Hindcast Evaluation Dashboard
Same-day streamflow estimates using day-T weather and river observations through T-1.
RESEARCH BENCHMARK
HINDCAST PEAK ESTIMATE
-- m3/s
Peak date
OBSERVED HINDCAST PEAK
-- m3/s
Includes training and validation periods
EVALUATION SCOPE
HOLDOUT
Used for checkpoint selection
HOLDOUT MAE
-- m3/s
Loading persistence baseline
Observed vs. Predicted Tracking
Architecture & Quality Benchmarks
Technical breakdown of how DeepFlood overcomes domain shifts and hydrological time lag.
Model Reliability (Predicted vs. Observed)
Pipeline Structure
DeepFlood utilizes a hybrid spatiotemporal neural network tailored for tropical catchments.
01
Input Windowing
Captures 7 days of historical meteorological variables.
02
1D-CNN Temporal Extraction
Extracts localized storm burst patterns and rate-of-change indicators.
03
Bidirectional LSTM
Processes forward and backward temporal sequences.
04
Temporal Attention Layer
Dynamically weights critical timesteps during extreme flood surges.
Technical Breakthroughs
Addressing standard time-series flaws in watershed hydrology.
1. Same-Day Meteorological Alignment
Prediction cutoff: Uses observed weather on day T and river observations through T-1 to estimate streamflow on day T.
Scope: This is hydrological nowcasting, not a demonstrated multi-day early-warning forecast.
Scope: This is hydrological nowcasting, not a demonstrated multi-day early-warning forecast.
2. Basin-Specific Calibration
Problem: Broad national scaling can suppress local-basin extremes.
Solution: V3.1 trains from scratch with scalers fitted only on the Long Dai training period and strictly lagged streamflow features.
Solution: V3.1 trains from scratch with scalers fitted only on the Long Dai training period and strictly lagged streamflow features.
Chronological Holdout Benchmark
MODEL MAE
--
30% chronological holdout
MODEL RMSE
--
NSE --
PERSISTENCE BASELINE MAE
--
Q(T) = Q(T-1)
Historical Data & Simulator
Inspect daily hydrological benchmarks and simulate extreme storm impacts.
What-If Storm Simulator
Test model response to simulated rainfall intensity increases over the Long Dai catchment.
Base Peak Forecast
-- m3/s
Simulated Adjusted Peak
-- m3/s
Risk Status
Safe
Exploratory frontend-only sensitivity illustration. It does not run the trained model and must not be interpreted as a flood warning.
Hydrological Log
| Date | Rainfall | Observed | Predicted | Abs Error |
|---|