Abstract: This paper puts forward a suitable sparsi fication scheme for L-band second-level radiosonde data. The sounding vertical profile may describe more accurately the vertical change of meteorological elements with increased number of the significant layers. The background field (ERA-Interim reanalysis) is applied for bias evaluation of the observation data. Sounding data from different sources are utilized for comparison and analysis. The bias, the root mean square error (RMSE) and other quantitative indicators are calculated to assess the effect of the sparsi fication scheme. The results show that this scheme can not only increase the numbers of the radiosonde pro file levels, but also achieve better Bias and RMSE results. This scheme may provide necessary guarantee for the assimilation applications of the high-resolution L-band radiosonde data in China for the future.