GJ 436
AXA Light Curves

Links internal to this web page
    Basic data
    Summary of transits
    Professional transit LCs
    Amateur transit LCs
    OOT LCs
    Finder image
    All-sky photometry

Comments on Transit LCs

It is now well-establisehd that the flurry of excitement about a possible second planet in the GJ 436 system as suggested by Ribas et al (2008a) has subsided dur to additional observations and analyses. Alonso et al (2008) simulated a range of hypothetical orbital parameters for "c" and imposed new observational constraints to show that it is very unlikely that "c" could exist without greater observational anomalies. At the 2008 IAU meeting in Boston Ribas et al (2008b) retracted the suggestion that "c" should be inferred on the basis of known observations. Prior to this retraction the TTV (transit timing variations) observations on this web page were interpreted as supporting "c" but at the suggestion of McLaughlin (private communication, January 2008) the TTV plot was viewed as merely evidence for the need of a period adjustment. As more observations were added to this web page it became clear that this interpretation was the correct one. For awhile, at least, there were grounds for observing GJ 436b intensively in order to refine the TTV plot; this served to "educate" the amateur community on a role we could play in exoplanet studies.

The star is red and all nearby reference stars are bluer, so all LCs will have a large "air mass curvature" systematic error (such that the star appears brighter at higher air mass). This will be especially pronounced for unfiltered observations. It's best to avoid observing when air mass >1.5. V-band should reduce this effect, R-band will be slightly worse (but will have better SNR), and both unfiltered and BB-band (blue-blocking) will have a large systematic effect. Since the depth is small the amateur observations will be "noisy." If the systematics are uncorrelated then it's legitimate to average (or even better, median combine) the LC parameters derived from the amateur measurements. "Eyeball" averages can be done using the plots, below. Eventually chi-squared solutions may be justified.
Basic Data

    RA = 11:42:11.1, DE = +26:42:24
    Season = Mar 15
    V = 10.68
, B-V = 1.5 2 (very red)
    My estimates (based on one night's all-sky photometry, 2008.02.12): B = 12.28 ± 0.04, V = 10.64 ± 0.03, R = 9.69 ± 0.03, I = 8.26 ± 0.03

Discovery Paper (Gillon et al, 2007):
HJDo = 4222.616
    P = 2.64385 (9) day
    Depth = ~6.5 ± 1.0 mmag (V)
    Length = not published (my measurement of the published LC yields 0.94
± 0.06 hr)

HST-based (Bean et al, 2008)
    HJDo = 4455.278241 (26)
    P = 2.643904 (5) days

Infrared (8 micron, SST obsns by Gillon et al, 2007; 2-planet analysis by Ribas et al, 2008):
    HJDo = 1551.78 (5) which must be a typographical error!
    P = 2.64384 (5) days
    Depth = 7.4
± 0.2 mmag
    Length = not published (my measurement of the published LC yields 0.977
± 0.005 hr )
Schneider's Extrasolar Planets Encyclopaedia listing (
    HJDo = 4280.78148 (15) & P = 2.643904 (5) day (using amateur & professional observations)

AXA-based (V,R,I,BB,C):
    HJDo = 4280.78238 (19) & P = 2.6438956 (6) day (using amateur & professional observations)
    Depth = 7.5 ± 0.2 mmag (average of V, R, BB, C)
    Length = 0.97
± 0.02 hr (average of V, R, BB, C)
    Fp = 0.48 ± 0.08, F2 = 0.88 ± 0.10

Summary of Transits

Amateur observations. (Downloading data files is unrestricted, but when use in a publication is underway please notify observer either directly or via webmaster. More description link.)

Transit depth and length appear to be constant, so far. Median depth = 7.5 ± 0.3 mmag, length = 0.97 ± 0.03 hour.

So far there's no evidence for a dependence of transit depth on filter band (which surprises me, given that this is a grazing transit). (Recent data has not been plotted.)

Amateur Transit Light Curves












9130MXI1  Long run after egress provides a very good check of stability and OOT model fit; good job!


8522roev  Three LC versions of the same data. The bottom one is the elast reliable because it was produced by an automatic reduction program that still has "bugs."

8506roev  Cloudy near end.


8428gary Windy most of the session.


First submission by Miguel Rodriguez, using a 6-inch Newtonian tlescope. Good agreement with consensus though a little noisy at high air mass end of observations.

8423nave (waiting for permission)

8415roeb  We're puzzled by the late arrival of this LC. Jim says the clock was checked and there were no saturation issues. This is the only B-band LC for GJ 436 on the AXA.


8325nave (waiting for permission) 


8322gary (data lost due to computer hard disk crash)

8317nave (waiting for permission) 




8301mulr (need permission) First submission; congratulations! Great job with an 8-inch aperture.

8301nave (waiting for permission)

8301gary Apologies for including this noisy egress only LC.





8216dufr (need permission) First submission by this observer. Note the quality and 8-inch aperture.

8217mrch (need permission)




8214wltr (need permission) First-time submission; congraulations! LC shape and timine are consistent with the consensus.

8208mend Amazingly good quality LC, especially for an 8-inch and first-time contributor.


8208greg  The early data had too long exposure times that saturated stars; when exposure was shortened behavior improved. Note that the egress was "late" by approximately the same amount as indicated by the Staels observations of the same transit.

8208stae (need permission) This LC "confirms" the recent trend to "lateness" - which is important evidence for a perturbing exoplanet proposed by Ribas et al, 2008. The high dependence upon air mass is due to observations being unfiltered and GJ 436 being very red and all reference stars being bluer.

8206gary Frequent uses of a hair dryer to remove frost from the corrector plate; temp = 28 F, Dew Pt = 22 F (RH = 78%). WWV check of time tags.

8201nave (waiting for permission)

8124srdc The clock was checked & found to be accurate.

7c31gary Air mass curvature is high due to use of a BB-filter and high air mass at the beginning.

7531gary Mid-transit = 4251.6992 (JD) = 4251.6997 (HJD). Depth = 8.5 mmag (R). Length = 0.78 hr.

7517vanm Good quality.

Out-of-Transit Light Curves

9222-13-FE2 OOT

The closest transit was at 23.06 UT on 2008.03.24 (i.e., 8.9 hrs before mid-observing session).

These observations were a test of a new optical configuration which accounts for the short duration. Nevertheless, there seems to be mild evidence for 1 mmag variations on an hourly timescale.

Professional Transit Light Curves

Gillon et al (2007) SST observations at 8 micron wavelength, reproduced from Ribas et al (2008). Lowest panel shoes effect of a hypothetical 0.1 degree inclination change that could be produced by perturbations from a 5-Earth mass
outer orbit planet in a 2:1 resonant orbit.

V-band, Observatory of Geneva 1.2-meter Euler telescope at La Silla Observatory, Chile
(Gillon et al, 2007). Mid-transit at 2007 May 02, 02:41 UT. My measurements of this LC yield depth ~6.5 ± 1.0 mag, length = 0.943 ± 0.064 hr.

Finder Image

 Figure F1. Finder image with identifier star numbers (above) and J-K colors (times 100, below) selected stars. GJ 436 has V = 10.68 and Rc = 9.66.

 Figure F2. This image has FOV = 16 x 11 'arc and FWHM ~2.5 "arc. Tentative magnitudes for these stars is summarized in the table below.


On the night of 2008.02.12 UT I conducted all-sky photometry measurements of the GJ 436 region. I used the Landolt star field at RA/DE 06:52/-00:27, which also has an extensive list of Henden all-sky observational results for many more stars than in the Landolt list. The Landolt field was observed before and after GJ 436, all at the same air mass. My recent experience with landolt stars is that some of them have changed over the decades, and not enough of them have masgnitudes for all four bands. Further, my culled list of Henden magnitudes (=> 4 observations per star) shows better internal consistency than the Landolt magnitudes. So for this all-sky analysis I adopted only Henden magnitudes for calibrating my telescope system and transferring this calibration to the GJ 436 star field. A fuller discussion of my all-sky photometry observing and analysis procedures are available at a web page (still under construction): http://brucegary.net/ASX/x.htm

The following table is based on 18 Henden stars (as many as 53 readings per band) for this one night's observations:

 Figure A1. Results of the all-sky photometry measurements of 2008.02.12. Star numbers correspond to the labels in Fig. F2. 

Normally I don't present all-sky photometry results without completing a second observing session and analysis and verify compatibility between the two results. In this case I have no plans for doing this since I doubt that anyone will use any of these results.

My favorite "reality check" for detecting the presence of a systematic error for one or more bands after performing an all-sky photometry session is to plot color/color scatter diagrams.


Figure A2 and A3. Color/color scatter diagram showing the location of GJ 436 (red square) and the 9 nearby stars (gray squares) in relation to 1259 landolt stars.

If the solution for one filter band had a systematic error it would show up in these plots as a group offset in the color/color scatter diagrams involving that filter. For example, if all B-magnitudes were high by 0.05 magnitude then the goup of gray squares and the red square would be offset to the right by 0.05 magnitude. It's possible that such an offset is present in Fig. A2, but it's clear that greater vaues fo such an offset are very unlikely. An alternative for explaining the rightward shift of Fig. A2 gray squares is for there to be instead a downward shift, or values for V that are to negative by about the same 0.05 magnitude amount. This is unlikely after inspecting Fig. A3, where there is no evidence of shifts. Presumably, all 3 bands (V, R and I) are free of calibration error offsets (unless by some unlikely circumstance there are offset errors in all 3 bands that excatly compensate to produce color/color agreement with the Landolot stars). If it is true that V, R and I are free of calibration offset errors greater than ~0.03 magnitude, then what should be make of the funny location for GJ 436 in Fig. A3? I claim that it is inescapable that GJ 436 has a much greater V-I color than the Landolt stars. Since V appears to be normal (e.g., Fig. A2), we must conclude that I is anomalous. In other words, these color/color scatter diagrams show that GJ 436 has an I-magnitude that is brighter than normal by ~0.5 magnitude! I'll leave it to others to explain how this could be the case.

Note: It won't matter what magnitude you assume for reference stars for the purpose of obtaining quality light curves. These estimated values are presented for the purpose of identifying star colors that "match" GJ 436's color which can be useful in minimizing extinction related systematic errors (i.e, LC curvature that's correlated with air mass). In this table the column for R-band will be the most accurate since it is based on observations. The other magnitudes for Stars 1 through 6 are based on JK magnitudes. For GJ 436 the B and I magnitudes are based on color/color correlations for main sequence stars. All stars in the table are compatible with main sequence color/color relationships.


Gillon et al, 2007, Astron, & Astrophys., "Detection of Transits of the Nearby Hot Neptune GJ 436" http://babbage.sissa.it/abs/0705.2219
Butler et al, 2004, Astrophys. J. Lett.,"A Neptune-Mass Planet Orbiting the Nearby M Dwarf GJ 436" http://adsabs.harvard.edu/abs/2004ApJ...617..580B
Ribas et al, 2008a, Astrophys. J. Lett., "A ~5_earth Super Earth Orbiting GJ 436?: The Poser of Near-Grazing Transits"  http://fr.arxiv.org/abs/0801.3230
Ribas et al, 2008b, IAU 253, Boston, MA, 2008 May 19-23
Alonso et al, 2008, "Limits to the planet candidate GJ 436c" http://arxiv.org/abs/0804.3030
Bean et al, 2008, arXiv:0806.0851v2, http://arxiv.org/abs/0806.0851
Coughlin et al, 2008, preliminary and final (ApJL, pay) 
Batygin et al, 2009, "A Quasi-Stationary Solution to Gleise 436b's Eccentricity," preprint: http://arxiv.org/abs/0904.3146

Return to calling web page AXA

WebMaster: Bruce L. Gary. Nothing on this web page is copyrighted. This site opened:  July 04, 2007 Last Update:  2009.08.02