A low footprint olive grove weather forecasting using a single-layered seasonal attention encoder-decoder model

Mohamed Abdelwahab, Hassan Mostafa, Ahmed K. F. Khattab

Abstract: Highlights•Deep learning-based framework for predicting weather variables for olive farming.•A low complexity encoder-decoder model with a single LSTM/GRU layer.•Seasonal attention exploits seasonality to improve accuracy and reduce footprint.•Up to 29.7% improvement in MAE and 27.4% in RMSE with only 37.6 kB of memory.•Realized on a resource-constrained microcontroller (Raspberry Pi Pico).