pzywczuk/n2o-ssa-models

Machine learning models (Random Forest, XGBoost, Feedforward Neural Network) for predicting N₂O emissions from Sub-Saharan African soils across forest, cropland, and grassland land-use types. # N2O Emissions Prediction Models - Open Access This repository contains three machine learning models for predicting N2O emissions from different land use types (Forest, Cropland, Grassland) using environmental and management variables. ## Models Included 1. **Random Forest Model** (`random_forest_n2o_model.py`) 2. **Artificial Neural Network Model** (`artificial_neural_network_n2o_model.py`) 3. **XGBoost Model** (`xgboost_n2o_model.py`) ## Author **P. Agredazywczuk** Date: 20245-05-22 Modified for open access: 2026 ## Requirements ### Python Dependencies ```bash # Core packages pandas>=1.3.0 numpy>=1.21.0 scikit-learn>=1.0.0 matplotlib>=3.3.0 seaborn>=0.11.0 scipy>=1.7.0 # Model-specific packages tensorflow>=2.8.0 # For Neural Network model xgboost>=1.5.0 # For XGBoost model joblib>=1.0.0 # For model saving/loading ``` ### Installation ```bash pip install pandas numpy scikit-learn matplotlib seaborn scipy tensorflow xgboost joblib ``` ## Data Requirements ### Input Data Format Your input data should be a CSV file with the following columns: #### Required for all land use types: - `date`: Date of measurement (YYYY-MM-DD format) - `lat`: Latitude - `lon`: Longitude - `site`: Site identifier - `landuse`: Land use type ('Forest', 'Cropland', 'Grassland', 'Wetland Forest') - `fert`: Fertilisation treatment indicator - `temp`: Air temperature (°C) - `tmax`: Maximum air temperature (°C) - `tmin`: Minimum air temperature (°C) - `rain`: Precipitation (mm) - `soil_moisture_1`: Soil moisture at depth 1 (m³/m³) - `soil_moisture_2`: Soil moisture at depth 2 (m³/m³) - `soil_moisture_3`: Soil moisture at depth 3 (m³/m³) - `soil_temp_1`: Soil temperature at depth 1 (°C) - `soil_temp_2`: Soil temperature at depth 2 (°C) - `soil_temp_3`: Soil temperature at depth 3 (°C) - `ccov`: Cloud cover (fraction) - `vpd`: Vapor pressure deficit (kPa) - `ssrd`: Surface solar radiation downwards (MJ/m²) - `ppfd`: Photosynthetic photon flux density (μmol/m²/s) - `days_si …