The soil heat flux determination method proposed by Gao (Boundary-Layer Meteorol 114:165-178, 2005) is discussed for (1) dry surfaces, (2) bare soil or sparse short-grass lands, and (3) dense-grass surfaces or forest. Our analysis shows that, when neglecting the contribution of soil vertical water movement to soil heat flux, the energy components measured independently will (1) still achieve balance over dry surfaces, and (2) be significantly in imbalance over bare soil or sparse short-grass lands. The mean of bare ground evaporation modeled by SiB2 is 1.58 x 10(-5) m(3) s(-1) m(-2), and the mean of soil water flux obtained by the method of Gao is 1.22 x 10(-5) m(3) s(-1) m(-2) for the Naqu site in the summer of 1998. Comparison of the bare ground evaporation with the mean of soil water flux shows a difference, the causes of which are investigated. Physically, the bare ground evaporation is equal to the sum of soil water flux and water content change in the soil surface layer. Because the bare ground evaporation is very limited for the dense-grass surfaces or forest, our analysis implies that the energy imbalance encountered over the dense-grass or forest is not caused by the fact that previous researchers neglected soil water movements in their energy budget analyses.