Diurnal cycle of low clouds (DCLC) is strongly modulated by regional meteorological conditions, leading to complex cloud diurnal patterns particularly in semi-arid regions. However, the mechanisms governing DCLC behavior in these regions remain poorly understood due to sparse high-precision observations and intricate meteorological conditions. In this study, we analyze 6 years (2014-2019) of ground-based Ka-band Zenith Radar (KAZR) observations from the Semi-Arid Climate and Environment Observatory of Lanzhou University (SACOL) to investigate DCLC variability. Using Fourier decomposition, we propose a Bimodal Prominence Index (BPI) to quantify DCLC features and examine their seasonal transitions linked with key meteorological factors. Our results show that low clouds exhibit prominent bimodal structure with greater coverage and amplitude in cold seasons, transitioning toward unimodal pattern during warm months. Key meteorological factors, including estimated inversion strength (EIS), mid-to-lower tropospheric wind shear (MLWS), 700 hPa relative humidity (RH700) and 700 hPa vertical velocity (omega 700), exhibit strong relationships with the DCLC variability. A multiple regression model incorporating these variables effectively reproduces the DCLC, achieving a high correlation coefficient of 0.71 with observations. RH700 and EIS primarily control low cloud formation while omega 700 and MLWS modulate seasonal patterns. We found noon-peaked RH700 and omega 700 drive unimodal cloud in summer while strong inversions and enhanced nocturnal MLWS support bimodal clouds in winter. Autumn pattern reflect enhanced nighttime humidity and weakened daytime EIS peak favoring bimodality. These findings provide an insight into the meteorological controls of DCLC in semi-arid environments and their role in affecting regional weather and climate.