models.py: Dataclasses for FBP inputs and outputs (FBPInput, FBPPrimaryOutput, FBPSecondaryOutput, FBPAllOutput)
All the rest of the modules match the R code in file and function names
Utilities:
- cffdrs/constants.py - Fuel type constants
- cffdrs/r_helpers.py - R compatibility helpers

Test Suite

Test coverage uses all of the same csv data from the R package
Nearly all tests are passing, with the exception of slope_calc.py, sdmc.py and a single fbp test. These failing tests have reasoning in their @pytest decorators.
slope_calc.py test failures are edge case failures seemingly from some difference between R and Python mathematical operations, but shouldn't have a practical effect on it's outputs (all tests that use the slope_adjustment function pass). The failures all occur when WSV is very small, <1e-11
- If you want to see the test results, you can comment out the @pytest.mark.xfail and run the test
I've left a comment on the single failing fbp test, I believe it might be due to a small error in the R code/test data
These failing tests are currently skipped in the test runner, all other tests run and pass

Configuration Updates

pyproject.toml: Added ruff for consistent code formatting.

R compatibility

Function names designed to match R package conventions for easy translation
r_helpers.py provides R-style math operations for specific cases like div by 0

brettedw · 2026-01-28T00:13:09Z

cffdrs/tests/test_fbp.py

+@pytest.mark.xfail(
+    reason="I believe default CFL should be 0, not 1 as it's currently written - https://github.com/cffdrs/cffdrs_r/blob/4e40dd3af841f3a708abd83d7f2d43fefb08649a/R/fire_behaviour_prediction.r#L38"
+)


I think it's possible there's a minor mistake in the fbp implementation and test data from the R package, that only surfaces if no input is provided

It is strange as it sets the CFL in the input definition but then re-establishes later on based on the fuel type? So a C-2 would have a CFL of .8 once the crown_fuel_load function runs. It will run if CFL is between 0 and 2, but having a fueltype defined would overwrite? Strange behaviour imo.

cffdrs/buildup_effect.py

cffdrs/constants.py

cffdrs/direction.py

dgboss · 2026-01-28T20:09:01Z

cffdrs/direction.py

+        dir = bearingT1T2 - ThetaAdeg
+    if dir is not None:
+        if dir < -180:
+            dir += 360


Are these + 360 and -360 correct? The R implementation has + 10 and -10.

Good catch, you're right, this implementation is slightly different than the R implementation. I was a bit confused by the +/- 10 and in my conversations with a higher power it also thought the +/- 10 was a bug. However, a scenario where these implementations differ never came up in the tests for pros/lros where this function is used (there are no unit tests for direction in the R library).

I just generated some random inputs and the majority of them are the same:

Python outputs

Test 1: bearingT1T2=10, bearingT1T3=5, ThetaAdeg=20 -> dir=-10 Test 2: bearingT1T2=5, bearingT1T3=10, ThetaAdeg=20 -> dir=25 Test 3: bearingT1T2=-10, bearingT1T3=-5, ThetaAdeg=20 -> dir=10 Test 4: bearingT1T2=-5, bearingT1T3=-10, ThetaAdeg=20 -> dir=-25 Test 5: bearingT1T2=45, bearingT1T3=-45, ThetaAdeg=30 -> dir=15 Test 6: bearingT1T2=-45, bearingT1T3=45, ThetaAdeg=30 -> dir=-15 Test 7: bearingT1T2=100, bearingT1T3=-100, ThetaAdeg=50 -> dir=150 Test 8: bearingT1T2=100, bearingT1T3=-100, ThetaAdeg=90 -> dir=-170 Test 9: bearingT1T2=-100, bearingT1T3=100, ThetaAdeg=50 -> dir=-150 Test 10: bearingT1T2=-100, bearingT1T3=100, ThetaAdeg=90 -> dir=170 Test 11: bearingT1T2=10, bearingT1T3=5, ThetaAdeg=200 -> dir=170 Test 12: bearingT1T2=5, bearingT1T3=10, ThetaAdeg=200 -> dir=-155 Test 13: bearingT1T2=120, bearingT1T3=-120, ThetaAdeg=70 -> dir=-170 Test 14: bearingT1T2=-120, bearingT1T3=120, ThetaAdeg=70 -> dir=170

R outputs

Test 1: bearingT1T2=10.0, bearingT1T3=5.0, ThetaAdeg=20.0 -> DIR=-10.0 Test 2: bearingT1T2=5.0, bearingT1T3=10.0, ThetaAdeg=20.0 -> DIR=25.0 Test 3: bearingT1T2=-10.0, bearingT1T3=-5.0, ThetaAdeg=20.0 -> DIR=10.0 Test 4: bearingT1T2=-5.0, bearingT1T3=-10.0, ThetaAdeg=20.0 -> DIR=-25.0 Test 5: bearingT1T2=45.0, bearingT1T3=-45.0, ThetaAdeg=30.0 -> DIR=15.0 Test 6: bearingT1T2=-45.0, bearingT1T3=45.0, ThetaAdeg=30.0 -> DIR=-15.0 Test 7: bearingT1T2=100.0, bearingT1T3=-100.0, ThetaAdeg=50.0 -> DIR=150.0 Test 8: bearingT1T2=100.0, bearingT1T3=-100.0, ThetaAdeg=90.0 -> DIR=-170.0 Test 9: bearingT1T2=-100.0, bearingT1T3=100.0, ThetaAdeg=50.0 -> DIR=-150.0 Test 10: bearingT1T2=-100.0, bearingT1T3=100.0, ThetaAdeg=90.0 -> DIR=170.0 Test 11: bearingT1T2=10.0, bearingT1T3=5.0, ThetaAdeg=200.0 -> DIR=-180.0 Test 12: bearingT1T2=5.0, bearingT1T3=10.0, ThetaAdeg=200.0 -> DIR=195.0 Test 13: bearingT1T2=120.0, bearingT1T3=-120.0, ThetaAdeg=70.0 -> DIR=-170.0 Test 14: bearingT1T2=-120.0, bearingT1T3=120.0, ThetaAdeg=70.0 -> DIR=170.0

Test 11 & 12 is where they differ, but I actually think the Python implementation is correct. For test 12, the direction would be calculated as 205, then adjusted to 195, which I don't quite understand.
Whereas in Python the output is -155 (equivalent to 205 on a 360-degree circle)

This may be a question for the CFS devs, but wanted to write all this out because I forgot to revisit it

An adjustment of 10 degrees does seem odd. At first glance it would seem that you would want to normalize the calculated direction to fit within 0-360 but neither implementation does that.

Looks like the original paper normalizes to -180 to 180, so your Python implementation is correct I'd say. Maybe you could include something in the doc string to clarify this?

Thanks for looking into the paper! Didn't realize it was in there. Good call, I'll put a note in

Docstring note added in 3fb0bd0

@jordan-evens I think this would be more in your wheelhouse to discuss than mine, do you have anything to add here?

dgboss

Strong work!

brettedw · 2026-01-29T18:46:59Z

@BadgerOnABike @jordan-evens This is quite a large PR, but it contains all the remaining CFFDRS R code translation, along with all tests ported over from the R repo. Tried to maintain all the same comments and references to the source formulas just like the R library.
The review might seem a bit daunting but hopefully the running and passing of all the tests helps. Please see the PR description for additional info on the testing

cffdrs/tests/test_distance_at_time.py

cffdrs/tests/test_pros.py

BadgerOnABike · 2026-02-13T17:24:36Z

cffdrs/tests/test_sdmc.py

+                Row 77: Calculated SDMC=0.0484, Expected=0.0484, Diff=-0.0000
+                Row 78: Calculated SDMC=5.9601, Expected=0.0484, Diff=5.9117
+                Row 79: Calculated SDMC=19.1639, Expected=3.0500, Diff=16.1139
+                Row 80: Calculated SDMC=32.3677, Expected=9.7560, Diff=22.6117"""


Hmm very odd

Maybe there's something I don't quite understand about how the R package is loading that data for tests

BadgerOnABike · 2026-02-13T18:03:39Z

A few comments and I reopened one of the comments to get the table in the comment.

Otherwise this is a huge lift well done folks, apologies for the delays!

BadgerOnABike · 2026-02-13T19:03:57Z

Worth keeping the discussion going to see if we need to adjust anything but I'd like to unblock further progress on this.

brettedw added 30 commits January 15, 2026 15:49

initial conversion

e12b35c

cfb_calc tests

1f39aa4

c6 tests

82cbc08

cbh tests

9a7d6d3

crown_fuel_load tweak

cc2197f

crown base height import unused

8d21719

r helpers and distance_at_time tests

5f56713

fire intensity test

7e92146

flank ros test

9395fb5

fmc test

b3d84f0

test grass fuel moisture

413810a

test length to breadth at time

383cdaf

test length to breadth

62014e6

test ros theta

078d3e6

test rate of spread at time

e168dad

rate of spread tests and cleanup

84683f2

test surface fuel consumption

4ab1a63

test total fuel consumption

f59a402

absolute errors

af51eb6

slope test

9a75a8c

comment

30bc3cb

r equivalent tweak

367de50

fbp fuel type strict

5a3d3cc

r behaviour tweak

241bb98

input validation and clamping

77e2cd9

fbp tests

b3cbe58

test file & field names = R

38d96e2

hffmc tests cleanup

4a63b50

function names and backward compatibility

0b5a83b

conftest cleanup

05bfe1e

combine classes

4d93297

brettedw commented Jan 28, 2026

View reviewed changes

conbrad reviewed Jan 28, 2026

View reviewed changes

cffdrs/buildup_effect.py Outdated Show resolved Hide resolved

brettedw added 3 commits January 28, 2026 11:15

snake case

bd0ace8

outputs & cleanup

1c217b1

rate_of_spread comments

42d45e4

dgboss reviewed Jan 28, 2026

View reviewed changes

cffdrs/constants.py Show resolved Hide resolved

dgboss reviewed Jan 28, 2026

View reviewed changes

cffdrs/direction.py Show resolved Hide resolved

dgboss reviewed Jan 28, 2026

View reviewed changes

brettedw added 5 commits January 28, 2026 15:22

BUIo and q description

8568fea

r behaviour doc string

ccfcbbe

remove depreciated

cc7e9ee

unused import

43bf077

FUEL_TYPE_ROS comments

be561d3

conbrad approved these changes Jan 29, 2026

View reviewed changes

dgboss approved these changes Jan 29, 2026

View reviewed changes

direction doc string

3fb0bd0

brettedw changed the title ~~Translate all remaining R package~~ Translate all remaining R functionality Jan 29, 2026

brettedw added 2 commits January 29, 2026 17:26

readme update

4b03c5f

readme cleanup

384dfca

BadgerOnABike self-assigned this Feb 13, 2026

BadgerOnABike reviewed Feb 13, 2026

View reviewed changes

cffdrs/tests/test_distance_at_time.py Show resolved Hide resolved

BadgerOnABike reviewed Feb 13, 2026

View reviewed changes

cffdrs/tests/test_pros.py Outdated Show resolved Hide resolved

BadgerOnABike reviewed Feb 13, 2026

View reviewed changes

brettedw added 2 commits February 13, 2026 10:15

pros abs error threshold

a83f808

reference table 7

34c0eab

BadgerOnABike merged commit 948c910 into cffdrs:main Feb 13, 2026
1 check passed

Conversation

brettedw commented Jan 27, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Key Changes

New Modules

Test Suite

Configuration Updates

R compatibility

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

dgboss Jan 29, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

dgboss left a comment

Choose a reason for hiding this comment

Uh oh!

brettedw commented Jan 29, 2026

Uh oh!

Uh oh!

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

BadgerOnABike commented Feb 13, 2026

Uh oh!

BadgerOnABike commented Feb 13, 2026

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

4 participants

brettedw commented Jan 27, 2026 •

edited

Loading

dgboss Jan 29, 2026 •

edited

Loading