As the increasing hub height of modern wind turbine, wind shear model is generally adopted as a useful tool to extrapolate wind speed available to higher levels for wind power assessment. This paper conducts a data-driven study to examine the power law model in assessing the wind power in forested regions featured with seasonally-varied roughness. Specifically, wind speeds are observed using a laser lidar mounted in forested regions in the northeast of China, ranging from 80 m to 200 m with a constant interval of 20 m. Subsequently, comparisons are made between the wind profiles established based on the power law model and field observations. The exponents used include the empirical exponents and the fitted exponents, the former are derived from the standard codes, and the latter are fitted based on the seasonal and annual data collections. Results indicate that the power law model is unable to fully capture the vertical distribution of wind speed in forested regions, it is the exponent that plays a decisive role influencing the reliability representing the real wind profile. Compared with the empirical exponents, the seasonally-fitted exponents exhibits much better suitability for wind speed extrapolation in forested regions, as well as for the wind power assessment.