Towards Spectroscopically Detecting the Global Latitudinal Temperature Variation on the Solar Surface

TL;DR

This study explores spectroscopic methods to detect tiny latitudinal temperature variations on the solar surface, aiming to improve measurement precision to confirm theoretical predictions of temperature gradients.

Contribution

It identifies optimal spectral lines for high-precision temperature variation detection and demonstrates the potential to measure temperature differences as small as 1 Kelvin.

Findings

Strong T-sensitive lines are effective for precise measurements.

Weak lines suffer large fluctuations and should be avoided.

Achievable precision could detect temperature differences of about 1 Kelvin.

Abstract

A very slight rotation-induced latitudinal temperature variation (presumably on the order of several Kelvin) on the solar surface is theoretically expected. While recent high-precision solar brightness observations reported its detection, confirmation by an alternative approach using the strengths of spectral lines is desirable, for which reducing the noise due to random fluctuation caused by atmospheric inhomogeneity is critical. Towards this difficult task, we carried out a pilot study of spectroscopically investigating the relative variation of temperature (T) at a number of points in the solar circumference region near to the limb (where latitude dependence should show up if any exists) based on the equivalent widths (W) of 28 selected lines in the 5367--5393A and 6075--6100A regions. Our special attention was paid to i) clarifying which kind of lines should be employed and ii) how much precision would be accomplished in practice. We found that lines with strong T-sensitivity (|log W/log T|) should be used and that very weak lines should be avoided because they inevitably suffer large relative fluctuations (Delta_W/W). Our analysis revealed that a precision of Delta_T/T ~ 0.003 (corresponding to ~15K) can be achieved at best by a spectral line with comparatively large |log W/log T|, though this can possibly be further improved if other more suitable lines are used. Accordingly, if many such favorable lines could be measured with sub-% precision of Delta_W/W and by averaging the resulting Delta_T/T from each line, the random noise would eventually be reduced down to <~1K and detection of very subtle amount of global T-gradient might be possible.