Accurate prediction of farmland soil moisture is crucial for determining crop water requirements and establishing effective irrigation standards. However, the high costs and potential disruption of soil structure make direct measurement of soil moisture across various depths challenging. In this study, the long short-term memory (LSTM) and decision tree (DT) models were proposed to predict soil moisture at 3-, 5-, 10- and 20-cm soil depths based on soil temperature data, and the accuracies of the two models in predicting soil moisture at different depths were evaluated at half-hour and daily scales. The results revealed that the accuracies of the LSTM and DT models in predicting soil moisture at different depths at the half-hour scale were greater than those at the daily scale. The accuracy of the LSTM model was better than that of the DT model at different depths. Both models performed best at the 20-cm soil depth, followed by the 10-, 5- and 3-cm soil depths, with R 2 values ranging from 0.90-0.95, 0.81-0.95, 0.84-0.95 and 0.86-0.95, respectively. Therefore, the LSTM model is recommended for the prediction of soil moisture at different soil depths, providing valuable references for farmland management, irrigation decision-making and the formulation of drought prevention and mitigation measures. Une pr & eacute;vision pr & eacute;cise de l'humidit & eacute; du sol des terres agricoles est cruciale pour d & eacute;terminer les besoins en eau des cultures et & eacute;tablir des normes d'irrigation efficaces. Cependant, les co & ucirc;ts & eacute;lev & eacute;s et la perturbation potentielle de la structure du sol rendent difficile la mesure directe de l'humidit & eacute; du sol & agrave; diverses profondeurs. Dans cette & eacute;tude, les mod & egrave;les de m & eacute;moire & agrave; long terme et & agrave; court terme (LSTM) et d'arbre de d & eacute;cision (DT) ont & eacute;t & eacute; propos & eacute;s pour pr & eacute;dire l'humidit & eacute; du sol & agrave; des profondeurs de 3 cm, 5 cm, 10 cm et 20 cm & agrave; partir des donn & eacute;es de temp & eacute;rature du sol, et la pr & eacute;cision des deux mod & egrave;les pour pr & eacute;dire l'humidit & eacute; du sol & agrave; des profondeurs diff & eacute;rentes a & eacute;t & eacute; & eacute;valu & eacute;e & agrave; des & eacute;chelles de demi-heure et de jour. Les r & eacute;sultats ont r & eacute;v & eacute;l & eacute; que la pr & eacute;cision des mod & egrave;les LSTM et DT dans la pr & eacute;vision de l'humidit & eacute; du sol & agrave; diff & eacute;rentes profondeurs & agrave; l'& eacute;chelle de demi-heure & eacute;tait plus grande que celle de l'& eacute;chelle de jour. La pr & eacute;cision du mod & egrave;le LSTM & eacute;tait meilleure que celle du mod & egrave;le DT & agrave; diff & eacute;rentes profondeurs. Les deux mod & egrave;les ont donn & eacute; des meilleurs r & eacute;sultats & agrave; une profondeur de 20 cm, suivis des profondeurs de 10 cm, 5 cm et 3 cm, avec des valeurs R2 allant de 0,90-0,95, 0,81-0,95, 0,84-0,95 et 0,86-0,95, respectivement. Par cons & eacute;quent, le mod & egrave;le LSTM est recommand & eacute; pour la pr & eacute;vision de l'humidit & eacute; du sol & agrave; diff & eacute;rentes profondeurs de sol, fournissant des r & eacute;f & eacute;rences pr & eacute;cieuses pour la gestion des terres agricoles, la prise de d & eacute;cisions en mati & egrave;re d'irrigation et la formulation de mesures de pr & eacute;vention et d'att & eacute;nuation de la s & eacute;cheresse.