Compare an ML wind prediction with a CFD result at the same points: VDI 3783-9 hit rate (D = 0.25, W = 0.05·U_ref; ≥ 0.66 passes), FAC2, Willmott's index of agreement, R, bias, MAE and RMSE per direction and overall, plus the per-point deviation. Write appends the compared points to a verification log (Dataset Curator columns + U_pred + U_cfd) that is retraining data and a training envelope for Prediction Confidence. ML screening prediction, not CFD: use it to compare options, then confirm anything that matters with a Wind Study (planning authorities do not accept surrogate results).
Input
Name
Nickname
Description
Type
Points
The points both results were sampled at (absolute Z).
Point
Predicted
ML wind speeds [m/s], one branch per wind direction (the Wind Predictor's Values output).
Number
CFD
CFD wind speeds [m/s] at the same points, one branch per direction in the same order (e.g. Probe |U|).
Number
Boundary Conditions
BC
The study's boundary conditions: U_ref sets the hit rate's absolute tolerance W = 0.05·U_ref and the log's normalised columns; the wind directions label the branches.
Generic Data
Buildings
Optional: building geometry, so the log carries SDF / Bldg_height exactly as the Dataset Curator writes them.
Geometry
Pedestrian Level
PedLevel
Pedestrian mounting height [m] for the log's features.
Number
Log
Verification log CSV to append to (created with its header when new).
Text
Case
A name for this comparison in the log.
Text
Write
Append the compared points to the log. Fires once per set of inputs, also when wired.
Boolean
Output
Name
Nickname
Description
Type
Metrics
Agreement metrics: overall first, then one block per direction.
Text
Hit Rate
HitRate
VDI 3783-9 hit rate per direction (then overall, last).
Number
Passes
Hit rate ≥ 0.66 per direction (then overall, last).
Boolean
Deviation
Dev
Predicted − CFD [m/s] per point, one branch per direction.
Number
Colors
Per point: green within the hit-rate tolerance, blue over-predicted, red under-predicted.