# 1/f Noise Analysis for FAST HI Intensity Mapping Drift-Scan Experiment

**Authors:** Wenkai Hu, Yichao Li, Yougang Wang, Fengquan Wu, Bo Zhang, Ming Zhu, Shifan Zuo, Guilaine Lagache, Yinzhe Ma, Mario G. Santos, Xuelei Chen  
**arXiv:** [2109.06447](https://arxiv.org/abs/2109.06447)  
**Date:** 2021-09-14  
**Summary page:** https://www.summarizepaper.com/en/paper/2109.06447v1/

## Key points

- The study investigates the 1/f noise of the Five-hundred-meter Aperture Spherical Telescope (FAST) receiver system using drift-scan data from an intensity mapping pilot survey.
- The FAST has a large aperture and is equipped with multiple feeds and receivers, making it ideal for conducting large surveys.
- The observations were carried out with the L-band 19-beam feed system and the associated cryogenic receiver.
- All 19 beams have 1/f fluctuations with similar structures, which can be important for HI intensity mapping.
- Data was collected by the FAST Hi intensity mapping pilot experiment during risk-shared observations made in May 2019.
- Both temporal and two-dimensional power spectrum densities were estimated, showing that correlations directly seen in the time series data at low frequency $f$ are associated with sky signal, likely due to a coupling between foregrounds and system response.
- To remove foregrounds, Singular Value Decomposition (SVD) was used to subtract the strongest components of measured noise power significantly reducing it by up to twenty modes subtraction; thus reducing knee frequency of noise in a ten MHz band to $1.8 \times10^{-3}\Hz$, well below thermal noise over a scale of five hundred seconds.
- A simple noise diode frequency-independent calibration of receiver gain at eight-second intervals did not affect results as the 1/f noise had a knee frequency of $(f_{k}) \sim$ 6 $\times$ 10$^{-4}$Hz and time and frequency correlation spectral indices $(\alpha) \sim 0.65$, $(\beta) \sim 0.8$ after SVD subtraction of thirty modes respectively .
- Residual 1/f noise can bias HI power spectrum measurement by up to ten percent; making it important to consider in HI intensity mapping experiments conducted using FAST receiver systems.
- The study provides valuable insights into characterization and mitigation of 1/f noise crucial for current optical surveys.

## Summary

This study investigates the 1/f noise of the Five-hundred-meter Aperture Spherical Telescope (FAST) receiver system using drift-scan data from an intensity mapping pilot survey. The FAST has a very large aperture and is equipped with multiple feeds and receivers, making it ideal for conducting large surveys. The observations were carried out with the L-band 19-beam feed system and the associated cryogenic receiver. All the 19 beams have 1/f fluctuations with similar structures, which can be important for HI intensity mapping. The data used in this work was collected by the FAST Hi intensity mapping pilot experiment during risk-shared observations made in May 2019. Five strips spanning 125◦ < R.A. < 325◦ and 25.4◦ < Dec. < 26.7◦ were drift scanned in five consecutive days; each strip observed for 13.3 hours with a frequency range from 1.0 GHz to 1.5 GHz. Both temporal and two-dimensional power spectrum densities were estimated, showing that correlations directly seen in the time series data at low frequency $f$ are associated with sky signal, likely due to a coupling between foregrounds and system response. To remove foregrounds, Singular Value Decomposition (SVD) was used to subtract the strongest components of measured noise power significantly reducing it by up to twenty modes subtraction; thus reducing knee frequency of noise in a ten MHz band to $1.8 \times10^{-3}\Hz$, well below thermal noise over a scale of five hundred seconds. The two-dimensional power spectra showed that variations restricted to a small region in time-frequency space could be suppressed with SVD mode subtraction; hence residual uncorrelated noise after SVD mode subtraction was uncorrelated in frequency as well. A simple noise diode frequency-independent calibration of receiver gain at eight second intervals did not affect results as the 1/f noise had a knee frequency of $(f_{k}) \sim$ 6 $\times$ 10$^{-4}$Hz and time and frequency correlation spectral indices $(\alpha) \sim 0.65$, $(\beta) \sim 0.8$ after SVD subtraction of thirty modes respectively . The study concluded that residual 1/f noise can bias HI power spectrum measurement by up to ten percent; making it important to consider in HI intensity mapping experiments conducted using FAST receiver systems . The results were discussed in detail in section five while section six summarized findings providing valuable insights into characterization and mitigation of 1/f noise crucial for current optical surveys .