ASTRA-DESI Early Data Release (EDR) This record bundles the ASTRA-DESI early data products for the 20 DESI rosettes (zones 00–19). Every catalogue is delivered as a compressed FITS binary table (.fits.gz) whose first extension (HDU 1) contains the data. File headers store the keywords RELEASE=EDR and the corresponding ZONE label for quick checks. Tracer Number of objects BGS 241,746 LRG 112,649 ELG 267,345 QSO 35,566 Directory layout The archive preserves the NERSC production tree. Top-level folders contain: raw/ – merged real and random catalogues per zone. classification/ – neighbour count summaries per tracer and random iteration. probabilities/ – web-type membership probabilities derived from the classification counts. Naming conventions raw/zone_XX.fits.gz – zone index XX is zero-padded (00–19). classification/zone_XX_classified.fits.gz – combined tracer summary for the zone. probabilities/zone_XX_probability.fits.gz – per-object web-type probabilities. Column reference Raw catalogues (raw/*.fits.gz) Column Type Description TARGETID int64 DESI target identifier. RA float64 Right ascension in degrees. DEC float64 Declination in degrees. Z float64 Spectroscopic redshift. XCART, YCART, ZCART float64 Comoving Cartesian coordinates (Mpc) computed with the Planck18 cosmology. TRACERTYPE string Base tracer with suffix _DATA or _RAND (BGS_ANY, ELG, LRG, QSO). RANDITER int32 -1 for real objects, otherwise the random iteration index (0–99). SED_SFR float64 Star formation rate derived from SED fitting. SED_MASS float64 Stellar mass derived from SED fitting. FLUX_G float64 Observed flux in the g photometric band. FLUX_R float64 Observed flux in the r photometric band. Each raw file merges all tracers for its zone. Random catalogues mirror the data counts for every tracer and random iteration. Additional columns SED_SFR, SED_MASS, FLUX_G, and FLUX_R were incorporated from the DESI EDR value-added catalogues associated with the emission-line sample: https://data.desi.lbl.gov/doc/releases/edr/vac/stellar-mass-emline/. These quantities are derived via SED fitting using CIGALE (Zou et al. 2024). Classification catalogues (classification/*.fits.gz) Column Type Description TARGETID int64 Identifier of the central object. RANDITER int32 -1 for data rows, otherwise the random catalogue index. ISDATA bool True for data rows; False for random iterations. NDATA int32 Number of neighbour pairs built from data tracers for this target. NRAND int32 Number of neighbour pairs contributed by random tracers. TRACERTYPE string Tracer family (BGS_ANY, ELG, LRG, QSO). Probability catalogues (probabilities/*.fits.gz) Column Type Description TARGETID int64 Identifier of the object. TRACERTYPE string Tracer family. PVOID float32 Probability of the object being a void class. PSHEET float32 Probability of the sheet class. PFILAMENT float32 Probability of the filament class. PKNOT float32 Probability of the knot class. Using the catalogues All FITS tables are compressed with gzip. Random catalogues use RANDITER values 0–99; filtering on RANDITER >= 0 isolates the random iterations. The Cartesian coordinates assume the Planck18 cosmology provided by astropy.cosmology.Planck18.