Improving z\sim7-11 Galaxy Property Estimates with JWST/NIRCam Medium-Band Photometry
Guido Roberts-Borsani, Tommaso Treu, Charlotte Mason, Kasper B. Schmidt, Tucker Jones, Adriano Fontana
DDraft version February 10, 2021
Typeset using L A TEX twocolumn style in AASTeX63
Improving z ∼ − Galaxy Property Estimates with
JWST /NIRCam Medium-Band Photometry
Guido Roberts-Borsani , Tommaso Treu , Charlotte Mason ,
2, 3
Kasper B. Schmidt , Tucker Jones , and Adriano Fontana Department of Physics and Astronomy, University of California, Los Angeles, 430 Portola Plaza, Los Angeles, CA 90095, USA Center for Astrophysics, Harvard & Smithsonian, 60 Garden St, Cambridge, MA 02138, USA Hubble Fellow Leibniz-Institut f¨ur Astrophysik Potsdam (AIP), An der Sternwarte 16, D-14482 Potsdam, Germany Department of Physics, University of California, Davis, 1 Shields Avenue, Davis, CA 95616, USA INAF – Osservatorio Astronomico di Roma, Via Frascati 33, I-00078 Monte Porzio Catone (RM), Italy
ABSTRACTThe past decade has seen impressive progress in the detection of z >
HubbleSpace Telescope , however little is known about their properties. The
James Webb Space Telescope willrevolutionise the high- z field by providing NIR (i.e., rest-frame optical) data of unprecedented depthand spatial resolution. Measuring galaxy quantities such as resolved stellar ages or gas metallicitygradients traditionally requires spectroscopy, as broad-band imaging filters are generally too coarse tofully isolate diagnostics such as the 4000 ˚A (rest-frame) break, continuum emission from aged stars,and key emission lines (e.g., [O II ], [O III ], H β ). However, in this paper, we show that adding NIRCamimages through a strategically chosen medium-band filter to common wide-band filters sets adoptedby ERS and GTO programs delivers tighter constraints on these galactic properties. To constrainthe choice of filter, we perform a systematic investigation of which combinations of wide-band filtersfrom ERS and GTO programs and single medium-band filters offer the tightest constraints on severalgalaxy properties at redshifts z ∼ −
11. We employ the JAGUAR extragalactic catalogs to constructstatistical samples of physically-motivated mock photometry and conduct SED-fitting procedures toevaluate the accuracy of galaxy property (and photo- z ) recovery with a simple star-formation historymodel. We find that adding > . µ m medium filters at comparable depth to the broad-band filters cansignificantly improve photo- z s and yield close to order-of-magnitude improvements in the determinationof quantities such as stellar ages, metallicities, SF-related quantities and emission line fluxes at z ∼ Keywords: galaxies: high-redshift, galaxies: ISM, galaxies: star formation, cosmology: dark ages,reionization, first stars INTRODUCTIONOne of the major endeavours of modern observationalcosmology is to paint a coherent picture of the history ofthe Universe. To this end, the final frontier remains theidentification and characterisation of the first sourcesthat appeared in the Universe, those which played a sig-nificant role in reionising the intergalactic medium froma neutral state to a fully ionised one over the first billion
Corresponding author: Guido [email protected] years (corresponding to redshifts of 6 (cid:46) z (cid:46) HST ) have yielded impressive gains in the num-ber of galaxy candidates at redshifts z = 7 −
10, withsamples reaching over 1000 objects, and revolutionisedour understanding of galaxy evolution therein. Comple-menting these observations, the spectroscopic confirma-tion (e.g., Finkelstein et al. 2013; Oesch et al. 2015;Zitrin et al. 2015; Roberts-Borsani et al. 2016; Hoag et a r X i v : . [ a s t r o - ph . GA ] F e b Roberts-Borsani et al. al. 2017; Stark et al. 2017; Hashimoto et al. 2018) andcharacterisation (e.g., Laporte et al. 2017a; Mainali etal. 2018; Endsley et al. 2021) of over a dozen sourceshas seen impressive advances with ground-based spec-troscopy (e.g., probing the rest-frame UV and FIR withKeck/MOSFIRE, VLT/X-Shooter and ALMA), partic-ularly for the brightest and rarest objects. For the rest-frame optical, however, the
Spitzer Space Telescope has,until now, afforded the only realistic means for statis-tical analyses. However, the Infrared Array Camera’s(IRAC) coarse spatial resolution and the limited depthprobed by many surveys makes robust and uncontam-inated constraints on galaxy properties a challengingfeat. Further advances with current facilities are chal-lenging owing to the limited wavelength coverage of
HST and the observed faintness of star-forming galaxies asone approaches redshifts of z >
10. The imminent ar-rival of the
James Webb Space Telescope ( JWST ) hasthe potential to detect galaxies well beyond the currentfrontier of z ∼
12 (e.g., Behroozi et al. 2020) thanksto the unprecedented resolution and sensitivity of itsNIR imaging and spectroscopic capabilities, and revolu-tionise our current understanding of galaxy evolution.The first observations with the observatory will becarried out through accepted ERS and GTO programsand will showcase each instrument’s working capabili-ties as well as set the benchmark for future science withCycle 1 and beyond. While the ERS and GTO datasets on their own will undoubtedly be extremely valu-able for the community (see photometric parameter re-covery analyses by e.g., Kemp et al. 2019 and Kauff-mann et al. 2020), the short life-span of
JWST makesextracting all possible data with the highest accuracyimperative, in order to build on the progress achieveduntil now. Most such observations will carry out wide-band imaging spanning wavelength ranges of ∼ z (cid:38) Spitzer /IRAC photome-try (e.g., Labb´e et al. 2013; Oesch et al. 2014; Smit etal. 2015; Roberts-Borsani et al. 2016; Castellano et al.2017). However, such observations lack the wavelengthresolution to make precise measurements of galaxy prop-erties linked to the state of the gas and underlying stellarpopulations (see e.g., Labb´e et al. 2013; Roberts-Borsaniet al. 2020). Even with
JWST , spectroscopy of a largenumber of sources is expensive and cannot reach thefaint limits of the photometric catalogs.In this paper, we demonstrate that medium-band pho-tometry provides a cost-effective complement to spec-troscopy, allowing for the determination of important physical parameters of galaxies for large and faint sam-ples of galaxies. We reach this conclusion by carryingout a systematic study of the medium-band NIRCam fil-ters and the gain in accuracy that each affords. We findthat the addition of a single > . µ m filter yields pre-cise photo- z s and galaxy properties from the rest-frameoptical for star-forming galaxies at z ∼ −
11. Thepaper is structured as follows: we describe the relevant
JWST /NIRCam data sets and construction of mock cat-alogs in §
2, the SED-fitting techniques and their resultsin § §
4. Where relevant, we as-sume H =70 km/s/Mpc, Ω m =0.3, and Ω ∧ =0.7. Allmagnitudes are in the AB system (Oke & Gunn 1983). JWST PROGRAMS & GALAXY CATALOGS2.1.
ERS/GTO programs and chosen filters
The main goal of the present paper is to determinethe combinations of
JWST /NIRCam medium-band fil-ters which, in addition to observations with wide-bandfilters yield the most precise recovery of galaxy proper-ties from the rest-frame optical at high redshift. We takeas a baseline ERS and GTO programs, which are likelyto yield the first large samples of new galaxy candidatesdue to their sky coverage and unprecedented depths inthe NIR with NIRCam imaging, however our conclu-sions apply to any similar collection of wide-band fil-ters. In constructing our fiducial broad-band setup, thefollowing extragalactic programs are of greatest inter-est to this analysis (due to the area of the sky probed):ERS 1324 (“
Through the Looking GLASS: A JWST Ex-ploration of Galaxy Formation and Evolution from Cos-mic Dawn to Present Day ”, PI: Treu), ERS 1345 (“
TheCosmic Evolution Early Release Science ”, PI: Finkel-stein), GTO 1176 (“
JWST Medium-Deep Fields ”, PI:Windhorst), GTO 1199 (“
The Metallicity of Galaxiesin the MAC J1149.5+2223 Field ”, PI: Stiavelli), GTO1208 (“
The CAnadian NIRISS Unbiased Cluster Sur-vey ”, PI: Willott) and GTO 1180/1181 (“
NIRCam-NIRSpec Galaxy Assembly Survey - GOODS-S/N ”, PI:Eisenstein). The primary objectives and instrumentsused in each program vary and are beyond the scopeof this paper, however their common denominator aremedium-to-deep ( m ∼ mproving z ∼ − galaxy properties with JWST /NIRCam Medium-Bands σ ) depth of ∼ HST and
Spitzer photometry, here we focus onlyon the
JWST observations, since they are expected tosupersede in both depth and resolution all prior data forgalaxies in the z ∼ −
11 redshift range. For simplicity,much of our discussion refers to integrated galaxy pho-tometry, however we emphasise that given sufficient an-gular resolution and signal-to-noise our results also ap-ply to spatially-resolved properties and thus the strate-gies discussed here are valid for both types of analyses.2.2.
JWST/NIRCam colours of high-z galaxies
The rest-frame optical portion of a galaxy spectrum(or SED) offers a plethora of useful features with whichto characterise the state of the underlying stellar pop-ulations and gas. Chief among these are nebular emis-sion lines from the Balmer series (e.g., H α , H β and H δ )and from ionic species of Oxygen (e.g., [O II ] λλ III ] λλ II ] λ II ] λλ III ] λ z > z (cid:38) III ] λ δ emission, these diagnostics can reveal the ageof the underlying stellar population (particularly starswith ages >
10 Myrs) and the timing of the most recentburst of star formation. Such diagnostics are far fewer in the rest-frame UV (e.g., a lack of strong emissionlines), which is in general well-sampled by
HST /WFC3and will be further by upcoming NIRCam wide-bandobservations.Each of these diagnostics contribute to and influencethe colours of galaxies at IR wavelengths (to varying de-grees), which can be useful in identifying galaxies withe.g., particularly low or high levels of star formationor that are likely to lie within a particular photomet-ric redshift range. For this latter consideration, sev-eral studies have demonstrated the effectiveness of us-ing IR colour cuts with deep
Spitzer /IRAC imaging toconstrain the photometric redshifts of high- z galaxies inthe absence of deep optical bands (see e.g., Labb´e etal. 2013; Smit et al. 2015; Roberts-Borsani et al. 2016).The contamination of the 3.6 µ m channel by H α emis-sion and later strong [O III ]+H β emission in the 4.5 µ mchannel produces particularly blue [3.6] − [4.5] colours at z ∼ . − [4.5]colours at z ∼ − ∼ ∼ III ]+H β at z > z ∼ − z (cid:38) − JWST /NIRCam medium-band filters offers useful alternatives: the reduced band-width of the medium-band filters offer a cleaner mea-surement of nebular emission lines and continuum emis-sion, thereby removing (in part) several degenerate solu-tions. In Figure 1 we plot the NIRCam colours betweenour fiducial set of wide-band filters (from F200W long-ward, since these are unaffected by the Lyman-breakuntil z (cid:38)
13) and each of the medium-band filters whichcould be contaminated by [O
III ]+H β at z > − F444W colours since these are virtually iden-tical to the first two
Spitzer /IRAC bands that are tra-ditionally used for this exercise. We find four redshiftranges not seen in wide-band colours where especiallyred medium-band colours could significantly aid in con-straining the photo- z . For each of the F410M, F430M,F460M and F480M filters, these ranges peak approxi-mately at z ∼ . z ∼ . z ∼ . z ∼ . Roberts-Borsani et al. have narrow ranges ∆ z ≈ .
0, ∆ z ≈ .
7, ∆ z ≈ . z ≈ .
85, respectively. However, we do not observeany major differences across the majority of our photo- z range with the choice of the wide-band filter as longas it remains unaffected by the Lyman-break and rest-frame optical. Within the z = 7 −
12 range, we do how-ever note far smaller and less constrained trends of redcolours at redshifts of z ∼ . z ∼ .
25 using theF430M and F460M filters, respectively. These are due tothe combination of a Balmer break and/or the presenceof less strong emission lines such as [O II ] λλ III ] λλ δ . Of course, the ampli-tudes of the curves are highly dependent on the EWof the nebular emission lines (in particular [O III ]+H β )and increasing them merely increases each of the afore-mentioned peaks. While such an exercise is typicallyconducted with line strengths of EW([O III ]+H β ) > III ]+H β ) ∼ ∼ z ∼ − Spitzer /IRACimaging (Labb´e et al. 2013; Roberts-Borsani et al. 2016;De Barros et al. 2019; Endsley et al. 2021). It is thusextremely encouraging that even less extreme and morerepresentative galaxies are able to produce much morepronounced ( ∼ z ∼ . z ∼ . z ∼ . z ∼ . z ∼ . z ∼ . ∼ µ m observations (but still in-cluding sufficient imaging for dropout selections) withwhich to constrain the location of the Lyman-break tohigh precision. NIRCam’s medium-band filters providean unprecedented opportunity to better sample photo-metric redshifts and galaxy properties through cleanermeasurements of emission lines and continua over rest-frame optical wavelengths at z > Mock catalogs of star-forming galaxies
To determine and evaluate the uncertainties on rest-frame optical galaxy properties (global and resolved)that the programs described in Section 2.1 are likelyto yield, a statistical approach is required, where photo-metric catalogs of representative galaxies are fitted with W i d e - F M log M * /M =7.8 log U=-2.8 Z/Z =0.3EW([OIII]+H ) 1300 Å [ O III ] + H [ O II ] F356W-F444WF200WF277WF356WF444W W i d e - F M [ O III ] + H [ O II ] W i d e - F M [ O III ] + H [ O II ] W i d e - F M [ O III ] + H Figure 1.
The
JWST /NIRCam wide- minus medium-bandcolours as a function of photometric redshift for the (from topto bottom) F410M, F430M, F460M and F480M filters, where[O
III ]+H β is likely to transit at z >
7. Additionally, theF356W − F444W colours are also plotted as a dashed blackline. The blue and magenta fill in each panel highlights theregion where particularly red colours - driven by [O
III ]+H β and a Balmer break plus [O II ]+[N III ]+H δ , respectively - iso-late a narrow photometric redshift range, providing a usefulconstraint in the absence of deep optical imaging where theLyman-break is likely to pass through. appropriate SED-fitting techniques over the filter com-binations of interest. Thus, we begin by constructingdistributions of representative global galaxy and star-formation history properties with which to randomly de-rive input parameters for the generation of mock SEDsand photometry at high redshift. mproving z ∼ − galaxy properties with JWST /NIRCam Medium-Bands a r b i t r a r y f l u x F W F W F W F W F W F W F W F M F M F M F M F M F M F M Balmer breakz = 7.2z = 7.7z = 8.4z = 8.7z = 10.4z = 11.25
Figure 2.
A mock star-forming galaxy SED at our fiducial z >
JWST /NIRCam filter response curves (bottom, gray) and red medium-band NIRCam filters (red, top) used throughout thisstudy. The dashed vertical line marks the position of the [O
III ] λ z = 7 . The galaxy properties are derived from each of the 10realisations of the v1.2 JAdes extraGalactic UltradeepArtificial Realizations (JAGUAR; Williams et al. 2018)mock catalogs, which make use of an empirical modeldriven entirely by observational trends at 0 . < z < z = 15.Each realisation covers an area of 121 arcmin , thus thetotal area probed by the combined samples spans 1210arcmin . Since rest-frame optical wavelengths remaininaccessible with even the reddest medium-band NIR-Cam filters at z >
12, we opt to carry out our analysisat a slightly lower redshift interval of 7 (cid:46) z (cid:46)
12, wheresuch wavelengths are still accessible. Specifically, giventhe improved photometric constraints with the F410M,F430M, F460M and F480M filters discussed in Section2.2, we opt for redshift intervals of z = 7 . z = 7 . z = 8 . z = 8 . z = 10 . z = 11 .
25. For il-lustration, the star-forming SED used in Section 2.2 isshown in Figure 2 at the various redshift intervals, alongwith the NIRCam filters of interest (i.e., all of our fidu-cial wide-band filters and the red, medium-band filters).Additionally, since even the deepest ERS and GTO pro-grams will reach depths up to m ∼
29 AB, we focus ongalaxies with m F W ∼ z =0.25 either side of our fiducialredshifts, except for z = 8 . z = 8 . z =0.15 in order to prevent overlap within theredshift ranges. Additionally, we also consider sourceswith log U < − Bagpipes (Carnall et al. 2018) SED-fitting code, whichis fed randomly drawn properties from the JAGUARdistributions. We opt for this method rather than us-ing the JAGUAR photometry itself, since the numberdensity of high- z sources in JAGUAR follows observedtrends and thus drops dramatically as a function of red-shift, making a statistical analysis challenging for ourpurposes. More specifically, we use as input propertiesthe stellar mass (M ∗ ), maximum stellar age, metallic-ity ( Z , assumed to be the same for both the stars andthe gas), ionisation parameter for nebular emission (log U ), dust attentuation (A v ; assuming a Calzetti et al.2000 attenuation law) and the star formation timescale( τ ) for a declining star formation history model. Thevelocity dispersion of absorption and emission lines isfixed at 150 km s − . To ensure the derived photometryis representative of the general galaxy population, we fiteach of the relevant JAGUAR distributions with a skew-normal (log M ∗ , stellar age, Z and log U ), uniform ( τ ),or exponential (A v ) profile and sample their resulting Roberts-Borsani et al.
Parameter Allowed RangeRedshift [0 , ,
10] M (cid:12)
Maximum stellar age [0 . ,
1] GyrMetallicity [0 −
5] Z (cid:12) log U [ − , − v [0 ,
3] mag τ [ > .
01] Gyr
Table 1.
The free parameters in our declining star forma-tion history model and allowed ranges, used to fit our mockNIRCam photometry with
Bagpipes . PDFs to generate the random values. Since stellar massand stellar age show some minor correlation, these aresampled jointly. We note here that since the distribu-tions are quite similar across our redshift bins, they areunlikely to introduce strong biases in our analyses. 100different mock galaxies SEDs are subsequently generatedfrom the randomly drawn parameters and the photom-etry is then extracted for the wide-band filters and allred, medium-band NIRCam filter. The procedure is re-peated for each redshift bin and F200W apparent mag-nitudes of m F W ∼ z = 7 . − . F W ∼ z = 8 . F W ∼ z = 10 . F W ∼
29 AB for z ∼ .
25 galaxies. The magnitude bin centers are lo-cated at each integer value of the considered range, insteps of 1 mag and widths of ± σ errors are derived by scaling and combining the 5 σ ERS 1324 depth to an additional 2% absolute calibra-tion uncertainty, for each filter. In total, accountingfor the redshift and magnitude bins the above results in1,800 mock galaxy SEDs from which to extract NIRCamphotometry. SED-FITTING AND RESULTS3.1.
The accuracy of galaxy properties
Each iteration and combination of galaxy photometryis subsequently fit using
Bagpipes , this time adoptinguniform priors on each of the parameters that were usedin the construction of the mock galaxy catalogs, includ-ing their redshifts. The allowed ranges are presentedin Table 1. While the observed properties of galaxiesare highly dependent on (and also degenerate with) thechosen star formation history, the goal here is not toretrieve the true star formation histories but rather todetermine the accuracy of each retrieved parameter asa function of filter combination and input parameter. https://jwst-docs.stsci.edu/data-processing-and-calibration-files/absolute-flux-calibration With this in mind, we present the main results of ourfits for representative, star-forming galaxies in Figure 4,where we plot the accuracy of recovered galaxy proper-ties as a function of wide+medium filter combination.Here we define the accuracy, A , as the median (over the100 iterations) absolute difference between the input andrecovered parameter: A = | log (param true ) − log (param fit ) | (1)We begin by exploring the accuracy of recovered prop-erties such as galaxy SFR, stellar mass (M ∗ ), maximumstellar age, metallicity, ionisation parameter ( U ), emis-sion line fluxes (specifically [O III ] λλ β ), and dust attenuation. Not all of these are inde-pendent quantities and will have some correlation - theSFR and related quantities are calculated as the aver-age of the preferred SFH over the last 100 Myrs. Con-sidering first results using the set of wide-band filtersonly (marked by the empty diamond symbols), we findthe recovery generally fares reasonably well, since ei-ther (or both) the F356W or F444W filter probe therest-frame optical. In particular, measurements of SFR,stellar mass, maximum stellar age, ionisation parame-ter, H β emission and dust contributions generally ap-pear accurate to less than ∼ III ] λλ ∼ ∼ ∼ z ∼
9, where therest-frame optical portion of the spectrum is still well-sampled by NIRCam. At z = 10 −
11, the gains in accu-racy become much more marginal, since the rest-frameoptical portion of the spectrum begins to drop out ofeven the reddest filter (see Figure 2). The gains in ac-curacy are best seen with bright galaxies, given theirhigher S/N, however the improvement remains signifi-cant even for their fainter counterparts. For most prop-erties (e.g., SFR, stellar mass, stellar age), the improve-ments span ∼ III ] flux, the improvements canbe more significant ( > . mproving z ∼ − galaxy properties with JWST /NIRCam Medium-Bands
20 18 M UV [mag] c o un t s / p o i n t i n g z~7.2z~7.7z~8.4z~8.7z~10.4z~11.25 log M * [M ] log max age * [yrs] log Z * [Z ] c o un t s / p o i n t i n g log U Av [mag] log [yrs]
Figure 3.
The distribution of integrated galaxy properties from the JAGUAR catalogs, extracted for all star-forming galaxiesat redshifts z ∼ . z ∼ . z ∼ . z ∼ . z ∼ . z ∼ . Bagpipes . The vertical lines denote the mean value of thedistribution at each redshift interval and each histogram is normalised by the combined pointing area ( ∼ ) of bothNIRCam modules (A and B) for red wavelengths, to illustrate the expected number of sources per pointing. tities such as the SFR, stellar mass and metallicity, sincesuch quantities are paramount towards characterisingthe mass-metallicity and “main sequence” relations athigh- z . Typical measurements using HST and
Spitzer are subject to large ( > ∼ ∼ ∼ ∼ m ) and redshift bin we define the accuracygain ( g ) of a galaxy property ( p ) between the fiducialset of wide-band filters (W) and an additional medium-band filter (M) as g p ( m, z ) = A p,W ( m, z ) A p,M ( m, z ) , (2)then we can define the total (median) gain for a galaxyin a given redshift bin ( G ( z ), summed over all magnitudebins and properties presented in Figure 4) as G ( z ) = (cid:88) m (cid:88) p g p ( m, z ) , (3) We display this gain factor as an integrated quantityper galaxy (i.e., the median value per galaxy summedacross all properties and over all magnitude bins) andas a function of redshift and medium-band filter in thetop panel of Figure 5. We find virtually all medium fil-ters return similar gains for the z = 7 . × z = 7 . z = 8 . z ∼
8; with gain factors of × (cid:62) . µ m bands: at z = 7 . z = 8 . z ∼ ∼ III ] emission line flux (increasing the accu-racy by ∼ . − . z ∼ z input = [7 . , . ∼ ∼ z (cid:38)
9, where only a handful of filtersprobe the bluest regions of the rest-frame optical and thenumber density of galaxies at such higher redshift dropsdramatically; we find no major difference in accuracygain between any of the medium-band filters at those
Roberts-Borsani et al. A [ d e x ] ~26 AB ~27 AB ~28 AB ~29 AB z=7.2 WB onlyWB+MB
SFR M * age * Z U [O III] H dust A [ d e x ] z=7.7 A [ d e x ] z=8.4 A [ d e x ] z=8.7 A [ d e x ] z=10.4SFR M * age * Z U [O III] H dust A [ d e x ] z=11.25 Figure 4.
The results of the SED-fitting procedure for
JWST /NIRCam photometry. The x -axis denotes a galaxy property/freeparameter of interest (from left to right: star formation rate, stellar mass, maximum stellar age, metallicity, ionisation parameter,[O III ] λ III ] λ β flux and dust extinction), while the y -axis denotes the median (and absolute) residualbetween the (log) true and (log) recovered parameter, according to each redshift, magnitude, and filter combination. Diamondsrepresent fitting results with the wide-band filters only, while each circle denotes a combination of wide filters plus an additionalred medium-band filter. The F200W magnitudes of the galaxy models are distinguished by the different colour maps (purple,blue, pink, and red shades), while the shade of each point in a map dictates the medium filter used, with darker shadescorresponding to redder wavelengths. mproving z ∼ − galaxy properties with JWST /NIRCam Medium-Bands z counterparts, due tothe declining density of galaxies as a function of red-shift. We note, however, that the F430M filter continuesto afford the largest gains for z ∼ ∼ ∼ z improvements from Figure 6,the F430M represents in our view the optimal medium-band filter for maximising accuracy gains in both photo- z and galaxy property estimates. Of course, calculationspresented here are for especially deep observations, how-ever significant gains can also be found with shallowerdata (as illustrated by the fainter galaxies in Figure 4)which would require shorter observation times.3.2. A note on systematic uncertaintes
We have performed the relatively simple exercise ofparameter recovery adopting SED-fitting of NIRCamphotometry and a certain SFH history model, and as-sessed the precision of the resulting best fit properties.However, such a feat glosses over important caveats thatmust be taken into consideration. The first is that wehave assumed a declining SFH history model. The mo-tivation for this was to be consistent with the modelsemployed by the JAGUAR catalogs, from which we de-rive our input galaxy property distributions. However,the star formation histories of high- z galaxies are un-constrained and the adoption of a SFH model naturallyintroduces significant uncertainties which any fitting ofphotometric data sets is likely to be limited to. Secondly,we have allowed the redshifts of our fitted galaxies tovary, meaning each photometric measurement will havesome inherent uncertainty due to the resulting photo-metric redshift. While in general the redshifts of z > Y - and J -bandobservations, we find the recovered redshifts of z > z s, however invirtually every redshift bin the recovered photometricredshift is improved with the addition of a medium-band filter; the maximum difference between photo- z sfrom the various filter combinations are ∆ z = 0 .
02 for z input = 7 .
2, ∆ z = 0 .
19 for z input = 7 .
7, ∆ z = 0 .
07 for z input = 8 .
4, ∆ z = 0 .
07 for z input = 8 .
7, ∆ z = 0 .
59 for z input = 10 . z = 0 .
65 for z input = 11 .
25. Forthe aforementioned redshifts, although the differencesare sometimes small (e.g., ∆ z = 0 .
02 for z input = 7 . z is F430M, F430M, F460M, F480M,F430M, and F430M, respectively. The preferred fil-ters for z input = 7 . − . III ]+H β emission lines. Clearly,the F430M filter provides not only the most significantgalaxy properties gains, but also samples well-definedspectral features that lead to the best photo- z con-straints across the majority of redshift bins.The improvement of the photo- z s is further illustratedin the bottom panel of Figure 6, where we plot the dis-tributions of all recovered photometric redshifts usingwide-band filters only (light shaded histograms) and thewide+medium-band combinations (darker shaded his-tograms) discussed in Section 2.2 (i.e., wide filters plusF410M, F430M, F460M, F480M, F430M and F460M for z input = 7 . z input = 7 . z input = 8 . z input = 8 . z input = 10 . z input = 11 .
25, respectively), wherethe medium-band straddles strong [O
III ]+H β emissionlines. As shown in the top panel of the same figure, wefind the latter distributions are peaked far more sharplythan the former, are particularly well constrained for z input <
10 sources, and fall exactly within the expectedphoto- z range given by the especially red colours dis-cussed in Section 2.2. The same cannot be said for the z input = 10 . z input = 11 .
25 fits, however, due to thesampling of weaker features than the [O
III ]+H β lineswhich result in a less well-defined red colour, and arethus less reliable. The approximate ranges of the redcolours (due to [O III ]+H β emission lines at z <
10 and[O II ]+Balmer break emission at z >
10) shown in Figure1 are also indicated in the bottom panel of Figure 6. Al-though not shown here, we note that the distributions ofphotometric redshifts considering all the medium-bandfilters mirrors those of the wide-band filter distributions,thus highlighting in general the constraining power ofthe Lyman-break but also the increased photo- z accu-racy from the isolation of strong nebular emission lines.We conclude that while there are significant uncertain-ties in any measurement of photometric data sets due tothe assumed SFH model, the impact of photo- z s on our z <
10 samples here is not significant and the photo-0
Roberts-Borsani et al. g a i n f a c t o r / g a l a xy z=7.2z=7.7z=8.4 z=8.7z=10.4z=11.25 all z F300M F335M F360M F410M F430M F460M F480M020040060080010001200 i n t e g r a t e d g a i n f a c t o r Figure 5.
Top: The total accuracy gain (integrated over all properties and magnitudes presented in Figure 4) per galaxyas a function of redshift and additional long-wavelength MB NIRCam filter, over results inferred with wide-band photometry(adopted by ERS and GTO programs) only. Additionally, the gray bars indicate the summed gains summed over all redshiftbins. Bottom: the same as the top but multiplied by the expected number of JAGUAR galaxies per NIRCam pointing in eachmagnitude and redshift bin. The (cid:62) µ m filters clearly provide the highest gain for z ∼ z input = 7 . z input = 8 . metric redshift themselves are generally improved withthe addition of a medium filter. OBSERVATIONAL STRATEGYThe medium-band filters of greatest interest here arethe > µ m filters, given the gains in accuracy they af-ford over the z ∼ −
11 population (the F300M, F335Mand F360M do not probe redward enough to affect theproperties of z ∼ −
11 galaxies), as illustrated in thetop panel of Figure 5. As discussed in previous sections,while each of these afford similar information gains for z ∼ . z ∼ z ∼
11 galaxies, however the numbercounts of these objects are expected to be comparativelylow (see e.g., Ellis et al. 2013; Oesch et al. 2018). Thus,for studies of the z ∼ . z ∼ . z ∼ z and galaxy property estimates.Considering the above, the choice of optimal filtermust also be cross-evaluated with the amount of timerequired to reach the required imaging depths. To doso, we use the JWST Exposure Time Calculator (ETC) mproving z ∼ − galaxy properties with JWST /NIRCam Medium-Bands m e d i a n z p h o t z input =7.2 z input =7.7 z input =8.4 z input =8.7 z input =10.4 z input =11.25 c o un t s z = 7.2F410M z = 7.7F430M z = 8.4F460M z = 8.7F480M z = 10.4F430M z = 11.25F460M Figure 6.
Top panel: The median (absolute) difference between the input redshift and recovered photometric redshift ( | z input − z output | ) as a function of wide-only (empty diamonds) and wide+medium (filled circles) filter combinations. The median is takenover all of the galaxies in a given redshift bin. The redshift colour scheme follows that of Figure 3, with darker shades of agiven colour map indicating redder-wavelength medium filters. Bottom panel: A comparison of the distributions of recoveredredshifts using wide filters only (light histograms) or wide+medium filters (darker histograms; F410M for z ∼ .
2, F430M for z ∼ . z ∼ .
4, F460M for z ∼ . z ∼ .
25, and F480M for z ∼ .
7; see Figure 1 and Section 2.2 for details).Clearly, the addition of a strategically-placed > µ m medium-band filter significantly improves the photo- z estimate at z < to estimate the amount of time required to get to a 5 σ depth of ∼ > µ m fil-ters. We perform the calculations assuming either 6 or8 groups per exposure and either 1 or 3 integrations perexposure and a “DEEP8” read mode. The resulting me-dian (interpolated) depths and 1 σ standard deviationsare presented in Figure 7. For each of the four > µ mfilters to reach the aforementioned depth, we find roughestimates of ∼ ∼ ∼ ∼ >
40 ks ( >
11 hrs) of required observing timefor the F410M, F430M and F460M/F480M filters, re-spectively. These results clearly illustrate the delicatebalance between opting for filters that yield the mostsignificant gains (i.e., F430M) or those which requireless exposure time (e.g., F410M); indeed, with a rela-tively constant F430M/F410M exposure time ratio of ∼ .
5, the difference in overhead and required expo-sure time for depths less than ∼ . µ m and thus is not suitable for a comparisonhere), we use the star-forming galaxy spectrum shownin Figure 2 at z = 8 . F W ∼
26 ABand with EW([O
III ]+H β ∼ ∼ (cid:48)(cid:48) × (cid:48)(cid:48) (cid:48)(cid:48) ) would be sufficient only to obtain ∼ . σ − . σ detections across the continuum bluewardof 4 µ m but ∼ σ detections of weak nebular emissionlines and > σ detections of strong lines such as [O II ],H β and both [O III ] lines. Fortunately, NIRSpec Prism2
Roberts-Borsani et al. d e p t h [ A B ] F410MF430MF460MF480M
Figure 7.
NIRCam imaging (5 σ ) depths estimated usingthe JWST
ETC, as a function of time and for a variety ofobservational settings. Each colour corresponds to a differentmedium-band filter. Circles mark the median depth betweeneach of the settings, while error bars mark their associated 1 σ standard deviation. Results for the F460M filter are virtuallyidentical to those for the F480M filter and thus lie behindthose points. capabilities offer far more sensitive spectroscopic obser-vations, with only ∼ > σ con-tinuum detections across all wavelengths blueward of 4 µ m, and ∼ σ − σ detections of the [O III ]+H β lines.However, it is important to note that while using fixedslit or MOS observations with NIRSpec may afford highS/N spectra, they do not afford the same sky cover-age as NIRCam imaging, and thus are better placed forthe characterisation of known objects rather than thediscovery and simultaneous characterisation of galaxypopulations across large (and potentially unknown) ar-eas of the sky. CONCLUSIONSWe have conducted an analysis of galaxy parameterrecovery and accuracy through mock
JWST /NIRCamgalaxy photometry and SED-fitting, at redshifts z ∼ −
11. While future wide-band imaging from ERS andGTO programs will undoubtedly offer unprecedenteddepth and quality at NIR wavelengths, their large band-widths prevent the isolation and precise characterisation of rest-frame optical galaxy properties. We find the ad-dition of a single, red medium-band NIRCam filter towide-band filters adopted by ERS and GTO programs(and at comparable depth) can significantly increase theaccuracy of a large number of galaxy parameters (globalor spatially-resolved), in particular galaxy metallicity,stellar age and mass, and emission line flux. The differ-ence in accuracy can span up to ∼ z < z ∼ − z ∼ −
11 populations. We argue the additionof a single, strategically-placed medium-band filter toexisting ERS and GTO data sets offers a valuable andrelatively inexpensive approach towards maximising thecharacterisation of the z ∼ −
11 galaxy populations,especially in light of the short lifespan of
JWST - whilethe final choice of filter will inevitably depend on thescience goal and target, we find on a per-galaxy basisthe F430M filter affords the most significant gains in ac-curacy for both z ∼ Bagpipes
SED-fittingcode.REFERENCES
Behroozi, P., Conroy, C., Wechsler, R. H., et al. 2020,MNRAS, 499, 5702. doi:10.1093/mnras/staa3164 Bradley, L. D., Trenti, M., Oesch, P. A., et al. 2012, ApJ,760, 108Bradley, L. D., Zitrin, A., Coe, D., et al. 2014, ApJ, 792, 76 mproving z ∼ − galaxy properties with JWST /NIRCam Medium-Bands13