Loading ...
Sorry, an error occurred while loading the content.
 

Re: 2012 DA14 a threat to geostationary satellites?

Expand Messages
  • wlodarczyk_i
    All virtual asteroids (VA) of 2012 DA14 computed according to 1-sigma variation go to the geosynchronous orbit not closer then about 27000 km. All 191
    Message 1 of 24 , May 21, 2012
      All virtual asteroids (VA) of 2012 DA14 computed according to 1-sigma variation
      go to the geosynchronous orbit not closer then about 27000 km.
      All 191 observations of the asteroid 2012 DA14 and the OrbFit software were used.
      Best regards,
      Ireneusz Wlodarczyk,
      553 Chorzow


      --- In mpml@yahoogroups.com, "Aldo Vitagliano" <alvitagl@...> wrote:
      >
      >
      > At 13:28 20/05/2012 -0000, wlodarczyk_i wrote:
      >
      > >All virtual asteroids (VA)
      > >of 2012 DA14 computed according to 1-sigma variation criterion
      > > goes closer on 2013, February 15th, to the Earth than the radius of
      > > the geosynchronous orbit
      > > i.e. about 42164 km from the center of the Earth.
      >
      > That's correct, but I was not mentioning the minimum distance of approach.
      > I was talking about the distance at which the asteroid crosses the Equator plane (where the geosynchronous satellites are located), which is a different thing.
      >
      > Regards
      > Aldo
      >
    • Bill J Gray
      ... It s true that this is all of purely intellectual interest, and that the greater danger does come from marbles (whether one-centimeter metric marbles or
      Message 2 of 24 , May 21, 2012
        wlodarczyk_i wrote:
        > All virtual asteroids (VA) of 2012 DA14 computed according to 1-sigma variation
        > go to the geosynchronous orbit not closer then about 27000 km.
        > All 191 observations of the asteroid 2012 DA14 and the OrbFit software were used.
        > Best regards,
        > Ireneusz Wlodarczyk,
        > 553 Chorzow

        It's true that this is all of purely intellectual interest, and that
        the greater danger does come from marbles (whether one-centimeter metric
        marbles or 3/8 inch "standard" marbles), and the whole "threat" to
        geosats from big rocks is entirely meaningless. Despite all that, I
        couldn't resist running the numbers.

        Using all observations and the Find_Orb software, I'm finding that there
        are "collisions" at about +1.03 and +3.02 sigmas. Use of all observations is
        probably not a good idea here; three of them look iffy. But dropping them
        doesn't change the results by much.

        OrbFit uses different weighting, so your results (if you extended the
        search out a few sigmas) would be slightly different.

        I hope the above doesn't persuade any journalists on this list to
        produce articles about the dire risk to geosats. Hence my adding [NO]
        to the subject line.

        -- Bill
      • Dave Tholen
        ... Only three? I threw out fifteen. The discovery observation is bad, as is the third observation on that same night. The entire set from code 950 on
        Message 3 of 24 , May 21, 2012
          > Use of all observations is probably not a good idea here; three of
          > them look iffy.

          Only three? I threw out fifteen. The discovery observation is bad, as
          is the third observation on that same night. The entire set from code
          950 on February 29 is systematically north by an arcsecond. There are
          a few other scattered observations that stick out. Our own observations
          from April 23 appear systematically west by a quarter arcsecond, which
          is annoying, because we had hoped the PPMXL catalog would be better than
          that.
        • Bill J Gray
          ... You re probably on the right track there. I will admit that I have no consistent policy on what observations I toss (*), but with 191 of them, it s hard
          Message 4 of 24 , May 23, 2012
            Dave Tholen wrote:
            >> Use of all observations is probably not a good idea here; three of
            >> them look iffy.
            >
            > Only three? I threw out fifteen. The discovery observation is bad, as
            > is the third observation on that same night.

            You're probably on the right track there. I will admit that I have no
            consistent policy on what observations I toss (*), but with 191 of them,
            it's hard to see how throwing out 8% of them will do actual harm.

            > The entire set from code 950 on February 29 is systematically north
            > by an arcsecond. There are a few other scattered observations that stick out.

            It looks as if (950) had about an 18-second clock error. If observations
            were scarcer, we'd want the capability to mark those as having a large
            sigma on the observed time, though still good (probably!) in position.
            And ideally, we'd be able to solve for a single clock error value for
            all five positions in question... of course, for this case, we might
            as well just throw out those five positions; but I've seen lots of cases
            where being able to say, "I know the object really was at this RA/dec;
            I just don't know exactly _when_ it was there" would have been a useful
            constraint on the orbit.

            > Our own observations from April 23 appear systematically west by a
            > quarter arcsecond, which is annoying, because we had hoped the PPMXL
            > catalog would be better than that.

            I just ran some Monte Carlo clones, and for April 23, they're scattered
            over a circle about .3 arcsec across. It could be you and PPMXL are right
            and the nominal position was wrong. It might be interesting to see what
            reduction against UCAC-4 gives you.

            If I weighted your observations heavily, as I'd normally do, it would
            probably both shrink that circle and pull it toward your observations. In
            this case, though, I'm assuming that your observations at mag 23.5 are
            not quite as solid as your usual data. (Though I'm glad to see you went
            after it, both in late April and on 12 May. I've not checked yet, but
            I assume this will bring down the uncertainty for the recovery next February.)

            -- Bill

            (*) I'm open to suggestions on such a policy. The usual one seems to be
            "toss anything with residuals higher than X", which isn't terrible, but
            does have drawbacks.
          • Aldo Vitagliano
            ... In this context it is interesting to compare the results obtained with and without adding to the observations the biases reported by the Neodys database.
            Message 5 of 24 , May 24, 2012
              --- In mpml@yahoogroups.com, Bill J Gray <pluto@...> wrote:
              ... but with 191 of them,
              > it's hard to see how throwing out 8% of them will do actual harm.
              >

              In this context it is interesting to compare the results obtained with and without adding to the observations the "biases" reported by the Neodys database. Normally, the shorter the observational arc, the heavier is the "weight" of the biases in affecting the result.
              In the case of 2012DA14, using the same observations "accepted" by Neodys (4 rejected), the same "weights" and adding the biases, I get a nominal minimum geocentric approach distance of 33950 km, quite consistent with the 0.0002265 AU (33880 km) given by Neodys.
              Without adding the biases, I get a nominal approach distance of 39300 km, larger than the former by about 1-sigma.

              Best regards
              Aldo
            • Dave Tholen
              ... We are not among the testers to whom the pre-release version of UCAC4 was sent, nor do we usually have enough of the brighter reference stars in our images
              Message 6 of 24 , May 24, 2012
                >> Our own observations from April 23 appear systematically west by a
                >> quarter arcsecond, which is annoying, because we had hoped the PPMXL
                >> catalog would be better than that.

                > I just ran some Monte Carlo clones, and for April 23, they're scattered
                > over a circle about .3 arcsec across. It could be you and PPMXL are right
                > and the nominal position was wrong. It might be interesting to see what
                > reduction against UCAC-4 gives you.

                We are not among the testers to whom the pre-release version of UCAC4
                was sent, nor do we usually have enough of the brighter reference
                stars in our images to use the less dense UCAC catalogs. The best
                we could likely do is to compare the PPMXL and UCAC4 catalogs over a
                sufficiently large field centered on each of our exposures, and we'd
                have to wait for the expected June release of the UCAC4 catalog.

                > If I weighted your observations heavily, as I'd normally do, it would
                > probably both shrink that circle and pull it toward your observations. In
                > this case, though, I'm assuming that your observations at mag 23.5 are
                > not quite as solid as your usual data. (Though I'm glad to see you went
                > after it, both in late April and on 12 May. I've not checked yet, but
                > I assume this will bring down the uncertainty for the recovery next February.)

                Our computed astrometric uncertainties are, in chronological order,
                0.17, 0.10, 0.07, 0.08, 0.10, 0.09, and 0.09 arcsec.

                > (*) I'm open to suggestions on such a policy. The usual one seems to be
                > "toss anything with residuals higher than X", which isn't terrible, but
                > does have drawbacks.

                We have been (slowly!) implementing the Peirce criterion for data rejection.
                It is superior to the Chauvenet criterion that we have been using, in that
                the Peirce criterion allows for multiple rejections. Marco describes it in
                our paper on 2009 BD, with references to some very old literature. Seems
                to have been largely forgotten; time to resurrect it!
              • wlodarczyk_i
                Using the OrbFit software I got similar results as Aldo Vitagliano for minimum geocentric close approache of the nominal orbit of asteroid 2012 DA14 with the
                Message 7 of 24 , May 24, 2012
                  Using the OrbFit software I got similar results as Aldo Vitagliano
                  for minimum geocentric close approache
                  of the nominal orbit of asteroid 2012 DA14 with the Earth on 15/Feb/2013:
                  with biases 0.00022646 AU = 33878 km
                  without biases 0.00025908 AU = 38758 km
                  It is still far from the MPC result:
                  0.0003012 AU = 45060 km
                  Best regards,
                  Ireneusz Wlodarczyk
                  553 Chorzow

                  --- In mpml@yahoogroups.com, "Aldo Vitagliano" <alvitagl@...> wrote:
                  >
                  >
                  > --- In mpml@yahoogroups.com, Bill J Gray <pluto@> wrote:
                  > ... but with 191 of them,
                  > > it's hard to see how throwing out 8% of them will do actual harm.
                  > >
                  >
                  > In this context it is interesting to compare the results obtained with and without adding to the observations the "biases" reported by the Neodys database. Normally, the shorter the observational arc, the heavier is the "weight" of the biases in affecting the result.
                  > In the case of 2012DA14, using the same observations "accepted" by Neodys (4 rejected), the same "weights" and adding the biases, I get a nominal minimum geocentric approach distance of 33950 km, quite consistent with the 0.0002265 AU (33880 km) given by Neodys.
                  > Without adding the biases, I get a nominal approach distance of 39300 km, larger than the former by about 1-sigma.
                  >
                  > Best regards
                  > Aldo
                  >
                • Bill J Gray
                  ... It occurred to me, after I wrote, that UCACx might not help much with an observation at mag 23.5. ... I first learned of the Peirce criterion from that
                  Message 8 of 24 , May 24, 2012
                    Dave Tholen wrote:

                    > We are not among the testers to whom the pre-release version of UCAC4
                    > was sent, nor do we usually have enough of the brighter reference
                    > stars in our images to use the less dense UCAC catalogs.

                    It occurred to me, after I wrote, that UCACx might not help much
                    with an observation at mag 23.5.
                    >
                    > We have been (slowly!) implementing the Peirce criterion for data rejection.
                    > It is superior to the Chauvenet criterion that we have been using, in that
                    > the Peirce criterion allows for multiple rejections. Marco describes it in
                    > our paper on 2009 BD, with references to some very old literature. Seems
                    > to have been largely forgotten; time to resurrect it!

                    I first learned of the Peirce criterion from that very paper. Just looked
                    into it a little further... it looks promising. The basic idea actually looks
                    pretty simple; it's the implementation of it that looks tricky. I can see why
                    you're implementing it "slowly".

                    -- Bill
                  • Brian Skiff
                    FWIW, Steve Levine here at Lowell made all-sky plots of mean proper motions for the fainter stars in UCAC2, UCAC3, and UCAC4 (the latter from a disc that
                    Message 9 of 24 , May 24, 2012
                      FWIW, Steve Levine here at Lowell made all-sky plots of
                      mean proper motions for the fainter stars in UCAC2, UCAC3, and UCAC4
                      (the latter from a disc that Norbert Zacharias sent him).
                      UCAC2 shows the bulk solar-reflex motion of the field stars
                      very clearly (and the solar apex and antapex locations); UCAC3 shows
                      gibberish with scatter much larger than UCAC2 (we already knew UCAC3
                      was rather poor); the UCAC4 plot is nearly identical to the UCAC2 plot.
                      So at least in terms of the precision of the proper motions, it looks
                      as though UCAC4 gets back to the quality of UCAC2. That's good news!
                      In use on a star-by-star basis, I have not found PPMXL to be
                      any good at all. Since it uses USNO-B1.0 as the 'seed' star-list,
                      if B1.0 is no good, generally neither is PPMXL. The GIGO law still
                      applies! Basically the linkage of the epochs involved with B1.0 and
                      2MASS is just as bad as with B1.0. An easy way to see this is to
                      grab a small chunk of sky using VizieR (PPMXL is item I/317) and look
                      down the list to see what the general run of proper motions is.
                      An amazing number of the stars have motions >> 50 mas/year, whereas
                      for the very faint stars (statistically very distant) involved
                      you expect essentially zero motion plus whatever the errors of the
                      catalogue are (the authors claim < 10 mas/year). Instead what you see
                      are zillions of stars that ought to be in catalogues of
                      large-proper-motion objects. Instead _none_ of them is, because the
                      PPMXL motions (and thus positions) are no good.
                      Occasionally I have found cases where PPMXL gets a star right
                      when B1.0 does not. I suppose for global astrometric solutions,
                      if you can iterate to throw out half or more of the stars, maybe things
                      would be okay, but one wonders whether this leaves you offset from the
                      nominal ICRS reference frame.
                      For very faint stars (only), probably still the best single-source
                      all-sky astrometry would come from USNO-B1.0 _without_applying_
                      _proper_motions_. This time of year, in the north, I would definitely
                      try to use SDSS DR7. (Avoid SDSS DR8 since both the astrometry and
                      the photometry is messed-up there.) I wonder if one could use
                      a subset of the WISE catalogue of objects with small internal errors
                      (probably 0".10 and smaller). That has been adjusted to be very close
                      to the ICRS and the mean epoch is about 2010.5. Differentially, at
                      least, I know that the astrometric errors are < 0".1 for the stars
                      with good S/N. Other recent-epoch sources that reach to faint limits,
                      but covering only small patches of sky, include the UKIDSS and
                      2MASS 6x catalogues.


                      \Brian



                      > > I just ran some Monte Carlo clones, and for April 23, they're scattered
                      > > over a circle about .3 arcsec across. It could be you and PPMXL are right
                      > > and the nominal position was wrong. It might be interesting to see what
                      > > reduction against UCAC-4 gives you.
                      >
                      > We are not among the testers to whom the pre-release version of UCAC4
                      > was sent, nor do we usually have enough of the brighter reference
                      > stars in our images to use the less dense UCAC catalogs. The best
                      > we could likely do is to compare the PPMXL and UCAC4 catalogs over a
                      > sufficiently large field centered on each of our exposures, and we'd
                      > have to wait for the expected June release of the UCAC4 catalog.
                    • Dave Tholen
                      ... We typically reject anything with a residual of 0.5 arcsec or larger. Yes, I know such a practice would make a statistician cringe, but as the saying
                      Message 10 of 24 , May 24, 2012
                        > Occasionally I have found cases where PPMXL gets a star right
                        > when B1.0 does not. I suppose for global astrometric solutions,
                        > if you can iterate to throw out half or more of the stars, maybe things
                        > would be okay, but one wonders whether this leaves you offset from the
                        > nominal ICRS reference frame.

                        We typically reject anything with a residual of 0.5 arcsec or
                        larger. Yes, I know such a practice would make a statistician
                        cringe, but as the saying goes, the proof of the pudding is in
                        the eating, and we've been getting decent results. The open
                        question is the level of bias, and is that bias random in
                        the all-sky sense? For now, until we have enough experience
                        to suggest otherwise, we're assuming the same level of bias
                        as for 2MASS, namely 20 mas. Random errors of the individual
                        stars are not any better than for USNO-B1.0, at least in the
                        magnitude range we normally use. Curiously, B1.0 had stated
                        uncertainties for star positions as large as 999 mas, but the
                        largest in the PPMXL catalog is 210 mas. Basically we ignore
                        the stated uncertainties.

                        Hang in there, Gaia's coming.
                      • Dave Tholen
                        ... Forgot to mention in my previous response that when I built a custom version of PPMXL for asteroid use, I compiled a list of anomalies. The catalog
                        Message 11 of 24 , May 24, 2012
                          > An amazing number of the stars have motions >> 50 mas/year

                          Forgot to mention in my previous response that when I built a
                          custom version of PPMXL for asteroid use, I compiled a list of
                          anomalies. The catalog contains 7226 coordinates with proper
                          motions in excess of 5000 mas/year. Barnard's Star is in the
                          catalog twice. PPMXL carries over the problem of bogus doubles
                          in the B1.0, but a little surprised that Barnard's Star is one
                          of them.
                        • John Menke
                          A number of us on the US East Coast attempted to observe/measure the June 13 Pluto occultation event. Many were clouded out, but here in MD I fortunately had
                          Message 12 of 24 , Jun 18 6:37 AM
                            A number of us on the US East Coast attempted to observe/measure the
                            June 13 Pluto occultation event. Many were clouded out, but here in MD
                            I fortunately had excellent conditions but observed a miss. My question
                            is whether anyone can comment on what happened to the shadow: where did
                            it go? Were the predictions in serious error?

                            John Menke
                            menkescientific.com


                            [Non-text portions of this message have been removed]
                          Your message has been successfully submitted and would be delivered to recipients shortly.