A new ozone dataset generated using the Complete Data Fusion algorithm applied to MIPAS and IASI observation is presented, and validated against ozonesondes from WOUDC stations across multiple latitude bands. The fusion propagates limb information from the stratosphere to the troposphere, across nadir observations. The resulting dataset has enhanced numbers of degrees of freedom and reduced total errors, providing improved capability for investigating ozone stratospheric intrusion events.

The Complete Data Fusion is an a posteriori method to combine remote sensing products from independent observations of the same air mass. In this study, we extended the algorithm’s applicability to two-dimensional products, testing it with simulated ozone datasets from nadir and limb measurements. We demonstrated that the exploitation of the tomographic capabilities of future atmospheric sensors maximizes the information extracted from complementary datasets.

New variables are proposed for the retrieval products that are used in subsequent data fusion activities. These variables, in the linear approximation of the forward model, are independent of the a priori information used in the retrieval. Furthermore, they reduce to about one third the stored data volume with respect to the use of the standard products and provide an optimal compromise between completeness and efficiency of the stored information.

A new formula of the complete data fusion that, differently from the original one, does not contain matrices that can be singular is discussed. We show that the new formula is a generalization of the original one and analytically and numerically, using a real IASI ozone measurement, derive the errors made with the old formula when the generalized inverse of singular matrices is used. An operational version of the new formula that includes interpolation and coincidence errors is also provided.

Synergistic retrieval (SR) and complete data fusion (CDF) methods exploit the complementarity of coinciding remote-sensing measurements. We assess the performance of the SR and CDF methods on the basis of synthetic measurements of the FORUM and IASI-NG missions. In the case of perfectly matching measurements, SR and CDF results differ by less than 1 / 10 of the error due to measurement noise. In the case of a realistic mismatch, the two methods show differences in the order of their error bars.