🔭 astrophysics

Updated Sensitivities of the Five STIS L-mode Gratings

This paper details the re-derivation of spectral sensitivities for the five STIS L-mode gratings following updates to standard star model atmospheres, resulting in new throughput tables that triggered a recalibration of historical datasets and demonstrated agreement within 1% across most spectral regions.

Original authors: Amy M. Jones, Svea Hernandez, Joleen K. Carlberg, Daniel Welty

Published 2026-02-17

📖 5 min read🧠 Deep dive

CC BY 4.0

Original authors: Amy M. Jones, Svea Hernandez, Joleen K. Carlberg, Daniel Welty

Original paper licensed under CC BY 4.0 (http://creativecommons.org/licenses/by/4.0/). ✨ This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer

The Big Picture: Updating the Telescope's "Ruler"

Imagine the Hubble Space Telescope as a giant, incredibly precise camera in space. Attached to it is a special tool called STIS (the Space Telescope Imaging Spectrograph). Instead of just taking pictures, STIS breaks light from stars into a rainbow (a spectrum) to tell us what those stars are made of and how bright they are.

To do this, STIS needs a perfect ruler (calibration) to measure the light accurately. For decades, this ruler was based on three "standard stars" (G191-B2B, GD 71, and GD 153). Astronomers treat these stars like the "gold bars" of the universe; if you know exactly how bright they should be, you can use them to measure everything else.

The Problem:
Scientists recently updated the theoretical models of these three "gold bar" stars. They realized their old math was slightly off—like realizing a gold bar is actually 2% heavier than they thought. Because the "gold bars" changed, the "ruler" (the sensitivity settings for the telescope) was now wrong. If they kept using the old ruler, every measurement of every other star in the universe would be slightly inaccurate.

The Solution:
This report describes the team's work to re-calibrate the ruler. They re-measured how the telescope responds to light using the new, more accurate models of the standard stars. They focused on five specific "gratings" (the prisms inside STIS that split the light), covering a huge range of colors from ultraviolet to near-infrared.

The Process: How They Fixed It

1. Gathering the Data (The "Test Drive")

The team went through the telescope's entire history, pulling up every observation they ever made of those three standard stars.

The Analogy: Imagine a car mechanic who wants to tune an engine. They don't just guess; they look at every test drive the car has ever taken over the last 20 years to see how it performed under different conditions.
They looked at data from different detectors (some sensitive to ultraviolet, some to visible light) and checked for "glitches" like cosmic rays (space dust hitting the sensor) or "hot pixels" (dead spots on the camera).

2. Cleaning the Mess (Data Preparation)

Space data is messy. The telescope moves, the temperature changes, and cosmic rays hit the sensors.

The Analogy: Think of trying to listen to a song on a radio while driving through a tunnel with static. The team had to "tune out" the static (cosmic rays), adjust for the temperature of the radio (detector temperature), and make sure the volume knob (gain) was set correctly.
They had to be very careful with one specific setting (G750L) because at very long wavelengths (red light), the light bounces around inside the detector like an echo in a cave, creating a "red halo" that messes up the measurement. They had to mathematically subtract this echo.

3. The Recalibration (The "New Ruler")

Once the data was clean, they compared what the telescope saw against what the new models said the stars should look like.

The Result: They created a new set of "Throughput Tables" (PHOTTAB). Think of these as the new instruction manual for the telescope's sensitivity.
The Precision: The new measurements from the three different stars agreed with each other to within 1%. That is incredibly precise. It's like three different people measuring a table with different rulers and all agreeing on the length within the width of a human hair.

4. The "Red Halo" Challenge (G750L)

The report highlights a specific difficulty with the G750L grating (which looks at red/infrared light).

The Metaphor: Imagine trying to catch a faint firefly in a dark room. If the firefly is too far away, it's hard to see. But if you look at it through a foggy window (the "red halo" effect), the light spreads out, making it look dimmer than it is.
The team had to be extra careful here because the standard stars are very faint in red light, and the "fog" (halo) makes the data noisy. Despite this, they managed to get a reliable calibration, though the "error bars" (uncertainty) were slightly larger at the very red end of the spectrum.

Why Does This Matter?

1. A New Standard for the Universe
Now that the "ruler" is fixed, every time Hubble looks at a star, a galaxy, or a nebula, the data is more accurate. This is crucial for the CALSPEC library, which is the master database of stellar brightness used by astronomers worldwide. If the master database is wrong, everyone's research is slightly off.

2. Historical Data Gets a Second Life
The report mentions that they didn't just fix future data; they went back and re-calibrated all the old data Hubble has ever taken with these modes.

The Analogy: It's like finding an old, blurry photo of your family, running it through a new AI upscaler, and suddenly seeing details you missed 20 years ago. Old observations of distant galaxies are now more accurate than they were when they were first taken.

3. Consistency
The team found that the new calibration works seamlessly across the different "prisms" (gratings). When they looked at the same star with different settings, the results matched up perfectly (within 1%). This means astronomers can stitch together data from different parts of the spectrum to get a complete picture of an object without worrying about "seams" or errors where the data overlaps.

The Bottom Line

This report is a "quality control" update. The scientists realized their reference stars had been updated by theory, so they went back, re-measured the telescope's sensitivity, and delivered a new, more accurate set of instructions. The result? Hubble's vision is sharper, its measurements are more precise, and our understanding of the universe's brightness is now more consistent than ever before.

1. Problem Statement

The absolute flux calibration of the Space Telescope Imaging Spectrograph (STIS) relies on theoretical models of three primary white dwarf standard stars: G191-B2B, GD 71, and GD 153.

The Trigger: Significant updates were introduced to the CALSPEC model atmospheres for these stars (specifically the release of CALSPEC v11). These updates included:
- Improved non-local thermodynamical equilibrium (NLTE) modeling.
- Updated stellar effective temperatures and surface gravities.
- Inclusion of metal line absorption (particularly in G191-B2B below ~2000 Å).
- Updates to the Vega flux normalization.
The Impact: The new predicted continuum fluxes differed from the models used in the original STIS calibration by up to 2–3%. Since the existing STIS sensitivity tables (PHOTTAB) were derived nearly two decades ago using older models (v4 and v5), they were no longer self-consistent with the updated CALSPEC library. This necessitated a re-derivation of the spectral sensitivities for the five STIS low-resolution (L-mode) gratings: G140L, G230L, G230LB, G430L, and G750L.

2. Methodology

The authors re-derived the sensitivities using a rigorous, multi-step process involving historical data and updated calibration pipelines:

Data Selection:
- Utilized high Signal-to-Noise (S/N) observations of the three primary standard stars taken over the entire lifetime of STIS.
- G191-B2B was observed only with CCD modes (G230LB, G430L, G750L) as it is too bright for the MAMA detectors.
- GD 71 and GD 153 were observed with all five L-modes.
- All observations used the widest long slit (52x2") to ensure photometric consistency.
Data Preparation & Correction:
- Data were processed using the CALSTIS pipeline.
- Time-Dependent Sensitivity (TDS) & Temperature (T): The pipeline was run twice (once with standard corrections, once with TDS/CTE corrections omitted) to derive observation-specific correction factors.
- Cosmic Ray Rejection: A strict threshold was applied; observations where >20% of pixels in the extraction region were rejected as cosmic rays were discarded (lowered to 12% for G750L/GD71 to remove outliers).
- Fringing (G750L): Spectra at wavelengths >7000 Å were corrected for fringing using contemporaneous fringe flats.
- Combination: Corrected net count rates were placed on a common wavelength scale and combined using a sigma-clipped mean to remove hot pixels and residual cosmic rays.
Sensitivity Derivation:
- The average corrected count rates ( $C_{\lambda}$ ) were divided by the updated CALSPEC v11 model fluxes ( $F_{\lambda,model}$ ) to generate sensitivity vectors ( $S_{\lambda}$ ).
- Spline Fitting: Initial sensitivities for each star were fitted with custom splines. Emission/absorption features (e.g., Lyman and Balmer series) were masked to avoid fitting artifacts where models do not perfectly match data.
- Averaging: The sensitivities from the three stars were averaged with equal weights to produce the final sensitivity curve for each grating.
- Conversion: These sensitivities were converted into integrated throughput curves ( $R_{\lambda}$ ) for the PHOTTAB reference files, accounting for the HST collecting area and infinite aperture/extraction box definitions.

3. Key Contributions

Updated PHOTTABs: Delivered new photometric throughput tables to the Calibration Reference Data System (CRDS) on April 7, 2022, and April 14, 2023.
Retriggered Recalibration: These updates triggered a re-calibration of all historical STIS datasets taken with L-mode gratings, ensuring the entire archive is consistent with the modern CALSPEC v11 flux scale.
Special Handling of G750L: Addressed specific challenges for the G750L grating, including fringing, lower S/N at long wavelengths, and the "red halo" effect (PSF wings), concluding that the standard 7-pixel extraction height remains robust despite these factors.
Comprehensive Validation: Provided a systematic validation of the new throughputs against the CALSPEC models and across overlapping wavelength regions of different gratings.

4. Results

Agreement with Models: The re-derived sensitivities show excellent agreement with the updated CALSPEC models.
- Overall: Residual discrepancies are typically < 1% across most of the spectral range.
- Discrepancies: Differences rise to ~1.5% in regions heavily impacted by hydrogen absorption (e.g., Balmer jump, Paschen jump) and up to ~4% at the extreme red end of G750L (>8000 Å) due to lower S/N and PSF width issues.
- Star-to-Star Consistency: Sensitivities derived from individual standard stars agree with each other to better than 1%, with larger discrepancies only in specific spectral features (e.g., G191-B2B metal lines, Balmer jumps).
Cross-Validation:
- Overlapping regions between gratings (e.g., G230L/G230LB, G430L/G750L) show agreement better than 1%.
- Independent validation using the helium white dwarf WD 0308-565 (not used in the derivation) confirmed the new sensitivities, with residuals generally within 2% outside of strong spectral lines and low S/N regions.

5. Significance

Flux Precision: This work ensures that STIS L-mode data products achieve the highest possible flux precision, aligning them with the state-of-the-art CALSPEC v11 flux scale.
Scientific Impact: The L-mode gratings provide continuous spectral energy distributions (SEDs) from the far-UV (1150 Å) to the near-infrared (1.027 µm). Accurate calibration is critical for:
- The CALSPEC flux standard library itself.
- Studies of stellar atmospheres, interstellar medium absorption, and extragalactic SEDs.
Archive Integrity: By updating the reference files, the entire historical archive of STIS L-mode data is now self-consistent, allowing for reliable comparisons between observations taken decades apart.
Future-Proofing: This report establishes a framework for handling future model updates, noting that while CALSPEC v12 has been released, the differences from v11 are minimal, and the current re-derivation remains valid.