Fig. 1. Results of spline curve fitting and two-stage retrieval of a Cn2 profile for simulated noisy r0 values mostly lower than the theoretical data: (a) spline curve fitting with 5% Gaussian noise; (b),(c) the first and second stage of the retrieval, (d) relative error of lgCn2 on the inversion.

Fig. 2. Results of spline curve fitting and two-stage retrieval of a Cn2 profile for simulated noisy r0 values mostly higher than the theoretical data: (a) spline curve fitting with 5% Gaussian noise; (b),(c) the first and second stage of the retrieval, (d) relative error of lgCn2 on the inversion.

Fig. 3. Cn2 profile retrieval with the error bars and corresponding RMSEs based on the 30 profiles in the simulation set. The red line represents the average retrieval values of the 30 profiles. The blue line (error bars) is the standard deviation of the 30 profiles.

Fig. 4. SNR curve based on the 30 profiles in the simulation set.

Fig. 5. Example of the measured r0 profile with a fitted spline curve (the red, dark yellow, and dark cyan curve) at three typical time periods.

Fig. 6. Comparison of Cn2 profiles using the DCIM lidar and radio-sounding balloons at three typical time periods corresponding to Fig. 5.

Fig. 7. Relative errors between the DCIM lidar and radio-sounding balloons corresponding to Fig. 6.

Fig. 8. SNR of the DCIM lidar at different altitudes corresponding to Fig. 6 (assuming the radio-sounding balloon measurements are pure signals).