Flood Forecast

Last updated: 2026-10-15 11:00:00 SAST
EN ZU XH
Home Lite View

Weather Models

Statistics

Water Level
Min—
Mean—
Max—
Rainfall
Total—
Max Hourly—
Mean Hourly—
Wet Hours—

Forecast Accuracy

MAE (m) by forecast horizon
Model 1-6h 7-12h 13-24h 25-48h 49-72h 73h+
Ensemble Mean 0.0138 0.0155 0.0150 0.0174 0.0134 0.0095
Best Match 0.0133 0.0148 0.0146 0.0159 0.0141 0.0130
CMA GRAPES 0.0126 0.0163 0.0148 0.0173 0.0143 0.0132
ECMWF AIFS 0.0127 0.0150 0.0155 0.0171 0.0156 0.0089
ECMWF IFS 0.0133 0.0148 0.0146 0.0159 0.0141 0.0130
ECMWF IFS 0.25° 0.0136 0.0155 0.0141 0.0158 0.0136 0.0132
GFS Global 0.0174 0.0181 0.0185 0.0218 0.0167 0.0136
GFS Seamless 0.0174 0.0181 0.0185 0.0218 0.0167 0.0136
MET Norway 0.0133 0.0148 0.0146 0.0159 0.0141 0.0130
NCEP AI-GFS 0.0143 0.0177 0.0212 0.0234 0.0157 0.0126
NCEP HGEFS 0.0136 0.0164 0.0165 0.0173 0.0136 0.0095
UKMO Global 0.0124 0.0153 0.0152 0.0168 0.0138 0.0131
Based on 128,892 prediction-observation comparisons (all-time). Updates every hour as new observations verify past predictions.
How is this calculated?

Each hour, the system generates water level predictions for the next 7 days using each weather model's rainfall forecast. Once real observations come in, the predicted values are compared against what actually happened. MAE (Mean Absolute Error) is the average difference between predicted and observed levels — lower is better. Columns group predictions by how far ahead they were made: a "1-6h" prediction is easier to get right than a "73h+" one. Values are colour-coded: green < 0.02m, yellow < 0.05m, orange < 0.1m, red ≥ 0.1m.