The KPF SURFS-UP Survey I: Transmission Spectroscopy of WASP-76 b

TL;DR

The study introduces the KPF SURFS-UP Survey and a public pipeline to perform high-resolution transmission spectroscopy of ultra-hot Jupiters, demonstrated on WASP-76 b. Using three SCI spectra from KPF, blaze/continuum normalization, order stitching, telluric correction, and PCA cleaning, the authors extract planetary signals via cross-correlation with petitRADTRANS atmospheric templates. They detect Fe I, Ca II, and Na I with strong signals and observe a clear ingress–egress blue-shift asymmetry for Fe I, indicating day-to-night winds in deeper atmospheric layers, while Ca II and Na I show little asymmetry, suggesting higher-altitude dynamics. The results support a two-layer atmospheric structure for WASP-76 b and showcase KPF’s capability for detailed atmospheric characterization, foreshadowing a large UHJ survey with transmission, emission, and phase-resolved spectroscopy for 3D atmospheric studies and refractory-abundance constraints.

Abstract

We introduce the KPF SURFS-UP (Spectroscopy of the Upper-atmospheres and ReFractory Species in Ultra-hot Planets) Survey, a high-resolution survey to investigate the atmospheric composition and dynamics of a sample of ultra-hot Jupiters with the Keck Planet Finder (KPF). Due to the unique design of KPF, we developed a publicly available pipeline for KPF that performs blaze removal, continuum normalization, order stitching, science spectra combination, telluric correction, and atmospheric detection via cross-correlation. As a first demonstration, we applied this pipeline to a transit of WASP-76 b and achieved some of the highest signal-to-noise detections of refractory species in WASP-76 b to date (e.g., Fe I is detected at a SNR of 14.5). We confirm previous observations of an asymmetry in Fe I absorption, but find no measurable ingress-egress asymmetry in Na I and Ca II. Together, these results suggest variations within different layers of the atmosphere of WASP-76 b: neutral metals such as Fe I trace deeper regions with stronger asymmetries, while Na I and Ca II probe regions higher in the atmosphere where the ingress-egress asymmetries are weaker. These findings provide new insights into the complex atmosphere of WASP-76 b and highlight the power of using KPF for atmospheric characterization. More broadly, the KPF SURFS-UP Survey will observe a large sample of UHJs, using transmission, emission, and phase-resolved spectroscopy to characterize their refractory abundances, upper atmospheres, and 3D dynamics.

Figures (6)

• Figure 1: KPF spectra of WASP-76 extracted by version 2.7.1 of the KPF Data Reduction Pipeline. This plot shows only one of the three science spectra (SCI2), although all three spectra have similar looking spectra. Each color corresponds to a single diffraction order, and each row displays a distinct wavelength region for visual clarity. Before performing high-resolution atmospheric characterization, we performed several additional reduction steps, including blaze removal, continuum normalization, order stitching, combining the three science spectra, and telluric correction (see Section \ref{['reduction']}).
• Figure 2: A general overview of the reduction steps for our pipeline. The three science spectra (SCI1, SCI2, and SCI3) are processed independently through blaze removal (Section \ref{['sec:blaze-correction']}), continuum normalization (Section \ref{['sec:cont-norm']}), and order merging (Section \ref{['stitching']}), and then combined (Section \ref{['threespectra']}) before telluric correction (Section \ref{['tellurics']}) and the subsequent atmospheric analysis (Section \ref{['cc']}). While the specific implementation of each step (e.g., continuum normalization) may evolve with future improvements, these steps will likely all remain important for preparing KPF data for atmospheric characterization.
• Figure 3: An example of the continuum normalization steps for a single order of a SCI2 spectrum of WASP-76 on the red CCD with significant fringing. First, the extracted 1D spectrum (blue) is divided by the smooth lamp pattern to produce the blaze-corrected spectrum (green). By comparing the blaze-corrected flux to a constant flux of 1.0 (top black dashed line), one can see a broad residual trend in the blaze corrected spectrum (i.e., the flux at bluer wavelengths is higher than the flux at redder wavelengths). To remove this trend, we implemented an iterative spline method on a master spectrum (constructed from the median of all exposures; see Section \ref{['sec:cont-norm']}) to produce the continuum normalized spectrum in purple.
• Figure 4: The results of our reduction pipeline for the first ten orders on the green chip. The top panel displays the raw extracted spectra from SCI2 as a function of the orbital phase of WASP-76 b, with each row representing a different exposure in the stellar rest frame. The bottom panel shows the same wavelength coverage after applying our reduction pipeline and dividing the flux by the out-of-transit average. These residuals contain the atmospheric absorption signal of WASP-76 b, which can be extracted using cross-correlation.
• Figure 5: Left: Residual cross-correlation map for Fe I, Cr I, Na I, and Ca II from a single KPF transit of WASP-76 b, shown in the planet rest frame (0 km s$^{-1}$ is the green dashed line). The signal (in ppm) is plotted as a function of orbital phase (vertical axis) and radial velocity (horizontal axis), with each cross-correlation function (CCF) normalized by the out-of-transit average. The horizontal dashed lines mark the start of ingress (pink) and the end of egress (dark blue). In the absence of atmospheric dynamics, the planetary signal would be centered exactly at 0 km s$^{-1}$. Right: Fe I exhibits a clear ingress–egress asymmetry, while Na I and Ca II remain nearly vertical in velocity space with no measurable asymmetry. This difference points to altitude-dependent circulation differences: Fe I samples regions with stronger day-to-night winds, whereas Na I and Ca II probe higher altitudes where winds are weaker.