Chinese Space Station Survey Telescope (CSST)

• CSST is a state-of-the-art Stage-IV space observatory that uses multi-band imaging and slitless spectroscopy to study cosmology, galaxy evolution, and time-domain phenomena.
• It integrates advanced instruments, including a Survey Camera, Multi-Channel Imager, and Integral Field Spectrograph, providing wide-field coverage with sub-arcsecond resolution.
• Its comprehensive surveys of 17,500 deg² enable precision measurements of dark energy, gravitational lensing, and stellar as well as Galactic structural dynamics.

The Chinese Space Station Survey Telescope (CSST) is a next-generation Stage‑IV space observatory designed to carry out simultaneous, high-precision multi-band imaging and slitless spectroscopic surveys from low Earth orbit. With a 2-meter primary mirror, a large field of view (~1.1 deg² per pointing), and advanced instrumentation—including the Survey Camera (SC), Multi-Channel Imager (MCI), Integral Field Spectrograph (IFS), Cool Planet Imaging Coronagraph (CPI‑C), and THz Spectrometer (TS)—CSST is configured to address cutting-edge questions in cosmology, galaxy evolution, stellar astrophysics, Solar System science, and time-domain astronomy. Its design enables the survey of 17,500 deg² of the sky with sub-arcsecond resolution, providing a data set for advances in the measurement of dark energy and dark matter properties, strong and weak gravitational lensing, galaxy and AGN demographics, Milky Way structure, stellar physics, exoplanet detection, and astrophysical transients.

1. Instrumentation and Survey Architecture

CSST integrates five primary scientific instruments, each tailored for specific observing modes and spectral ranges (Collaboration et al., 7 Jul 2025):

• Survey Camera (SC): The main survey instrument, offering both photometric imaging (NUV, u, g, r, i, z, y; 255–1000 nm) and slitless grating spectroscopy (GU, GV, GI bands; R ≳ 200) over large fields. The central imaging field reaches a pixel scale of 0.074″ and an 80% encircled energy radius of <0.15″.
• Multi-Channel Imager (MCI): Simultaneous NUV (255–430 nm), optical-blue (430–700 nm), and optical-red (700–1000 nm) imaging on separate 9216×9232-pixel CCDs, all with 0.05″ pixels and extensive filter options (30 per channel). The MCI excels in extreme-deep field imaging and precise flux standards establishment (Zheng et al., 18 Sep 2025).
• Integral Field Spectrograph (IFS): Provides 3D spectroscopy over a 6″×6″ field at R ≥ 1000, resolved spatially at 0.2″/pixel, for kinematic and chemical mapping of compact sources.
• Cool Planet Imaging Coronagraph (CPI‑C): Achieves contrasts down to 10⁻⁸ in the optical (IWA ~0.55″ at 633 nm) for direct imaging of exoplanets and circumstellar environments.
• THz Spectrometer (TS): Covers 0.41–0.51 THz (λ ≈ 0.6–0.7 mm) for ISM studies, with >1 GHz IBW and <100 kHz frequency resolution.

The telescope operates on a TMA off-axis optical scheme (aperture = 2 m, focal length = 28 m), supporting focal-plane layouts that provide simultaneous imaging, multiplexed spectroscopy, and high-contrast starlight suppression.

Major survey modes consist of:

• Wide Survey: ~17,500 deg², deep in seven photometric bands (5σ to g ≈ 26.3 mag), with matching slitless spectroscopy (limiting mag ≈ 23–24).
• Deep Fields: ~400 deg² to one magnitude deeper.
• Ultra-Deep Fields: <10 deg², targeting faintest limits (multi-visit, potentially ≃28 mag in some bands).

2. Photometric and Spectroscopic Techniques

CSST’s photometric performance is built on high-resolution, stable PSFs (R_EE80 < 0.15″), well-controlled backgrounds, and a comprehensive multi-band filter system. The combination of broad, medium, and narrow-band filters (notably in the MCI: F275W, F336W, F375M, F450M, F500M, F630M, F763M, F845M, F960M) enables both wide-area SED fitting and optimized photometric redshift estimation (Cao et al., 2021).

For photometric redshifts, template fitting (e.g., with LePhare) using the medium-band MCI filters achieves σ_z ~ 0.015 and catastrophic outlier fractions f_c ~ 1.5%, meeting the requirements for precision cosmological studies (Cao et al., 2021). Filter design (band center, width, steepness) is informed by simulations using realistic SED libraries and instrumental models.

Slitless spectroscopic observations employ transmissive gratings for dispersing the full field onto dedicated areas of the detector with minimal overlap. Instrumental challenges, including spectral overlap, self-broadening (extended galaxy morphology effects), and contaminant fluxes, are addressed via tailored simulation frameworks such as CESS (Wen et al., 8 Jan 2024), which account for morphological parameters (Sérsic index, effective radius, position angle, axis ratio) and estimate both the SNR of emission/continuum features and cross-object contamination rates. Typical spectral resolution R ≳ 200 is well-matched to galaxy and emission-line science needs, with a completeness (secure redshifts) of ∼25% for m_z < 21 mag galaxies out to z ∼ 1 in simulated surveys (Wen et al., 8 Jan 2024).

Star-based methodologies ensure high-precision flux and wavelength calibration. For instance, the UaRA-net neural network translates normalized LAMOST spectra (R = 2000) into CSST SEDs mimicking slitless output (R = 200), with SED precisions as good as 1–4% depending on band, temperature, and SNR (Yang et al., 24 Jan 2024); Bayesian transfer-learning neural nets establish robust redshift estimation accuracies σ_NMAD ≈ 0.00063 suitable for BAO measurements (Zhou et al., 19 Jul 2024). Wavelength calibration leverages velocity-standard stars observed in normal survey modes, achieving GU-band precision of ≲2 km s⁻¹ with ≲400 standards (Yuan et al., 2020).

3. Cosmological and Structural Probes

CSST’s survey configuration enables the simultaneous execution of major cosmological probes (Gong et al., 25 Jan 2025, Gong et al., 2019, Miao et al., 2022, Lin et al., 2022):

• Weak Lensing (WL): Shape measurements of billions of galaxies across four or more photo-z bins, with PSF-calibrated systematics, provide convergence and shear power spectra. Multiplicative calibration (m_i), additive errors (N_add), and intrinsic alignment modeling (A_IA, ηIA) are parameterized and marginalized during joint analyses. Forecasts indicate percent-level constraints (σΩₘ ≈ 1%, σ_σ₈ ≈ 1%, σ_w₀ < 5%, σ_wₐ < 20%) for joint 3×2pt (shear, clustering, and galaxy–galaxy lensing) (Miao et al., 2022, Gong et al., 2019).
• Galaxy Clustering: Both angular (broad-band photometric) and three-dimensional (spectroscopic) clustering data are obtained. Redshift-space power spectrum is decomposed into multipoles, including redshift-space distortions (β = f/b_g), the AP effect, and small-scale velocity dispersion dampings.
• Cluster Counts: Mass-selected cluster catalogs derived from photometric and spectroscopic data provide an independent constraint on the amplitude of fluctuations and matter density via the mass function dn/dM and comoving volume.
• Cosmic Voids: Void size functions add complementary sensitivity, breaking degeneracies with other probes.
• Type Ia Supernovae: Ultra-deep fields (and deep temporal coverage) provide high-cadence light curves for SNe Ia, essential for luminosity distance moduli,

$\mu = m_B + \alpha x_1 - \beta c - M_0$

and cosmological expansion history tests.

• Baryon Acoustic Oscillations (BAO): Precise photometric and spectroscopic redshifts enable detection of radial ($\alpha_{\parallel}$) and transverse ($\alpha_{\perp}$) BAO signatures, with standard ruler calibrations on both galaxies and AGN.
• Neutrino Cosmology: The multi-probe approach supports tight upper limits on $\Sigma m_{\nu}$ (e.g., $\lesssim0.36$ eV, 68% C.L.) (Lin et al., 2022).
• Strong Lensing: Simulations predict discovery of ∼160,000 galaxy–galaxy strong lenses and ∼1000 strongly lensed SNe, sharply expanding samples for precision cosmography and time-delay measurements (Cao et al., 2023, Dong et al., 15 Jul 2024).

4. Stellar, Galactic, and Time-Domain Science

CSST’s coverage and sensitivity facilitate:

• Stellar Science: Dedicated pipelines recover radial velocities (σ_RV ≈ 2–4 km s⁻¹ for FGKM stars at SNR=100; GU-band is optimal owing to dense absorption features). Typical SED predictions for flux standards achieve <2% errors in the visible (Sun et al., 2020, Yang et al., 24 Jan 2024).
• Milky Way Studies: Stellar mock catalogs (TRILEGAL-based, ~12.6 billion simulated stars to g=27.5 mag) establish a benchmark grid for the inner and outer Galaxy, and crowding analyses confirm CSST’s unique capacity to survey low-latitude fields with high photometric precision (Chen et al., 2023). Crowding limits are analyzed analytically (σₘ = f(LF, PSF, stellar density)) with the CSST’s 0.15″ PSF.
• Astrometry: Simulations demonstrate parallax and proper motion accuracy of ≲1 mas/yr for 18–22 mag stars, enabling kinematic mapping beyond Gaia’s faint limit (Fu et al., 2023).
• Morphological Studies: Non-parametric galaxy indicators (C, A, Gini, M₍₂₀₎, A_O, D_O) are measured with bias–correction functions, aligning CSST wide/deep field measurements with HST-based morphological frameworks (Luo et al., 4 May 2025). Analytical correction: $P_{cal} = P_{obs} + a \cdot SB_H + b \cdot z + c$ for parameter P.
• Time-Domain Science: Forecasts predict ∼5 million observed SNe over 10 yr, with well-defined “gold” samples (e.g., ~7400 SNe Ia at 91% classification precision) and early–alert capabilities (~15,500 SNe Ia identified pre-max) (Liu et al., 18 Jul 2024). Hundreds of shock-cooling events/year will be detected in the NUV, enhancing constraints on SN progenitors and explosion mechanisms. Strongly lensed SN yields are also significant (~1000 in the wide survey), with meaningful fractions caught before peak, facilitating time-delay cosmography (Dong et al., 15 Jul 2024).

5. Calibration, Data Processing, and Systematics Control

Calibration—flux, wavelength, astrometric, and photometric—is implemented via:

• Flux calibration: UaRA-net leverages large sets of LAMOST standards for robust SED prediction, achieving band-dependent precisions as low as 0.3% (GI) at SNR=80 (Yang et al., 24 Jan 2024).
• Wavelength calibration: Velocity-standard stars observed regularly enable field-wide, star-based solutions at GU-band precisions of 2 km s⁻¹ (Yuan et al., 2020).
• Astrometric calibration incorporates survey schedule optimization, with simulations highlighting the role of observation cadence spread for minimizing parameter correlations (Fu et al., 2023).
• Photometric redshift calibration: MCI’s nine medium-band filters are near-optimal (σ_z ≈ 0.015, f_c ≈ 1.5%), with small corrections possible for further optimization (Cao et al., 2021).

Instrumental systematics (PSF modeling, multiplicative/additive errors, photo-z stretch and bias, spectral overlap, contamination in dense fields) are formally modeled and marginalized (e.g., via MCMC or Fisher matrix approaches). For the multi-probe cosmology programs, typical requirements include

• $|\Delta z_i| < 0.02$
• Stretch uncertainty $s_z^i$ < 10%
• Multiplicative error $|m_i| < 0.02$ (Gong et al., 2019)

6. Synergies, Data Products, and Legacy Impact

CSST’s data products follow standard processing levels:

• Level-0: Raw data (images, spectra), telemetry, engineering metadata
• Level-1: Calibrated frames (including flux and wavelength calibration, PSF modeling, astrometry)
• Level-2: Catalog products (photometric redshifts, morphologies, emission-line measurements, stellar parameters, time-domain tables)

With a cloud-based processing infrastructure, the data release strategy is designed for regular domestic and international community dissemination.

CSST’s surveys are architected for complementarity and synergy with other Stage IV facilities such as Euclid and Rubin Observatory. Its combination of high-resolution, wide-area NUV–NIR imaging, and large multiplex spectroscopic data fills gaps in depth, wavelength, and sky coverage, offering opportunities for cross-survey calibration, joint cosmological analyses, and legacy studies extending from high-redshift structure formation to Galactic archaeology and time-domain astrophysics.

7. Forward Outlook and Limitations

While simulations forecast transformative constraint improvements (order-of-magnitude tighter parameters over current surveys for Ωₘ, σ₈, w₀, wₐ, Σm_ν), it is acknowledged that these analyses involve optimistic or idealized assumptions; real-world instrumental effects may degrade achievable precisions (Gong et al., 2019, Miao et al., 2022). Planned survey operation will iteratively incorporate schedule optimizations, more realistic noise and background models, and machine-learning enhancements for large-scale data vetting.

CSST's design serves as a reference for how to integrate multi-instrument, multi-wavelength, and multi-modal (photometric, spectroscopic, time-domain) surveys in a single facility, both for maximizing scientific return and for offering a flexible, modular platform for future upgrades and mission extension. Its legacy data set will underpin the next decade’s advances in cosmological, Galactic, and time-domain astrophysics.