Improving ATMS Remapping Accuracy Using Adaptive Window and Noise-Tuning Method in Backus-Gilbert Inversion

One of the data fusion issues for observations from multiple spaceborne microwave sensors is the nonuniform spatial resolution. Although the Backus-Gilbert inversion (BGI) algorithm has long been used for the Advanced Technology Microwave Sounder (ATMS) antenna pattern matching, previous studies showed that it has difficulty in accurate remapping from the coarser to the finer observations. Since BGI tends to enhance the data's high spatial frequency components including both information and noise, it is a challenge to increase the spatial resolution while maintaining an acceptable noise level. This study unveils that the main cause of this issue is the insufficiency of the information provided by the conventional fixed reconstruction window. An adaptive window method is applied to provide sufficient information for the reconstruction at each scan position. In addition, a new noise tuning method is proposed to eliminate the scan-angle-dependent features in the noise caused by the sensor's cross-track scanning manner. Results from simulations and NOAA ATMS data show that compared to the fixed window, the new method can significantly reduce the bias stemming from the resolution difference. The issue of the deterioration of the resolution enhancement capability near the scan edge in the fixed window method has been largely ameliorated. The overall root-mean-square error is declined by 30%. The new noise tuning method is capable of suppressing the noise level at around 0.6 K over scan.

Automated summary

One of the challenges in fusing observations from multiple spaceborne microwave sensors is the nonuniform spatial resolution. This study proposes an adaptive window method to provide sufficient information for reconstruction and a new noise tuning method to eliminate noise caused by the sensor's scanning manner. Results show significant reduction in bias and improvement in resolution enhancement capability.