Timely acquisition of soil moisture content (SMC) of edible roses is crucial for achieving precision irrigation. Unmanned aerial vehicle (UAV) multispectral technology was adopted, and through field experiments, SMC data at different soil depths and corresponding UAV multispectral images were collected during the flowering period of edible roses, with vegetation indices and texture features having strong correlations with crop parameters established. Grey relational analysis (GRA) was used to evaluate the influence degree of vegetation indices and texture features on SMC at each soil depth, and parameters with significantly correlated coefficients to SMC at each depth were selected as model input variables (Combination 1: vegetation indices;Combination 2: texture features;Combination 3: vegetation indices and texture features). Random forest (RF), extreme gradient boosting (XGBoost), and gradient boosting decision tree (GBDT) models were employed to model SMC at each soil depth respectively. The results showed that the inversion effect of the 0~10 cm soil layer was overall better than that of the 10~20 cm and 20~30 cm soil layers, the mean coefficient of determination (R2) of the validation sets of each model for the 0~10 cm layer was 0.12~0.21 higher than that of the deeper soil layers, while the root mean square error (RMSE) and mean relative error (MRE) were reduced by 0.8~1.5 percentage points and 3~5 percentage points respectively. At the optimal soil depth of 0~10 cm, the GBDT model with Combination 3 as input performed the best, with R2=0.8363, RMSE of 1.28%, and MRE of 5.06%, which was superior to the RF and XGBoost models. Finally, the SHapley Additive exPlanations (SHAP) analysis method was used to reveal the importance of spectral vegetation indices (SVI) and mean (MEA) in constructing the prediction model and clarify the influence of the top-ranked SHAP values, and the results can provide a basis for UAV multispectral monitoring of SMC in edible roses and a reference for the rapid evaluation of crop growth under the condition of integrated water and fertilizer management.