<p>The mixed layer height (MLH) plays a key role in the atmospheric numerical simulation and environmental assessment. Although the medium threshold (MT) method for retrieving the MLH from lidar turbulence data is promising, the retrieval suffers from the interferences of the low-level jets close to the ground under weak thermal convection conditions and the boundary layer clouds topped on a well-mixed layer under strong thermal convection conditions. Here, a robust threshold (RT) method is developed and demonstrated to be robust against the interferences. It helps recalibrate the MLH by taking account into the influences of the low-level jets and the boundary layer clouds. To evaluate the performance of the RT method, a field experiment is conducted with a coherent Doppler wind lidar. In the experiment, the typical turbulence profiles and their MLH retrieval results under different thermal convection conditions are analyzed. The diurnal variations of the MLH retrieval results by the MT method and the RT method are compared qualitatively and quantitatively. The experiment results indicate that the developed method is more robust to the interferences and thus more suitable for retrieving the MLH under different thermal convection conditions.</p>