Site database for national strong motion stations in mainland China

China site database (CNSDB) contains site metadata for 1450 strong motion stations with recordings in the China Flatfile project. The stations are from China National Strong Motion Observation Network System (NSMONS), in 27 provinces of mainland China. The principal site parameters in CNSDB are time-averaged shear wave velocity in the upper 30m (V_S30) and site classification results according to China seismic design code. V_S30 values are derived or extrapolated when reliable velocity profiles or field survey results are available. The extrapolation relationship is developed according to statistical properties of 6179 engineering boreholes, which is separated into four subregions in China mainland. Besides measurement-based site parameters, CNSDB consists of site parameters derived from the earthquake horizontal-to-vertical spectral ratio (HVSR) curve, including predominant period and amplitude. Our previously proposed machine learning-based HVSR site classification schemes are also utilized to estimate V_S30 and China site classifications. For stations without velocity profiles and enough ground motion recordings for HVSR computation, we utilize geology age/genesis, ground surface slope, and terrain category as site description proxies to estimate V_S30. We analyze the performance of these proxies in relation to the measured V_S30 values and provide the recommended V_S30 value and its dispersion. We present protocols for V_S30 estimation and China site classification from proxies that emphasize methods minimizing bias and dispersion relative to data. Except for the recommended site parameters results, site characterization proxies for each site and corresponding site parameters are also provided in the open-source site table of CNSDB. This can facilitate the search for the optimal site parameter(s) for the prediction of site amplification in different application occasions, like GMM development, scenario ground motion simulation, and seismic hazard/risk assessment.