Estimating analysis error statistics using the TIGGE database

Analyses are often used for forecast verification. The validity of using analyses for verification is explored by comparing a year's worth of global operational analyses from 5 different operational centers, using the TIGGE data base. It is found that for many common fields, there are very large differences between the analyses, and these errors are both random and systematic. Statistical methods are proposed that may use multi-center analyses to determine when there is a general consistency between analyses such that they can safely be used for forecast verification.