islam-shahboun/AI-BEE-Weather-Data-Platform.-Libya

Hybrid ML + ERA5 reanalysis platform for spatial climate prediction and EnergyPlus weather file (EPW/DDY) generation in data-scarce regions — validated on Libya's 28-station network with independent statistical bias correction. # AI-BEE Weather Data Platform **A hybrid machine-learning and reanalysis-corrected climate prediction platform for Libya**, built on 28 real TMYx weather stations (2011–2025) and ERA5 reanalysis. Predicts climate metrics (HDD, CDD, solar radiation, temperature, humidity, wind) for **any location in Libya** and generates ready-to-use EnergyPlus `.epw` weather files and ASHRAE-style `.ddy` design-day files — no measured station required at the target site. **Created by:** Eng. Islam K Shahboun & Prof. Dr. Samah K Alghoul --- ## Live Demo Open `platform/AI-BEE_Weather_Platform.html` directly in any browser — no installation, no server needed. All computation runs client-side in JavaScript. ## Features - Interactive map with the 28 real reference stations (exact GPS coordinates extracted from real EPW headers) - Click anywhere in Libya, or enter coordinates manually, for an instant hybrid ML + ERA5-corrected climate prediction - **Hybrid bias-correction model**: for each climate variable, automatically selects the best-performing approach (baseline geography-only, ERA5 feature augmentation, or ERA5 residual correction), validated via Leave-One-Out Cross-Validation - Location-specific confidence metrics based on spatial distance decay and agreement with the nearest real station - Full 8,760-hour synthetic `.epw` file generation for any point, using a physically-corrected hybrid method (temperature lapse rate, Magnus-Tetens humidity, solar radiation scaling, barometric pressure correction) - **ASHRAE-style `.ddy` design-day file generation** (0.4th/99.6th percentile heating/cooling design conditions) - Downloadable Excel export of predictions and the full 28-station reference dataset - Statistical charts: station ranking, elevation vs. cooling degree-days, climate zone distribution ## Repository Structure ``` ├── platform/ │ └── AI-BEE Weather Data Platform.html # Self-contained interactive platform (hybrid model, EPW+DDY generation) ├── scripts/ │ …