Solar Acoustic Modes: Seeing Order in “Noise”

This visualization is a small, personal example of why spectral analysis is powerful: the raw signal looks messy, but in the right representation, the hidden structure becomes undeniable.

What I Computed

Starting from a Doppler-velocity data cube \(v(x,y,t)\), I computed a 3D Fourier transform across both spatial dimensions and time, and then formed a power spectrum:

\[ V(k_x,k_y,\omega)=\mathcal{F}\{v(x,y,t)\}, \qquad P(k_x,k_y,\omega)=|V(k_x,k_y,\omega)|^2 \]

The animation shows slices through this spectral cube. As \(\omega\) changes, rings and ridges emerge in \((k_x,k_y)\)-space. Those are the Sun’s acoustic resonances: coherent standing-wave power hiding inside data that otherwise feels like shimmering noise.

---

The Story: How Features “Appear”

If you only look at the Doppler maps in real space, everything seems complicated—granulation, evolving patterns, and fluctuations everywhere. But when you transform into frequency and wavenumber space, the physics becomes legible. Suddenly, the Sun looks less like chaos and more like an instrument: many modes superposed at once, each leaving a geometric fingerprint in the spectrum.

In the left panel, the 1D spectrum \(P(\omega)\) highlights dominant frequencies; in the right panel, the 2D spatial spectrum shows how power organizes by spatial scale. Watching the animation feels like turning a dial until the “right” notes come into focus. That moment—when the rings become obvious—is exactly the feeling that made me love applied math.

Two Practical Signal-Processing Lessons

(1) Windowing reduces spectral leakage. Sharp edges in the spatial domain inject artificial high-frequency power. Applying a smooth, radially symmetric window suppresses those boundary artifacts so the true modal structure stands out.

\[ w(r)= \begin{cases} 1-(r/R)^2, & r<R\\ 0, & r\ge R \end{cases} \qquad (r^2=x^2+y^2) \]

(2) Aggregation boosts signal-to-noise. By averaging/bucketing power in a way that respects symmetry (e.g., annuli in \((k_x,k_y)\)-space), weak but coherent features become robust. In industry terms: you reduce variance by summarizing along the right invariances.

---

Project Details

Author: Sajal Gupta

Year: 2020

Context: Math Methods homework (Problem 2)

Tech: Python, NumPy/SciPy/AstroPy/Bash, FFTs, Matplotlib

What it demonstrates: 3D spectral estimation \(P(k_x,k_y,\omega)\), leakage control via windowing, and feature discovery by transforming a noisy-looking signal into a representation where structure is visible.