YoussefGlb/climate-prediction-morocco-2050

# 🌍 Climate Prediction — Morocco (2025–2050) Machine learning pipeline to predict future climate variables across **Morocco**, trained on historical TerraClimate data (1981–2024) and visualized through Köppen-Geiger classification maps. --- ## 📁 Project Structure ``` climate-prediction-morocco/ │ ├── ML/ # XGBoost-based approach │ ├── projectionML.py # Main training & prediction script │ ├── koppenviwer.py # Tkinter GUI — Köppen-Geiger map viewer (basic) │ ├── koppen_viewer_extended.py# Tkinter GUI — extended viewer with trends & animation │ └── data/ # ⚠️ Local only — not tracked by Git │ └── {year}/ │ └── {variable}/ │ └── morocco_{variable}_{year}_{month:02d}.tif │ ├── DL/ # Deep Learning approach │ ├── projectionDL.py # LSTM model — same pipeline, DL counterpart │ └── data/ # ⚠️ Local only — not tracked by Git │ └── {year}/ │ └── {variable}/ │ └── morocco_{variable}_{year}_{month:02d}.tif │ ├── .gitignore └── README.md ``` --- ## 🔬 What It Does | Step | Script | Description | |---|---|---| | Train & Predict (ML) | `ML/projectionML.py` | Trains one XGBoost model per climate variable, generates annual + monthly GeoTIFF predictions for 2025–2050 | | Train & Predict (DL) | `DL/projectionDL.py` | Trains a single multi-output LSTM model for all 6 variables jointly | | Visualize | `ML/koppen_viewer_extended.py` | Tkinter GUI — load historical or predicted data, explore variables, compare periods, animate over time | --- ## 🌡️ Climate Variables | Variable | Description | Unit | Aggregation | |---|---|---|---| | `tmax` | Maximum temperature | °C | Monthly mean | | `tmin` | Minimum temperature | °C | Monthly mean | | `prec` | Precipitation | mm | Monthly sum | | `vap` | Vapor pressure | kPa | Monthly mean | | `ws` | Wind speed | m/s | Monthly mean | | `def` | Climate water defici …