Testing strategy (desktop baseline)

This page documents a testing approach that can be applied to HDAT’s calculation engine even if the UI is desktop-based.

Test layers

• Unit tests: pure formulas and helpers (Tc methods, ARF clamp, SCS excess rainfall, Manning helpers).
• Component tests: hydrology engine end-to-end for one catchment; culvert/channel/gully/BOQ module runs.
• Regression / golden tests: fixed projects with stored expected outputs and tolerance checks.

Golden dataset recommendations

Create a case matrix that covers:

• small/medium/large catchments
• Rational-only and SCS-only scenarios
• multiple culvert shapes and slopes
• multiple channel shapes/materials
• gully grade + sag cases

What to snapshot

• inputs (settings + catchment properties + IDF rows)
• intermediate series (for SCS: rainfall increment, excess increment, UH, convolution)
• final outputs (peaks, H/D, depths, quantities)

Release gate (practical)

Block release if:

• parity drift exceeds tolerance
• critical computational branch changes without documented reason
• inputs/outputs cannot be reproduced from stored project artifacts