Supervised evaluation
Test the evidence-to-forecast workflow
Use synthetic Simulation mode to test or demonstrate the workflow. Real source connections require approved access and validation. This prototype is not an operational flood-warning service.
Inspect the synthetic workflow
Open the workbench. Confirm “Simulation,” the illustrative Kratie location, daily units and synthetic source label. Change the selected day and check the chart values.
Follow a single snapshot
Read the source evidence and export the labeled manifest. The chart, map point and calculated summary must refer to that same run.
Test uncertainty and failure
Switch to Simulation. Try the missing-value and unavailable-source cases. Through-flow volume becomes unknown when a day is missing; no failure becomes a low-risk label.
Rehearse a citizen report
Use the built-in fixture, confirm its limited consent and simulate a review. Evidence status changes, while publication stays private and scientific use stays blocked.
Check protected storage after deployment
Use two authorized test accounts to confirm owner isolation, media protection and separately assigned reviewer access. These hosted tests remain pending in the unpublished build.
Readiness questions for a real pilot
Which basin, decision and useful lead time? Who reviews reports? Which observations and local thresholds can a partner provide? Are data rights, privacy procedures and costs approved? Establish these before asking a client to rely on flood predictions.
Discuss pilot readiness