Although the thermal dissipation probe (TDP) method is prominent in forest transpiration studies, the accurate estimation of sap flow density in particular trees with different wood structures remains highly variable. To verify the applicability of the TDP method in the measurement of the transpirational water consumption of shrub communities in degraded karst areas in southern China, Granier’s original formula was validated and corrected using the isolated stem segment method with Celtis sinensis, Xylosma congesta, Triadica sebifera, Paliurus ramosissimus, Pyracantha fortuneana, Sageretia thea, and Phanera championii. Finally, the correction coefficients were analyzed in relation to wood structural parameters. Our study reveals that the sap flow density estimated using Granier’s original formula exhibited varying degrees of overestimation or underestimation compared to the measured sap flow density. The differences ranged from −78.9% to 114.2%, highlighting the need for formula correction. Consequently, we adjusted the coefficients α and β of the original formula (n = 3–5), resulting in a revised range of 0.0047–0.0509 g cm−2 s−1 and 0.5882–1.9821 (R2 = 0.81–0.97), respectively. The conduit diameter was significantly correlated with the correction coefficient α. It is currently unknown whether the results obtained for a particular species can be applied to different growth environments and more prominent individuals of the same species. In conclusion, it is necessary to verify the applicability of Granier’s original formula when estimating water consumption through transpiration in trees.