Interpolation of Missing Values in AMSR-E Soil Moisture Data Using Modified SSA

Soil moisture data available from the Advanced Microwave Scanning Radiometer-Earth Observation System (AMSR-E) onboard the National Aeronautic and Space Administration's Aqua satellite have many inherent gaps. For a region in the Southeast United States, data are collected for spring 2006. This data set has nearly 28% missing data due to radio interference and instrument errors, just to mention a few. To address this issue, an improved singular spectral analysis (SSA)-based interpolation scheme is presented, where a lag covariance matrix is computed for a smaller spatial-temporal subset, as opposed to using the entire spatial grid for the computation of the lag covariance structure. An AMSR-E soil moisture data set with a size of 28 × 22 × 90 is used, and the corresponding results are compared with the ones obtained from the original SSA gap-filling method to validate the applicability of this method. It is shown that our approach provides an improvement over the existing method by utilizing the local variations in the observations for estimating the missing values and thus significantly improving the computational efficiency of the algorithm. It is also found that a spatiotemporal block of 11 × 11 × 28 is optimal for interpolation, where the resulting optimal block information is used to fill the real gaps in the experimental data set.