I. Introduction
In passive microwave remote sensing, compared with aperture synthesis (AS), mirrored aperture synthesis (MAS) has higher spatial resolution [1], [2]; therefore, MAS has an application prospect for earth observation in geostationary orbit [3]. The existing methods for the brightness temperature (BT) reconstruction in MAS include the inverse cosine transform method [1], the impulse matrix method [4], [5], and the method with deep convolutional neural network (MAS-CNN) [6]. The impulse matrix method and MAS-CNN utilize the spatial impulse response matrix ( matrix) to establish a mapping relationship between BT and correlation of antenna outputs. Based on this mapping relationship, BT can be reconstructed. The quality of the matrix directly affects the quality of the BT reconstructed by these two methods. Therefore, the in-depth analysis of the matrix is crucial for improving BT reconstruction quality.