In order to scientifically evaluate water resource reserves and the regulation mechanisms of vegetation in a hydrological cycle, it is necessary to improve the inversion accuracy of aboveground vegetation water storage (AVWS). To calculate more accurate AVWS, Sentinel-1 (S1) and Sentinel-2 (S2) imagery were used to map the AVWS of the Jiuzhaigou Nature Reserve. Predictors extracted from S1 and S2 imagery were divided into three modelling groups: S1 data (S1G), S2A data (S2G) and S12G data (combining S1G and S2G). The best linear model (LM) and Random Forest (RF) regression model of AVWS were established by predictors of the three groups; finally, RF regression models based on S2G (10-fold cross-validation determination coefficients [R²] = 0.67, relative root mean square error [rRMSE] = 13.78 %) and S12G (R² = 0.72, rRMSE = 11.94 %) were selected for mapping AVWS in the study area. The results show that the aboveground vegetation water storage (AVWS) of the study area could be effectively estimated and mapped based on synergistic predictors derived from S1 and S2, as well as S12G predictors (cooperating with S1G and S2G).