The structure of Chariklo's rings from stellar occultations

TL;DR

This study analyzes stellar occultation data from 2014-2016 to characterize Chariklo's rings, revealing their widths, edge sharpness, and pole position, confirming the circular ring structure with detailed physical properties.

Contribution

It provides detailed measurements of Chariklo's ring widths, edge sharpness, and pole position using multiple occultation observations, improving understanding of ring structure and properties.

Findings

Ring widths vary from 5 to 7.5 km for C1R.

C2R's width may range from 0.1 to 1 km.

Ring edges are consistent with sharp boundaries within 1 km.

Abstract

Two narrow and dense rings (called C1R and C2R) were discovered around the Centaur object (10199) Chariklo during a stellar occultation observed on 2013 June 3. Following this discovery, we planned observations of several occultations by Chariklo's system in order to better characterize the physical properties of the ring and main body. Here, we use 12 successful occulations by Chariklo observed between 2014 and 2016. They provide ring profiles (physical width, opacity, edge structure) and constraints on the radii and pole position. Our new observations are currently consistent with the circular ring solution and pole position, to within the $\pm 3.3$ km formal uncertainty for the ring radii derived by Braga-Ribas et al. The six resolved C1R profiles reveal significant width variations from $\sim 5$ to 7.5 km. The width of the fainter ring C2R is less constrained, and may vary between 0.1 and 1 km. The inner and outer edges of C1R are consistent with infinitely sharp boundaries, with typical upper limits of one kilometer for the transition zone between the ring and empty space. No constraint on the sharpness of C2R's edges is available. A 1$\sigma$ upper limit of $\sim 20$ m is derived for the equivalent width of narrow (physical width <4 km) rings up to distances of 12,000 km, counted in the ring plane.

Tables

Table 1. Table 1: Circumstances of positive detections (main body and/or rings)

Date
Rmag (NOMAD catalog), $(\alpha ,\delta )$ star coordinates, ${\theta }_{\star }$ stellar diameter(a)
$({\alpha }_{Ck},{\delta }_{Ck})$ derived Chariklo’s geocentric coordinates at specified date
Site	Longitude	Telescope	Instrument	Observers	Results
	Latitude		Exposure Time (s)
	Altitude (m)
June 3, 2013
$R=12.070$, $\alpha =16h56m06.4876s$, $\delta =-40°31\prime 30.205″$, ${\theta }_{\star }=2.18$ km
at 06:25:30 UT: ${\alpha }_{Ck}=16h56m06.3202s$, ${\delta }_{Ck}=-40°31\prime 30.2803″$
See details in Braga-Ribas et al. (2014)
February 16, 2014
$R=16.$, $\alpha =17h35m55.3333s$, $\delta =-38°05\prime 17.184″$, ${\theta }_{\star }=0.265$ km
at 07:45:35 UT: ${\alpha }_{Ck}=17h35m54.980s$, ${\delta }_{Ck}=-38°05\prime 17.449″$
Paranal	24 37 31. S	UT4 8.2 m	HAWK-I	F. Selman, C. Herrera	C1R and C2R
Chile	70 24 07.95 W	H-filter	0.25	G. Carraro, S. Brillant	partially
	2635.43			C. Dumas, V. D. Ivanov	resolved
San Pedro Atacama	22 57 12.3 S	50 cm	APOGEE U42	A. Maury	Main body
Chile	68 10 47.6 W		10	N. Morales
	2397
March 16, 2014
$R=15.45$, $\alpha =17h40m39.8690s$, $\delta =-38°25\prime 46.887″$, ${\theta }_{\star }=0.121$ km
at 20:31:45 UT: ${\alpha }_{Ck}=17h40m39.7743s$, ${\delta }_{Ck}=-38°25\prime 46.4198″$
Doi Inthanon	18 34 25.41 N	TNT 2.4 m	ULTRASPEC	P. Irawati	C1R and C2R
Thailand	98 28 56.06 E	R’-filter	3.3	A. Richichi	unresolved
	2450
April 29, 2014
$R_A^{\left(b\right)}=12.72$, ${\alpha }_A^{\left(b\right)}=17h39m02.1336s$, ${\delta }_A^{\left(b\right)}=-38°52\prime 48.801″$, ${\theta }_{\star A}^{\left(b\right)}=0.199$ km
at 23:14:12 UT: ${\alpha }_{Ck}=17h39m01.7943s$, ${\delta }_{Ck}=-38°52\prime 48.858″$
SAAO	32 22 46.0 S	1.9 m	SHOC	H. Breytenbach	C1R and C2R
Sutherland	20 48 38.5 E		0.0334	A. A. Sickafoose	resolved
South Africa	1760				Main body
Gifberg	31 48 34.6 S	30 cm	Raptor Merlin 127	J.-L. Dauvergne	Grazing C2R
South Africa	18 47 0.978 E		0.047	P. Schoenau
	338
Springbok	29 39 40.2 S	30 cm	Raptor Merlin 127	F. Colas	C1R and C2R
South Africa	17 52 58.8 W		0.06	C. de Witt	sharp and resolved
	900
June 28, 2014
$R=13.65$, $\alpha =17h24m50.3821s$, $\delta =-38°41\prime 05.609″$, ${\theta }_{\star }=0.167$ km
at 22:24:35 UT: ${\alpha }_{Ck}=17h24m50.2954s$, ${\delta }_{Ck}=-38°41\prime 05.7445″$
Hakos	23 14 11 S	50 cm AK3	Raptor Merlin 127	K.-L. Bath	C1R and C2R
Namibia	16 21 41.5 E		0.2		unresolved
	1825
Kalahari	26 46 26.91 S	30 cm	Raptor Merlin 127	L. Maquet	Main body
South Africa	20 37 54.258 E		0.4
	861
Twee Rivieren	26 28 14.106 S	30 cm	Raptor Merlin 127	J.-L. Dauvergne	Main body
South Africa	20 36 41.694 E		0.4
	883
April 26, 2015
$R=12.04$, $\alpha =18h10m46.1450s$, $\delta =-36°38\prime 56.368″$, ${\theta }_{\star }=0.361$ km
at 02:11:58 UT: ${\alpha }_{Ck}=18h10m45.9676s$, ${\delta }_{Ck}=-36°38\prime 56.608″$
Los Molinos	34 45 19.3 S	OALM	FLI CCD	S. Roland	CR and C2R
Uruguay	56 11 24.6 W	46 cm	0.8	R. Salvo	unresolved
	130			G. Tancredi
May 12, 2015
$R=15.93$, $\alpha =18h08m29.2962s$, $\delta =-36°44\prime 56.814″$, ${\theta }_{\star }=0.219$ km
at 17:55:40 UT: ${\alpha }_{Ck}=18h08m29.2447s$, ${\delta }_{Ck}=-36°44\prime 56.7965″$
Samford Valley	27 22 07.00 S	35 cm	G-star	J. Bradshaw	Main Body
Australia	152 50 53.00 E		0.32		Emersion of unresolved
	80				rings only
July 25, 2016
$R=14.02$, $\alpha =18h20m35.3645s$, $\delta =-34°02\prime 29.590″$, ${\theta }_{\star }=0.234$ km
at 23:59:00 UT: ${\alpha }_{Ck}=18h20m35.3640s$, ${\delta }_{Ck}=-34°02\prime 29.0378″$
Liverpool Telescope	28 45 44.8 N	2 m	RISE	J.-L. Ortiz	Main Body
Canary Islands	17 52 45.2 W		0.6	N. Morales
	2363
August 08, 2016
$R=13.67$, $\alpha =18h18m03.6927s$, $\delta =-33°52\prime 28.392″$, ${\theta }_{\star }=0.204$ km
at 19:57:00 UT: ${\alpha }_{Ck}=18h18m03.8297s$, ${\delta }_{Ck}=-33°52\prime 28.181″$
or ${\alpha }_{Ck}=18h18m03.8449s$, ${\delta }_{Ck}=-33°52\prime 28.196″$
Windhoek (CHMO)	22 41 54.5 S	35 cm	ZWO / ASI120MM	H.-J. Bode	Main Body
Namibia	17 06 32.0 E		1		C1R and C2R
	1920				unresolved
August 10, 2016
$R=16.53$, $\alpha =18h17m47.3492s$, $\delta =-33°51\prime 02.516″$, ${\theta }_{\star }=0.053$ km
at 14:23:00 UT: ${\alpha }_{Ck}=18h17m47.3089s$, ${\delta }_{Ck}=-33°51\prime 02.478″$
Murrumbateran	34 57 31.50 S	40 cm	WATEC 910BD	D. Herald	Main Body
Australia	148 59 54.80 E		0.64
	594
August 10, 2016
$R=16.22$, $\alpha =18h17m46.4827s$, $\delta =-33°50\prime 57.826″$, ${\theta }_{\star }=0.083$ km
at 16:43:00 UT: ${\alpha }_{Ck}=18h17m46.4457s$, ${\delta }_{Ck}=-33°50\prime 57.523″$
Les Makes	21 11 57.4 S	60 cm	Raptor Merlin 127	F. Vachier	Main Body
La Réunion	55 24 34.5 E		2
	972
August 15, 2016
$R=14.64$, $\alpha =18h17m06.2228s$, $\delta =-33°46\prime 56.315″$, ${\theta }_{\star }=0.103$ km
at 11:38:00 UT: ${\alpha }_{Ck}=18h17m06.1638s$, ${\delta }_{Ck}=-33°46\prime 56.513″$
Darfield	43 28 52.90 S	25 cm	WATEC 910 BD	B. Loader	Main Body
New Zealand	172 06 24.40 E		2.56
	210
October 1, 2016
$R=15.36$, $\alpha =18h16m20.0796s$, $\delta =-33°01\prime 10.756″$, ${\theta }_{\star }=0.119$ km
at 10:10:00 UT: ${\alpha }_{Ck}=18h16m20.0324s$, ${\delta }_{Ck}=-33°01\prime 10.841″$
Rockhampton	23 16 09.00 S	30 cm	WATEC 910BD	S. Kerr	Main Body
Australia	150 30 00 E		0.320		C1R and C2R
	50				unresolved
Adelaide	34 48 44.701 S	30 cm	QHY 5L11	A. Cool	Main body
Heights School	138 40 56.899 E		1	B. Lade	C1R and C2R
Australia	167				unresolved
(a) projected at Chariklo’s distance (using van Belle 1999, except for the April 29, 2014, see text for details).
(b) The index $A$ refers to the primary of the binary star.

Table 2. Table 2: Circumstances of observations that detected no event or during which no data was acquired

Site	Longitude	Telescope	Instrument	Observers
	Latitude		Exposure Time (s)
	Altitude (m)
February 16, 2014
Cerro Tololo	30 10 03.36 S	0.4 m	PROMPT	J. Pollock
Chile	70 48 19.01 W	4 telescopes	6.0/2.0
	2207
La Silla	29 15 16.59 S	TRAPPIST	FLI PL3041-BB	E. Jehin
Chile	70 44 21.82 W	60 cm	4.5
	2315
La Silla	29 15 32.1 S	NTT 3.55 m	SOFI	L. Monaco
Chile	70 44 01.5 W	H-filter	0.05	+ visitor team
	2375
April 29, 2014
Hakos	23 14 50.4 S	50 cm AK3	Raptor Merlin 127	K.-L. Bath
Namibia	16 21 41.5 E		0.075
	1825
Hakos	23 14 50.4 S	50 cm RC50	i-Nova	R. Prager
Namibia	16 21 41.5 E		1.
	1825
Windhoek (CHMO)	22 41 54.5 S	35 cm	Raptor Merlin 127	W. Beisker
Namibia	17 06 32.0 E		0.1
	1920
June 28, 2014
Les Makes	21 11 57.4 S	60 cm	WATEC 910HX	A. Peyrot
La Réunion	55 24 34.5 E		0.4	J-P. Teng
	972
April 26, 2015
Bigand	33 26 11 S	15 cm	Canon Ti	S. Bilios
Provincia Santa Fé	61 08 24 W		5
Argentina	90
Bigand	33 26 11 S	15 cm	Canon EOS	J. Nardon
Provincia Santa Fé	61 08 24 W		3.2
Argentina	90
La Silla	29 15 16.6 S	TRAPPIST	FLI PL3041-BB	E. Jehin
Chile	70 44 21.8 W	60 cm	4.5
	2315
Bosque Alegre	31 35 54.0 S	76 cm	QHY6	R. Melia
Argentina	64 32 58.7 W		1.2	C. Colazo
	1250
Santa Rosa	36 38 16 S	20 cm	Meade DSI-I	J. Spagnotto
Argentina	64 19 28 W		3
	182
Santa Martina	33 16 09.0 S	40 cm	Raptor Merlin 127	R. Leiva Espinoza
Chile	70 32 04.0 W		0.5
	1450
Buenos Aires (AAAA)	34 36 16.94 S	25 cm	ST9e	A. Blain
Argentina	58 26 04.37 W		4
	0
May 12, 2015
Reedy Creek	28 06 30.4 S	25 cm	WATEC 120N+	J. Broughton
Australia	153 23 52.90 E		0.64
	66
July 25, 2016
Granada	37 00 38.49 N	60 cm	Raptor Merlin 127	S. Alonso
Spain	03 42 51.39 W		0.4	A. Román
	1043
Albox	37 24 20.0 N	40 cm	Atik 314L+	J.-L. Maestre
Spain	02 09 6.5 E		3
	493
August 10, 2016 - 14h UT
Blue Mountains	33 39 51.9 S	30 cm	WATEC 910BD	D. Gault
Australia	150 38 27.9 E		5.12
	286
Samford Valley	27 22 07.00 S	35 cm	WATEC 910BD	J. Bradshaw
Australia	152 50 53.00 E		0.64
	80
Rockhampton	23 16 09.00 S	30 cm	WATEC 910BD	S. Kerr
Australia	150 30 00.00 E		1.28
	50
Dunedin	45 52 20.83 S	36 cm	Raptor Merlin 127	F. Colas
New Zealand	170 29 29.90 E		2.	A. Pennell
	154			P.-D. Jaquiery
Sydney	33 48 35.04 S	36 cm	Raptor Merlin 127	H. Pavlov
Australia	150 46 36.90 E		2.2
	37
August 15, 2016
Canberra	35 11 55.30 S	40 cm	WATEC 910BD	J. Newman
Australia	149 02 57.50 E		2.56
	610
Murrumbateran	34 57 31.50 S	40 cm	WATEC 920BD	D. Herald
Australia	148 59 54.80 E		0.32
	594
Greenhill Observatory	42 25 51.8 S	1.3 m	Raptor Merlin 127	K. Hill
Tasmania	147 17 15.8 E		0.5	A. Cole
	641
Rockhampton	23 16 09.00 S	30 cm	WATEC 910BD	S. Kerr
Australia	150 30 00.00 E		1.28
	50
Linden Observatory	33 42 27.3 S	76 cm	Grasshopper	D. Gault
Australia	150 29 43.5 E		Express with ADVS	R. Horvat
	574		0.533	R.A. Paton
				L. Davis
WSU Penrith Observatory	33 45 43.31 S	62 cm	Raptor Merlin 127	H. Pavlov
Sydney	150 44 30.30 E		2	D. Giles
Australia	60			D. Maybour
				M. Barry
October 1, 2016
Blue Mountains	33 39 51.9 S	30 cm	WATEC 910BD	D. Gault
Australia	150 38 27.9 E		0.64
	286
Linden Observatory	33 42 27.3 S	76 cm	Grasshopper	M. Barry
Australia	150 29 43.5 E		Express with ADVS
	574		0.27
Miles	26 39 20.52 S	25 cm	WATEC 120N+	D. Dunham
Australia	150 10 19.44 E		0.64	J. Dunham
	277
Reedy Creek	28 06 30.4 S	25 cm	WATEC 120N+	J. Broughton
Australia	153 23 52.90 E		1.28
	66
Samford Valley	27 22 07.00 S	35 cm	WATEC 910BD	J. Bradshaw
Australia	152 50 53.00 E		0.16
	80
The following stations were cloudy or had technical failure, no data were acquired:
February 16, 2014: Santa Martina (Chile), Bosque Alegre (Argentina)
April 29, 2014 : Rodrigues, Sainte Marie, Les Makes (La Réunion) ; Calitzdorp and LCOGT (South Africa)
April 26, 2015 : Cerro Tololo (Chile)
July 25, 2016: TRAPPIST Nord (Marocco), TAD (Canary Islands), Teide Observatory (Canary Islands)
August 8, 2016 : Les Makes (La Réunion)
August 15, 2016: Mount John Observatory, Dunedin, Bootes-3, Wellington (New Zealand)
October 1, 2016: Murrumbateran, Canberra (Australia)

Table 3. Table 3: Sharpness of C1R edges, Δ ​ w r Δ subscript 𝑤 𝑟 \Delta w_{r} , from April 29, 2014 events

Event	Inner edge (km)	Outer edge (km)
	($1\sigma$ level)
Springbok Ingress	1.1	1.1
Springbok Egress	1.2	1.5
SAAO Ingress	0.6	0.9
SAAO Egress	0.8	0.4

Table 4. Table 4: Ring occultation timings and derived physical parameters (resolved events)

Date	Event	$t_0$ UT(a)	$v_{⟂}^{\left(b\right)}$	$v_r^{\left(b\right)}$	$L^{\left(c\right)}$	$W_r^{\left(d\right)}$	$E_p^{\left(e\right)}$	$p_N^{\left(e\right)}$
			(km/s)	(km/s)	(deg.)	(km)	(km)
C1R
Jun 3, 2013	Danish ingress(f)	06:25:21.166$\pm$0.0007	20.345	36.113	341.76	$6.16\pm 0.11$	$1.90\pm 0.022$	$0.308\pm 0.003$
	Danish egress(f)	06:25:40.462$\pm$0.0012	22.031	36.504	124.38	$7.14\pm 0.04$	$1.73\pm 0.023$	$0.24\pm 0.004$
Feb 16, 2014	VLT ingress	07:45:25.541${}_{}{}^{+0.010}{}_{-0.004}{}^{}$	19.532	28.794	183.37	${5.316}_{-1.916}^{+0.868}$	${1.996}_{-0.031}^{+0.092}$	${0.375}_{-0.025}^{+0.125}$
	VLT egress	07:45:45.133${}_{}{}^{+0.313}{}_{-0.332}{}^{}$	21.293	29.602	300.99	${4.833}_{-0.476}^{+1.667}$	${2.04}_{-0.14}^{+0.36}$	${0.443}_{-0.103}^{+0.078}$
Apr 29, 2014	Springbok ingress	23:14:25.884$\pm$0.007	13.432	16.493	287.42	$5.575\pm 0.398$	${1.80}_{-0.143}^{+0.122}$	${0.3125}_{-0.027}^{+0.024}$
	Springbok egress	23:15:04.362$\pm$0.006	10.720	16.655	157.83	${6.75}_{-0.21}^{+0.48}$	${2.595}_{-0.166}^{+0.148}$	${0.33}_{-0.033}^{+0.017}$
	SAAO ingress	23:13:56.191$\pm$0.007	12.756	13.895	266.656	$5.68\pm 0.2$	${1.88}_{-0.12}^{+0.22}$	${0.32}_{-0.021}^{+0.037}$
	SAAO egress	23:14:28.964$\pm$0.008	9.260	14.249	198.899	$6.625\pm 0.2$	${1.695}_{-0.115}^{+0.175}$	${0.241}_{-0.022}^{+0.024}$
C2R
Jun 3, 2013	Danish ingress(f)	06:25:20.765$\pm$0.011	20.412	36.283	341.76	${3.380}_{-1.797}^{+1.424}$	$0.168\pm 0.02$	${0.05}_{-0.01}^{+0.05}$
	Danish egress(f)	06:25:40.847$\pm$0.006	22.029	36.632	124.38	${3.231}_{-1.124}^{+0.899}$	$0.228\pm 0.02$	${0.07}_{-0.01}^{+0.03}$
Feb 16, 2014	VLT ingress	07:45:25.285${}_{}{}^{+0.057}{}_{-0.033}{}^{}$	19.532	28.794	183.37	${5.053}_{-2.385}^{+1.000}$	${0.491}_{-0.227}^{+0.445}$	${0.091}_{-0.00}^{+0.495}$
	VLT egress	07:45:45.473${}_{}{}^{+0.037}{}_{-0.053}{}^{}$	21.293	29.602	300.99	${3.333}_{-1.333}^{+1.667}$	${0.522}_{-0.050}^{+0.078}$	${0.119}_{-0.118}^{+0.609}$
Apr 29, 2014	Springbok ingress	23:14:24.990$\pm$0.020	13.430	16.460	287.42	${0.34}_{-0.24}^{+1.37}$	${0.125}_{-0.064}^{+0.076}$	${0.368}_{-0.288}^{+0.632}$
	Springbok egress	23:15:5.324$\pm$0.019	10.722	16.620	157.832	${0.6}_{-0.1}^{+1.7}$	${0.253}_{-0.069}^{+0.079}$	${0.582}_{-0.45}^{+0.32}$
	Gifberg ingress(g)	23:14:30.109${}_{-0.008}{}^{}{}_{}{}^{+0.015}$	$\left(g\right)$	$\left(g\right)$	227.190	${0.522}_{-0.399}^{+0.227}$	${0.090}_{-0.000}^{+0.039}$	${0.186}_{-0.043}^{+0.814}$
	Gifberg egress(g)	23:14:33.750$\pm$0.008	$\left(g\right)$	$\left(g\right)$	217.761	${0.181}_{-0.091}^{+0.008}$	${0.129}_{-0.039}^{+0.000}$	${0.814}_{-0.671}^{+0.186}$

Table 5. Table 5: Ring occultation timings and derived physical parameters (unresolved events)

Date	Event	$t_0$ UT†	$v_{⟂}$(km/s)†	$v_r$ (km/s)†	$L$(deg)†	$E_p$ (km)†
June 3, 2013	Iguacu ingress	06:24:17.5$\pm$1.7(a)	18.059	28.899	2.44	${7.602}_{-5.195}^{+2.198}$
	Iguacu egress	06:24:34.1$\pm$2.0(a)	21.246	30.446	104.20	${2.580}_{-1.713}^{+3.712}$
	Bosque Alegre 154 egress	06:25:11.44$\pm$0.14(a)	18.889	32.663	176.11	${3.806}_{-2.198}^{+1.199}$
	Ponta Grossa ingress	06:23:58.6$\pm$2.5(a)	19.781	34.398	348.04	${9.600}_{-4.795}^{+0.400}$
	Ponta Grossa egress	06:24:18.0$\pm$2.5(a)	21.965	35.520	120.30	${4.605}_{-3.340}^{+3.596}$
	PROMPT ingress	06:25:20.0.46$\pm$0.011(a)	21.373	38.269	326.56	${2.208}_{-0.200}^{+3.196}$
	Santa Martina ingress	06:25:21.03$\pm$0.29(a)	17.556	18.537	264.53	${2.208}_{-0.200}^{+2.997}$
	Santa Martina egress	06:25:31.811$\pm$0.025(a)	14.605	22.124	200.07	${2.408}_{-0.200}^{+5.394}$
	SOAR ingress	06:25:18.8$\pm$1.3(a)	21.444	38.320	325.16	${2.208}_{-0.400}^{+4.196}$
	SOAR egress	06:25:38.4$\pm$1.4(a)	21.660	38.310	140.37	${5.205}_{-3.596}^{+0.799}$
	Bosque Alegre C11 ingress	06:24.55.45$\pm$1.85(a)	21.522	30.340	287.37	${4.206}_{-2.597}^{+3.297}$
	Bosque Alegre C11 egress	06:25:09.45$\pm$1.75(a)	17.882	30.062	183.50	${4.206}_{-2.597}^{+3.396}$
	TRAPPIST ingress	06:25:20.9$\pm$1.9(a)	20.293	36.229	341.31	${4.605}_{-2.198}^{+3.796}$
March 16, 2014	Thailand ingress	20:31:37.640$\pm$1.33	3.656	3.821	95.06	${1.856}_{-1.197}^{+0.948}$
	Thailand egress	20:31:53.885$\pm$0.175	3.990	4.290	60.85	${1.856}_{-0.801}^{+0.150}$
June 28, 2014	Hakos ingress	22:24:25.796$\pm$0.041	$19.127$	$28.619$	5.064	${1.472}_{-0.517}^{+0.455}$
	Hakos egress	22:24:44.218$\pm$0.035	20.971	29.744	117.061	${1.983}_{-0.508}^{+0.598}$
April 26, 2015	Los Molinos ingress	02:11:45.707$\pm$0.058	3.503	3.513	238.857	${2.914}_{-0.149}^{+0.151}$
	Los Molinos egress	02:12:09.195$\pm$0.070	2.957	3.989	199.749	${2.400}_{-0.320}^{+0.28}$
May 12, 2015	Brisbane egress	17:55:56.823$\pm$0.012	11.823	16.567	357.23	${2.707}_{-1.198}^{+2.398}$
Aug 8, 2016	Windhoek (CHMO) ingress	19:57:18.209 $\pm$ 0.249	15.920	21.878	332.963(b)	${2.043}_{-0.734}^{+2.762}$
	Windhoek (CHMO) egress	19:57:51.870 $\pm$ 0.382	15.216	21.950	180.196(b)	${3.806}_{-2.297}^{+2.598}$
Oct 1, 2016	Rockhampton ingress	10:12:26.284 $\pm$ 0.072	10.795	13.121	123.960	${2.523}_{-0.615}^{+3.481}$
	Rockhampton egress	10:13:22.928 $\pm$ 0.049	12.573	13.146	278.911	${2.586}_{-0.778}^{+3.818}$
	Adelaide ingress	10:10:19.826$\pm$ 0.186	12.421	12.597	91.347	${1.867}_{-0.539}^{+4.073}$
	Adelaide egress	10:11:14.558 $\pm$ 0.218	9.942	12.651	311.914	${2.047}_{-0.719}^{+3.534}$

Table 6. Table 6: Occultation timings for the main body.

Date	Event	$t_{ingress}$ UT	$t_{egress}$ UT
June 3, 2013	Danish	06:25:27.861$\pm$0.014	06:25:33.188$\pm$0.014
	PROMPT	06:25:24.835$\pm$0.009	06:25:35.402$\pm$0.015
	TRAPPIST	06:25:27.893$\pm$0.019	06:25:33.155$\pm$0.007
	SOAR	06:25:24.34$\pm$0.59	06:25:34.597$\pm$0.009
February 16, 2014	San Pedro de Atacama	07:45:27.450$\pm$0.6	07:45:31.125$\pm$0.57
June 28, 2014	Kalahari	22:24:07.383$\pm$0.126	22:24:14.854$\pm$0.096
	Twee Rivieren	22:24:06.689$\pm$0.093	22:24:16.481$\pm$0.105
April 29, 2014	Springbok	23:14:30.02$\pm$0.075	23:14:48.03$\pm$0.075
May 12, 2015	Brisbane	17:55:35.530$\pm$0.010	17:55:44.135$\pm$0.075
July 25, 2016	Liverpool Telescope	23:59:05.494 $\pm$0.054	23:59:12.310 $\pm$ 0.054
August 8, 2016	Windhoek (CHMO)	19:57:28.469 $\pm$ 0.042	19:57:41.886 $\pm$ 0.045
August 10, 2016 - 14h UT	Murrumbateran	14:18:35.030 $\pm$ 0.3	14:18:45.145 $\pm$ 0.125
August 10, 2016 - 16h UT	Les Makes	16:42:51.305 $\pm$ 0.530	16:43:07.917 $\pm$ 0.848
August 15, 2016	Darfield	11:38:27.465 $\pm$ 0.385	11:38:38.019 $\pm$ 0.873
October 1, 2016	Rockhampton	10:12:44.664 $\pm$ 0.041	10:13:03.199 $\pm$ 0.051
	Adelaide	10:10:41.818 $\pm$ 0.118	10:10:54.102 $\pm$ 0.064
The error bars quoted are given at $1\sigma$ level.

Equations

$${\mu }_{\alpha }$$

$${\chi }^2=\sum _i\frac{({\Phi }_{i,\mathrm{obs}}-{\Phi }_{i,\mathrm{calc}}{)}^2}{{\sigma }_i^2}$$

$$E_p(i)=\frac{|\sin (B)|}{2}[1-\varphi (i)]\Delta r(i)$$

$$E_p(W_r)=W_r(1-{\mathrm{e}}^{-2\overline{A_{\tau }}/W_r})$$

$$E_p={\int }_{W_r}(v_r{p}_N)dt$$

$$A_{\tau }={\int }_{W_r}(v_r{\tau }_N)dt,$$

$$E_p=|\sin (B)|\cdot v_r{\int }_{\mathrm{profile}}(1-\sqrt{1-p^{\prime }})dt$$

$$A_{\tau }=-\frac{|\sin (B)|}{2}\cdot v_r{\int }_{\mathrm{profile}}\ln (1-p^{\prime })dt$$

$$E_p=|\sin (B)|\frac{v_r}{v_{\perp }}\cdot \frac{E^{\prime }}{2-p}$$

$$|\sin (B)|\cdot \frac{v_r}{v_{\perp }}\cdot \frac{E^{\prime }}{2}\le E_p\le |\sin (B)|\cdot \frac{v_r}{v_{\perp }}\cdot E^{\prime },$$

Full text

Rings - Centaur objects: individual (Chariklo) - Stellar Occultations

††slugcomment: Last updated

1 Introduction

The asteroid-like body (10199) Chariklo is a Centaur object orbiting between Saturn and Uranus. It probably moved recently ($\sim$ 10 Myr ago) from the Trans-Neptunian region to its present location, and will leave it on a similar short time scale, due to perturbations by Uranus (Horner et al., 2004). With a radius of $119\pm 5$ km, estimated from thermal measurements (Fornasier et al., 2014), it is the largest Centaur known to date, but still remains very modest in size compared to the telluric or giant planets. On June 3, 2013, a ring system was discovered around this small object during a stellar occultation. Two dense and narrow rings, 2013C1R and 2013C2R (C1R and C2R for short) were detected. They are separated by about 15 km and orbit close to 400 km from Chariklo’s center (see Braga-Ribas et al. (2014) for details).

Until 2013, rings were only known around the giant planets. This discovery was thus surprising, and is a key to better understanding of the planetary rings, since they now appear to be more common than previously thought. In particular, the two rings being dense, narrow and (at least for C1R) sharp-edged, they look like several of the dense ringlets seen around Saturn and Uranus (Elliot and Nicholson, 1984; French et al., 1991, 2016). In that context, there was a strong incentive for planning more occultation campaigns, first to unambiguously confirm the existence of Chariklo’s rings and second, to obtain more information on their physical properties.

While the discovery occultation of June 3, 2013 provided the general ring physical parameters (width, orientation, orbital radius, optical depth,…), several questions are still pending, some of them being addressed in this work: do the rings have inner structures that give clues about collisional processes? How sharp are their edges? What are the general shapes of C1R and C2R? Do they consist of solidly precessing ellipses like some Saturn’s or Uranus’ ringlets? Do they have more complex proper modes with higher azimuthal wave numbers? Are there other fainter rings around Chariklo? What is the shape of the object itself and its role for the ring dynamics? Based on new results, what can we learn about their origin and evolution, which remains elusive (Sicardy et al., 2016)?

This study is made in a context where material has also been detected around the second largest Centaur, Chiron (again using stellar occultations). The nature of this material is still debated and it could be interpreted as a ring system (Ortiz et al., 2015) or a dust shell associated with Chiron’s cometary activity (Ruprecht et al., 2015). Since Chariklo is presently moving close to the galactic plane, stellar occultations by this body are much more frequent than for Chiron, hence a more abundant amount of information concerning its rings. The spatial resolution achieved during occultations reaches the sub-km level, impossible to attain with any of the current classical imaging instruments. This said, the very small angular size subtended by the rings (0.08 arcsec tip to tip, as seen from Earth) have made occultation predictions difficult in the pre-Gaia era.

In spite of those difficulties, we could observe 13 positive stellar occultations (including the discovery one) between 2013 and 2016, from a total of 42 stations distributed worldwide (in Brazil, Argentina, Australia, Chile, La Réunion Island, Namibia, New Zealand, South Africa, Spain, Thailand and Uruguay). Here, we focus on the ring detections (a total of 11 chords recorded after the discovery). We also obtained a total of 12 occultation chords by the main body from 2014 to 2016. Their timings are derived here, but their implications concerning Chariklo’s size and shape will be presented elsewhere (Leiva et al., 2017, in preparation). In Section 2, we present our observations and data analysis. In Section 3 we concentrate on the ring structures (width, inner structures, edge sharpness) and geometry (radius and orbital pole). The ring integral properties (equivalent width and depth) are derived in Section 4, before concluding remarks in Section 5.

2 Observations and Data Analysis

Following the ring discovery of June 3, 2013, we predicted and observed 12 positive stellar occultations by Chariklo and/or its rings between 2014 and 2016. In the following list, we mark in italic the events that led to multi-chord ring detections (thus providing constraints on the ring orientation, as discussed latter). Four occultations were observed in 2014, on February 16 (rings), March 16 (rings), April 29 (rings and body) and June 28 (rings and body). In 2015, only two positive detections were recorded on April 26 (rings) and May 12 (body), while six occultations were recorded in 2016: July 25 (body), August 8 (rings and body), August 10 near 14h UT (body), August 10 near 16h UT (body), August 15 (body) and October 1 (rings and body).

2.1 Predictions

Predicting stellar occultations by Chariklo and its rings is a difficult task, as the main body subtends about 25 milliarcsec (mas) as seen from Earth, while the rings have a span of about 80 mas. Thus, to be effective, predictions require accuracies of a few tens of mas on both Chariklo’s ephemeris and the star position. To meet this requirement, we used a bootstrapping approach, in which each new detection of occultation is used to improve Chariklo’s ephemeris, thus providing a better prediction for the next occultation. This continuous update results in the so-called NIMA ephemeris (Numerical Integration of the Motion of an Asteroid, Desmars et al. 2015) accessible online222see http://lesia.obspm.fr/lucky-star/nima/Chariklo/.

The candidate stars for events in 2014 and in 2016 were identified during a systematic search for occultations by TNOs using the Wide Field Imager (WFI) at the ESO/MPG 2.2m telescope (Camargo et al., 2014), with typical accuracies of $\sim 30$ mas. However, for the 2015 season, the candidate stars were observed using only the IAG 0.6m telescope at OPD/LNA in Brazil, with lower accuracy than WFI, resulting in a larger number of missed events (two successes out of six attempts).

In the majority of the cases, the occulted star was imaged a few days or weeks prior the event in order to improve the astrometry. If possible the observations were made when Chariklo and the star were in the same Field Of View in order to cancel systematic errors. In those cases, accuracy of the predictions was estimated down to $\sim 20$ mas.

The last occultation in our list (October 1st, 2016) is special as its prediction was based on the new GAIA DR1 catalog released on September 15, 2016 (Gaia Collaboration et al., 2016). However, the J2000 DR1 star position $\alpha=18\rm h16\rm m20.0796\rm s$, $\delta=-33\arcdeg 01\arcmin 10.756\arcsec$ (at epoch 2015.0) does not account for proper motion. We estimated the latter by using the UCAC4 star position (under the name UCAC4 285-174081) at epoch 2000 and obtained proper motions in right ascension (not weighted by $\cos(\delta)$) and declination of

[TABLE]

This provides a star position of $\alpha=18\rm h16\rm m20.0789\rm s$, $\delta=-33\arcdeg 01\arcmin 10.760\arcsec$ at the epoch of occultation. Combining this result with the NIMA ephemeris (version 9) finally provided a prediction that agreed to within 5 mas perpendicular to the shadow track and 20 seconds in terms of timing, and lead to a multi-chord ring and body detection.

2.2 Observations

The circumstances of the observations (telescope, camera, set up, observers, site coordinates, star information) that lead to ring or main body detections are listed in Table 1. Conversely, the circumstances of negative observations (no event observed) are provided in Table 2. Note that observations were made with both small portable telescopes and larger, fixed instruments. Each detection will be designated herein by the name of the station or by the name of the telescope, if well known.

From the timings of the star disappearance (or “ingress”) and re-appearance (“egress”) behind Chariklo and/or the rings, the geometry of each occultation was reconstructed, as illustrated in Fig. 1. Currently, Chariklo’s size and shape are not known well enough to reconstruct the occultation geometries from the events involving the main body. So, we used instead the ring events (even single-chord) to retrieve those geometries. As a starting point, we assume that the rings are circular with fixed orientation in space, and with the orbital parameters derived by Braga-Ribas et al. (2014), namely a J2000 pole position of $\alpha_{p}=10\rm h05\rm m11.0016\rm s$, $\delta_{p}=+41\arcdeg 28\arcmin 32.4891\arcsec$ and respective radii $a_{C1R}=390.6$  km and $a_{C2R}=404.8$ km for the two rings. The reconstructed geometry allows us to derive the observed position of Chariklo center (reported in Table 1). If the star position was perfect, this derived position must coincide with the occulted star position. The difference between the two positions is the offset between the predicted and the observed Chariklo’s position. This offset is implemented in NIMA after each occultation, in order to improve Chariklo’s ephemeris.

If the rings are not circular, this will impact their pole position and will eventually be visible as discrepancies between observations and predictions. The pole position problem is discussed further in Section 3.4.

Note that some stations did not detect any ring occultations, whereas they should have considering the occultation geometry, see Reedy Creek on May 12, 2015 and Sydney on August 10, 2016. Data analysis shows that those non-detections are actually consistent with the low signal-to-noise-ratio (SNR) obtained at those stations. Thus, secondary events have always been detected if SNR was high enough. This leads us to conclude that C1R (which always dominates the profile) is continuous. Same conclusion on C2R is more ambiguous as C2R was usually blended together with C1R. Nevertheless, we will assume that C2R is continuous in this paper.

2.3 Data Reduction

After a classical data processing that included dark subtraction and flat fielding, aperture photometry provided the stellar flux as a function of time (the date of each data point corresponding to mid-exposure time), the aperture being chosen to maximize SNR. The background flux was estimated near the target and nearby reference stars, and then subtracted, so that the zero flux corresponds to the sky level. The total flux from the unocculted star and Chariklo was normalized to unity after fitting the light curve by a third or forth-degree polynomial before and after the event. In all cases, a reference star (brighter than the target) was used to correct for low frequency variations of the sky transparency.

The light curves are displayed in Fig. 3 and 4, each of them providing a one-dimensional scan across Chariklo’s system, as projected in the sky plane. In some cases, the readout time between two frames caused a net loss of information as photon acquisition was interrupted during those “dead time” intervals. The flux statistics provides the standard deviation of the signal, which defines the $1\sigma$ error bar on each data point which was used latter for fitting diffraction models to ingress and egress events. Note that during an occultation by the main body, the stellar flux drops to zero, but the flux in the light curve is not zero, as it contains Chariklo’s contribution, and in one case, the flux from a nearby companion star, see below.

2.4 The case of the double star of April 29, 2014

This event, observed from South Africa (see Table 1), revealed that the occulted star was a binary. As seen from Springbok, the primary star (“A”) was occulted by C1R and C2R (but missed the main body), while the fainter companion star (“B”) disappeared behind Chariklo along an essentially diametric chord at Springbok (Fig. 1). Because the component B was about 9 times fainter than A (see below), and considering the drop of A caused by C1R at Springbok, we expect a short drop in the light curve of only 8% due to the disappearance of component B behind C1R. This is too small to be detected, in view of the SNR of about 7 per data point obtained at that station (Fig. 4).

Meanwhile in Gifberg, we obtained only a grazing occultation of the primary star by C2R (Fig. 1). This provides the best profile of that ring ever recorded (see Section 3.3). Finally, at the South African Astronomical Observatory (SAAO), only the component B was occulted by the rings, while the main star missed both the rings and the main body (Fig. 1). However, due to the high SNR obtained at that station, the partial drop caused by the rings on component B has about the same useful SNR as the drop of component A as seen from the smaller telescope at Springbok.

For the Springbok light curve, we can estimate the flux ratio $\Phi_{A}/\Phi_{B}$ between the two stars by considering the drop of component B caused by Chariklo. In doing so, we can neglect Chariklo’s contribution to the total flux. From Chariklo’s absolute magnitude, $H_{V}=7.0$ in 2014 (Duffard et al., 2014), and heliocentric and geocentric distances of 14.8 au and 14.1 au during the event, respectively, we obtain a Chariklo apparent magnitude $\sim 18.6$. This is 5.6 magnitudes fainter than the star, which has V=13.0 (NOMAD catalog333See http://vizier.u-strasbg.fr/viz-bin/VizieR), meaning that Chariklo contributed to the total flux of less than 0.6%, a negligible value at our level of accuracy.

The fractional drop observed during the occultation of B by Chariklo provides its partial contribution to the total stellar flux, $\Phi_{B}=0.1036\pm 0.0075$ (Fig. 5). This implies a flux ratio $(\Phi_{A}/\Phi_{B})_{\rm TC247}=8.65\pm 0.65$, as measured by the Texas Instruments TC247 array used at Springbok (in broad band mode, no filter). This directly provides the baseline level for the occultations of A by the rings (Figs. 5 and 8), i.e. the level that corresponds to a total disappearance of component A.

A similar calibration is not possible for the SAAO ring events, as that station did not record an occultation by the main body. Moreover, the ratio $(\Phi_{A}/\Phi_{B})_{\rm TC247}$ cannot be used, as the SHOC instrument (see Coppejans et al. 2013) used at SAAO (also in broad band mode) has a different spectral response, so that the ratio depends on the color of the two stars.

To proceed forward, we have used the B, V, K magnitudes of the star (taken from the Vizier page, in NOMAD catalog). We have generated combined synthetic spectra energy distribution of the two components, and using various (and separate) effective temperature $T_{\rm eff}$ for A and B. The effect of interstellar reddening has been parametrized using the color excess $E(B-V)$. We adopted the classical total to selective extinction parameter $R_{V}=3.1$ for Milky Way dust from Fitzpatrick (1999). The relative contributions of each component were adjusted in order to fit both the observed magnitude of the star and the flux ratio as observed with the TC247 array. Finally, accounting for the spectral response of the Andor array, we can then estimate the ratio $(\Phi_{A}/\Phi_{B})_{\rm Andor}$ for that detector.

A difficulty stems from the fact that there is a degeneracy between the effective temperatures assumed for the two components, $T_{\rm eff}(A)$ and $T_{\rm eff}(B)$. The star B cannot be much cooler than A, otherwise its diameter would be larger and strong signatures in the near IR would appear in the composite spectrum. We have opted for a difference $T_{\rm eff}(A)-T_{\rm eff}(B)\sim 1000$ K, and assume that the two stars are on the main sequence. We find a good fit to the observed magnitudes with $T_{\rm eff}(A)=5000$ K and $T_{\rm eff}(B)=4000$ K, and then a ratio $(\Phi_{A}/\Phi_{B})_{\rm Andor}=7.66$, corresponding to a contribution to the total flux of $\Phi_{A}=0.885\pm 0.025$ for component A, where the error bar is estimated from the typical possible ranges for $T_{\rm eff}(A)$ and $T_{\rm eff}$.

Finally, we can estimate the apparent diameter of each component projected at Chariklo’s distance: $\theta_{A}=0.199\pm 0.015$ km and $\theta_{B}=0.092\pm 0.015$ km. Those values will be used latter when fitting the ring profiles with models of diffracting, semi-transparent bands.

Assuming the ring radii and pole orientation of Braga-Ribas et al. (2014), see also Section 2.2, and using the ring detections in Springbok, Gifberg and SAAO, we deduce that star B was at angular distance 20.6 mas from star A as projected in the sky plane, with position angle $P=209.8\arcdeg$ relative to the latter (where $P$ is counted positively from celestial North towards celestial East).

3 Ring events analysis

3.1 Profiles fitting

In order to determine accurate and consistent timings of the ring occultations, we use a “square-well model” in which each ring is modeled as a sharp-edged, semi-transparent band of apparent opacity $p^{\prime}$ (along the line of sight) and apparent width (in the sky plane) $W_{\perp}$. We use the numerical schemes described in Roques et al. (1987) to account for Fresnel diffraction, stellar diameter projected at Chariklo’s distance, finite bandwidth of the CCD, and finite integration time of the instrument. Finally, considering projection effects, we can derive the ring physical parameters (radial width, normal opacity, etc…) and orbital elements, see Appendix for details.

For sake of illustration, we give various parameters of interest in the case of the April 29, 2014 occultation. The Fresnel scale $F=\sqrt{\lambda D/2}$ for Chariklo’s geocentric distance at epoch, $D=2.11\times 10^{9}$ km is 0.83 km, for a typical wavelength of $\lambda=0.65$ $\mu$m. The projected stellar diameters have been estimated above to $0.199\pm 0.015$ km and $0.092\pm 0.015$ km for the primary star and secondary star, respectively (see Section 2.4). The smallest cycle time used during that campaign was 0.04 s (at SAAO), corresponding to 0.5 km traveled by the star relative to Chariklo in the celestial plane. Consequently, the light curves are dominated by Fresnel diffraction, but the effects of stellar diameters and finite integration time remain comparable. Similar calculations for the other twelve occultations show that the effect of finite integration time dominated in all those cases.

The synthetic ring profiles are then fitted to observations so as to minimize the classical $\chi^{2}$ function:

[TABLE]

where $\Phi$ is the flux, $i$ refers to the $i^{\rm th}$ data point, “obs” refers to observed, “calc” refers to calculated, and $\sigma$ the $1\sigma$-level error of the $i^{\rm th}$ data point. The free parameters of the model are described in the next subsection. The $1\sigma$ error bar on each parameter is estimated by varying this particular parameter to increase $\chi^{2}$ from the best value $\chi^{2}_{min}$ to $\chi^{2}_{min}+1$, the others parameters are set free during this exploration.

3.2 Mid-times and widths of the rings

The best fitting square-well model described above provides relevant parameters that depend on the occulting object. Three cases are possible: occultations by (1) main body; (2) resolved rings; (3) unresolved rings. The relevant parameters in each case are respectively (1) the times of ingress and the egress of the star behind the body; (2) the mid-time of the occultation $t_{0}$, the radial width reprojected in the plane of the rings, $W_{r}$, and the local normal opacity $p_{N}$ for each ring (see Appendix for details); (3) the mid-time of the occultation. Those parameters are listed in Table 4 (resolved ring events), Table 5 (unresolved ring events), and in Table 6 (main body events). The best fits for each occultation are plotted in Fig. 3 and 4 (ring occultations) and in Fig. 5 (main body occultations).

The grazing occultation by C2R recorded in Gifberg (Fig. 1) requires a special analysis. In this geometry, the radial velocity of the star relative to the ring changes significantly during the event (while it is assumed to be constant for all the other events). To account for this peculiarity, we first converted the light curve (time, flux) into a profile ($\Delta r$, flux), where $\Delta r$ is the radial distance to the point of closest approach to Chariklo’s center (in the sky plane). Then we can apply the square-well model as explained in section 3.1, except that the flux is now given in terms of $\Delta r$, instead of time. Best fits for ingress and egress are plotted in Fig. 6.

Table 4 summarizes the values of $W_{r}$ for each resolved profile. Fig. 7 shows $W_{r}$ vs. the true longitude $L$ counting from the ascending node. Accounting for the most constraining events, $W_{r}$ varies between 5 km and 7.5 km in C1R and between 0.05 km and 1 km in C2R (at $1\sigma$-level). Fig 7 could constrain the rings proper mode. Unfortunately, the true longitude $L$ plotted in Fig. 7 (and latter in Fig. 11) is not the correct quantity to use in order to detect $m=1$ proper modes (the true anomaly $L-\varpi$ should be used instead of $L$, where $\varpi$ is the longitude of periapse). As the precession rates of the rings are unknown, no conclusion can be made. Nevertheless, those width variations are observed both for a given occultation at different longitudes and for different occultations at different dates, see Fig. 7. Implications are discussed in Section 6.

3.3 Ring inner structures

Fig. 8 shows the best radial profiles of the rings that we have obtained so far, taken from the discovery observation of June 3, 2013 and the April 29, 2014 event. They are currently the only profiles that clearly resolve C1R from C2R, and in the case of the April 29, 2014 event, the only profiles that resolve C1R. A W-shape structure inside C1R is clearly seen at egress in the Springbok and SAAO profiles, and marginally detected in the Springbok ingress profile, while being absent (to within the noise) in the SAAO ingress profile.

Note that small (2-4 km) variations of radial distances between the two rings are visible in Fig. 8. The average gap distance between the two rings on the six profiles is thus 14.8 km.

Since the origin of radial distance has been fixed arbitrarily on the center of C2R, it is not possible to attribute those variations to an eccentricity of C1R, C2R or both. Note also that the April 29 profiles are montages obtained by juxtaposing the profiles of C1R recorded at Springbok and SAAO and the profile of C2R recorded in Gifberg. So, they scan different rings longitudes, and conclusions based on this plot can only be qualitative.

3.4 Ring pole

By analogy with their Uranian counterparts, we expect that Chariklo’s ring orbits have essentially elliptical shapes, corresponding to a normal mode with a $m=1$ azimuthal harmonic number. Moreover, other modes with higher values of $m$ are possible and the two rings may not be coplanar. However, data on Chariklo’s rings are currently too scarce to reach those levels of details. Instead, we have to simplify our approach, considering the observational constraints at hand.

The simplest hypothesis is to assume that the two are circular, concentric and coplanar. Then, their projections in the sky plane are ellipses characterized by $M=5$ adjustable parameters: the apparent semi-major axis $a^{\prime}$, the coordinates of the ellipse’s center $(f_{c},g_{c})$, the apparent oblateness $\epsilon^{\prime}=(a^{\prime}-b^{\prime})/a^{\prime}$ (where $b^{\prime}$ is the apparent semi-minor axis), and the position angle $P$ of the semi-minor axis $b^{\prime}$. For circular rings, $\epsilon^{\prime}=1-\sin(B)$, where $B$ is the ring opening angle ($B=0$ and $B=90\arcdeg$ corresponding to edge-on and pole-on geometries, respectively).

Note that $(f_{c},g_{c})$ is related to the offsets in right ascension and declination between the predicted and observed positions of the object, relative to the occulted star. The positions of Chariklo deduced from $(f_{c},g_{c})$ – at prescribed times and for given star positions – are listed in Table 1. They can be used to improve Chariklo’s ephemeris, once the star positions are improved, using the DR1 Gaia catalog and its future updates.

This circular ring model requires at least $N\geq M=5$ data points in order to provide a unique solution for the ring radius $a$ (coincident with $a^{\prime}$) and its J2000 pole position $(\alpha_{p},\delta_{p})$. Only the June 3, 2013 discovery observation with 7 chords (and thus $N=14$ data points corresponding to the chord extremities) has sufficient constraints to provide unambiguous ring orbits. More precisely, as only one instrument (Danish telescope) could resolve the rings C1R and C2R in 2013, this multi-chord event mainly determines the orbit of C1R, which largely dominates the usually blended ring profiles. Then we assumed that C2R is coplanar with C1R and separated radially from it by a constant distance $\Delta a=14.2\pm 0.2$ km (Braga-Ribas et al., 2014).

The April 29, 2014 event provides two chords ($N=4$ data points) on C2R. This allows us to definitely eliminate one of the pole positions derived from the 2013 event. Actually, determining the angles $B$ and $P$ at a given date provides two possible pole positions, 1 and 2, depending on which part of the rings, as seen in the sky plane, is the “near arm” or the “far arm”, see Braga-Ribas et al. (2014) for details. The C2R chord observed at Springbok turned out to be longer than the longest possible length allowed by solution 2, thus confirming that the preferred solution 1 of Braga-Ribas et al. (2014), based on the long-term photometric behavior of Chariklo (see also below), was actually the correct one.

In order to constrain the pole position, even with $N<M$, we vary the couple ($P,B$) in a predetermined grid, while the other three parameters are adjusted in order to minimize the radial residuals in the sky plane relative to the ring center. Since the pole position is given by two parameters $(\alpha_{P},\delta_{P})$, the 68.7% confidence domain (called $1\sigma$-level here) is obtained by allowing variations of the $\chi^{2}$ function from $\chi^{2}_{\rm min}$ to $\chi^{2}_{\rm min}+2.3$ (Press et al., 1992), and by selecting values of $a$ to $\pm 3.3$ km, the nominal error on the C1R and C2R radii: $a_{C1R}\sim 391$ km and $a_{C2R}\sim 405$ km (Braga-Ribas et al., 2014) . The pole position derived from the April 29, 2014 occultation is displayed in Fig. 9. Note that it is consistent with but less accurate than the pole determined in 2013.

Finally, the October 1st, 2016 event also provided two chords ($N=4$ data points) across the rings, but without resolving C1R from C2R (Fig. 4). Thus, we assumed that the profiles are dominated by C1R, and derived the pole position displayed in Fig. 9. It is again consistent with the poles of 2013 and 2014, but with larger error bars due to the ill-configured chord geometry (nearly diametric) that permits more freedom on the pole position (Fig. 1).

Further constraints are in principle provided by the long-term photometric behavior of Chariklo’s system between 1997 and 2014, as compiled by Duffard et al. (2014), see their Fig. 1. The observed photometric variations can be explained by the changing viewing geometry of the rings, linked itself to the pole orientation. Contrary to the occultation data, the photometric variations do not depend on the particular shape of the rings (e.g. circular vs. elliptic). Fitting for the pole position and accounting for the error bars taken from Duffard et al. (2014), we obtain the possible domain shown in Fig. 9. Note that it is consistent with but less accurate than all our occultation results.

From Fig. 9 we can conclude that our current data set (spanning the 3-year interval 2013-2016) is consistent with circular rings that maintain a fixed pole in space, and to within the current formal error bar on the semi-major axis $a$ ($\pm 3.3$ km). Note that the extensions of the error domains for the pole position (colored regions in Fig. 9) are dominated by the errors in the data (i.e. the timings of the ring occultations), not by the formal error for $a$ quoted above. In other words, even if the ring shape were known perfectly, the pole position would not be significantly improved compared to the results shown in Fig. 9. A Bayesian approach could be used to estimate the probability that the rings are elliptic, considering the data at hand and assuming a random orientation for the ring apsidal lines. Considering the paucity of data and the large number of degrees of freedom, this task remains out of the scope of the present paper. In any case, new observations will greatly help into this approach by adding more constraints on the ring shapes and orientations.

For all the other ring single-chord detections ($N=2$ data points), neither the rings radii nor their pole position can be constrained. Instead, assuming the pole orientation of Braga-Ribas et al. (2014), we determined the ring center, also assumed to coincide with Chariklo’s center of mass. Having only one ring chord introduces an ambiguity as two solutions (North or East of the body center) are possible. However, in all cases but one (August 8, 2016) it was possible to resolve this ambiguity as the absence of detections made by other stations eliminated one of the two solutions. For the August 8, 2016 event, the ambiguity remains, and we give the two possible Chariklo’s positions, see Table 1.

None of the single chords are longer than the longest chord expected from the Braga-Ribas et al. 2014’s solution, and thus remain fully consistent with that solution.

3.5 Sharpness of C1R edges

A striking feature of the resolved C1R profiles from the April 29, 2014 event is the sharpness of both its inner and outer edges. This is reminiscent of the Uranian rings (Elliot et al., 1984; French et al., 1991), and might stem from confining mechanisms caused by nearby, km-sized shepherding moonlets (Braga-Ribas et al., 2014). In order to assess the sharpness of C1R’s edges, we use a simple model, where each edge has a stepwise profile, as illustrated in Fig. 10. Instead of having an abrupt profile that goes from apparent opacity 0 to $p^{\prime}$, we add an intermediate step of radial width in the ring plane $\Delta w_{r}$ and opacity $p^{\prime}/2$ around the nominal ingress or egress times, as deduced from the square-well model described before, see also Table 4. With that definition, $\Delta w_{r}$ is a measure of the typical edge width, i.e. the radial distance it takes to go from no ring material to a significant optical depth.

We explored values of $\Delta w_{r}$ by varying the $\chi^{2}$ function (Eq. 1) from its minimum value $\chi_{\rm min}^{2}$ to $\chi_{\rm min}^{2}+1$. The results are listed in Table 3 and illustrated in Fig. 10. Note that all edges are consistent with infinitely sharp edges ($\Delta w_{r}=0$) to within the $1\sigma$ level and that upper limits for $\Delta w_{r}$ are typically 1 km. No significant differences are noticeable between the inner and the outer edges, contrary to, e.g., some Uranian rings (French et al., 1991).

Note finally that the width of C2R, as derived from the grazing event in Gifberg (Fig. 6) is slightly smaller ($\sim 0.7$ km) than the Fresnel scale ($\sim 0.8$ km). As such, it is not possible to assess the sharpness of its edges.

4 Integral properties of rings: equivalent width and depth

We now turn to the measure of the ring’s equivalent width $E_{p}$ and equivalent depth $A_{\tau}$, two quantities defined and discussed by Elliot et al. (1984) and French et al. (1991), as detailed in the Appendix. Those quantities are physically relevant, as they are related to the amount of material present in a radial cut of the ring, in the extreme cases of monolayer and polylayer rings, respectively.

The values of $E_{p}$ are given in Table 4 (resolved events) and in Table 5 (unresolved events). For the resolved profiles, we have plotted $E_{p}$ against the radial width $W_{r}$ in Fig. 11. Implications in terms of mono- versus polylayer models will be discussed in Section 6. For the profiles that resolve C1R from C2R (and where both rings were detected), and those where the two profiles are blended (the majority of our observations), we have plotted the integrated $E_{p}(1+2)$ against the true longitude $L$ (counted from the J2000 ring plane ascending node) in Fig. 11. From that figure, we see that the values of $E_{p}(1+2)$ lie in the interval 1-3 km, with no significant differences between the various measurements. In other words, no significant variations of $E_{p}(1+2)$ with time and/or longitude are detected in our data set.

In this preliminary study, the rings are considered as one entity C1R + C2R but further studies should treat them independently to derive conclusions on the structure of each of them.

5 Search for faint ring material

The best light curve available in terms of photometric quality is from the Danish Telescope. It was acquired at a rate of 10 frames per second during the 30 minutes bracketing the occultation of June 3, 2013 (Braga-Ribas et al., 2014). It can be used to search for additional material orbiting Chariklo, assuming semi-transparent, uninterrupted, and permanent rings coplanar to C1R and C2R.

For this purpose, we consider the equivalent width $E_{p}(i)$ of the putative ring material intercepted during the acquisition interval $\Delta t(i)$ corresponding to the $i^{\rm th}$ data point, and counted radially in the ring plane. Using the results of the Appendix (see also Boissel et al. 2014 for details), we obtain

[TABLE]

where $\Delta r(i)$ is the radial interval travelled by the star during $\Delta t(i)$ (projected in the ring plane), and where $\phi(i)$ is the normalized stellar flux. Due to projection effects, the value of $\Delta r(i)$ varied between the extreme values of 3 to 4 km during the acquisition interval, which sets the radial resolution of this particular data set.

The values of $E_{p}(i)$ vs. radial distance $r$ is displayed in Fig. 12. Note that the light curve probes radial distances of up to $\sim 12,000$ km, about 30 times the ring radii. Using bins of width 60 km, we evaluate the variance of the difference between two consecutive points in each box, thus eliminating low frequency variations of $E_{p}(i)$. Dividing this variance by two (to account for the fact that the data points are uncorrelated) and taking the square root, we obtain the $1\sigma$ level, standard deviation of $E_{p}(i)$, denoted $E_{p}(1\sigma)$, see the red line in Fig. 12. The value of $E_{p}(1\sigma)$ remains stable in the entire range considered here, with typical values of $20$ m. Thus, at the $1\sigma$ level, we do not detect narrow ($W_{r}<$ 3-4 km) rings coplanar with C1R and C2R with equivalent width larger than about 20 meters. This is about ten times fainter than the equivalent width of C2R (Fig. 11). Note that this limit corresponds to extreme cases of either opaque rings with width $\sim$ 20 m, or semi-transparent rings of width $\sim$ 3-4 km and normal opacity 0.007-0.005, and all the intermediate solutions that keep $E_{p}(i)$ at 20 m.

6 Concluding remarks

We detected Chariklo and/or its rings during a total of thirteen stellar occultations between 2013 and 2016. They demonstrate beyond any doubt that this Centaur is surrounded by a system of two flat rings, C1R and C2R. All the observations at hand are consistent with the circular ring solution of Braga-Ribas et al. (2014), with C1R orbiting at $391\pm 3$ km from Chariklo center and with C2R orbiting outside C1R at an average distance of 14.8 km (Fig. 8). This definitely rules out interpretations of the initial observation of June 3, 2013 by a 3D dust shell, or a set of cometary-type jets being ejected from the surface of the body. In fact, the changing aspect of the rings seen during the occultations is entirely attributable the changing position of Chariklo relative to Earth, with a ring pole position that remains fixed in space (Fig. 9).

Our best resolved observation (April 29, 2014) reveals a W-shaped structure inside the main ring C1R (Fig. 8). Moreover, the radial width $W_{r}$ of C1R measured on the best profiles exhibits significant variations with longitude, with a peak to peak variation of $\delta W_{r}\sim 2.5$ km between 5 and 7.5 km, see Table 4 and Fig. 7. All the resolved profiles of C1R exhibit edges that are consistent with infinitely sharp boundaries, once diffraction and star diameter effects are accounted for. The typical 1$\sigma$ upper limits for the edge transition zones is about one kilometer (Table 3 and Fig. 10). Note finally that none of our observations permits to resolve the profile of ring C2R, whose width is constrained between 100 m to 1 km (Fig. 11).

Remarkably, C1R properties (W-shaped profile, variation of width with longitude and sharp edges) are reminiscent of the narrow eccentric ringlets found around Saturn (French et al., 2016) or Uranus (Elliot and Nicholson, 1984; French et al., 1991). The maintenance of apse alignment could be due to self-gravity (Goldreich and Tremaine, 1979), viscous effects at the edges (Chiang and Goldreich, 2000), or a combination of self-gravity and viscous effects (Mosqueira and Estrada, 2002). If validated, those models may provide insights into the ring physical parameters. For instance, the overdensities of material at some 100’s m from the edges (as seen in Fig. 8) is predicted by viscous models and deserve more detailed observational support in the case of Chariklo. Also, the measure of the eccentricity gradient across the rings, $q_{e}$, could be related to the surface density of the ring material, once Chariklo’s dynamical oblateness $J_{2}$ is known (Pan and Wu, 2016). However, our current data set is too fragmentary for drawing any reliable conclusions in that respect, since both a comprehensive ring orbit model and the knowledge of Chariklo’s $J_{2}$ are missing.

In their simplest forms, the Saturn or Uranus ringlets are described as sets of nested elliptical streamlines, with a width that varies as $W_{r}=[1-q_{e}\cos(f)]\delta a$, where $f$ is the true anomaly, $q_{e}=a\delta e/\delta a$ measures the eccentricity gradient across the ring, $\delta a$ and $\delta e$ being the changes of the semi-major axis $a$ and eccentricity $e$ across that ring. Consequently, the interpretation of Fig. 7 remains ambiguous, since only the true longitude corresponding to the events is currently known, while the true anomaly $f$ is unknown. In fact, any (expected) apse precession between observations impairs a correct interpretation of that figure. At this point, only the total eccentricity variation across the ring can be estimated, i.e. $\delta e=~{}\delta W_{r}/2a\sim~{}0.003$ from the estimations of $W_{r}$ and $a$ given above. This sets a lower limit of the same order for $e$, close to the eccentricity of Uranus’ $\epsilon$ ring, 0.008 (French et al., 1991).

A much better case for modeling the rings would be to derive $W_{r}$ vs. the ring radial excursion $r-a$ relative to the mean radius $r$. The formula above predicts a linear behavior, u. Unfortunately, the ring center is currently undetermined: we assume on the contrary a circular ring to derive it, and determine its pole. The fact that the circular hypothesis provides satisfactory fits to our data, to within the accuracy of C1R’s radius determination (some $\pm 3$ km), suggests that $r-a$ should also vary by a few kilometers at most. In any case, the degeneracy between the ring eccentricity and its pole position can be lifted by obtaining several multi-chord occultations and more accurate pole positions than shown in Fig. 9 (and thus distinguish between projection and eccentricity effects). Also, as expected apsidal precession rates are of the order of a couple of months (Sicardy et al., 2016), observations closer than that in time should be done to derive Chariklo’s $J_{2}$.

Turning now to the integral properties of the rings, we have determined the equivalent widths $E_{p}$ of C1R and C2R, when resolved, and the sum of the two when unresolved (Fig. 11). We see that C1R, with $E_{p}(C1R)\sim$ 2 km, contains about ten times more material than C2R, $E_{p}(C2R)\sim$ 0.2 km. On one hand, if the equivalent width is constant within radial width, the ring can be considered as monolayer (French et al., 1986), as no shadowing by neighboring particules occurs (except in nearly edge-on view). On the other hand, if the ring is polylayer, the equivalent depth is independent from $W_{r}$. In that latter case, the equivalent width can be expressed as a function of ring width $W_{r}$ and the constant value of equivalent depth $\bar{A}_{\tau}$:

[TABLE]

(this equation, based on the work of French et al. 1986 has been corrected by the factor 2 in optical depth due to the diffraction by ring particules - see Appendix). Fig. 11 shows $E_{p}$ vs. $W_{r}$ assuming several values of $\bar{A}_{\tau}$ between 1.15 and 2 km for C1R and between 0.15 and 0.4 km for C2R (no real measurement of this parameter has been made in this work, the lines show the expected trends - see Appendix). Contrary to French et al. (1986), the data do not allow any discrimination between $E_{p}$ or $A_{\tau}$ constant within the radial width. Thus, no choice between the mono- or polylayer models can be made.

Finally, we have searched for faint material ring around the already discovered rings. The best data set at hand provides 1$\sigma$ upper limits of $\sim 20$ m for the equivalent width of narrow ($<$ 3-4 km physical width) rings coplanar with C1R and C2R, up to distances of $12,000$ km (counted in the ring plane). Note that in 2015, direct images of Chariklo have been recorded using HST and SPHERE (Sicardy et al. 2015a, b). The goal was to image the rings and/or look for possible shepherd satellite(s) and jets. Considering material of same albedo as the rings (p=0.1), following limits have been inferred: (1) no satellite bigger than $\sim 2$ km (being brighter than $V\sim 26.1$) up to 6400 km ($\sim 8$ times the ring size) from Chariklo center. (2) no satellite bigger than $\sim 1$ km ($V\sim 27.5$) up to 8 arcsec. For comparison the Hill radius is 7.5 arcsec. (3) no jet, coma or material brighter than $V\sim 28$ corresponding to jets of width $\sim 10$ km or material of optical depth of around $2\times 10^{-5}$ per pixel. Note that HST resolution did not allow to look closer than 1000 km from Chariklo’s center, so the rings were not detected.

Future observations will benefit greatly from the Gaia catalog. A flavor of it has been provided by the Gaia-based prediction of the October 1st, 2016 occultation, which turned out to be correct to within 5 mas in declination (respectively 9 mas in right ascension), corresponding to about $50$ km (respectively 90 km). The improvement of Chariklo’s orbit stemming from successful occultation observations and the sub-mas accuracy of forthcoming Gaia catalogs will provide predictions accurate to the few-kilometer level. This will allow a much better distribution of stations (using portable instruments), with an optimal ring longitude coverage aimed at improving the ring orbital models. It will also be possible to plan multi-wavelength observations to constrain the ring particle sizes. Multi-wavelength instruments are rare, and difficult to obtain unless a strong case is made, based on reliable predictions. Higher SNR light curves will also be obtained in order to calculate the equivalent depths of both rings and definitely answer if the rings are monolayer or polylayer. Finally, the Gaia catalog will allow a much better coverage of Chariklo’s limb, which is currently poorly mapped. The general shape and local irregularities of the body will in turn have important consequences for a better understanding of the ring dynamics.

Appendix A Appendix: Equivalent width and equivalent depth definitions

We define $p^{\prime}$ as the apparent opacity of the ring. It measures the fractional drop of stellar flux $1-I/I_{0}$ as observed from Earth (where $I_{0}$ and $I$ are the incident and transmitted fluxes, respectively). Thus, $p^{\prime}=0$ means a transparent ring and $p^{\prime}=1$ means an opaque ring. By “apparent”, we mean here as observed from Earth in the plane of the sky. The apparent quantities will be primed hereafter to distinguish them for the actual quantities at the level of the ring, see below. The apparent ring optical depth is defined as $\tau^{\prime}=-\ln(1-p^{\prime})$.

Appropriate transformations, accounting for the ring opening angle $B$ and distance $D$ to the ring, must be applied to derive the opacity $p_{N}$ and optical depth $\tau_{N}$ at the ring level, where “N” means normal to the ring plane. Once this is done, one may define the equivalent width $E_{p}$ and equivalent depth $A_{\tau}$ of the ring as the integrals of $p_{N}$ and $\tau_{N}$, respectively, over the ring radial profile of width $W_{r}$ (measured radially in the plane of the ring):

[TABLE]

[TABLE]

where $v_{r}$ is the radial velocity of the star relative to Chariklo in the ring plane.

The quantities $E_{p}$ and $A_{\tau}$ are relevant for two extreme cases of ring structures. One is a monolayer ring, in which case $p_{N}=|\sin(B)|\cdot p$ (for $|\sin(B)|\leq 1/p$), where $p$ is the ring opacity as seen under an opening angle $B$. The other model is a polylayer ring (where the ring thickness is much larger than the particle sizes), in which case $\tau_{N}=|\sin(B)|\cdot\tau$, where $\tau$ is the ring optical depth, seen again under an angle $B$, see details in Elliot et al. (1984).

In principle, $E_{p}$ and and $A_{\tau}$ can be determined by numerically performing the integrations $|\sin(B)|\cdot\int(v_{r}p)dt$ and $|\sin(B)|\cdot\int(v_{r}\tau)dt$ over the observed profiles. Since the convolutions of the profiles by both Fresnel diffraction and stellar diameter conserve energy, those integrations provide the correct values of $E_{p}$ and and $A_{\tau}$. Those two quantities are eventually measures of the amount of material (per unit length) contained along a radial cut of the ring, in their respective domains of validity (monolayer vs. polylayer), see French et al. (1991).

However, complications arise because of two effects: (1) the ring is not an uniform screen of opacity $p$, but rather a set of many particles that cover a fractional surface area $p$ of the ring, while individually diffracting the incoming wavefront, and (2) in several cases, the ring profiles are not resolved, i.e. the entire stellar drop occurs inside an individual acquisition interval, thus “diluting” the opacity $p$ over that interval. We now comment these points in turn.

First, individual ring particles of radius $r$ diffract the incoming wave (with wavelength $\lambda$) over an Airy scale $F_{A}\sim(\lambda/2r)D$, as seen by the observer at distance $D$ from the rings. With $r\sim$ of a few meters and $D\sim 2\times 10^{9}$ km, and using wavelengths in the visible range, we obtain $F_{A}>\sim 500$ km, which is significantly larger than typical values of a few kilometers for $W_{\perp}$, the width of ring as seen in the sky plane. This results in a loss of light in the occultation profiles, making the rings appear more opaque than they actually are. It can be shown that the ring apparent optical depth $\tau^{\prime}$ (in the sky plane) is actually twice as large as its actual value $\tau$, i.e. the one would have for an observer close to the ring: $\tau^{\prime}=2\tau$, see Cuzzi (1985). An equivalent way to describe that effect is to note that the actual ring opacity $p$ is related to $p^{\prime}$ by $(1-p)^{2}=1-p^{\prime}$. Thus, the ring acts as a screen of amplitude for the incoming wave, instead of screen of intensity, see details in Roques et al. (1987).

If the ring profile is resolved, it is enough to estimate numerically the integrals:

[TABLE]

[TABLE]

The second point to examine is the fact that the ring profile may not be resolved during the integration time $\Delta t$. In this case $p^{\prime}$ is not known, and the integrals above cannot be evaluated without an independent piece of information. Let us consider the simple case of a uniform opacity $p$ across the ring profile (square-well model). Then, the apparent equivalent width $E^{\prime}=p^{\prime}W_{\perp}$ (where $W_{\perp}$ is the width of the ring as observed in the sky plane) can be evaluated from energy conservation by $E^{\prime}=f^{\prime}v_{\perp}\Delta t$, where $v_{\perp}$ is the velocity of the star normal to the ring in the sky plane, and $f^{\prime}$ is the fractional stellar drop during $\Delta t$. From the definition of $E_{p}$ above (Eq. A3) and from $(1-p)^{2}=1-p^{\prime}$, one obtains:

[TABLE]

Since $0\leq p\leq 1$, we have:

[TABLE]

i.e. a uncertainty factor of two, depending on the assumption on $p$.

For unresolved events, the fit of the best square-well model to the data allows measurements of $E_{p}$. The problem is that $p$ is badly constrained ($0\leq p\leq 1$) by the fits. Eq. A6 shows that error bars will be much larger than for resolved events. It could be possible to solve that problem by noting that $p^{\prime}=E^{\prime}/W_{\perp}=(E^{\prime}/W_{r})(v_{r}/v_{\perp})$. As we know $W_{r}$ we can constrain $p^{\prime}$, and thus $E_{p}$. Assuming that $W_{r,C1R+C2R}$ lies between 3 and 14 km (see Table 4), the error bars values of $E_{p}$ remain similar to those without the width constraint. As we are not certain that 3 and 14 km are the width minimum and maximum, we choose not to use this constraint.

Note that the case of $A_{\tau}$ is in general harder to solve. Even when the profile is resolved, the densest parts of the ring have high opacities $p^{\prime}\sim 1$, and thus large uncertainties on $\tau^{\prime}=-\ln(1-p^{\prime})$ stemming from the data noise and uncertainties on the baseline levels (Fig. 8). Consequently, we have not attempted to derive $A_{\tau}$ for our current data set.

Acknowledgments: The authors acknowledge support from the French grants “Beyond Neptune” ANR-08-BLAN-0177 and “Beyond Neptune II” ANR-11-IS56-0002. Part of the research leading to these results has received funding from the European Research Council under the European Community’s H2020 (2014-2020/ERC Grant Agreement no. 669416 “LUCKY STAR”). This work is partly based on observations performed at the European Southern Observatory (ESO), proposals 092.C-0186(B) and 092.C-0186(C), and on observations made at the South African Astronomical Observatory (SAAO). This work has made use of data obtained at the Thai National Observatory on Doi Inthanon, operated by NARIT. Technical support was provided by G. Hau and P. Kabath for the February 16, 2014 observation at ESO/VLT, and by G. Román for the July 25, 2016 observation with the Dobson 60-cm telescope at Granada. A. Maury acknowledges the use of Caisey Harlingten’s 50cm telescope for the February 16, 2014 occultation. The 50 cm telescopes used for the Hakos observations belong to the IAS observatory at Hakos/Namibia. E. Jehin is a FNRS Research Associate. TRAPPIST is a project funded by the Belgian Fund for Scientific Research (Fonds National de la Recherche Scientifique, F.R.S.-FNRS) under grant FRFC 2.5.594.09.F A. Pennell and P.-D. Jaquiery thank the Dunedin Astronomical Society. E. Meza acknowledges support from the Contrato de subvención 205-2014 Fondecyt - Concytec, Perú. M. Assafin thanks the CNPq (Grants 473002/2013-2 and 308721/2011-0) and FAPERJ (Grant E-26/111.488/2013). G.Benedetti-Rossi acknowledges for the support of the CAPES (203.173/2016) and FAPERJ/PAPDRJ (E26/200.464/2015 - 227833) grants. R.Vieira-Martins thanks grants: CNPq-306885/2013, Capes/Cofecub-2506/2015, Faperj: PAPDRJ-45/2013 and E-26/203.026/2015. The research leading to these results has received funding from the European Union’s Horizon 2020 Research and Innovation Programme, under Grant Agreement No. 687378, project SBNAF. The authors acknowledges the use of Sonja Itting-Enke’s C14 telescope and the facilities at the Cuno Hoffmeister Memorial Observatory (CHMO), the use of the Skywatcher 16” telescope of the Deutsche Höhere Privatschule (DHPS) in Windhoek and the use of the Meade 14 telescope of Space Observation Learning (Rob Johnstone). Funding from Spanish grant AYA-2014-56637-C2-1-P is acknowledged, as is the Proyecto de Excelencia de la Junta de Andalucía, J. A. 2012-FQM1776. J.I.B. Camargo acknowledges CNPq grants 308489/2013-6 and 308150/2016-3. The research leading to these results has received funding from the European Union?s Horizon 2020 Research and Innovation Programme, under Grant Agreement N. 687378, project SBNAF.