A method for climate signal estimation from incomplete data

Typical approaches to climate signal estimation from data are susceptible to biases if the instrument records are incomplete, cover differing periods, if instruments change over time, or if coverage is poor. Here, a method is presented for obtaining unbaised, maximum-likelihood estimates of means, trends, or other desired climate signals given the available data from an array of fixed observing stations that report intermittently. The conceptually straightforward method follows a spatio-temporal mixed-model approach, making use of data analysis concepts that are well-known in the geophysical sciences. It performs well in the face of missing data problems, and is also helpful in dealing with common data heterogeneity issues and gross errors. Perhaps most importantly, the method facilitates quantitative error analysis of the actual signal being sought, which is often not available from typical approaches based on purely spatial analysis of the data. The method is used to estimate from rawinsonde data weak wind signals in the tropical lower stratosphere that are relevant to troposphere-stratosphere transport.