APOGEE DR12 Status/Discussion
Schedule
• Nominal DR12 deadline is July 31, i.e. tomorrow
• SDSS-III upper level management was consulted
about a small delay to take advantage of feedback
from this meeting. Adam Bolton:
– “From July 31 to August 31, the schedule impact of this
goes more or less linearly from "no big deal" to "we're
totally hosed". By "we're totally hosed", I mean that we
will be forced to choose between either leaving APOGEE
out of DR12 CAS or delaying the whole of DR12
significantly.”
• Practical deadline perhaps 8/15
DR12 products to be delivered
• SAS
– All raw and intermediate data files
– Final summary: allStar, allVisit files
• Data model for parameters/abundances:
– FPARAM, PARAM, TEFF, LOGG
– FELEM, ELEM (in native FERRE) + X_H
• SASDB (allows web/API interface to SAS files)
– Populated from apField, aspcapField files
• CAS
– Datamodel for abundances: translate arrays into prefexed parameters, have
named parameters for TEFF, LOGG, abundances
• FPARAM_TEFF, FPARAM_LOGG, FPARAM_M_H, etc., PARAM_..., TEFF, LOGG
• FELEM_X_[MH], ELEM_X_[MH], X_H, ….
– Populated from allStar, allVisit files
• DR12 web pages
• Datamodel complete and up-to-date
Focus: key issues
• Current set of files exists
– Reduction/star combination final (?!)
– ASPCAP will be rerun, but test files exist for proposed final run
– Files need to be checked for datamodel, implementation
errors/issues
• Is there anything significantly wrong with basic contents of
current files: bar needs to be high to modify significantly (OK
to modify calibration decisions)
• Issues
– Calibrations to apply
– Flags to apply
– Discussion/documentation issues:
• What do we say about quality/usability for individual elements
• Unexpected abundance trends in some elements
• More S/N sensitivity than expected
Scope/analysis
• New reductions of all data taken to date
– Reduction version r5 has improved telluric, etc.
– Includes apo25m commissioning/survey + apo1m
• ASPCAP analysis
– ASPCAP configuration l23
• December 2013 linelist (not quite latest: incorrect treatment of Sun
when deriving astrophysical gf)
• ASSET synthesis using solar isotope ratio
• Same grid dimensions/spacing as DR10
• lsfcombo5 LSF: average of fibers 50,100,150,200,250
• speclib bundling of libraries
– 7d analysis of calibration subsample (parameters only), from
which logg-vmicro relation is derived to construct 6d library
– 6d library run on calibration subsample
– Calibrations derived from 6d results of calibration sample
– 6d library run on full sample, applying calibrations
microturbulence
• Proposed relation: linear fit as f(log g)
– Similar to DR10 relation
– Dwarfs are hitting bottom edge of
microturbulence dimesion
– Fit derived from giants only
• Cluster parameters appear very similar using
adopted relation
Calibration
• Raw ASPCAP data probably needs to be “calibrated” for
some quantities
– DR10: calibrated Teff, logg, [M/H]
• DR12 proposal
– Teff: calibrate using photometric (GHB) temperatures for stars
with well-measured extinctions/temperatures: functional
dependence?
– Logg: calibrated using Kepler sample of RGB stars, as f(log g)
– Other “intermediate” parameters: no calibration
– Elemental abundances:
• “internal” calibration: correction as f(Teff) to minimize trends of
abundance of each element within clusters (no C or N)
• “external” calibration
– Attempting to have all data/routines used for calibration
archived within SDSS software product idlwrap
Teff calibration
• TBD
Log g calibration via Kepler
asteroseismic gravities
Abundances: internal calibration
• Small trends of abundance with temperature
seen for some elements within clusters
• Propose to fit this trend with metallicitydependent slope
– Linear relation
– Slope frozen outside of specified metallicity range
• Is this a good idea?
– Is linear trend OK?
Cluster plots
Full sample plots
Uncertainties in abundances
• Propose to fit scatter within clusters as
f(Teff,[M/H],S/N) with linear relations,
truncated outside of calibration range
• Needs checking
• Need decision on “raw” uncertainty
computation from FERRE; do any options
provide meaningful errors in regimes of large
uncertainty (e.g. non-detections)?
Abundances: external calibration
• Challenge: lack of external calibration samples for
individual elements
• [Fe/H] suggests trends similar to those seen in
[M/H] for DR10: relatively accurate around solar
but off by ~0.2 dex at low metallicity
– Should we correct this?
– Issue: if we correct [Fe/H], then when people look at
[X/Fe], they will all be shifted at low metallicity
– What to do for other elements?
• External cal plots
DR12 products
• SAS
– All raw and intermediate data files
– Final summary: allStar, allVisit files
• Data model for parameters/abundances:
– FPARAM, PARAM, TEFF, LOGG
– FELEM, ELEM (in native FERRE) + X_H
• SASDB (allows web/API interface to SAS files)
– Populated from apField, aspcapField files
• CAS
– Datamodel for abundances: translate arrays into prefexed parameters, have
named parameters for TEFF, LOGG, abundances
• FPARAM_TEFF, FPARAM_LOGG, FPARAM_M_H, etc., PARAM_..., TEFF, LOGG
• FELEM_X_[MH], ELEM_X_[MH], X_H, ….
– Populated from allStar, allVisit files
• DR12 web pages
• Datamodel complete and up-to-date
Work session
• Task list for checks
• Summary files: allStar, allVisit
– Compare uncalibrated vs calibrated (calibrations applied
correctly?)
– Play with all quantities!
• DR12 documentation
– All pages up-to-date
– Description/discussion of abundances
• Datamodel
– Complete and up-to-date?
– All SAS files should be in datamodel: but start with most
important ones!
Other open issues
• Hot stars: hot grid not currently implemented;
F grid goes to 8000, but with 6D fits
• Low metallicity stars
• Dwarfs: generally poor results
Other versions
• V304 (DR10)
• V402 (DR11): extend DR10 to year 1+2, flat
files have first cut at 9 elements
• V500: first cut 15 elements
• casset: new version synthetic libraries with
self-consistent atmospheres
• l22_6d: new reductions
•  v600 (DR12): with calibrationsdr12