Skip to content

Change log#

Unreleased (feat/phase-4)#

Hydrology — new stream-network surface (W-1 → W-5)

  • StreamRaster.order(method="hack") — Hack 1957 main-stem-first ordering.
  • StreamRaster.order(method="topological") — Kahn-sort sequential numbering.
  • StreamRaster.to_vector(...) now carries a sinuosity column per link.
  • StreamRaster.main_stem(flow_direction, outlet=None) — public API for the longest source-to-outlet trace.
  • StreamRaster.prune_short(flow_direction, min_length_m) — drop headwater links below a threshold.

Hydrology — watershed metrics (W-7 → W-9)

  • FlowDirection.isobasins(streams, accumulation, target_area_km2) — equal-area sub-basin partition.
  • WatershedRaster.statistics(...) gained longest_flow_path_m (via the shared kahn_max_upslope_length Kahn pass). Triggers on flow_direction alone after the M1 review fix.
  • FlowDirection.upslope_flowpath_length() — per-cell longest upslope flow path.

Hydrology — indices and Monte-Carlo (W-10 → W-14, W-20)

  • DEM.hand(..., method="euclidean") — nearest-stream-by-2D-distance HAND variant; drops stream cells on no-data DEM positions with a UserWarning (M2 / L2 review fixes).
  • DEM.stochastic_depressions(sigma, n_runs, seed) — Monte-Carlo per-cell depression-occurrence probability (Lindsay & Creed 2006).
  • DEM.twi(accumulation, slope_deg=None) — Beven & Kirkby 1979 wetness index.
  • DEM.spi(accumulation, slope_deg=None) — stream power index.
  • DEM.sti(accumulation, slope_deg=None) — Moore & Burch 1986 sediment transport index.
  • DEM.full_hydro_pipeline(...) — composite fill → flow_direction → accumulate (+ optional streams) returning a typed-results dict.

LiDAR (W-15 → W-19)

  • LasPoints dataclass + read_las / write_las via soft-imported laspy.
  • classify_ground(method="zhang") — Zhang 2003 morphological tophat ground filter.
  • grid_lidar_points(aggregate=...) extended with idw / nn / tin (linear barycentric) / rbf interpolation plus a count density aggregator.
  • clip / merge / filter_classes — vector clip + concatenation + ASPRS-class subset on LasPoints.
  • detect_trees(chm, min_height_m, radius_fn) — variable-window local-maxima tree detection on a canopy height model.

Terrain attributes (W-21 → W-28)

  • DEM.tpi(window) — Guisan 1999 Topographic Position Index.
  • DEM.deviation_from_mean(window) — standardised TPI.
  • DEM.elev_std(window) — focal-window standard deviation of elevation.
  • DEM.ruggedness(window) — Riley et al. 1999 terrain ruggedness index.
  • DEM.curvature(kind="plan"/"profile"/"total"/"mean"/"gaussian") — Zevenbergen-Thorne 1987 curvature family.
  • DEM.normal_vector_deviation(window) — surface-normal angular deviation roughness metric.
  • DEM.openness(search_radius, kind="positive"/"negative") — Yokoyama 2002 topographic openness.
  • DEM.sky_view_factor(search_radius) — Zakšek et al. 2011 sky-view factor.

Shared private helpers: _focal_window_stats (no-data-aware focal mean / SD) and _numba.horizon_walk_kernel (numba-accelerated 8-azimuth horizon scan).

Phase-4 backfill (P29 – P34)

  • cloud_io.tile_windows / cloud_io.write_cog — chunked-tile iteration and Cloud-Optimized GeoTIFF writes.
  • topobathy_fusion(method=...) — 4-mode topobathy blend (max / min / topo_above / bathy_below).
  • Mesh — triangle-mesh container with Laplacian smoothing and aspect-ratio quality metrics.
  • grid_lidar_points (foundation that W-17 extended).
  • DEM.anudem_interpolate — Hutchinson 1989 biharmonic ANUDEM.
  • FlowDirection.subbasins_pfafstetter(accumulation, streams, level=N) — multi-level Pfafstetter coding (levels 1 and 2; deeper recursion deferred).

Dependencies

  • cleopatra removed (now sourced transitively through pyramids-gis[viz]).
  • pyramids-gis bumped from 0.9.1 → 0.18.0.
  • gdal >=3.10 and numba >=0.60,<1 added as hard deps to keep PyPI and pixi installs in sync.

Documentation

  • New guides: hydrology pipeline, watershed analytics, terrain indices, LiDAR pipeline.
  • New reference pages: FlowDirection, Accumulation, StreamRaster, WatershedRaster, LiDAR, Mesh, Cloud I/O, Topobathy fusion.

0.1.0 (2025-09-**)#

  • First release on PyPI.