--- title: 'MINERVA: JWST Treasury Medium-Band Survey' url: https://www.emergentmind.com/topics/minerva-survey type: topic --- # MINERVA: JWST Treasury Medium-Band Survey MINERVA is a JWST Cycle 4 treasury imaging survey that combines NIRCam medium-band and MIRI imaging across four CANDELS extragalactic fields—UDS, COSMOS, AEGIS, and GOODS-N—with the explicit aim of extending existing deep-field datasets into a regime of denser spectral sampling and larger areal coverage [2507.19706]. The program uses 259.8 hours of prime time and 127 hours of parallel time to obtain eight NIRCam medium-band images and two MIRI images, reaching a $5\sigma$ depth of 28.1 mag in F300M over $\sim 542\,\mathrm{arcmin}^2$ and 23.9 mag in F1280W over $\sim 275\,\mathrm{arcmin}^2$ [2507.19706]. In combination with prior JWST and HST imaging, the survey is designed to produce photometric catalogs with 20–26 JWST filters and 26–35 filters total, enabling more robust photometric redshifts, stellar masses, and resolved stellar-population measurements than broadband-only configurations [2507.19706]. ## 1. Survey design and filter architecture MINERVA deploys eight NIRCam medium-band filters and two MIRI broad-band filters. The NIRCam set is F140M, F162M, F182M, F210M, F250M, F300M, F360M, and F460M, with central wavelengths and approximate bandwidths of $1.40\,\mu\mathrm{m}$ ($\Delta\lambda\approx0.04\,\mu\mathrm{m}$), $1.62\,\mu\mathrm{m}$ ($0.04\,\mu\mathrm{m}$), $1.82\,\mu\mathrm{m}$ ($0.05\,\mu\mathrm{m}$), $2.10\,\mu\mathrm{m}$ ($0.05\,\mu\mathrm{m}$), $2.50\,\mu\mathrm{m}$ ($0.05\,\mu\mathrm{m}$), $3.00\,\mu\mathrm{m}$ ($0.10\,\mu\mathrm{m}$), $3.60\,\mu\mathrm{m}$ ($0.10\,\mu\mathrm{m}$), and $4.60\,\mu\mathrm{m}$ ($0.10\,\mu\mathrm{m}$), respectively [2507.19706]. The parallel MIRI observations are obtained in F1280W and F1500W, centered at $12.8\,\mu\mathrm{m}$ and $15.0\,\mu\mathrm{m}$. Typical on-sky integration times per pointing are $\sim 0.5$ h for the NIRCam medium bands and $\sim 0.6$ h for the MIRI bands. Survey depth is estimated by placing a large number ($>10^3$) of empty apertures of diameter $0.32''$ on the final mosaics, measuring the RMS of enclosed fluxes, and converting the result to AB magnitudes with correlated-noise corrections derived from the same empty-aperture RMS scaling. The reported $5\sigma$ depths are $m_{5\sigma}\approx27.5\text{--}28.3$ AB mag for NIRCam, including $m_{5\sigma}(\mathrm{F300M})=28.1$ mag, and $m_{5\sigma}\approx23.5\text{--}24.2$ AB mag for MIRI, including $m_{5\sigma}(\mathrm{F1280W})=23.9$ mag [2507.19706]. The design is explicitly cumulative rather than standalone. The targeted fields were previously observed in Cycle 1 with 7–9 NIRCam filters by the PRIMER, CEERS, and JADES programs, so MINERVA functions as a medium-band augmentation layer over already deep broadband imaging. This configuration is central to the survey’s role in transforming existing deep fields into higher-dimensional photometric datasets [2507.19706]. ## 2. Footprint, tiling strategy, and coverage expansion MINERVA tiles four canonical CANDELS fields with distinct NIRCam and MIRI footprints. The NIRCam coverage is 234 arcmin$^2$ in UDS, 144 arcmin$^2$ in COSMOS, 96 arcmin$^2$ in AEGIS, and 68 arcmin$^2$ in GOODS-N, for a total of $$ A_{\rm NIRCam}=234+144+96+68=542\,\mathrm{arcmin}^2. $$ The MIRI prime pointings cover 125 arcmin$^2$ in UDS, 111 arcmin$^2$ in COSMOS, 23 arcmin$^2$ in AEGIS, and 16 arcmin$^2$ in GOODS-N, giving $$ A_{\rm MIRI}=125+111+23+16=275\,\mathrm{arcmin}^2. $$ These values define the survey’s effective areal scale [2507.19706]. | Field | NIRCam area | MIRI area | |---|---:|---:| | UDS | 234 arcmin$^2$ | 125 arcmin$^2$ | | COSMOS | 144 arcmin$^2$ | 111 arcmin$^2$ | | AEGIS | 96 arcmin$^2$ | 23 arcmin$^2$ | | GOODS-N | 68 arcmin$^2$ | 16 arcmin$^2$ | A central quantitative feature of MINERVA is its expansion of JWST medium-band coverage. Prior medium-band programs attained only $\sim 70$ arcmin$^2$ in eight NIRCam medium bands, and MINERVA increases that area to 542 arcmin$^2$, corresponding to $$ \frac{542}{70}\gtrsim 7\times. $$ This enlarged footprint is intended to improve volume coverage and mitigate cosmic variance in population studies [2507.19706]. ## 3. Photometric catalog construction and inference framework When combined with ancillary JWST imaging from PRIMER, CEERS, COSMOS-Web, and JADES, together with deep HST data, each MINERVA field yields $$ N_{\rm JWST}=20\text{--}26,\qquad N_{\rm total}=26\text{--}35\,\mathrm{filters}. $$ This filter density is the basis for the survey’s catalog strategy [2507.19706]. The source-extraction workflow consists of PSF homogenization to the largest PSF, typically MIRI or F444W, using convolution kernels computed from empirical star stacks; aperture photometry in matched $0.32''$ apertures corrected to total flux via encircled-energy curves; local background subtraction and correlated-noise correction through empty-aperture statistics; and error estimation that includes photon noise, read noise, and background noise. These steps define the survey’s photometric homogenization and uncertainty model [2507.19706]. Photometric redshifts are computed with `EAZY` using a template set optimized for JWST medium bands. Simulations matched to MINERVA depth yield an expected $$ \sigma_{\rm NMAD}\!\left(\frac{\Delta z}{1+z}\right)\simeq0.9\%,\qquad f_{\rm outlier}\simeq5.5\%. $$ The summary attributes this to an improvement of $\sim 3.7\times$ in scatter and $\sim 2.6\times$ in outlier rate relative to broadband-only surveys. Stellar masses and related parameters are fit with the `DenseBasis` nonparametric SED-fitting code, giving mass uncertainties of $\sigma_{\rm NMAD}(\log M)\sim0.06\text{--}0.08$ dex [2507.19706]. The catalog design is therefore not limited to photometric detection. It is structured to support redshift inference, stellar-mass estimation, and resolved physical characterization within a unified PSF-matched, multi-instrument framework. ## 4. Science goals and predicted source populations MINERVA is designed to reveal source classes that are difficult to isolate in broadband catalogs. The stated targets include pristine $z>13$ Lyman-break galaxies; galaxies with strong Balmer breaks, described as quiescent or “napping” sources, at $z>4$; extreme emission-line galaxies with ${\rm EW}\gtrsim2000\,\text{\AA}$; and heavily obscured “dark” galaxies visible only in MIRI [2507.19706]. The survey also aims to reduce systematics in stellar-mass functions and number densities by $\gtrsim 3\times$ through improved $z_{\rm phot}$ and $M_\star$, and to map stellar mass and star-formation rate within galaxies over $16$ and $\sim 1500$ at $413$ candidates over 542 arcmin$^2$, enabling UV luminosity-density constraints to a factor of $\sim 4$ precision [2507.19706]. Completeness curves derived by injecting artificial sources of known magnitude and size into the mosaics show 80–90% completeness down to the $5\sigma$ depth in all medium bands, with small redshift-dependent variations. These targets collectively define MINERVA’s role as a survey of populations that are either rare, spectrally ambiguous in broadband data, or heavily affected by photometric-redshift and stellar-mass systematics. ## 5. Reduction pipeline, validation, and control of systematics Data reduction follows the JWST Calibration Pipeline, including Stage 1 ramp fitting, Stage 2 flat-fielding and background subtraction, and Stage 3 astrometric alignment plus drizzle mosaicking. The MIRI component uses a three-point small-grid dither to improve bad-pixel masking and background uniformity. Noise modeling includes both white and correlated noise terms informed by blank-sky measurements [2507.19706]. Catalog validation is multi-pronged. The procedure includes cross-matching to existing deep spectroscopic redshifts from 3D-HST, MOSDEF, NIRSpec GTO/WIDE, CEERS, and RUBIES; injection–recovery simulations to quantify photometric biases and completeness; and systematic checks based on comparisons between medium-band and broadband fits. The summary states that such comparisons demonstrate a factor $\sim 2\text{--}3$ reduction in biases on $M_\star$ and $z_{\rm phot}$ [2507.19706]. This methodology places completeness, bias control, and redshift validation at the center of the survey definition. A plausible implication is that MINERVA is intended not only to discover uncommon source classes, but also to regularize the statistical foundations of deep-field population measurements by explicitly addressing correlated noise, PSF heterogeneity, and broadband degeneracies. ## 6. Legacy role within deep-field astronomy Upon release to MAST, MINERVA data products are planned to include PSF-matched images, photometry tables, $z_{\rm phot}$, $M_\star$, and SFR measurements. The survey is framed as a long-duration infrastructure dataset rather than a narrowly targeted observing campaign: it is intended to serve as a cornerstone deep-field resource and to facilitate spectroscopic follow-up with JWST/NIRSpec, ELTs, and ALMA for decades [2507.19706]. The program is explicitly positioned within a larger ecosystem of extragalactic surveys. The summary notes synergies with COSMOS-Web, CEERS, JADES, PRIMER, UNCOVER, SPAM, MEGA, and SMILES, and emphasizes that, once complete, MINERVA will become an integral part of the treasury deep field imaging datasets [2507.19706]. In operational terms, its importance derives from a conjunction of three features already quantified elsewhere in the survey design: medium-band coverage in eight NIRCam bands, a footprint of 542 arcmin$^2$, and dense multiwavelength catalog products spanning 20–26 JWST filters and 26–35 filters total. Within that framework, MINERVA’s distinctive contribution is the conversion of existing flagship deep fields into medium-band-enhanced survey volumes with better-controlled completeness, stronger photometric-redshift performance, tighter stellar-mass constraints, and direct support for resolved studies of galaxy growth, quenching, and obscured activity from $1