#   Comments can be added freely as lines starting with '#'
#
#   'sigma.txt' is the replacement for 'weights.txt',  and renders it
# obsolete.  It is similar to 'weights.txt',  but extends its capabilities.
#
#   'sigma.txt' allows you to set up rules describing the default
# uncertainties in position,  time,  and magnitude for astrometric data.
# The idea is that you can tell Find_Orb that all observations from a
# given observatory over a given time span and magnitude range have,
# by your estimate,  specific uncertainties in those three quantities.
#
#   The resulting uncertainties are simply default values,  used when
# there's no other way to know the uncertainty.  If you use the
# AstDyS/NEODyS .rwo format or the ADES format,  the data have
# astrometric uncertainties given with each observation.  And
# Find_Orb provides ways to specify uncertainties via special keywords
# within the observations.  See
#
# https://www.projectpluto.com/find_orb.htm#sigmas
#
#  for details.
#
#   Also,  note that if you revise this file,  it's recommended to rename
# it and to put the new name on the SIGMAS_FILE line in 'environ.dat'.
# If you just edit this file without renaming,  it will get clobbered
# the next time you update.
#
#   This is a revised version of what used to be 'weight.txt'.  That
# file was quite similar,  except that instead of storing uncertainties,
# weights were stored.  These basically just boiled down to 1/sigma_posn,
# with sigma_posn expressed in arcseconds.  Also,  only sigmas for
# position could be defined;  'sigma.txt' adds sigmas for magnitude
# and time as well,  and allows you to specify a project code.
#
#   Find_Orb looks through this file from top to bottom.  Non-comment
# lines give an observatory code,  an optional program code,  and a date
# range and magnitude range.  (The ranges can be blank,  to specify "any
# date" and/or "any magnitude",  or to indicate that there is no lower
# or upper limit on that range.)  The idea is that such a line defines
# what the sigmas are for any observation from that code,  taken within
# that date range,  and within that magnitude range.
#
#   Assigning uncertainties this way isn't a great idea.  Just because
# two observations were made at around the same time from the same place
# with comparable magnitudes doesn't mean they'll have similar uncertainties.
# This is a hack to get around the absence of observer-supplied sigmas.
#
#   When Find_Orb finds a line describing the observation in question,
# it extracts any non-blank sigmas.  It keeps going,  possibly to
# the bottom of the file,  until sigmas have been found for all
# three quantities. That's why there is a final line that will
# assign sigmas to _any_ observation that hasn't already been handled.
#
#    Comments can be added at the end of each line.  I've just listed
# the station names in that slot.
#
#   Some comments on how to assign sigmas are at the bottom of this text.
# I've noticed that the following stations get unusually good astrometry.
# These uncertainties roughly reflect how well I've seen them do.  Note
# that (413) made some pre-CCD observations that ought to just get the
# default "old-era",  larger uncertainties.
#
#Obs P  <--start-> <--end -->           Pos   Mag  Time
#COD C  yyyy mm dd yyyy mm dd mag1 mag2 sig   sig  sig    Comment
 689                                    .1                Flagstaff
 422                                    .3                Loomberah
 413    1993 01 01                      .3                Siding Spring
 E12                                    .25               Siding Spring #2
 568 2                                  .1          .2    David Tholen et. al., Mauna Kea
 T09                                    .1          .2    David Tholen et. al., Subaru
 T12                                    .1          .2    David Tholen et. al., 2.24-m
 T14                                    .1          .2    David Tholen et. al., CFHT
 568 ~                                  .1          .2    TNO survey folks, Mauna Kea
 J95                                    .2          .2    Great Shefford
 J75                                    .2          .1    La Sagra
 Y00                                                .2    SONEAR
 F51                                    .2                PanSTARRS
 F52                                    .2                PanSTARRS 2
#   (J95), (Y00) and Tholen at (568) have demonstrated,  via observations of
# artsats,  that they really have gotten timing right.  (J75) has had
# cause to observe very fast-moving artsats;  their actual time sigma
# is probably quite a bit better than the above .1 second.
#
#    SOHO and STEREO,  though,  were not intended to do astrometry and
# produce noisy data.  "Program codes" A and 2 = SOHO C2; B and 3=
# SOHO C3.
#    For SOHO,  we check for "program codes" A or 2,  meaning C2
# was used.  Everything else from (249) is assumed to use the
# lower-accuracy C3.  (People sometimes mangle the codes;  we
# want to default to the lower accuracy in such cases.)
 249 A                         8.7      36                SOHO
 249 2                         8.7      36                SOHO
 249 A                         8.2  8.8 24                SOHO
 249 2                         8.2  8.8 24                SOHO
 249 A                         7.2  8.2 12                SOHO
 249 2                         7.2  8.2 12                SOHO
 249 A                         4.2  7.2  6                SOHO
 249 2                         4.2  7.2  6                SOHO
 249 A                         2.7  4.2 12                SOHO
 249 2                         2.7  4.2 12                SOHO
 249 A                         1.7  2.7 18                SOHO
 249 2                         1.7  2.7 18                SOHO
 249 A                         0.2  1.7 24                SOHO
 249 2                         0.2  1.7 24                SOHO
 249 A                        -1.8  0.2 36                SOHO
 249 2                        -1.8  0.2 36                SOHO
 249 A                             -1.8 48                SOHO
 249 2                             -1.8 48                SOHO
 249                           9.7      280               SOHO
 249                           9.2  9.7 224               SOHO
 249                           7.7  9.2 168               SOHO
 249                           7.2  7.7 112               SOHO
 249                           6.7  7.2 84                SOHO
 249                           5.7  6.7 56                SOHO
 249                           2.7  5.7 28                SOHO
 249                           1.2  2.7 56                SOHO
 249                          -0.2  1.2 84                SOHO
 249                          -1.2 -0.2 112               SOHO
 249                               -1.2 168               SOHO

# Karl also guesstimates sigmas for HI-1 to be pretty bad,  maybe
# two 70" pixels.  COR-2 is much better by comparison,  10".
# G or 4 = STEREO/SECCHI HI1-A; H or 5 = COR2-A (for code C49);
# O or 4 = STEREO/SECCHI HI1-B; P or 5 = COR2-B (for code C50).
# These are also from an e-mail from Karl Battams,  2019 Nov 1.
 C49 H                                  10                STEREO-A
 C49 5                                  10                STEREO-A
 C49                                    140               STEREO-A
 C50 P                                  10                STEREO-A
 C50 5                                  10                STEREO-B
 C50                                    140               STEREO-B
# Still puzzling out how Solar Orbiter astrometry should be handled.
# Based on exactly one object,  let's say 20".
 SoO                                    20                Solar Orbiter
# TESS data has (I assume) beautiful photometry,  but the astrometry
# appears to be approximate at best.
 C57                                     5                TESS
# MPC appears to be weighting (C51) WISE data lightly,  which appears to
# be... well... wise.  I am also informed by Joe Masiero that the mag
# data from WISE (which is an R magnitude guesstimated from W1 and W2)
# is very approximate.
 C51                                     1      3         WISE
# ...but Hipparcos and HST get incredibly good astrometry:
 248                                    .01               Hipparcos
 250                                    .01               HST
# The radar folks don't really have 'uncertainties' in RA/dec or mag;
# they aren't measuring those quantities.  The quantities they _do_
# measure -- round-trip time and Doppler shift -- have sigmas given
# for them,  provided along with the data;  we don't need to specify
# them in 'sigma.txt'.  The only thing we need to specify is the
# uncertainty in the timing.  That sigma really should be effectively
# zero.  We can't actually do that,  but a nanosecond is good enough.
 251                                     7         1e-9   Arecibo
 252                                     7         1e-9   Goldstone DSS 13, Fort Irwin
 253                                     7         1e-9   Goldstone DSS 13, Fort Irwin
 254                                     7         1e-9   Haystack, Westford
 255                                     7         1e-9   Evpatoria
 256                                     7         1e-9   Green Bank
 257                                     7         1e-9   Goldstone DSS 25, Fort Irwin
 259                                     7         1e-9   EISCAT Tromso UHF
#  Marco Micheli tells me that (Z84) Calar Alto has been tested
#  to be good to about 0.05 seconds.  He usually provides positional
#  uncertainties with the data,  so I'll just specify a 0.1-s timing
#  uncertainty here.
 Z84                                                .1     Calar Alto
#  Gaia data will carry its own sigmas,  except for time.  We'll set that
# to be effectively zero,  same as for radar.
 258                                     7         1e-9   Gaia
# Occultations give extremely precise positions.  (Unfortunately,
# the 80-column astrometric format won't give us all of that.)
# I assume accuracy in time to a video frame = 1/60 second.
 244                                    .001       .017   Geocentric occultation
#  You MUST have some final "catchall" lines that assign sigmas to all
# observations not otherwise catalogued.  Here,  we'll say that if the
# observation predates 1993,  we'll assume a sigma of 3",  no matter
# which observatory it came from.  (Older observations are usually
# photographic,  and reduced with older catalogues such as the SAO;
# the assumption that they're just a little bit dodgy is usually right.)
                   1993 01 01            3
# ...which will just leave everything after that date,  and we'll
# assume a default sigma for those (more "modern" CCD obs) of a half
# arcsecond,  half a magnitude,  and one second.  This is the "catch-all"
# line mentioned above,  to assign uncertainties to anything not already
# found.  It's also used if you have turned off weighting (i.e.,  all
# observations are to be treated equally,  with identical uncertainties).
                                        .5     .5    1    Default
#
# HOW TO ASSIGN SIGMAS:  I don't really know of a simple way to do this.
# The above numbers reflect what I've observed in the data,  and are
# better than no sigmas at all,  but they are somewhat ad hoc.
#
#   One _can_ analyze data from an individual observatory and get a feel
# for what their RMS residuals are and base the sigmas on that.  It helps
# if the observatory in question gets a _lot_ of data,  to get some
# meaningful statistics.  .rwo files give uncertainties assigned in
# this manner.
