LAMOST Spectroscopic Data Overview

Updated 10 November 2025

LAMOST spectroscopic data are a vast collection of flux- and wavelength-calibrated spectra obtained from a multi-object fiber-fed 4-meter Schmidt telescope, essential for modern astrophysics.
They encompass low and medium-resolution surveys across a broad spectral range, enabling accurate determination of stellar parameters, redshifts, and astrophysical classifications.
Robust data reduction pipelines and value-added catalogs ensure precise calibration, quality control, and facilitate extensive research in Galactic and extragalactic astronomy.

The Large Sky Area Multi-Object Fiber Spectroscopic Telescope (LAMOST) has, since 2009, produced one of the largest spectroscopic data troves in astrophysics, with low and medium resolution surveys spanning Galactic stars, galaxies, and quasars. LAMOST spectroscopic data consist of flux- and wavelength-calibrated one-dimensional spectra acquired via a multi-object fiber-fed instrument on a 4-meter Schmidt telescope configured for high-multiplexing wide-area surveys. Across a polychromatic spectral baseline (370–900 nm for low-resolution, narrower for medium-resolution), dedicated pipelines deliver derived astrophysical parameters, stellar and extragalactic classifications, and value-added catalogues well suited to stellar astrophysics, Galactic archaeology, extragalactic studies, and time-domain science.

1. Instrumentation, Survey Modes, and Spectral Characteristics

LAMOST is a 4-meter effective aperture active-optics Schmidt telescope located at Xinglong Observatory. It is equipped with a 5-degree-diameter focal plane, populated by 4000 individually positionable fibers, each feeding one of 16 spectrographs. In its low-resolution mode (LRS, R ≈ 1800), each spectrograph covers 370–900 nm divided into blue (370–590 nm) and red (570–900 nm) arms. In medium-resolution mode (MRS, R ≈ 7500), the blue arm is 4950–5350 Å and the red arm is 6300–6800 Å. The system can simultaneously observe several thousand objects per field, with multiplexing and high areal coverage key for Galactic structure mapping and statistical studies (Yan et al., 2022).

Typical signal-to-noise ratios for 1000 s exposures are SNR ≳ 50 for r ≲ 15 mag and SNR ≈ 10 at r ≈ 17.5 for LRS; for MRS, SNR ≳ 50 at G ≲ 14 mag, SNR ≈ 20–30 at G ≈ 15 mag. Over a decade, LAMOST has released spectra for over 10 million stars, ~220,000 galaxies, and ~71,000 quasars up to DR8 (Yan et al., 2022).

2. Data Reduction, Quality Assessment, and Parameter Pipelines

Raw CCD data flow through comprehensive automated reduction pipelines. The 2D pipeline handles bias/dark subtraction, flat-fielding, fiber tracing/extraction, sky subtraction using dedicated sky fibers, wavelength calibration using arc lamps and sky emission lines (with residuals ≲ 0.02 Å), and inverse-variance combination of sub-exposures. Relative flux calibration is achieved via F-type standard stars per plate, yielding calibration accuracy to ≈5–8% (Yan et al., 2022).

The 1D pipeline applies template-matching for object classification (STAR/GAL/QSO/UNKNOWN), emission/absorption line detection, and radial velocity or redshift estimation via cross-correlation. For stars, the LAMOST Stellar Parameter Pipeline (LASP) employs χ²-minimization against empirical and/or synthetic spectral libraries to determine effective temperature (T_eff), surface gravity (log g), metallicity ([Fe/H]), and radial velocity (v_r) (Luo et al., 2015, Zong et al., 2018). For LRS with SNR > 50, typical uncertainties are ΔT_eff ≈ 100–150 K, Δlog g ≈ 0.10–0.20 dex, Δ[Fe/H] ≈ 0.10–0.20 dex, and Δv_r ≲ 5 km/s (Yan et al., 2022, Zong et al., 2018).

Additional pipelines such as MKCLASS allow automated MK classification (Cat et al., 2014, Liu et al., 2019), and for medium-resolution spectra, recent codes like LAMA utilize template matching to infer T_eff, log g, [Fe/H], v_r, and projected rotation (v sin i) with ΔT_eff ≲ 75 K, Δ[Fe/H] ≲ 0.12 dex for SNR > 10 (Li et al., 18 Jul 2024). Data-driven models, including neural networks, are leveraged for estimation of absolute magnitudes and binarity in specific subpopulations (e.g., OB stars) (Xiang et al., 2020).

Error estimation and internal/external calibration

Internal errors are evaluated from repeat observations; external accuracy is cross-validated using comparisons to high-resolution surveys such as APOGEE and GALAH, where typical cross-survey biases and scales for T_eff, log g, [Fe/H], and v_r are quantified (Qin et al., 26 Jul 2025, Wang et al., 2020, Li et al., 22 Nov 2024). Systematic errors are further controlled using matching with external astrometry (Gaia), known photometric catalogs, or additional RV standards. Quality metrics and masks are provided (e.g., SNR per band, data flags, bitmask arrays), and caution is advised for spectra with low SNR or flagged reduction artifacts (Zong et al., 2018, Lu et al., 2021).

3. Data Products, Value-Added Catalogues, and Access

LAMOST data releases (DR1–DR8+) provide:

1D, wavelength- and flux-calibrated FITS spectra (blue/red, or both arms merged for LRS)
Catalogs of derived parameters (T_eff, log g, [Fe/H], v_r, v sin i, extinction, MK spectral types, etc.)
Special-purpose value-added tables: absolute magnitudes (from NN or spectroscopic parallax), distances, binarity flags, variability statistics, and chemical abundances (up to 16 elements for LRS)
For extragalactic targets: redshifts, emission line diagnostics, and velocity dispersion measurements (e.g., via pPXF for σ_e) (Napolitano et al., 2020)

Each spectrum is indexed by object ID, observing plan, fiber ID, sky coordinates, SNR, and flags. Data are delivered as machine-readable FITS tables, with web-based and TAP/ADQL/SQL query interfaces (He et al., 2016). Large-scale datasets are accessible via dedicated web portals per data release (e.g., http://dr8.lamost.org), with the ability to batch-download or cross-match with external catalogs (Luo et al., 2015).

Value-added catalogues increase in sophistication with later data releases, featuring, for instance, neural-network-predicted absolute magnitudes and granular multi-band photometry, spectro-photometric distances, high-fidelity binarity diagnostics, and dedicated outlier catalogs for quality control (Lu et al., 2021).

4. Scientific Applications

LAMOST spectroscopic data are used extensively for:

Galactic structure: Probabilistic distance and extinction mapping, metallicity/kinematics of disk and halo, constraints on thin/thick disk formation and evolution (Yan et al., 2022, Qin et al., 26 Jul 2025)
Stellar astrophysics: Empirical HR diagrams in multiple bands, mass-radius determinations, asteroseismic modeling with contemporaneous Kepler/K2/TESS photometry (Zong et al., 2018, Wang et al., 2020, Wang et al., 2021)
Stellar populations: Systematic discovery and characterization of OB stars, AFGKM stars, very metal-poor stars ([Fe/H] < –2), and emission-line objects (Liu et al., 2019, Zong et al., 2018)
Exoplanet host characterization: Improved radii, masses, and activity diagnostics for transiting planet hosts (Zong et al., 2018, Wang et al., 2020)
Extragalactic science: Galaxy velocity dispersions (σ_e), red sequences, mass-σ relations, AGN, and quasar science (Dong et al., 2018, Napolitano et al., 2020)
Time-domain astrophysics: Binarity, rotation, activity cycles, and outbursting stars using multi-epoch spectroscopy (Qin et al., 26 Jul 2025, Wang et al., 2021)
Outlier detection and data quality assessment: Systematic identification of pathological spectra, quality flagging, and sample curation (Lu et al., 2021)

5. Quality Control, Limitations, and Outlier Analyses

Data quality is affected by seeing, fiber positioning, sky background, instrumental throughput variation across fibers, flux calibration (which is relative, due to the absence of a dedicated photometric telescope), and template mismatch in parameter pipelines. Systematic issues include limited sensitivity to very metal-poor ([Fe/H] < –2.5) and very hot/cool stars, sky-subtraction residuals (especially at λ > 720 nm), and known bugs (e.g., A-star RV calibration errors in DR1) (Luo et al., 2015, Lu et al., 2021, Yan et al., 2022).

Dedicated statistical analyses, such as principal component analysis (PCA) and local outlier factor (LOF) metrics, are used to identify, cluster, and catalog spectra with reduction defects or astrophysical anomalies, supporting robust science by enabling users to filter or flag problematic instances (Lu et al., 2021). Cross-validation with high-resolution surveys (e.g., APOGEE) demonstrates LASP's reliability except in cases with strong nebular contamination or significant data reduction artifacts.

6. Specialized Surveys and Advanced Analyses

Large thematic sub-surveys exploit LAMOST's strengths for particular domains:

LAMOST-Kepler/K2 fields: Systematic spectroscopic follow-up for Kepler and K2 targets enabling asteroseismology, exoplanet host assessment, activity monitoring, and time-domain studies (Cat et al., 2014, Zong et al., 2018, Wang et al., 2020, Qin et al., 26 Jul 2025, Wang et al., 2021)
OB star catalogs: Extensive selection and manual/MKCLASS verification of ~16,000 OB-type stars with completeness assessment and classification accuracy metrics; identification of high-latitude, low-metallicity, and peculiar population features (Liu et al., 2019)
Binary detection: Human-AI hybrid pipelines leveraging cross-correlation functions and deep neural classifiers for detection of double-line and triple-line spectroscopic binaries, yielding a tripling of existing medium-resolution samples (Li et al., 22 Nov 2024)
Advanced parameter estimation in specific populations: Data-driven neural net models for spectroscopic parallaxes and binarity in OB stars; machine-learning models in stellar parameter pipelines (DD-Payne, SLAM, LAMA) (Xiang et al., 2020, Li et al., 18 Jul 2024, Wang et al., 2021)

Dedicated extragalactic efforts (e.g., LEGAS, LaCoSSPAr) supplement the stellar-focused infrastructure with redshift catalogs, nebular diagnostics, and velocity dispersion measurements (Yang et al., 2017, Napolitano et al., 2020).

7. Data Access, Long-term Curation, and Future Prospects

LAMOST data releases are indexed, archived, and distributed via robust infrastructure, including Java Spring Framework frontends, PostgreSQL+pgSphere backends, and version-controlled software under China-VO GitLab. Catalogs and spectra are served via web portals, public FTP, and VO-compliant interfaces, with tools for API-based retrieval and bulk querying. Data cycle management includes backup to geographically separated sites and systematic versioning (He et al., 2016).

Future LAMOST releases are expected to further expand the volume of spectra, depth, sky coverage, and temporal baseline. Advancements in machine learning–driven analysis, time-domain astrophysics, and cross-survey calibration (e.g., Gaia synergies) will increase the scientific returns for Galactic and extragalactic astronomy. Known limitations (e.g., in sky subtraction and template libraries) are being addressed in ongoing development of updated pipelines and expanded calibration datasets (Yan et al., 2022, Li et al., 18 Jul 2024).

LAMOST spectroscopic data provide a comprehensive, statistically homogeneous resource for modern astrophysics, with mature pipelines, deep ancillary product sets, and wide uptake across the stellar and extragalactic research communities. Their continued development and public availability ensure LAMOST's central role in multi-dimensional, multi-epoch, and data-intensive studies of the cosmos.