Lidar detection requires strict calibration and stabilization on transmitted wavelength. The backlash errors, which occurs in wavelength control of the tunable laser, seriously reduce the detection accuracy. In order to solve the backlash problem of the tunable laser, a algorithm which can offset the positive and negative backlash error by jumping back and forth is proposed and firstly applied to the frequency stabilization of the high detection of vertical profiles of atmospheric CO2 concentration. The experiment result indicates that the algorithm not only can improve the adjusting precision of the tunable laser′s wavelength, but also eliminate the backlash error of the laser′s wavelength adjustment caused by multiple reverse. This method reduces the shortage of the wavelength adjustment caused by mechanical rotating and has great significance for the development of hyperspectral laser technology.