Electrical imaging of soil water availability to grapevine: a benchmark experiment of several machine-learning techniques

摘要: Electrical resistivity (ER) can be used to assess soil water in the field. This study investigated possibility of extending use ER measure plant available variables, i.e. (ASW), total transpirable SW (TTSW), and fraction (FTSW) using a pedotransfer approach. In vineyard, 224 electrical tomography (ERT) transects 672 time domain reflectometry (TDR) profiles were acquired over 2 years. Soil physical–chemical properties measured on 73 samples from eight different sites. To estimate amount plants, grapevine (Vitis vinifera L.) status was monitored by means leaf potentials. A benchmark experiment carried out compare four machine-learning techniques: multivariate adaptive regression splines (MARS), k-nearest neighbours (KNN), random forest (RF), gradient boosting machine (GBM). Model interpretation led deeper understanding relationships between when predicting availability for plant. The models assessed had good predictive performance therefore map ASW, TTSW FTSW vineyard. coupled algorithms shown proxy quantification visualisation with low disturbance.