In this study, we applied thermal dissipation probe technology to examine sap flow in various directions (east, south, west, and north) and at different depths (0-2, 2-4, 4-6 cm) within the stem of natural Picea mongolica trees in the eastern of Otindag Sandy Land to provide a scientific basis for accurately quantifying water consumption of P. mongolica forests through transpiration and to enhance the understanding of water relations. The results showed that the diurnal variation of sap flow in different directions displayed a unimodal curve, with the sap flow sequence being south>east>west>north. The sap flow at different sapwood depths exhibited an obvious unimodal curve, with a significant decrease as sapwood depth increased. Compared with that calculated from the mean sap flux density in four directions (23.57 kg·d-1), water consumption calculated using the mean value in south-east, east-west, south-west, north-east, north-south, and north-west was overestimated by 10.2%, 5.5%, 14.5%, and underestimated by 12.3%, 8.2%, 9.8%, respectively. The water consumption calculated using the values from the east, south, and west was overestimated by 6.1%, 14.4%, and 15.4%, respectively, and underestimated by 30.7% in the north. In addition, compared with the water consumption calculated from the mean value in three sapwood depths (48.51 kg·d-1), that calculated using sap flux density at sapwood depths of 0-2, 2-4, and 4-6 cm were overestimated by 18.8%, underestimated by 1.7%, and underestimated by 62.9%, respectively. These results indicated that sap flow of P. mongolica had significant azimuthal and radial variations, which considerably influence the estimation of tree water consumption. Installing probes at 0-2 cm simultaneously in both the north and east of the trunk could effectively reduce the estimation error of whole-tree water consumption by 4.2%. This approach enabled the accurate quantification of water consumption of individual P. mongolica trees in sandy areas, thereby improving the precision of transpiration water consumption estimates scaling up from individual level to stand level.