Quick Start¶

Installation¶

pip install git+https://github.com/cgodoyle/losneomrade.git

For development (clone and install in editable mode):

git clone https://github.com/cgodoyle/losneomrade.git
cd losneomrade
pip install -e ".[dev]"

Breaking changes¶

If you are updating older scripts, the main changes to watch for are:

bounds now always use (xmin, ymin, xmax, ymax)
clip_to_msml was removed in favor of mask=...
mask helpers now live under losneomrade.masks

Basic usage¶

All bounds use standard GIS order (xmin, ymin, xmax, ymax) in EPSG:25833.

Terrain Criteria (pixel-based)¶

Slope is computed for each pixel in the DEM relative to the source points. Results are reclassified into slope interval classes.

import numpy as np
import geopandas as gpd
from shapely.geometry import LineString
from losneomrade import terrain_criteria, masks, utils

# Define bounds and source line (EPSG:25833)
bounds = (268463.9, 6651396.2, 270007.6, 6652564.4)

source_line = gpd.GeoDataFrame(
    geometry=[LineString([
        (268883.5, 6651785.9),
        (268917.3, 6651614.7),
        (269098.9, 6651692.1),
        (269266.9, 6651685.2),
        (269430.1, 6651718.2),
        (269573.4, 6651894.1),
    ])],
    crs=25833,
)

# Fetch MSML mask from NVE
mask = masks.get_msml_mask(bounds)

# Run terrain criteria
tc = terrain_criteria.run_terrain_criteria(
    bounds=bounds,
    source=source_line,
    source_depth=0.5,
    mask=mask,
    h_min=5,
)

# tc is a GeoDataFrame with classified slope polygons
print(tc.head())

You can also pass source as a numpy array of coordinates or a GeoDataFrame of Points.

Retrogression (step-wise propagation)¶

Propagates a release area outward following slope criteria (1:15 by default):

from losneomrade import retrogression

retro, animation = retrogression.run_retrogression(
    bounds=bounds,
    rel_shape=source_line,
    point_depth=0.5,
    mask=mask,
    min_slope=1 / 15,
    min_height=5,
    min_length=75,
    return_animation=True,
)

print(f"Release area: {retro.area.sum():.0f} m²")

Retrogression with initial landslide (erosion-triggered)¶

For erosion-triggered landslides, use a steeper initial slope (e.g. 1:4) to model the first failure, then continue with the full retrogressive slope (1:15):

retro, animation = retrogression.run_retrogression_with_initial_landslide(
    bounds=bounds,
    rel_shape=source_line,
    point_depth=0.5,
    mask=mask,
    ini_slope=1 / 4,
    retro_slope=[1 / 15],
    min_height=0,
    min_length=20,
    return_animation=True,
)

This two-phase approach models a scenario where an initial steep erosion event triggers subsequent retrogressive failures at the gentler 1:15 slope.

Using a custom raster¶

All methods accept custom_raster to bypass the Høydedata API:

tc = terrain_criteria.run_terrain_criteria(
    bounds=None,
    source=source_line,
    custom_raster="path/to/dem.tif",
)

Profile Retrogression¶

Analyze retrogression along cross-section profiles:

from losneomrade.profile_retrogression import retrogression_from_profiles

result, debug = retrogression_from_profiles(
    profiles=profiles_gdf,
    dem_path="path/to/dem.tif",
    mask_array=mask_raster,
    source_line=source_line,
)