Three different statistical interpolation schemes respectively based on the principal component regression(PCR) of empirical orthogonal function(EOF),the iteration of EOF(IEOF) and the canonical variable regression(CVR) of principal component are adopted to interpolate the same missing data fields.Contrast experiments show that the interpolation precision of all kinds of schemes is associated with its parameters.If the pencentage of missing data station numbers is less than 60%,no matter what distributions these stations show,all schemes,especially CVR exhibit a satisfactory accuracy,and this character becomes more significant when data missing goes a longer time.