*3.2. Extraction h*

3.2.1. Multi-GNSS and Multi-Frequency SNR Sequence Extraction

In the article, multi-GNSS and multi-frequency SNR sequences are extracted by RTK-LIB software. Figure 7 shows the rising SNR sequences of G10 (S1C, S2L, and S5Q), R17 (S1C and S2C), E12 (S1C, S5Q, S6C, S7Q, and S8Q), and C14 (S2I, S6I, and S7I) satellites in the four GNSS systems on DOY 024 in 2017. Through data preprocessing, it is found that the SNR values of G10 (S1W and S2W) are the same, and that the SNR difference with other signal frequencies is significant. Therefore, these two types of SNR data are not used for the experimental in the article. Figure 7 shows the SNR sequence change in the elevation angle from low to high.

**Figure 7.** Multi-GNSS and multi-frequency SNR sequences: (**a**) DOY 024: GPS SNR sequence; (**b**) DOY 024: GLONASS SNR sequence; (**c**) DOY 024: Galileo SNR sequence; (**d**) DOY 024: BDS SNR sequence.

Figure 7 shows that the SNR sequence has strong oscillation at a low elevation angle. It is greatly affected by multipath at a low elevation angle, and the interference caused by the direct signal and the reflected signal is obvious. When increasing the elevation angle, the interference degree gradually decreases. Therefore, after removing the direct signal from the SNR sequence, this article mainly selects the SNR sequence of the reflected signal in the range of elevation angles of 5–30 degrees for snow depth retrieval. At the same time, it can be seen that the SNR sequences of the four GNSS systems at different frequencies have specific differences.

#### 3.2.2. SNR Sequence Data Processing

In the article, the composite SNR sequence is linearized first, and then the linearized SNR sequence is fitted by a cubic low-order polynomial to obtain the direct signal part. The SNR sequence of the reflected signal is obtained by subtracting the composite SNR sequence from the direct signal part. Figure 8 shows the processing of GPS S1C SNR data.
