## Changelog ### 3.0 — major correctness and workflow release **Pour-point workflow.** Set one pour point and the upstream catchment, longest flow path and channel slope are all delineated from the DEM. Delineation uses pointer jumping over the D8 flow graph — 1.44M cells in 4.8 s. No catchment polygon is required anywhere in the plugin. **Flood extent corrected.** Depression filling was seeded from the array edge after nodata had been replaced with the catchment mean elevation. On a clipped catchment that constant ring acts as a dam: the entire lower basin was raised to a flat surface, HAND against it was ~0, and roughly 66% of the catchment flooded *regardless of discharge* — a 1.9× increase in Q moved the extent from 65.8% to 65.9%. Extents are now channel-scale and respond to discharge. **Volume budget.** Mapped flood water can no longer exceed the runoff the storm produced. Depth comes from Manning's normal depth at regime width `W = 3√Q`; at small Q that width collapses while the valley it is spread across does not, so a trickle was being rendered as a metres-deep flood. The consistency check existed and warned every time, and nothing acted on it. **Channel slope.** Measured longitudinally as S1085 along the traced main stem, not from the DEM gradient magnitude — which has no sign, is dominated by the cross-section, and was returning ~30× the true slope. Velocity goes as √S, so that alone was inflating velocity about five-fold and driving the hazard classification. **HAND.** Computed by walking each cell's own D8 flow path down to the channel it drains into, rather than taking the geometrically nearest stream cell, which reached across divides and put 16% of cells at zero. **Sub-basin routing.** Sub-basins are clipped to the analysis catchment, so basins draining elsewhere are no longer summed into the hydrograph. **Method-range warnings.** The SCS lag equation reports when it is being extrapolated past its ~8 km² validity range, and when time to peak exceeds the design storm duration (the storm ends before the UH peaks, heavily attenuating the result). **Rainfall location check.** Warns when the rainfall point falls outside the delineated catchment. **CN / LULC extent.** The CN tab takes an area of interest — layer extent, canvas extent, or a rectangle drawn on the map — instead of demanding a catchment polygon. **Tabs 1 and 2.** Catchment boundary now optional. Slope is computed before any clip, so boundary cells keep real neighbours in Horn's method instead of taking their slope from missing data. **Interface.** Rebuilt with a contrast-checked light theme (every text/background pair verified against WCAG AA), new raster colour ramps for DEM, slope, CN, LULC and Manning's n, and QGIS 4 / PyQt6 compatibility.
yes
rahulpandey696
2026-09-20T10:26:37.226896+00:00
3.28.0
4.99.0
None
no
Plugin Tags