Revisiting Ca II Activity Indices in FGK Stars: Systematic Biases in Infrared Triplet Measurements

Synthetic-template subtraction methods used to measure chromospheric activity in FGK stars often yield systematically negative Ca II infrared triplet (IRT) residual indices, which this study investigates to clarify their origin.

How it works

The researchers investigate systematic negative biases in R+ IRT by measuring activity indices (R+) for both the Ca II H&K and IRT lines across solar-like stars from LAMOST DR9, MaStar, and XSL DR3 using a uniform framework. The core methodology involves defining the line-core index as:

The line-core index is defined as R = 1/2∆λ λ0+∆λ / (λ0−∆λ) F(λ) C(λ) dλ, where λ0 is the line center, ∆λ = 0.5 ˚A, F(λ) is the flux spectrum, and C(λ) is the fitted local pseudocontinuum.

The template-subtracted residual core-flux index (R+) is then defined as:

The template-subtracted residual core-flux index is defined as R+ = 1/2∆λ λ0+∆λ / (λ0−∆λ) [Fobs(λ) / Cobs(λ) − Ftem(λ)/Ctem(λ)] dλ.

The study employs three distinct observational data sets:

1. LAMOST DR9 Low-Resolution Spectra, providing the primary large-sample data set.

2. SDSS DR17 MaStar Stellar Library (MaStar), offering an independent survey-level comparison.

3. XSL DR3 (XSL), providing higher-resolution spectra for examining observed–template mismatches in line profiles.

Synthetic spectra are constructed using one-dimensional stellar atmospheres under the local thermodynamic equilibrium (LTE) assumption and very high intrinsic spectral resolution (λ/∆λ ≥ 300,000). These templates span broad grids in Teff, log g, [Fe/H], and [α/Fe] through various synthesis configurations. The A1 template set is adopted as the fiducial configuration.

The analysis systematically assesses several potential sources of the negative bias:

1. Atmospheric-parameter offsets (Teff, log g, [Fe/H], [α/Fe]).

2. Treatment of the instrumental line-spread function (LSF).

3. Propagated measurement uncertainties (including RV uncertainties and SNR).

The researchers found that observational effects contribute to scatter but do not explain the systematic negative bias in R+ IRT. The results suggest that the negative bias most likely arises because photospheric templates underestimate the depth of the IRT cores, likely owing to missing chromospheric structure and, to a lesser extent, NLTE effects.

The study employed an empirical approach to mitigate this mismatch:

An empirical increase in the adopted microturbulent velocity deepens the synthetic IRT cores and partially mitigates the negative offset.

This adjustment was tested by increasing Vmic from Vtmic to Vtmic + 2 km s−1. This empirically shifts nearly all XSL R+8542 values above zero, moving the median R+8542 for LAMOST up by 0.016 and for MaStar up by 0.010, while R+HK decreases slightly. The authors emphasize that this is an empirical mitigation rather than a physical solution intended to replace chromospheric or NLTE modeling.

The investigation into template dependence revealed that the absolute scale of R+ depends on the synthesis configuration:

The largest variations arise from the radiative transfer code, followed by the model atmosphere, while the solar abundance scale and line list have comparatively smaller effects.

Specifically, the radiative transfer code is the dominant source of configuration-dependent variation in the derived indices, followed by the model atmosphere. The solar abundance scale and line list produce comparatively small biases and dispersions.

The study concludes that negative IRT residual indices should not be interpreted literally as negative chromospheric emission, but rather as evidence that the adopted photospheric template does not fully reproduce the observed inactive line core. This discrepancy is linked to structural differences in model atmospheres, where models lacking a chromospheric temperature inversion may underpredict the Ca II population, leading to weaker line-core absorption in the synthetic templates and making the observed line cores appear deeper by comparison. NLTE effects are also noted as contributing to this mismatch.

The final conclusion is that while observational systematics contribute to scatter, they cannot account for the systematic negative offset. The persistence of negative R+8542 values across surveys indicates that "template physics and synthesis assumptions may play an important role.

Here are specific improvements to AI systems derived from this scientific paper, along with descriptions of what those improved systems could achieve:

) 1. Improve Data Interpretation and Bias Detection in Spectroscopic Surveys:

The core finding is that negative Ca II IRT residual indices are not purely physical but arise from template-dependent systematic biases (template mismatch).

"The results instead suggest that the negative bias most likely arises because photospheric templates underestimate the depth of the IRT cores, likely owing to missing chromospheric structure and, to a lesser extent, NLTE effects."

"The absolute scale of R+ depends on the adopted synthesis configuration. The largest variations arise from the radiative transfer code, followed by the model atmosphere, while the solar abundance scale and line list have comparatively smaller effects."

AI System Improvement: Develop an advanced spectroscopic analysis pipeline capable of performing a multi-stage diagnostic check on any activity index measurement derived via synthetic-template subtraction.

Capabilities of Improved AI System:

1. Automatic Source Identification: When analyzing survey data (LAMOST, MaStar, XSL), the AI system would automatically flag indices with negative residuals as potentially systematically biased rather than physically impossible.

2. Configuration Sensitivity Mapping: The system would quantify the sensitivity of an index (e.g., R+8542) to specific synthesis parameters (Code, Model Atmosphere, Line List). It could output a confidence score for any derived value based on how far the current template configuration deviates from the fiducial A1 set.

3. Cross-Survey Calibration Recommendation: The AI could analyze the systematic offsets between different surveys (e.g., LAMOST vs. MaStar) and recommend specific correction factors or scaling adjustments based on which physical parameters ([Fe/H], log g) are most strongly driving the discrepancy, rather than just applying a uniform shift.

2. Improve Template Generation and Synthesis Fidelity:

The paper emphasizes that the choice of synthesis configuration (RT code, atmosphere model) significantly impacts the absolute scale of the derived index.

The dominant source of configuration-dependent variation is the radiative transfer code, followed by the model atmosphere, while the solar abundance scale and line list have comparatively smaller effects.

Maintaining a consistent synthesis setup is therefore important when comparing activity indices within or across large spectroscopic surveys.

AI System Improvement: Implement a modular, configurable synthetic spectrum generator that allows for on-the-fly testing of different physical modeling choices.

Capabilities of Improved AI System:

1. Automated Configuration Selection: The AI could ingest observed spectral data and automatically test the performance of multiple template configurations (e.g., comparing SPECTRUM/MARCS vs. MOOG/ATLAS9) to determine which yields the lowest residual scatter or best correlation with empirical activity indices, effectively self-calibrating the template set for that specific data regime.

2. Parameter Sensitivity Profiling: The system could generate a sensitivity map showing exactly how much a change in one parameter (e.g., [Fe/H] uncertainty) propagates into the final R+ index, allowing researchers to prioritize which atmospheric parameters need the most precise constraints during template construction.

3. Improve Empirical Mitigation Strategy for Survey Data:

The paper introduces an empirical fix using microturbulent velocity as a pragmatic solution, acknowledging its limitations but highlighting its utility for survey analysis.

An empirical increase in the adopted microturbulent velocity deepens the synthetic IRT cores and partially mitigates the negative offset.

"This empirical correction is therefore primarily applicable to the IRT cores of solar-like FGK stars and, in its current atmospheric-parameter-independent implementation, is less suitable for the H&K."

AI System Improvement: Integrate a learned, data-driven empirical adjustment module that predicts optimal parameter tweaks based on survey characteristics.

Capabilities of Improved AI System:

1. Adaptive Parameter Tuning: The system would learn from large sample statistics (LAMOST DR9) which specific parameter adjustments (like Vmic = Vtmic + 2 km/s) most effectively shift the distribution of negative indices toward zero, providing a best-guess empirical correction tailored to the specific survey instrument or data release being used.

2. Mitigation Strategy Recommendation: Instead of just applying a fixed correction, the AI could analyze whether the mitigation strategy is more effective for H&K lines versus IRT lines (as suggested by Figure 6), guiding users on which index to prioritize when dealing with negative residuals.

4. Improve Interpretation of Physical vs. Artifactual Results:

The paper moves beyond simply stating a bias, suggesting deeper physical causes (chromospheric structure/NLTE) and clearly delineating artifacts from physics.

Negative R+ 8542 values are consistent with a mismatch between the observed and synthetic line cores.

"This discrepancy is unlikely to be explained solely by the choice of convolution kernel or by resolution mismatch, as reproducing the XSL core depth would require degrading the templates to approximately twice the XSL resolution (∼ 20000). Therefore, the discrepancy in the line cores may stem from deeper underlying physical causes."

AI System Improvement: Develop a Physics-Artifact Classifier module that assesses residual patterns against known physical models.

Capabilities of Improved AI System:

1. Residual Pattern Classification: The AI would analyze residuals (R+ vs. Teff, log g, [Fe/H]) and classify them into categories: (a) Noise/Uncertainty dominated, (b) Template Mismatch (suggesting missing chromospheric structure), or (c) NLTE effects.

2. Physical Hypothesis Generator: Based on the classification of a negative residual trend in the IRT lines, the AI could generate a prioritized list of physical hypotheses for human review, such as Hypothesis: Absence of Chromospheric Temperature Inversion or Hypothesis: NLTE effect due to low metallicity.

5. Improve Robustness Against Survey-to-Survey Differences:

The paper highlights that negative R+ values are found across all surveys (LAMOST, MaStar, XSL), suggesting the problem is systemic to template subtraction rather than individual datasets.

Negative R+ 8542 values are present in all three surveys, indicating that the phenomenon is not confined to a single survey or dataprocessing pipeline.

AI System Improvement: Implement a meta-analysis layer that explicitly models and quantifies inter-survey systematic offsets.

Capabilities of Improved AI System:

1. Systematic Offset Quantification: The AI could use the cross-matched samples (Figure 5) to calculate and report the precise systematic offset between any two survey pipelines (e.g., LAMOST vs. MaStar), providing a quantitative metric for combining results safely, rather than relying on simple linear fits of individual datasets.

Synthetic-template subtraction is widely used to measure chromospheric activity in large spectroscopic surveys. However, many solar-like FGK stars show systematically negative Ca II infrared triplet (IRT) activity indices. We investigate this effect using solar-like stars from LAMOST DR9, MaStar, and XSL DR3, measuring activity indices (R+) for both the Ca II H&K and IRT lines within a uniform framework. We find that observational effects contribute to the scatter but do not explain the systematic negative bias in R+ IRT. This bias mainly arises because radiative-equilibrium photospheric models lack the chromospheric layers, which exhibit a temperature rise and provide additional Ca II IRT opacity. This additional opacity makes the observed IRT cores deeper than those in the synthetic photospheric profiles. NLTE effects may contribute but are unlikely to dominate. The different behavior of the H&K and IRT diagnostics can be related to their different source-function behavior: the stronger chromospheric source-function enhancement of H&K makes their cores more likely to appear in emission. Increasing the adopted microturbulent velocity partially mitigates the negative offset, serving as an empirical compensation for missing atmospheric physics. In addition, R+ values derived from different synthesis configurations show systematic offsets but generally preserve strong linear correlations, indicating that they can be cross-calibrated. These results clarify the origin of negative Ca II IRT residual indices and help interpret template-dependent systematics in chromospheric activity measurements.