Abstract:Data missing is a common problem in soil survey and related researches. When this problem proposed, the common solution in most studies was to neglect it or remove records that have missing data due to the lack of the understanding of the importance of data missing. Obviously, this solution could not to satisfy the needs of practical studies. The application of pedotransfer functions (PTFs) provides a broad prospect for the interpolation of missing data of soil database in a simple, rapid and batch processing way. At present, few studies were carried to analyze or interpolate the missing data of soil database in China. More importantly, the method to interpolate the missing data of soil database needs to be standardized. In this study, the characteristics of missing data in Chinese Soil Database from the Second National Soil Survey were analyzed, and the interpolations of serious missing date of soil properties were tried in order to provide knowledge for future researches. Results showed, the data of soil texture (sand, silt and clay contents), pH, organic matter, total nitrogen, total phosphorus and total potassium were most complete for they are the basic survey factors in soil survey. The data of available phosphorus, available potassium and cation exchange capacity had a certain miss. The data of alkaline nitrogen, bulk density and iron oxides were missed seriously. Considering that the stability of prediction model is essential, so soil properties with complete data would be used with top priority in the interpolation of data missing. The existing Chinese Soil Database is in short of spatial attribution data, so it is better to use regression model than use spatial interpolation in the interpolation of missing data of soil database. In this study, PTFs from regression analysis could meet the requirement of data interpolation of bulk density, alkaline nitrogen and partial cation exchange capacity. Besides, some soil properties such as available potassium could be time limited, so the historical background of data sources should be considered in the application of PTFs.