OPTIMAL - cOPernicus irrigaTION mAnagement tooLkit - environmental parameters forecast demonstration
OPTIMAL for cOPernicus irrigaTION mAnagement tooLkit is a combination of state-of-the-art Machine Learning techniques that is designed to forecast environmental parameters which enable the delivery of irrigation needs intelligence.
The model presented is based in an Long short-term memory model used to generated forecasts of the several environmental parameters while given 7 days of inputs and towards delivering 7 days of outputs.
The parameters used are:
◦ Average Temperature (degrees Celsius);
◦ Maximum Temperature (degrees Celsius);
◦ Minimum Temperature (degrees Celsius);
◦ Maximum Temperature - Next Day Forecast (degrees Celsius);
◦ Minimum Temperature - Next Day Forecast (degrees Celsius);
◦ Average Humidity (%);
◦ Maximum Humidity (%);
◦ Minimum Humidity (%);
◦ Radiation (MJ/m²)
◦ Precipitation (mm)
◦ NDVI
◦ Irrigation (m³);
The model was trained using the data of an almonds plantation located in the region Salamanca, Spain.
The dataset used in the training is composed of 6 years of real data and 8 years of synthetic data, generated via a process of data augmentation, giving a total of 14 years.
The model is supplied in the TensorFlow format and a Scaler is provided in the PKL (pickle) file.
Its demonstration is supported by a Python Notebook.