Non-differential water vapor estimation from SBAS-InSAR

Abstract

Water vapor is the most variable constituent in the atmosphere and plays an important role in climate studies, mesoscale meteorology modeling and numerical weather forecasting. Being able to penetrate clouds, interferometric synthetic aperture radar (InSAR) shows great potential in atmospheric water vapor mapping. But InSAR can only measure differential water vapor between two acquisitions. In this paper, we formulate a general framework by constructing the Gauss-Markov model and developing the estimation method to retrieve the non-differential water vapor from Small BAseline Subset InSAR (SBAS-InSAR). To address the rank-deficiency in the Gauss-Markov model, we propose a new constraint, i.e., the temporal mean of water vapor being invariant. Simulated and real data experiments are conducted to validate the effectiveness of the framework and the advantages of the proposed constraint. The results show that the new constraint can offer an estimation more robust than the two traditional ones, i.e., the temporal mean of water vapor being zero and single or multiple epoch water vapor referencing. In addition, we found that there exists a constant bias, which equals to the temporal mean of water vapors, between the solutions under the new constraint and that under the constraint of the temporal mean of water vapor being zero. Finally, the possible methods to evaluate the temporal mean of water vapor are discussed.