Worldmaps were obtained with OCCULT 4.2.4.4 (D. Herald), using the astorb.dat version of 2016 Nov. 15 (by L.H.Wasserman, Lowell Observatory) and the star catalogs Tycho-GAIA, GAIA14 (by D.Herald), URAT1 and UCAC4.

USED CRITERIONS

I first selected all the trans-neptunian objects (TNO) of absolute magnitude H < 5.7 according to the Minor Planet Center (MPC). Among them I only kept those having at the date of the 2017 solar opposition the one standard deviation (1 s.d.) minor semiaxis of their uncertainty ellipse smaller than 140 mas (main criterion) and the major semiaxis smaller then 500 mas (secondary criterion). I preferred this not classic selection to the traditional "U parameter" of orbit quality (from 0 to 9) calculated by the MPC. My sources of uncertainty ellipses were Horizons from the JPL and AstDys-2 at Pisa University, not the MPC. The uncertainty limits plotted as dashed lines by OCCULT have a fourth source: the "current ephemeris uncertainties" (CEU) provided by astorb.dat. Those various uncertainty estimations are in rough agreement, but never in detail.
[Recall: The terrestrial diameter seen from the trans-neptunian belt is between 370 and 500 mas typically, but drops a bit under 200 mas from 136199 Eris or 2014 UZ224, the currently known most distant bodies.]

Very limited override of above criterions was finally allowed, for various reasons, to ~10 % bodies lying slightly off limits.
Especially I decided to reinstate six bodies from the Rio & Meudon program which, for obscure reason, were falling off limits. So the entire Rio & Meudon program (the Centaurs excepted) now is covered by my predictions.

So has resulted a first list of 161 trans-neptunian objects. Then I added 134340 Pluto, a body in my selection inner core, but computed not with astorb.dat as all the asteroids, rather with DE435 (JPL Development Ephemeris) like a normal planet. In spite of small diameters (Charon excepted) I added the satellites of Pluto to my list. Then I also included Neptune, discarding many unobservable occultations of faint stars by the planet itself, but obviously keeping his major satellite Triton.
[Triton, although formally he is not a TNO, is comparable to many TNOs, is even larger than Pluto and Eris, has an atmosphere like Pluto, and will offer in Europe on 2017 Oct.05 a very interesting occultation of a mag. 12 star.]
Finally I also added te small TNO 2014 MU69, that the "New Horizons" spacecraft will fly-by on 2019 Jan. 01. But as its position uncertainty still is as large as 4 Earth diameters, I have added it only for the record.
[So, please, don't be obsessed by the 2017 Jul. 17 event I propose!]

Eventually my special file AsteroidElements.dat used by OCCULT contains 164 entries: 93 numbered bodies (among them 78 still with a provisional name), plus 71 not yet numbered.


A MOSTLY EUROPEAN-ORIENTED SELECTION

My prime purpose was to show only occultations of stars brighter than the V magnitude +16.5, and potentially observable from Europe and/or Canaries (whole Europe or only a part). Bright twilight always, close horizon often, were exclusion factors I systematically applied to discard many events. In contrast I was not careful of potential Full Moon neighbourhood; so users have to evaluate the moonshine impact by themselves.

In second lecture I added southern tropical regions from where European people observe occultations more and more often, either in remote mode or on site during stays: north of Chile and adjacent countries (especially north of Argentina and Brazil), plus oversea to the east: Namibia and Réunion island.

NB: I also plan a more worldwide selection including a large Pacific area (Australia + New Zealand, Japan, Hawaii...) with target stars up to the V magnitude +18.5. That would be distributed to interested people on request, through e-mailing of an OCCULT occelmnt file, i.e. a simple textfile of 540 bytes per occultation item, to be processed by OCCULT.

Please well realize that in matter of trans-neptunian events, possible large shift of the shadow path has to be anticipated everytime. For example a path predicted (and plotted) as crossing Africa in the vicinity of terrestrial Equator might end into observed occultation, either 100 mas to the south from South-Africa, or 100 mas to the north from southern Europe!
Similarly any path missing Earth at seemingly impressive distance has 50 % chance to be shifted inward. Sometimes it happens that the TNO occults the star, at low elevation in countries lying along the limb.


ABOUT THE ASTROMETRIC STAR CATALOGS

GAIA is just beginning a revolution, still very far from achievement. Of course I have used Tycho-GAIA and GAIA14 in absolute priority. However they are tiny provisional subsets of the future GAIA database, and only countain stars of magnitude R < 14.0. Unfortunately big trans-neptunian bodies occulting stars of R magnitude 13 or 12 are just a handful every year, and moreover occurrences for V < 12 are quite vanishing. Due to the almost homogeneous distribution of stars around the Sun, if we take complete inventory of all the V < 16.5 stars, only 17 % from them are brighter than R= 14.0, or 15 % brighter than V= 14.0. So joining some more complete old catalog was an absolute necessity for my 2017 trans-neptunian predictions.


Now what about the URAT1 and UCAC4?
[I thank D.Herald, for useful discussion about this thema]

UCAC4 has a faint stars technical limit R ~ 16.0 or R ~16.3, i.e. V ~ 16.5 or so.
URAT1 does not exist south from the declination -15 degrees, excepted around Pluto and 2014 MU69 until 2017.0.

Among all the V < 16.5 stars in the sky, 50 % stand between V=16.5 and V=15.6. As moving trans-neptunian bodies made perfect random drawing through the stars they encounter, it results that 50 % from all the potential occultations I have compiled are crowded within less than a 0.9 magnitude range under the UCAC4 technical limit ...

As now we reliably know - thanks to GAIA-DR1 and VizieR - the precise 2015.0 position of many V < 16.5 stars, then for a small (N ~ 40) sub-sample of my ~2100 V < 16.5 targets which are present in both UCAC4 and URAT1, I was able to compare how far from the GAIA reference are falling the UCAC4 and URAT1 calculated positions.

Thanks to this sub-sample, which, because of my V < 16.5 fixed condition, is automatically set around V ~ 16.0, I have discovered that UCAC4 is clearly under-performing relative to URAT1. In contrast, around the magnitude R ~ 13, URAT1 and UCAC4 have shown more equal performances, though I guess that URAT1 is still keeping some advantage.

So it turns out that around V= 16.0, UCAC4 is undoubtly working too close to its technical limit. In consequence I clearly had to choose URAT1. But I must replace URAT1 by UCAC4 in all the star fields more south than declination -15 (special case of Pluto excepted).

The URAT1 possess a technical limit higher by about 2 magnitudes than the UCAC4, as using a modern back-illuminated CCD definitely outperforming its predecessor. URAT1 also uses longer exposures, plus a redder filter than the UCAC4 (by ~100 nm), which a bit reduces the atmospheric turbulence and probably improves the image sharpness of the five elements lens. I guess that the URAT positions of mean epoch were nicely close to GAIA's ones, and, as now we only are three years or so after the mean epoch, that the positional blurring due to provisional proper motions have not yet erased the URAT initial advantage.


ABOUT THE STAR MAGNITUDES

The "Mv" magnitude reported in the worldmap header sometimes will seem contradicting my V < 16.5 rule. Actually one cannot accept as a valid V neither the G-magnitude of GAIA (a very wide 650 nm band), nor the instrumental UCAC4 and URAT1 magnitudes, as all these three magnitudes actually are approximations of R.
[NB: When OCCULT prints the same value for Mv, Mp and Mr, this is R again].
So, when possible, I took V from the most reliable source, the AAVSO APASS (accessed from VizieR or through Guide9).
In GAIA-DR1 the value "phot_g_mean" of the magnitude G is very interesting, as more accurate than anything else and probably homogeneous on the whole sky. G is a red magnitude, but knowing from any source the star colour (for example from 2MASS the infrared (J-Ks) colour index), one may derive thanks to G a refined approximation of V and Rc.
Solving the faint stars photometric puzzle by combining several catalogs will disappear with the publication of the GAIA multi-colour data.