The Korean Space Telescope (KST): A Dedicated Space Observatory for the Dark Universe

Author(s)

Jee, M. James, Hwang, Ho Seong, Lim, Dongwook, Kim, Minjin, Im, Myungshin, Hong, Sungwook E., Lee, Ho-Gyu, Yang, Soung-Chul, Parkinson, David, Shafieloo, Arman, Jiang, Jun-Qian, Heo, Haeng Pal, Ryu, Dongok, Kihm, Hagyong, Noh, Hyung Yun, Kang, Wooyong, Kim, Youngsun

Abstract

The Korean Space Telescope (KST) is a next-generation 1.4-meter wide-field space observatory, a dedicated probe of the dark sector. Its single instrument provides diffraction-limited visible imaging (~0.15 arcsec in the optical) together with simultaneous near-infrared (R ~ 100) slitless spectrophotometry, so that every detected source is observed in the NIR by the survey itself, with spectral information delivered at a resolution set by its brightness (full R ~ 100 typing at the bright end, synthetic medium-band or broadband photometry below it) and with no separate follow-up, a capability decisive for time-domain science. With it, KST addresses the three outstanding questions of dark physics in concert: the dynamics of dark energy, the nature of dark matter, and gravitational-wave dark sirens as independent cosmological rulers. KST combines measurements of expansion and structure growth to test dark-energy models. A two-tiered "Wedding Cake" strategy (2,000 deg^2 Wide plus 10 deg^2 cluster-targeted Deep Survey) combines weak lensing, baryon acoustic oscillations, and a near-infrared-standardized Type Ia supernova Hubble diagram to constrain dark energy at its pivot redshift (sigma(w_p) ~ 0.03); maps dark matter from cluster cores to galactic substructure to test cold dark matter against warm and self-interacting alternatives; and builds a deep near-infrared host catalog for a distance-ladder-independent H_0 from dark sirens. The same data yield a Legacy Science portfolio (galaxy evolution, active galactic nuclei, exoplanets, and the Milky Way) at no additional observing cost. As Korea's first dedicated astrophysics space mission, KST occupies a niche complementary to the Vera C. Rubin Observatory, Euclid, and Roman.

Figures

Caption

Conceptual rendering of the Korean Space Telescope (KST) in its operational orbit. The spacecraft design leverages the high-resolution imaging and orbital stability heritage of the Korean Multi-Purpose Satellite (KOMPSAT) series \citep{Lee2021KOMPSAT,Seo2020KOMPSATIQ} to support a 1.4-meter wide-field astronomical surveyor.

Caption

\textbf{Aperture optimization for the imaging channel.} The survey grasp (etendue $\propto D^{-1}$; grey dashed) \textit{decreases} with aperture, while the density of cleanly resolved and deblended galaxies in the visible (VIS) imaging channel (blue) \textit{increases} as the PSF shrinks below the faint-galaxy size and then saturates. Their product, the rate at which science-usable galaxies are surveyed (red), has a genuine interior optimum at $D\approx1.4$~m: the aperture whose $\sim0.15''$ visible PSF matches the $\theta_g\sim0.15''$ scale that must be resolved for faint-galaxy shape measurement and deblending. Below this the PSF blends the faint population and degrades shapes; above it, survey grasp is spent resolving already-resolved sources. Curves are peak-normalized; the construction is an illustrative design rationale, with the model and its inputs given in Section~\ref{sec:etendue}. The near-infrared channel provides $R\sim100$ spectrophotometry (photometric redshifts and SEDs), not imaging, and is not what sets the imaging optimum.

Caption

\textbf{Schematic of the KST dichroic beam splitter and detector mosaics.} The common incident beam is separated into simultaneous VIS and NIR channels. \textbf{(a)} Four CMOS detectors form a $4\times1$ mosaic, with one fixed $g$, $r$, $i$, or $z$ filter per detector. \textbf{(b)} Twenty-seven square H2RG detectors form a $9\times3$ mosaic beneath an LVF whose transmission wavelength varies continuously from 1.1 to 1.8~$\mu$m along the scan direction, providing $R\approx100$ spectral mapping. Green outlines mark the identical $1^{\circ}\times0.25^{\circ}$ science fields; hatched regions indicate detector area outside the adopted field. Both mosaics are drawn on a common scale using the 19~m effective focal length and the specified pixel pitches. Detector gaps and H2RG reference-pixel borders are omitted. Beam routing and channel orientation are schematic, and relay optics are not shown.

Caption

Illustrative comparison of spatial sampling in KST and SPHEREx spectral cubes. The unobscured front planes show the same illustrative galaxy with fine and coarse sampling, while the offset planes and arrows indicate the wavelength dimension. The labels give the nominal detector pixel scales: $0.195''$/pixel for KST's NIR channel and $6.2''$/pixel for SPHEREx. Colors distinguish schematic wavelength slices, with the same underlying morphology used throughout each stack. The images provide a qualitative comparison of spatial sampling; they do not incorporate instrument-specific PSF convolution, noise, or wavelength-dependent galaxy morphology.

Caption

\textbf{KST sensitivity.} \textbf{(a)} $5\sigma$ limiting AB magnitude versus wavelength for a 1-hr exposure. Filled circles show KST point-source depths in $g,r,i,z$ from a first-principles exposure-time calculator (ETC) calibrated to reproduce the tabulated depths (Table~\ref{tab:kst_specs}); open circles show the corresponding $0.3''$-galaxy depths. The orange bar marks the NIR LVF per-resolution-element ($R\sim100$) continuum depth across $1.1$--$1.8~\mu$m, which is naturally shallower than broadband imaging. Grey squares show the Euclid Wide survey for comparison (filled: point source; open: extended VIS). \textbf{(b)} Point-source $5\sigma$ depth versus integration time for the four VIS bands; the dotted line and filled points mark the 1-hr survey depth. The ETC assumes a 1.4-m aperture with central obscuration, CDS read noise $<5\,e^-$, dark current, and a conservative zodiacal$+$Earthshine sky model (bounding the low-Earth-orbit case); the effective per-band throughput is calibrated to the point-source depths of Table~\ref{tab:kst_specs}.

Caption

\textbf{Spectroscopy without external follow-up.} Schematic kilonova light curve illustrating the finite spectroscopic window (green band: the source is brighter than the spectroscopic limit, dashed line). The survey detects the transient near peak (green point), and its $R\sim100$ spectrum accumulates during the same KST scan (blue circle). An externally triggered spectrum (orange) arrives only after trigger, scheduling, and weather latencies of days, by which time the kilonova has faded far below the spectroscopic limit and the opportunity is lost. The brightness scale is schematic.

Caption

\textbf{LVF dispersion orientation.} Two alternative design orientations, fixed at integration, for the LVF's wavelength gradient on the $1^\circ\times0.25^\circ$ focal plane (color shows the $1.1$--$1.8~\mu$m mapping; white lines, the $9\times3$ NIR detector mosaic; arrow, the scan direction). \textbf{(a)} Dispersing along the long axis (baseline) gives a gentle gradient and keeps the vignetted edges in the cross-scan direction, at the cost of a long scan to complete each spectrum. \textbf{(b)} Dispersing along the short axis completes a spectrum in a $4\times$ shorter scan and crosses fewer detector gaps, but steepens the gradient and places the vignetted edges on the extreme wavelength channels.

Caption

\textbf{Time-delay $H_0$ from cluster-lensed supernovae.} Projected fractional precision on $H_0$ versus the number of multiply imaged SNe~Ia with measured time delays, adopting the per-system precision of $1$--$2\%$ enabled by the standard-candle breaking of the mass-sheet degeneracy (shaded band), with an illustrative $0.4\%$ systematic floor added in quadrature; the dotted curve shows the purely statistical $2\%/\sqrt{N}$ scaling. Horizontal lines mark current published precisions: the six-system lensed-quasar measurement \citep{Wong2020}, the SH0ES distance ladder \citep{Riess2022}, and the \textit{Planck} CMB inference \citep{Planck2020}. At the benchmark sample of 20 systems (vertical line), KST reaches $\sigma(H_0)/H_0\approx0.5$--$0.6\%$, a sub-percent, distance-ladder-independent measurement sufficient to adjudicate the Hubble tension.

Caption

\textbf{The error budget, not the statistics, sets the constraint.} Forecast $68\%$ confidence contours in the $w_0$--$w_a$ plane for the \textit{same} KST survey under three error budgets, from the systematics-marginalized Fisher analysis described in the text: statistical precision only (blue); the KST error budget, with its internally measured shear calibration, mean-redshift, and NIR supernova systematics (red); and an otherwise identical survey carrying a conventional broadband systematics chain (orange dashed). Contours are centred on the $\Lambda$CDM fiducial (star); the prior widths defining the two chains are stated in the text. Marginalizing over systematics costs the conventional chain a factor $\sim$3 in figure of merit relative to KST's. The KST-error-budget contour corresponds to $\sigma(w_0)=0.096$, $\sigma(w_a)=0.36$, FoM $=79$ ($\sigma(w_p)\approx0.035$ at the pivot), computed for the current survey ($5{,}000$ SNe to $z\approx1$, $n_{\rm eff}=57$, $(1+z)$-scaled redshift-bias priors); the bridge to the full-probe headline forecasts is given in the text.

Caption

\textbf{The double-radio-relic constraint on dark-matter self-interactions.} Posterior probability density of the self-interaction cross-section $\sigma/m$ from eleven merging clusters hosting symmetric double radio relics (thin curves), and their stacked constraint (thick curves; the smoothed stack in orange). The dashed and dash-dotted lines mark the resulting upper limits, $\sigma/m<0.22~\mathrm{cm^2\,g^{-1}}$ ($68\%$) and $<0.63~\mathrm{cm^2\,g^{-1}}$ ($95\%$), at the $\sim$1000--2000~km\,s$^{-1}$ collision velocities characteristic of clusters. KST's diffraction-limited lensing will extend this method from eleven systems to the $\gtrsim$1000 merging clusters of the Deep and Wide Surveys, projecting to $\sigma/m\lesssim0.02~\mathrm{cm^2\,g^{-1}}$. Adapted from \citet{Jee2026}.

Caption

\textbf{Primordial-black-hole dark matter.} Microlensing upper limits on the fraction of dark matter in compact objects, $f_\mathrm{PBH}=\Omega_\mathrm{PBH}/\Omega_\mathrm{DM}$, as a function of PBH mass; the region above each curve is excluded. Grey curves and shading show approximate current constraints from microlensing surveys toward the Magellanic Clouds and the Galactic bulge (EROS/MACHO, OGLE, Subaru/HSC), which already exclude PBHs as the entirety of the dark matter across the planetary-to-stellar range, while the asteroid-mass window (green) remains microlensing-transparent. The blue curve shows the projected KST reach: its high-cadence, space-based monitoring of $>$$10^{8}$ bulge and Magellanic stars tightens the limit to $f_\mathrm{PBH}\lesssim10^{-2}$ across $\sim10^{-4}$--$10\,M_\odot$ and extends coverage toward lower masses. The current limits are indicative only; a quantitative comparison requires the published exclusion curves.

Caption

\textbf{Dark-siren forecast for the Hubble constant.} Projected fractional precision on $H_0$ versus the number of standard-siren events, under the idealized $\sigma(H_0)/H_0\propto N^{-1/2}$ scaling anchored to the GWTC-3 combined result ($\sim$15\% from a few tens of events; \citealt{AbbottGWTC3cosmo2023}); the single bright siren GW170817 is shown for reference. KST enables the dark-siren channel by providing the complete NIR host catalogue: $\sim$hundreds of events in the O4/O5 era reach the few-percent level, and the next-generation-detector era approaches the $\sim$1\% needed to arbitrate the Hubble tension. The scaling is a statistical idealization; see Section~\ref{sec:forecast} for the forecasting methodology.

Caption

The cosmic star formation history; the cyan-shaded region is the redshift range where KST plays an important role. Adapted from \citet{Madau2014}.

Caption

Example of 2D imaging and spectrophotometric decomposition of a high-$z$ AGN using \texttt{GalfitS}. Combining the high spatial resolution of NIR imaging with the spectrophotometric capabilities of the LVF enables robust disentanglement of the host galaxy from the central nucleus, providing detailed constraints on host galaxy properties \cite[][]{li2026}.

Caption

Comparison between the relative strength of \text{Fe\,\textsc{ii}} ($R_{\rm Fe\,II}$) and the FWHM of broad H$\beta$ for low-$z$ QSOs, adapted from \citet{shen2014}. The color coding denotes the equivalent width of [\text{O\,\textsc{iii}}], indicating that it traces EV1 along with $R_{\rm Fe\,II}$.

Caption

Histogram of confirmed exoplanets by discovery year and detection method. The annual number of confirmed exoplanets is shown, color-coded by detection method. Transit surveys dominate the overall discovery yield, with radial-velocity surveys providing the second-largest contribution, while microlensing, imaging, astrometry, and other methods probe complementary regions of parameter space. The rapid increase in confirmed planets in the mid-2010s illustrates the transformative role of space-based missions such as Kepler and K2. Adapted from the NASA Exoplanet Archive.

Caption

Light curve of the planetary microlensing event OGLE-2016-BLG-1195. The brief anomaly near peak magnification indicates the presence of a planetary companion. Adapted from \citet{Bond2017}.

Caption

Distribution of known exoplanets in the planet-mass versus semimajor-axis plane, color-coded by detection method. Microlensing occupies a complementary discovery space relative to transit and radial-velocity surveys by probing planets at wider separations, particularly cold planets located near and beyond the snow line. This sensitivity makes microlensing a key method for studying planetary populations in the regime most relevant to giant-planet core formation. Adapted from \citet{Mroz2023}.

Caption

Projected KST microlensing planet yield as a function of planet mass (illustrative; per logarithmic mass interval), scaled from the demonstrated yields of ground-based surveys and space-based bulge-survey predictions. KST's high cadence and stable space-based photometry extend sensitivity down to approximately an Earth mass, while deeper observations and reduced blending in crowded bulge fields raise the overall detection efficiency above current ground-based surveys; an additional $\lesssim100$ free-floating planets are expected from short-timescale events.

Caption

Density contours of RC stars in the $(K_{\rm s}, u-g)_{0}$ color--magnitude diagram (CMD), illustrating the population-dependent structure of the Galactic bulge. The bluer RC population, associated with a more metal-poor component, shows a single clump consistent with a relatively spheroidal distribution, while the redder RC population, associated with a more metal-rich component, shows a double RC feature that traces the X-shaped bulge. This separation was enabled by the broad wavelength baseline of optical-to-NIR photometry. The optical-to-NIR space-based observations of the KST will be critical for extending such bulge population and structure studies to more crowded and highly reddened regions of the inner Milky Way. Adapted from \citet{Lim2021}.

Caption

The KST two-tier ``wedding-cake'' survey. Tiers are ordered by sky area (widest at the base) and by observing cadence (increasing upward): a $2{,}000~\text{deg}^2$ single-epoch Wide Survey for weak lensing and large-scale structure, a $10~\text{deg}^2$ cluster-targeted Deep Survey at 5-day cadence (in $\sim$8-month seasons) for supernova cosmology, and a $\sim1~\text{deg}^2$ Galactic-Bulge Microlensing campaign at 15-minute cadence. Call-outs list the primary science from each tier. The bottom bar shows the allocation of the 6-year prime mission across the four programs (Wide, Deep, Microlensing, and Target-of-Opportunity/calibration), which run in parallel and seasonally rather than strictly in series.