Forecasting soil temperature based on surface air temperature using a wavelet artificial neural network

Summary

This study developed and validated wavelet-enhanced artificial neural network models for forecasting soil temperature at multiple depths based solely on surface air temperature observations. Using four years of hourly data from a synoptic station in Iran, the authors demonstrated that wavelet preprocessing significantly improved forecasting accuracy, with WANN models capable of reliable 1–7 day ahead predictions at agricultural depths of 5–30 cm. The approach offers a practical, data-minimal tool for agricultural meteorology and crop management planning.

UK applicability

The methodology is transferable to UK conditions, though UK soil temperature dynamics and air-temperature coupling may differ from the continental climate of Iran. Recalibration and validation using UK meteorological station data would be necessary to establish local applicability for UK crop sowing schedules and frost protection.

Key measures

Soil temperature forecasting accuracy at 0300, 0900 and 1500 GMT; comparison of ANN vs WANN model performance; effect of temporal increment on forecasting errors

Outcomes reported

The study demonstrated that wavelet-transformed artificial neural network (WANN) models improved forecasting accuracy of soil temperature compared to standard ANN models. WANN models were validated to forecast soil temperature 1–7 days ahead at depths of 5–30 cm using only surface air temperature data.

Topic tags