Multi-view multi-depth soil temperature prediction (MV-MD-STP): a new approach using machine learning and time series methods

Estimation of soil temperature is of great importance because of its great effects on plant development, yield increase, chemical and biological activities in the soil. This paper proposes a novel multi-view multi-depth learning framework for soil temperature prediction. Under the proposed framework, soil temperature prediction at various soil depths is performed using multivariate time series and machine learning methods. This is the first study that two different views to represent antecedent soil data and past meteorological data were designed to capture different features. According to the experimental results, when support vector regression (SVR) is applied with the multi-view multi-depth learning framework, the predicted soil temperature values approached to the real values at most, and it outperformed other methods. It is the first time that multi-view multi-depth learning has been provided a very powerful opportunity to estimate soil temperature when both time series and machine learning methods are used as base learners.