OGLE-2016-BLG-1195Lb: A Sub-Neptune Beyond the Snow Line of an M-dwarf Confirmed by Keck Adaptive Optics Aikaterini Vandorou 1,2,3 , Lisa Dang 4,5 , David P. Bennett 1,2,3 , Naoki Koshimoto 6 , Sean K. Terry 1,2,7 , Andrzej Udalski 8 , Jean-Philippe Beaulieu 9,10 , Christophe Alard 10 , Aparna Bhattacharya 1,2 , Joshua W. Blackman 9 , Ian A. Bond 11 , Tarik Bouchoutrouch-Ku 4 , Andrew A. Cole 9 , Nicolas B. Cowan 4,12 , Jean-Baptiste Marquette 10,13 , Clément Ranc 10,14 , Natalia E. Rektsini 9,10 , Sylvain Cetre 15 , Jim Lyke 15 , Eduardo Marin 15 , and Peter Wizinowich 15 1 Laboratory for Exoplanets and Stellar Astrophysics, NASA/Goddard Space Flight Center, Greenbelt, MD 20771, USA 2 Department of Astronomy, University of Maryland, College Park, MD 20742, USA 3 Center for Research and Exploration in Space Science and Technology, NASA/GSFC, Greenbelt, MD 20771, USA 4 Department of Physics, McGill University, 3600 University Street, Montréal, QC H3A 2T8, Canada 5 Université de Montréal, Institut de Recherche sur les Exoplanètes, 1375 Avenue Thérèse-Lavoie-Roux, Montréal, QC H2V 0B3, Canada 6 Department of Earth and Space Science, Graduate School of Science, Osaka University, Tokyonaka, Osaka, 560-0043, Japan 7 Department of Astronomy, University of California Berkeley, Berkeley, CA 94701, USA 8 Astronomical Observatory, University of Warsaw, Aleje Ujazdowskie 4, 00-478 Warszawa, Poland 9 School of Natural Sciences, University of Tasmania, Private Bag 37 Hobart, Tasmania, 7001, Australia 10 Sorbonne Université, CNRS, Institut d’Astrophysique de Paris, IAP, F-75014, Paris, France 11 Institute of Natural and Mathematical Sciences, Massey University, Auckland 0745, New Zealand 12 Department of Earth & Planetary Sciences, McGill University, 3450 rue University, Montréal, QC H3A 0E8, Canada 13 Laboratoire d’Astrophysique de Bordeaux, Univ. Bordeaux, CNRS, B18N, allée Geoffroy Saint-Hilaire, 33615 Pessac, France 14 Zentrum für Astronomie der Universität Heidelberg, Astronomisches Rechen-Institut, Mönchhofstr. 12-14, 69120 Heidelberg, Germany 15 W. M. Keck Observatory, 65-1120 Mamalahoa Hwy, Kamuela, HI 96743, USA Received 2022 November 28; revised 2025 April 15; accepted 2025 April 17; published 2025 May 20 Abstract We present the analysis of high-resolution follow-up observations of OGLE-2016-BLG-1195 using Laser Guide Star Adaptive Optics with Keck, seven years after the event’s peak. We resolve the lens, measuring its flux and the relative source-lens proper motion, thus finding the system to be a Mp = 10.08 ± 1.18M⊕ planet orbiting an M-dwarf, ML = 0.62 ± 0.05Me, beyond the snow line, with a projected separation of r⊥ = 2.24 ± 0.21 au at DL = 7.45 ± 0.55 kpc. Our results are consistent with the discovery paper, which reports values with 1σ uncertainties based on a single mass–distance constraint from finite source effects. However, both the discovery paper and our follow-up results disagree with the analysis of a different group that also present the planetary signal detection. The latter utilizes Spitzer photometry to measure a parallax signal claiming the system is an Earth-mass planet orbiting an ultracool dwarf. Their parallax signal though is improbable since it suggests a lens star in the disk moving perpendicular to or counter to the Galactic disk rotation. Moreover, microlensing parallaxes can be impacted by systematic errors in the photometry. Therefore, we reanalyze the Spitzer photometry using a pixel level decorrelation model to detrend detector systematics. We find that we cannot confidently recover the same detrended light curve that is likely dominated by systematic errors in the photometric data. The results of this paper act as a cautionary tale that a careful understanding of detector systematics and how they influence astrophysical constraints is crucial. Unified Astronomy Thesaurus concepts: Exoplanet detection methods (489); Gravitational microlensing (672); Adaptive optics (2281) 1. Introduction Gravitational microlensing is a unique method that can detect Earth to super-Jupiter mass exoplanets in wide orbits beyond the snow line (S. Mao & B. Paczynski 1991; A. Gould & A. Loeb 1992; D. P. Bennett & S. H. Rhie 1996). The snow line refers to a region just beyond the habitable zone where water can only be sustained in the form of ice, and it is the predicted region where more massive planets form (J. J. Lissauer 1993; S. Ida & D. N. C. Lin 2004; G. M. Kennedy et al. 2006). Statistical studies have shown that the frequency of planets beyond the snow line (especially ice and gas giants) is about 7 times greater than smaller separation systems (A. Cumming et al. 2008; A. Gould et al. 2010). Contrary to other planetary detection techniques, microlen- sing can detect planets without using any light from the host star. This allows for a range of stellar type hosts to be probed, e.g., Solar-type, M dwarfs, and white dwarfs (J. W. Blackman et al. 2021). In addition, this method can investigate more distant planetary systems in our galaxy toward the Galactic Bulge, thus complementing the other detection techniques that typically probe the Solar neighborhood. The advantages of microlensing, however, also bring certain caveats. Not directly detecting the light from the host star means that the physical parameters of the system, such as mass and distance, cannot be determined easily. Instead, the method accurately constrains the host-planet mass ratio, which may be considered a more fundamental property of planetary systems than the planetary mass (I. Pascucci et al. 2018; A. Lecavelier des Etangs & J. J. Lissauer 2022). However, the planet The Astronomical Journal, 169:319 (15pp), 2025 June https://doi.org/10.3847/1538-3881/adcec2 © 2025. The Author(s). Published by the American Astronomical Society. Original content from this work may be used under the terms of the Creative Commons Attribution 4.0 licence. Any further distribution of this work must maintain attribution to the author(s) and the title of the work, journal citation and DOI. 1 https://orcid.org/0000-0002-9881-4760 https://orcid.org/0000-0002-9881-4760 https://orcid.org/0000-0002-9881-4760 https://orcid.org/0000-0003-4987-6591 https://orcid.org/0000-0003-4987-6591 https://orcid.org/0000-0003-4987-6591 https://orcid.org/0000-0001-8043-8413 https://orcid.org/0000-0001-8043-8413 https://orcid.org/0000-0001-8043-8413 https://orcid.org/0000-0003-2302-9562 https://orcid.org/0000-0003-2302-9562 https://orcid.org/0000-0003-2302-9562 https://orcid.org/0000-0002-5029-3257 https://orcid.org/0000-0002-5029-3257 https://orcid.org/0000-0002-5029-3257 https://orcid.org/0000-0001-5207-5619 https://orcid.org/0000-0001-5207-5619 https://orcid.org/0000-0001-5207-5619 https://orcid.org/0000-0003-0014-3354 https://orcid.org/0000-0003-0014-3354 https://orcid.org/0000-0003-0014-3354 https://orcid.org/0000-0001-5860-1157 https://orcid.org/0000-0001-5860-1157 https://orcid.org/0000-0001-5860-1157 https://orcid.org/0000-0002-8131-8891 https://orcid.org/0000-0002-8131-8891 https://orcid.org/0000-0002-8131-8891 https://orcid.org/0000-0003-0303-3855 https://orcid.org/0000-0003-0303-3855 https://orcid.org/0000-0003-0303-3855 https://orcid.org/0000-0001-6129-5699 https://orcid.org/0000-0001-6129-5699 https://orcid.org/0000-0001-6129-5699 https://orcid.org/0000-0002-7901-7213 https://orcid.org/0000-0002-7901-7213 https://orcid.org/0000-0002-7901-7213 https://orcid.org/0000-0003-2388-4534 https://orcid.org/0000-0003-2388-4534 https://orcid.org/0000-0003-2388-4534 https://orcid.org/0000-0002-1530-4870 https://orcid.org/0000-0002-1530-4870 https://orcid.org/0000-0002-1530-4870 https://orcid.org/0000-0002-1646-442X https://orcid.org/0000-0002-1646-442X https://orcid.org/0000-0002-1646-442X http://astrothesaurus.org/uat/489 http://astrothesaurus.org/uat/672 http://astrothesaurus.org/uat/2281 https://doi.org/10.3847/1538-3881/adcec2 https://crossmark.crossref.org/dialog/?doi=10.3847/1538-3881/adcec2&domain=pdf&date_stamp=2025-05-20 https://crossmark.crossref.org/dialog/?doi=10.3847/1538-3881/adcec2&domain=pdf&date_stamp=2025-05-20 https://creativecommons.org/licenses/by/4.0/ formation process is thought to depend sensitively on the host star and planet masses (G. D. Mulders et al. 2015), so both masses and mass ratios are important. The individual masses of a planetary microlensing event can be determined if both second-order effects on the light curve are measured (e.g., K. Furusawa et al. 2013; A. Udalski et al. 2018): finite source effects (features due to the source radius) and the microlensing parallax, πE. Most planetary microlensing events exhibit finite source effects; however, observable parallax effects are uncommon. Using just the finite source effects, the physical parameters can be estimated using a Bayesian analysis with a Galactic model as a prior (J. P. Beau- lieu et al. 2006; B. S. Gaudi et al. 2008). However, this method typically yields poorly constrained parameters, since it only uses a single constraint. Parameters can be estimated with greater precision when additional constraints on the mass– distance relation are added. A complementary mass-measurement method that has been successful is high angular resolution follow-up observations from 8 to 10 m class ground telescopes, or the Hubble Space Telescope (HST). Follow-up observations can take place several years after the microlensing event has occurred, which means the source and lens can be resolved and their fluxes measured individually (e.g., A. Bhattacharya et al. 2018; D. P. Bennett et al. 2020; A. Vandorou et al. 2020; S. K. Terry et al. 2021, 2022). With this method, it is also important to identify the lens star, since companions to either the lens and source can also be mistaken for the lens system. The source star can be identified by comparing the candidate lens-source separations to the relative proper motion predicted by the light- curve model. In this paper, we present the Keck follow-up observations of OGLE-2016-BLG-1195 (henceforth, OB161195) used to resolve the source and lens. Section 2 discusses the previous analysis of this event. Section 3 discusses the Keck adaptive optics observations and our measurements of the flux ratio and separation of the lens and source, and constrain the relative lens-source proper motion to constrain the relative source-lens proper motion, thus obtaining accurate physical properties of the planetary system. Section 4 describes the image-constrained light-curve modeling used (D. P. Bennett et al. 2024) to find light-curve models that fit both the light curve and high angular resolution follow-up data. Section 5 presents the inferred properties of the OGLE-2016-BLG- 1195L planetary system, and we present our conclusions in Section 6. 2. Microlensing Event OGLE-2016-BLG-1195 The microlensing event OB161195 was alerted by the Optical Gravitational Lens Experiment (OGLE, A. Udalski et al. 1994; A. Udalski 2003) on 2016 June 27. The event was also independently alerted by the Microlensing Observations in Astrophysics (MOA, I. A. Bond et al. 2001) collaboration on 2016 June 28, where a 2.5 hr photometric perturbation was discovered (the event has the alternative name of MOA-2016- BLG-350). Observations were taken in the I and V bands with OGLE, and R and V bands with MOA. OB161195 is located toward the Galactic Bulge at (R.A., decl.)= (17:55:23.50, −30:12:26.1) and Galactic coordinates (l, b)= (−0.004, −2.475). The initial data, modeling, and analyses of the microlensing light curve are presented in I. A. Bond et al. (2017) (henceforth, B17). The event exhibited a planetary anomaly with a mass ratio, q = 4.2 ± 0.7 × 10−5, which at the time of detection was the lowest mass ratio found in microlensing. The Einstein ring radius (θE) was derived from the microlensing parameters (Einstein radius crossing time, tE, and the source radius crossing time, t*) and the angular size measurement of the source star, θ*. The physical parameters of the lensing system were estimated by employing a Bayesian technique, as described in D. P. Bennett et al. (2008). A Galactic model was used, which comprised of a double-exponential disk (I. N. Reid et al. 2002) and a bar model (C. Han & A. Gould 1995). The Bayesian estimate of the lens mass was found to be = - +M M0.37L 0.21 0.38 , at a distance of = - +D 7.20L 1.02 0.80 kpc. These values imply a planet mass of = - + ÅM M5.10planet 2.85 5.5 . The high uncertainties of these values are due to there only being a single mass–distance constraint, the θE value, therefore the Bayesian result is highly dependent on the prior assumption that stars of all masses are equally likely to host a plant at the measured q and s. Therefore, these estimates could be wrong if this assumption is wrong. From the light-curve model, B17 find the apparent color and magnitude of the source star to be (V − I, I)s = (2.113, 19.581) ± (0.020, 0.008). Using extinction values AI = 1.627 ± 0.018, E (V − I) = 1.450 ± 0.007 and AK = 0.256 ± 0.017 (S. Nishiyama et al. 2006; F. Surot et al. 2019), and the color relation V − K = 1.53 (M. S. Bessell & J. M. Brett 1988), we find a predicted source magnitude of Ks,pred = 17.34 ± 0.04mag. This can be compared at a later stage to our measure Keck K-band value for the source. In parallel, the event was also observed in the paw-print of the Korean Microlensing Telescope Network (KMTNet, S.- L. Kim et al. 2016), which operates three 1.6 m telescopes equipped with four square degree wide-field imagers. Their data were taken in survey mode with the majority in I band, and only a few in V band, and it was not influenced by the OGLE and MOA alert. Their data confirmed a planetary perturbation in the light curve, which was presented in Y. Shvartzvald et al. (2017) (henceforth, S17). Their source star characteristics, extinction determination, mass ratio, and projected separations from the light-curve model are in agreement with B17. The light-curve parameters from both detection papers can be found in Table 1. Subsequent Spitzer observations were also obtained with a cadence of one observation per day. A strong microlens parallax of πE ∼ 0.45 was retrieved by fitting the KMTNet and Spitzer light curves simultaneously; however, the low signal-to-noise Spitzer data (presented in Figure 1 of S17) is seen to systematically differ from the model. In addition, S17 report a lens-source relative proper motion of μhel,(N,E) ≡ (± 4.0, −7.5)mas yr−1 measured from their parallax value, which suggests a counter-rotating lens in the disk. Combining this Spitzer parallax with the light-curve model parameters from the KMTNet observations, S17 found the system to be an Earth-mass planet, = - + ÅM M1.43planet 0.32 0.45 , orbiting an ultracool dwarf, = - +M M0.078L 0.012 0.016 , at a distance of = - +D 3.19L 0.46 0.42 kpc. The large difference between the conclusions of B17 and S17 are entirely due to the Spitzer data and their constraint on the parallax measurement. This measurement, and the reason for the discrepancy, is investi- gated in Section 5.1. 2 The Astronomical Journal, 169:319 (15pp), 2025 June Vandorou et al. 3. Keck Follow-up of OGLE-2016-BLG-1195 We observed OB161195 with Keck II’s laser guide star AO system (P. L. Wizinowich et al. 2006) and the NIRC2 science instrument on 2018 August 3. We used the wide camera with a plate scale of ∼39.69 mas pixel−1. Images were taken using the Kshort (henceforth, K ) passband, where 15 good images were obtained. The exposure time of the images was 30 s with a dither of ∼2″. We reduced the images using the standard techniques described in J. P. Beaulieu et al. (2016) and V. Batista et al. (2014). This procedure includes dark current and flat-field correction. After sky correcting the wide-frame images, they were stacked using SWARP (E. Bertin 2010) and the stellar fluxes were measured using aperture photometry with SExtractor (E. Bertin & S. Arnouts 1996). We cross identify 71 stars in our stacked wide-frame with the Vista Variables in the Via Lactea survey data (D. Minniti et al. 2010) as described in J.-P. Beaulieu (2018). This allowed us to calibrate our K-band Keck photometry. Thus, the total measured combined K-band magnitude of both source and lens is: = K 16.38 0.04 mag.SL On 2023 August 12, OB161195 was observed with Keck’s OSIRIS instrument, and made use of the new Keck I real-time controller system as part of the KAPA upgrade (J. C. Y. Chin et al. 2022; P. Wizinowich et al. 2022). OSIRIS has a narrow camera with plate scale of 9.942 mas pixel−1. In total, 15 usable images were obtained in Kprime (hereafter, Kp) band and reduced using the Keck AO Imaging (KAI) data reduction pipeline (J. R. Lu et al. 2021). The full width at half-maximum (FWHM) across each frame was ∼53 mas with an average Strehl ratio of 0.368. The 14 frames were stacked using a jackknife method (M. H. Quenouille 1949, 1956; J. W. Tukey 1958), where N− 1 frames are stacked, where N is the number of total frames (for this event N= 15). Therefore, we produce 15 different stacks of 14 frames, since one frame is removed each time. See Section 3.2.3 for more details on the jackknife routine and getting error estimates using this method. The predicted separation between the source and lens is estimated to be ∼ 67 mas in 2023 (∼7.12 yr after peak magnification) with a predicted geocentric source-lens relative proper motion of ∼μGeo,rel = 9.38 ± 0.75 mas yr−1 (I. A. Bond et al. 2017). Since this separation is less than 1 FWHM, we expect the source and lens to be partially resolved, and thus found it necessary to use a point-spread function (PSF) fitting photometry routine to measure both stars individually. Additional epochs of data for this event were also collected in 2019, 2020, and 2021. The quality of the 2019 images was very low, with a FWHM of >100 mas. At a predicted separation of ∼33 mas the lens would be undetectable, and therefore the data were not used. The quality of the 2020 data was good, with a FWHM of ∼60 mas, and the lens was faintly detected. The quality of the 2021 data was also not very high with a FWHM of ∼75 mas. The lens was not visibly detected by eye; however, a Markov Chain Monte Carlo (MCMC) revealed a tentative lens detection. The additional epochs of data are presented in Section 3.3 The 2018 wide-frame images were used for calibration only, whereas the 15 jackknife frames from 2023 (and additional data from 2020 and 2021) were used to resolve the source and lens. The detailed analyses of the narrow images are described in Section 3.2. Table 1 Best-fit Light-curve Model Parameters from B17 and S17, and the New Values Found from Remodeling Using a 2L1S (Two Lens One Source) Model and a 2L2S (Two Lens Two Source) Model Parameters Bond17 Shvartzvald17 2L1S 2L2S tE (days) 10.16 ± 0.25 9.95 ± 0.11 10.09 ± 0.11 9.72 ± 0.28 t0 (HJD’) 7568.772 ± 0.002 7568.769 ± 0.001 7568.771 ± 0.001 7568.771 ± 0.001 tfix (HJD’) L L 7569.0 7569.0 u0 0.0514 ± 0.0014 0.0532 ± 0.0007 0.0522 ± 0.0007 0.0543 ± 0.0018 s 1.0698 ± 0.0078 - +1.0862 0.0075 0.0080 1.0788 ± 0.0461 1.0798 ± 0.0446 α (rad) 0.966 ± 0.005 0.969 ± 0.002 2.174 ± 0.002 2.173 ± 0.002 t* (days) 0.0336 ± 0.0023 - +0.0283 0.0038 0.0034 0.0310 ± 0.0017 0.0311 ± 0.0016 q/10−5 4.25 ± 0.67 - +5.68 0.72 0.80 4.80 ± 0.49 4.95 ± 0.49 θE (mas) 0.261 ± 0.020 - +0.286 0.038 0.053 0.289 ± 0.008 0.290 ± 0.013 πE,N L −0.3016 ±0.0074 0.0187 ± 0.0059 0.0188 ± 0.0057 πE,E L −0.377 ± 0.032 0.0552 ± 0.0052 0.0536 ± 0.0052 Is,0 17.82 ± 0.02 17.84 ± 0.04 17.95 ± 0.04 17.90 ± 0.04 (V − I)s,0 0.71 ± 0.02 0.68 ± 0.03 0.66 ± 0.06 0.66 ± 0.05 Is2,0 L L L 18.07 ± 0.25 (V − I)s2,0 L L L 0.40 ± 0.47 t0,2 (HJD’) L L L 7584.089 ± 11.073 u0,2 L L L 2.4520 ± 0.4844 f2rI L L L 0.701± 0.226 f2rV L L L 0.884 ± 0.302 dtE21 (days) L L L 0.243 ± 0.219 dα (rad) L L L 0.00244 ± 0.022 DS (kpc) L L 8.75 ± 0.62 8.74 ± 0.63 χ2 19580.4 10213 28180.23 28165.64 d.o.f 19587 L 28170 28175 Note. The 2L1S and 2L2S models use image-constrained modeling from the 2023 Keck data. The S17 model has eight degenerate solution, we present the best-fit wide solution. Parameter descriptions can be found in Section 4. 3 The Astronomical Journal, 169:319 (15pp), 2025 June Vandorou et al. 3.1. Target Identification The initial analysis conducted with the 2020 data was with the target being misidentified as the star directly southeast of the true target (see Figure 2 panel (A)—the correct target is identified with a red dot). This problem was first brought to the author’s attention by A. Gould et al. (2023). The misidentified target star did reveal a residual feature in the 2020 data, which was thought to be the lens. This faint feature was again seen in the 2023 data; however, the proper motion was not consistent with it being the lens (see Appendix for more details.). The misidentification was also confounded by the fact that the false target had the predicted source brightness, while no other star in the vicinity did. However, given that an xallarap model was evaluated (but ultimately rejected) in the discovery paper by I. A. Bond et al. (2017), we should have considered the possibility of a binary source, and therefore the possibility that the target would be “brighter” than predicted. To confirm which star is the true OB161195 target, we conducted a reanalysis of the MOA data. This gave inconclusive results for the identification between the false and correct target because different reference images favored different stars. Finally, the correct target was identified in the OGLE data using multiple reference images, and the result was conclusive (see Figure 1). 3.2. PSF Fitting on 2023 Data We used a PSF fitting routine on each of the 15 jackknife frames to detect the lens, which we expect to be partially resolved and relatively faint compared to the source. The analysis is done using a modified version of the DAOPHOT-II package (P. B. Stetson 1987). We call this DAOPHOT- MCMC, since it runs Markov Chain Monte Carlo (MCMC) sampling on the target star. Additional details of this MCMC routine are outlined in S. K. Terry et al. (2021). The first step in the DAOPHOT-MCMC routine is to build an empirical PSF model using single bright stars in the field of the 15 jackknife frames. The stars chosen are located toward the center of the field where image distortion is the least. Using the same reference stars, a different PSF model is built for each of the jackknife frames because the shape of the PSF varies from image to image. These variations contribute to the astrometric and photometric errors described further in Section 3.2.3. We tested three PSF models, each of which had a different combination of reference stars. The selected stars had to be bright but not saturated, single, and close to the area of the target star which is at the center of the frame. This narrowed our list down to three to five stars, therefore there is some overlap between the models. To determine which PSF model to use, we compared their “chi” values, which is the normalized rms of the residuals that are left. PSF model 1 had a chi= 0.0543, model 2 a chi= 0.0533, and model 3 a chi= 0.0553. Therefore, we continued with model 2, which had the lowest value. The PSF model is then fitted to the target star, which is a blend of the source and lens. This produces corresponding residual frames for each of the 15 jackknife images, and can be seen in panel (B) of Figure 2, which we call the “1-star” residual. In this residual frame, all stars have been fitted with the PSF model. Our residual frames indicate a signal which we identify as the lens star (Figure 2, panel (B)). We then rerun the routine on the 15 jackknife frames using the 2-star fitting mode on our target star, and this produces a featureless residual, as shown in panel (C) of Figure 2. Once we have identified the lens star, we use the MCMC portion of the DAOPHOT-MCMC to constrain the (x,y) pixel positions of the source and lens star, as well as their flux ratio. We do this initially using a 2-star fit which assumes one source and one lens. However, this scenario implies that the source star would be almost 1 mag brighter than the predicted source K-band magnitude from the discovery papers B17 and S17 of Ks,pred = 17.34 ± 0.04 mag. This implies that there is very likely to be a binary companion to the source star. Therefore, we also explore a scenario where a third star is contributing Figure 1. OGLE identification of OB161195. Red dots on the OGLE image are OGLE detected stars. The target is shown as a blue dot, which is the centroid measured at the peak of magnification. In the OGLE reference image, it is a blend of faint stars. The Keck image shows the OGLE stars transformed to the Keck grid, in red, and the target in blue. The Keck OSIRIS frame is 9.942 mas pixel−1, and the OGLE frame is 260 mas pixel−1. 4 The Astronomical Journal, 169:319 (15pp), 2025 June Vandorou et al. flux, repeating our MCMC run using a 3-star fit which assumes two sources and one lens. The DAOPHOT-MCMC fitting routine outputs the best-fit values from each jackknife frame for lens and source star position, and their flux ratio. From these parameters, we can obtain the source-lens relative proper motion (direction and magnitude) and calibrated K-band magnitudes for each star (see Table 2 for the 2-star scenario and Table 3 for the 3-star). 3.2.1. 2-star Scenario In the 2-star scenario (1 lens and 1 source), we find the direction of relative heliocentric proper motion to be (μHel,East, μHel,North) = (10.393, 3.592) ± (0.351, 0.383)mas yr−1, and the vector m = 10.995 0.355rel,Hel mas yr−1. We can calculate the flux of the lens using the flux ratio fL/fS = 0.044 ± 0.003 and the total flux that we measure from the wide-frame Keck images (KSL = 16.38 ± 0.04). We do this by using the following equations: ( ) ( )/- = -K K f f2.5 log 1L S 10 L S ( ) ( )= - +- -K 2.5 log 10 10 , 2K K SL 10 0.4 0.4L S where the notations “L” and “S” represent the lens and source, respectively, and KKeck is the K-band magnitude of the target. Solving Equations (1) and (2) simultaneously, we find a new lens and source K-band magnitudes of: =  =  K K 19.801 0.101 mag 16.443 0.075 mag. L S We notice that the KS value is almost 1 mag brighter than the predicted KS,pred from B17’s light-curve model parameters— hence, motivating us to investigate a companion to the source. Figure 2. Keck OSIRIS K-band jackknife image from the 2023 observing run. The images have a pixel size of ∼9.942 mas pixel−1. Panel (A) marks the target as the pink dot and the approximate position of the residual with a green dot. Panel (B) shows the residuals from a 1-star fit, and panel (C) shows the smooth residuals from a 2-star fit. All stars in the frame are fitted with the PSF model and subtracted to produce the residual frame. Table 2 Separation, Relative Proper Motion, and Flux Ratio from the 2023 Keck Jackknife Images Parameter Value S - L Sep (mas) 78.291 ± 2.525 S - L μrel,Hel,East (mas yr−1 ) 10.393 ± 0.351 S - L μrel,Hel,North (mas yr−1 ) 3.592 ± 0.383 L flux / S flux 0.0447 ± 0.0027 χ2 343 dof 431 Note. Values represent the MCMC best-fit values and errors from DAOPHOT- MCMC 2-star fit analysis (Section 3.2). 5 The Astronomical Journal, 169:319 (15pp), 2025 June Vandorou et al. 3.2.2. 3-star Scenario In the 3-star scenario we consider a lens and two source stars, which we name S1 and S2. The separation between these two source stars is 6.318 ± 1.737 mas. This is less than 1 pixel width (9.942 mas); therefore, it is unclear which is the primary source in the microlensing model. Both S1 and S2 have a similar separation to the lens; however, since S1 has the better constrained X and Y pixel positions, we proceed with these values. Therefore, we find the direction of relative heliocentric proper motion to be (μHel,East, μHel,North) = (10.505, 3.676) ± (0.437, 0.600)mas yr−1, and the vector μrel,Hel = 11.131 ± 0.458mas yr−1. Using the flux ratio fL/fS1+S2 = 0.044 ± 0.003 and the total flux that we know from the wide-frame Keck images (KSL = 16.38 ± 0.04), we can calculate the flux of the lens and the flux of the individual sources following the method described above. Therefore, we find: =  =  =  K K K 19.817 0.077 mag 16.958 0.141 mag 17.453 0.180 mag. L S1 S2 The large error bars on the S1 and S2 magnitudes relative to the lens are due to the very small separation between them, and not being resolved. A result of the very small separation (less than 1 pixel width) is that flux values are exchanged during the MCMC run since the position of each source is also exchanged. The lens star position and flux, however, are unaffected by the unconstrained positions and fluxes of the two source stars. The best-fit MCMC contours for the X and Y pixel positions for the lens and source stars is shown on Figure 3. Figure 3. Best-fit MCMC contours (68.3%, 99.5%, 99.7%) of stellar X and Y pixel positions. Top panels: Lens (green) and one source star (pink). Bottom panels: Lens (green) and two source stars (pink and blue). These contours are overplotted on the 2023 jackknife K-band OSIRIS image (left), and the residuals of that image (right). Each frame is ∼0.35 × 0.35. Table 3 Separation, Relative Proper Motion, and Flux Ratio from the 2023 Keck Jackknife Images Parameter Value S1 - S2 Sep (mas) 6.318 ± 1.737 S1 - L Sep (mas) 78.926 ± 3.136 S2 - L Sep (mas) 78.831 ± 3.746 S1 - L μrel,Hel,East (mas yr−1 ) 10.505 ± 0.437 S1 - L μrel,Hel,North (mas yr−1 ) 3.676 ± 0.600 L flux / S1+S2 flux 0.0440 ± 0.0028 χ2 342 dof 428 Note. Values represent the MCMC best-fit values and errors from DAOPHOT- MCMC 3-star fit analysis (Section 3.2). 6 The Astronomical Journal, 169:319 (15pp), 2025 June Vandorou et al. 3.2.3. Errors with MCMC + Jackknife A major source of uncertainty in the Keck images comes from optical distortion effects in the atmosphere. This can affect the PSF of individual stars in each frame and across a single image. By stacking the 2023 narrow-frames using the jackknife method, and running the PSF fitting routine on each, we can capture the errors associated with PSF variations in the individual images. This routine must be included to account for the systematic errors in the PSF model shapes. The jackknife method involves taking N images and forming a series of stacked images that consist of N − 1 images. Therefore, since we have 15 images in total, we produce 15 stacks that consist of of N − 1 = 14 frames. Finally, the best-fit values and errors from each MCMC run are combined using the jackknife error calculation, where the standard error of parameter x can be found using: ( ) ( ¯) ( )å= - -x N N x xSE 1 , 3i 2 where N is the number of stacked images, xi is the parameter measured in each of the stacked images, and x is the mean of x for all N stacked images. Although the jackknife method takes into account the variations in PSF shapes throughout the different frames, it does not necessarily account for photon noise since we only have 15 images. Therefore, the uncertainty estimate is not more precise that / N1 . The errors from each MCMC can also be combined by taking their average. Since this combination method yields a different error value compared to combining with the jackknife equation, we add these errors in quadrature for the most conservative result. This process is applied to both the 2-star scenario and the 3-star scenario. 3.3. Additional Keck Epochs The Keck data from 2020 and 2021 were analyzed using the same PSF fitting method described for the 2023 data. Using the DAOPHOT-MCMC code, we fit a 2-star PSF model. Since fitting the second source does not impact the relative proper motion or the lens flux, we do not find it necessary to fit a 3-star PSF model. The 2020 data were taken on August 8, 4.12 yr after the peak of the microlensing event. The data consists of 11 images with the Keck I OSIRIS instrument, with a FWHM of ∼60 mas across the frames and an average Strehl ratio of 0.343. We build a PSF model using seven nearby stars in the frame, and then fit this model to the target star. We produce a 1-star residual where the source star is subtracted and see a very faint residual in the predicted position of the lens. We run an MCMC to determine the best-fit source-lens separation and flux ratio. From the source-lens relative proper motion we measure in the 2023 data, we predict a separation of ∼45 mas in the 2020 data and use this as a constraint while running the MCMC. A constrained MCMC was additionally run on the 2021 data. This data was also taken with the OSIRIS Keck I instrument, on August 14, 5.13 yr after the event’s peak. Across the five frames used, we find a FWHM of ∼70 mas and a Strehl ratio of 0.223. A predicted separation of ∼55 mas is expected to be seen in the 2021 data; however, no visible detection of the lens is made upon inspecting the 1-star residual frame. A constrained MCMC is still run on this data using the predicted separation, with the results shown in Figure 4. The separation, relative propoer motion, and flux ratio, from the 2020 and 2021 epochs compared to 2023 are presented in Table 4. It should be noted that the constraint used on the 2020 and 2021 data is just on the magnitude of the separation—not on the direction. In addition, we do not place any constraints on the source or lens flux. 4. Image-constrained Light-curve Models Our detection of the exoplanet host star in the high angular resolution follow-up observations can be used to determine the masses and distance of the host star and planet only when the constraints from microlensing light-curve models are also included. One common approach (D. P. Bennett et al. 2015) is to generate a collection of model parameters that are consistent with the light curve using MCMC calculations, and then to exclude those models that are inconsistent with the high angular resolution follow-up observation results. In cases where the modeling includes the microlensing parallax effect (πE), the fraction of light-curve models in the Markov Chain that are consistent with the high-resolution follow-up data can be very small. This is largely because the light-curve measurement of the source-lens relative proper motion constrains the direction of the πE vector to be parallel to the relative proper motion. Although πE should be present in every microlensing event observed by ground-based tele- scopes, it is often not included in light-curve modeling analyses since its parameters cannot be measured very precisely. This approach can be problematic, however, because even when πE parameters cannot be measured precisely, they can still influence some of the other model parameters. Therefore, it is prudent to include πE values that are consistent with the high angular resolution follow-up data when exploring πE models. Furthermore, the event in this paper, OB161195, has a claimed πE signal from Spitzer observations, so it is helpful to compare the πE signal implied by the high angular resolution follow-up observations to the claim made based on Spitzer images. The vast majority of models from an MCMC run (including πE models) are unlikely to be consistent with the follow-up data, which means that very long and time consuming MCMC runs will be needed to get reasonable sampling of the light- curve models consistent with the data (e.g., A. Bhattacharya et al. 2018; D. P. Bennett et al. 2020). To address this problem, we have modified our fitting code, eesunhong16 (D. P. Bennett & S. H. Rhie 1996; D. P. Bennett 2010), to include the constraints from the position and brightness of the lens star identified in the Keck AO images, as described by D. P. Benn- ett et al. (2024). However, to determine the mass of the host star based on the source-lens relative proper motion (which determines θE) we need to know the distance to the source star, DS, so that the mass–distance relation, Equation (6), can be used. This requires us to include DS as a light-curve model parameter, and we use N. Koshimoto et al.’s (2021) Galactic model to generate a prior DL distribution based on the tE measurement from a more conventional light-curve model without the high angular resolution follow-up observation results. Since a companion to the source is likely, we tested two light-curve model configurations, a 2L1S (two lens one source) and a 2L2S (two lens two source) with constraints on μrel and lens flux from the 2023 Keck data. The best-fit parameters can 16 https://github.com/golmschenk/eesunhong 7 The Astronomical Journal, 169:319 (15pp), 2025 June Vandorou et al. https://github.com/golmschenk/eesunhong be found in Table 1. The 2L2S model improves the fit by Δχ2 ∼ 15; however, there is no significant difference on the planetary system parameters, which are detailed in Section 5. This table shows the Einstein radius crossing time, tE; time of closest approach, t0; the impact parameter, u0; instantaneous projected separation (scaled to θE), s; the angle between the lens axis and the source trajectory, α; source star crossing time, t*; the mass ratio, q; the Einstein angular ring radius, θE; and the source color and magnitude. Since we introduce a source companion to the primary source, additional parameters need to be included, such as the time of closest approach (t0,2) and impact parameter (u0,2) of the second source, and the source 2 to source 1 flux ratios in I and V (f2rI and f2rV, respectively). The two source stars are unlikely to have the same velocity due to orbital motion; therefore, their tE and α values will be different. The difference in these values can be described by dtE21 implied by tE2 − tE1, and dα implied by αs2 − αs1, where tE1 = tE and αs1 = α (D. P. Bennett et al. 2010, 2018b). These parameters describe a circular orbit with no constraint on orbit orientation. For each model configuration, 2L1S and 2L2S, we find a close and wide fit with a difference of Δχ2 ∼ 1. For our analysis, we continue with the 2L2S model since the χ2 is slightly favored and we know from our Keck data that there is excess flux at the location of the source. The remodeled light curve using the full data set (MOA, OGLE, and KMTNet) is shown in Figure 5. We note that an alternative binary source model was considered since there is a companion of the source, and a short-term planetary perturbation can be mimicked by a binary source system with an extreme flux ratio (B. S. Gaudi 1998). This involved the high magnification microlensing of a very faint source companion, which would have been undetectable in our Keck images. Such models are excluded by the light- curve data, but we did find a model, with a more extreme flux ratio of 0.0023, that was a better fit to the MOA and OGLE data than the extreme flux ratio model presented by I. A. Bond et al. (2017). However, this model is excluded more strongly by the combined MOA, OGLE, and KMTNet data with a Δχ2 > 110. 4.1. Source Colors and CMD We plot the positions of both sources and lens on the color– magnitude diagram (CMD) in Figure 6. The MCMC distribu- tions for the source’s color and magnitude from the 2L2S Figure 4. Best-fit MCMC contours of the source and lens pixel positions (X, Y) for 2020, 2021, and 2023. The left panels show the target, and the right panels show the 1-star residual with the source subtracted. The lens best-fit positions are in green and the source in red. Table 4 Separation, Relative Proper Motion, and Flux Ratio from the 2020 and 2021 Keck Data Compared to the 2023 Data for the 2L1S Model Parameter 2020 Data 2021 Data 2023 Data Sep (mas) 48.625 ± 4.022 44.661 ± 18.652 78.291 ± 2.525 μrel,Hel (mas yr−1 ) 11.802 ± 0.482 10.864 ± 3.305 10.393 ± 0.351 L flux / S flux 0.039 ± 0.004 0.029 ± 0.007 0.044 ± 0.003 Lens K mag 19.926 ± 0.108 20.259 ± 0.251 19.801 ± 0.101 χ2 1367 1156 343 dof 684 238 431 Figure 5. The OB161195 best-fit 2L2S light curve with constraints from the high angular resolution Keck data—using the full data set (MOA, OGLE and KMTNET). 8 The Astronomical Journal, 169:319 (15pp), 2025 June Vandorou et al. remodeled light curve are: ( ) ( ) ( ) ( ) ( ) ( ) - =  - =  V I I V I I , 2.111, 19.525 0.053, 0.031 , , 1.851, 19.697 0.469, 0.251 . s1 s2 Using extinction values of AI = 1.762 ± 0.018 and E (V − I) = 1.408 ± 0.007 (S. Nishiyama et al. 2006; F. Surot et al. 2019), we find the dereddened magnitudes: ( ) ( ) ( ) ( ) ( ) ( ) - =  - =  V I I V I I , 0.703, 17.763 0.053, 0.036 , , 0.443, 17.935 0.469, 0.252 . s1,0 s2,0 The dereddened magnitudes can be used to calculate the angular source radius, θ*, using the following relation from T. S. Boyajian et al. (2014): [ ( )] ( ) ( ) / * q = + - -V I I log 2 1mas 0.5014 0.4197 0.2 . 4 10 s0 s0 Therefore, we find θ* = 0.877 ± 0.054 μas for the primary source in the 2L2S model. We can estimate the relative proper motion in the geocentric reference frame from the light-curve parameters using μrel,G = θ*/t* = 10.23 ± 0.77 mas yr−1. This value cannot be directly compared to our measured values from the Keck images, since those are in the heliocentric reference frame. Further details about converting from heliocentric to geocentric frames are given in Section 5. The MCMC distribution of the lens V and I measurements from the remodeled 2L2S light curve are (V − I, I)L = (3.211, 22.803) ± (1.265, 0.809)mag. For the source distance we find DS = 8.74 ± 0.63 kpc. The source distance for the 2L2S model can be used to estimate the physical separation (in astronomical unit) between the two sources when multiplied by the separation in arcsecond units (S1 − S2 = 6.318 ± 1.737 mas). Therefore, we calculate the physical separation between the two sources to be ∼53 ± 15 au. Although the physical separation between the two sources is such that it can barely be detected in the light-curve model, it can be estimated from the length of the separation vector between the two sources multiplied by θE and DS, thus ( ) ( )q - + - ´ t t t u u D . 5 0,2 0 E 2 0,2 0 2 E S⎜ ⎟⎛ ⎝ ⎞ ⎠ This calculation gives a separation of 7.017± 1.895 au at the peak of the microlensing event. If we assume the relative transverse velocity of the two source stars to be ∼10 km s−1 = 2 au yr−1, this would imply the two sources could have separated an additional ∼14 au at the time of the follow-up observations in 2023. Even though this is not in full agreement with the value we find in the 2023 Keck data, it should be noted that the positions of the two sources in the Keck data are not well constrained (due to their small separation which is <1 pixel= 9.942 mas), and the uncertainty on their separation is likely underestimated. However, this does not affect the physical parameters of the planetary system. 5. Planetary System Parameters Without additional constraints (e.g., the lens flux from Keck) the physical parameters of a lensing system can be derived by combining two second-order effects on the microlensing light curve. These are finite source effects and the microlensing parallax effect. The former provides a measurement for the source radius crossing time, t*, which can be used to determine the angular Einstein radius, θE. Therefore, a mass–distance relation can be obtained, ( ) q kp p= = -M D D , au 1 1 , 6L E 2 rel rel L S ⎜ ⎟⎛ ⎝ ⎞ ⎠ where k = -M8.144 mas 1. However, this equation has two unknowns: lens mass and distance. If a microlensing parallax effect, πE, is measured from the light curve, then an additional mass–distance relation can be obtained (A. Gould 1992), which when combined with Equation (6), yields: ( ) p kp q kp = =M . 7L rel E 2 E E Many microlensing events do not have a measurable parallax effect, though, and at best only a projection of the parallax vector parallel to the Earth’s orbit is measured, leaving room for degeneracies. Thus, additional constraints are needed for accurate estimates of the physical parameters to be made, such as high angular resolution follow-up observations with Keck. These can be implemented from an empirical mass–luminosity relation with the measured Keck apparent magnitude. The empirical mass–luminosity relations (also used in D. P. Bennett et al. 2015, 2016, 2018a) are from T. J. Henry & D. W. McC- arthy (1993) (ML � 0.66Me), X. Delfosse et al. (2000) (0.54Me � ML � 0.12Me), and T. J. Henry et al. (1999) (0.10Me � ML � 0.07Me). The empirical mass–luminosity relation merges these three relations, smoothly interpolating between them. The foreground extinction is also considered for the mass– luminosity relation. Following R. Drimmel & D. N. Spergel (2001), we use the following equation to estimate the Figure 6. CMD for the OB161195 field. In black are the OGLE-III stars within 90″ of the target, and with the HST CMD of the Stanek window in green (K. Z. Stanek et al. 1994; S. K. Terry et al. 2020). The red spot is the Red Clump centroid, the orange and blue are source 2 and source 1, respectively. In purple is the location of the lens. 9 The Astronomical Journal, 169:319 (15pp), 2025 June Vandorou et al. foreground extinction for the lens: ∣ ∣ ( ) ∣ ( ) ∣ ( ) / / = - - - - A e e A 1 1 , 8i L D b h D b h i S, sin sin , L S dust dust where hdust is the dust scale height, b is the Galactic latitude of the target (b = −2.475), and i is an index refering to the passband. In addition, since Keck can measure the source-lens proper motion, θE can be recalculated using θE = tEμrel. However, the microlensing light-curve model is in the geocentric reference frame, so we must convert our heliocentric relative proper motion, using: ( )m m p = - Åv au , 9rel,G rel,Hel rel where v⊕ is the projected velocity of the earth relative to the Sun at the time of peak magnification. For OB161195, this was ( ) ( )= - = -Å - -v 28.861, 0.991 km s 6.085, 0.209 au yrE N, 1 1 at HJD’= 7569. With πrel as the relative parallax, we can rewrite Equation (9) as: ( ) ( ) m m= - - ´ - D D 6.085, 0.209 1 1 . 10 rel,G rel,Hel L S ⎜ ⎟⎛ ⎝ ⎞ ⎠ μrel,G and μrel,Hel are in units of mas yr−1, and DL and DS are in units of kpc. This relation is used in the Bayesian analysis of the light curve with the high-resolution follow-up constraints from Keck. Therefore, we find μrel,G = 10.91 ± 0.26 mas yr−1 and θE = 0.290 ± 0.013 mas. These values are the weighted sum of the values from the Keck images and the light-curve parameters (t* and θ*). This is within 1σ agreement with μrel,G = θ*/t* = 10.23 ± 0.77 mas yr−1, calculated previously. For OB161195 we combine constraints from the empirical mass–luminosity relation and Equation (6), plotting them on a mass–distance diagram shown in Figure 7. These two relations are shown in different shades of red, with their intersection (black cross) indicating the result we find by summing over the MCMC runs of the 2L2S model using a Galactic model (discussed in Section 4). It should be noted that Figure 7 is a simplified presentation of the results, whereas Figure 8 and the results shown in Table 5 include the effects of the different correlations with other model parameters from the MCMC analysis. We run two Galactic models with different planet hosting priors (M∧0 and M∧1) and sum over the MCMC results. We compare the results of both priors, finding that the mass measurement does not vary significantly between the two. The M∧0 prior is most commonly used in Bayesian analyses of microlensing events that lack mass measurements (D. P. Benn- ett et al. 2014). However, more recent studies (N. Koshimoto et al. 2021; D. P. Bennett et al. 2024) indicate that the probability of hosting a planet with a fixed mass ratio increases linearly with the host mass, i.e., M∧1 model. For OB161195, the final values that we use are from the M∧1 model. However, because we have a mass estimate from the Keck follow-up data, there is no significant difference between the two models. These results are presented in Table 5 and Figure 8, where the blue histograms indicate the parameters without Keck constraints, and red histograms with Keck constraints (these are generated using the model with the M∧1 prior). Therefore, we find that the OB161195 system is a cold sub-Neptune with a mass of Mp = 10.08 ± 1.18M⊕ orbiting a star with ML = 0.62 ± 0.05Me, at a distance of DL = 7.45 ± 0.55 kpc toward the Galactic Bulge. The projected separation between planet and host is r⊥ = 2.24 ± 0.21 au, placing it beyond the snow line. In Figure 7 we can see a graphical summary of the various conclusions for OB161195, where B17 is represented with a blue cross, S17 with gray and for this work, black. We immediately notice the drastic difference that the added parallax measurement from Spitzer has made to the physical parameters. In general, parallax measurements can be difficult to constrain since they require adequate sampling and event coverage for their effects to be seen on a light curve. Even parallax measurements obtained with space-based telescopes, such as Spitzer, can be ambiguous if there are systematic errors in the photometry that are difficult to detrend due to poor cadence and signal coverage. In S17, the Spitzer light-curve data have low signal-to-noise because the target is faint and covers only the trailing end. In addition, it could be affected by low level systematic errors in the photometry (N. Koshimoto & D. P. Bennett 2020), and therefore it is necessary to investigate the Spitzer data further. 5.1. Reanalysis of the Spitzer Data The large discrepancy between S17’s constraints on the planetary system’s physical parameters and those reported by B17 and this work can be explained by the microlensing parallax direction claimed in S17. The value they present, πE,N, πE,E = (−0.3016, −0.3770) ± (0.0074, 0.032), is highly unlikely when compared to the stellar population of the Galactic disk (see Figure 9). In fact, all their Spitzer solutions yield unlikely parallax values, such as πE,N, πE,E = (−0.2158, −3820) ± (0.0066, 0.032), πE,N, πE,E = (0.2350, −0.413) ± (0.008, 0.031), and πE,N, πE,E = (0.1491, −0.404) ± (0.0075, 0.032). In S17, they explain that the unlikely parallax direction (and therefore a counter-rotating lens) is a direct result of the Figure 7. Mass–distance relations for OGLE-2016-BLG-1195, obtained using the Einstein ring radius (θE) and parallax (πE) from remodeling the light curve with the measured 2023 Keck μrel constraint, and also the 2023 Keck K-band flux constraint (KL). The planet-host properties are derived from the intersection of the two most reliable mass–distance relations. The host system properties estimated by B17 and S17 are shown as crosses in blue and green, respectively. The Spitzer parallax constraint obtained by S17 is shown in gray. This is a simplified presentation of the results that include errors measured on the host star magnitudes and average errors on the θE values. 10 The Astronomical Journal, 169:319 (15pp), 2025 June Vandorou et al. magnification seen by Spitzer, and that high-resolution imaging can resolve this issue. Our high-resolution Keck observations indicate that the apparent Spitzer parallax signals are not real. Using constraints from Keck, the measured parallax direction is πE,N, πE,E = (0.0194, 0.0551) ± (0.0057, 0.0052). Counter- rotating low-mass star (e.g., brown dwarfs) detections have been claimed before with the microlensing method (e.g., OGLE-2017-BLG-0896 Y. Shvartzvald et al. 2019 and MOA- 2015-BLG-231 S.-J. Chung et al. 2019). Both these events however, have large parallax measurements from Spitzer, which have not yet been validated with high angular resolution follow-up observations. Counter-rotating brown-dwarf solu- tions are likely evidence for systematic errors. Therefore, we decided to reassess the Spitzer light-curve data of OB161195. Observations of this event were obtained on two occasions with Spitzer/IRAC (G. G. Fazio et al. 2004) in the 3.6 μm channel as part of the 2016 Spitzer microlensing campaign during 2016 July 12–24 at a cadence of one epoch per day. This corresponds to a total of 17 epochs with six dithered frames each. Additionally, seven epochs of baseline observations of OB161195 were collected during 2019 July 12–24 at the same cadence. Figure 8. Bayesian posterior probability distributions for the host and planetary companion mass, their separation, and lens system distance. Using only light-curve constraints are shown in blue, and with additional constraints from the follow-up Keck data in red. The center of the distributions (68.3%) are shaded in dark red and blue. The remaining 95.4% are shaded with lighter hues. The dark vertical line indicates the median of the probability distribution of each parameter. Table 5 Physical Parameters from B17 and S17 Compared to Those Found in this Paper from the Best-fit 2L2S MCMC Values from Both Planet Hosting Priors, M∧0 and M ∧1 M ∧0 M ∧1 B17 S17 ML (Me) 0.62 ± 0.05 0.62 ± 0.05 - +0.37 0.21 0.38 - +0.078 0.012 0.016 Mp (M⊕) 10.04 ± 1.18 10.08 ± 1.18 - +5.10 2.85 5.50 - +1.43 0.32 0.45 DL (kpc) 7.42 ± 0.56 7.45 ± 0.55 - +7.20 1.02 0.80 - +3.19 0.46 0.42 r⊥ (au) 2.23 ± 0.21 2.24± 0.21 - +2.01 0.32 0.27 - +1.16 0.13 0.16 Note. We adopt the M∧1 values. Figure 9. Two-dimensional parallax distribution for the lens in OB161195. The blue cross indicates the value obtained from this work, with constraints from Keck, and the gray crosses are the values presented in S17 using the Spitzer data set. The color-scale shows the relative probability based on a Galactic model. 11 The Astronomical Journal, 169:319 (15pp), 2025 June Vandorou et al. Following L. Dang et al. (2020), we perform aperture photometry and then model the detector systematics using pixel level decorrelation (PLD). We first identify the six dithered regions of the 256× 256 pixel detector where the target is positions and perform aperture photometry by using a 3× 3 square aperture. We then calculate the relative flux measured by each pixel and use them as regressor for our PLD noise models. We define the measured flux as a combination of the astrophysical model and PLD systematics model as a multi- plicative term. Assuming a 1S1L model for the astrophysical model, we use an MCMC to simultaneously evaluate the astrophysical parameters and PLD coefficients. We find that with such a small number of epochs, the detector systematics at each dither position are poorly sampled, so a self-calibration approach such as the PLD tends to overfit the data, as suggested by the overly optimistic residuals. Consequently, the PLD analysis is inconclusive, which motivates a second look at the Spitzer frames. Nonetheless, we can compare our raw photometry with that presented in S17 which uses S. Calchi Novati et al.’s (2015) PSF fitting procedure. After close inspection of our initial photometry, there is no apparent downward trend as reported by S17, which suggests that the magnification of the source is not significantly greater than the instrumental noise and the potential contamination from nearby sources. We compared the original 2016 observations with the 2019 baseline data and we note that the scatter between dithers in 2016 is systematically larger than that observed in the 2019 observations. This could be indicative that the 2016 observations either exhibit stronger instrumental systematics or that the time-variable contamina- tion from nearby sources was more prominent. Hence, given that the source is faint and that the Spitzer magnification could be as large as the instrumental noise, if not modeled properly, we suggest that the discrepancy between the physical parameters of OB161195 inferred in this work and the results reported by S17 could be due to a poor characterization of the instrumental systematics of this particular event. While PLD tends to overfit OB161195’s Spitzer observa- tions, we investigate the discrepancy physical parameters inferred from S17 and our new Keck constrains by fitting three different light-curve models to our raw photometry with a PLD correction: (1) a flat line, a single-lens model with t0 fixed to that reported by S17, and (3) a single-lens model with t0 fixed to that predicted from our Keck constrains. These fits can be seen in Figure 10. In this analysis, we find that the model with t0 from S17 is disfavored, with the highest χ2 compared to the other models. The model that is favored when comparing χ2 is that from the Keck follow-up results, although the difference is small and not significant enough to conclude which of the two is better. Therefore, our ultimate deduction is that the Spitzer data set is inconclusive, and the parallax yielded from such sparsely sampled data is not a reliable avenue for a mass-measurement to be made. 6. Discussion and Conclusion Our 2023 high angular resolution follow-up observations of the microlensing event OB161195 have allowed us to resolve the source and the lens with a separation of ∼ 79 mas, 7.12 yr after the peak of the event. By directly measuring the lens’s K- band flux, and the amplitude and direction of the source-lens relative proper motion, we have estimated the mass and distance of this planetary system. We explore a scenario where the source star has a companion since we find excess flux at the location of the source, and we rerun the microlensing light- curve model with a 2L2S configuration which improves the model by Δχ2 ∼ 15. We combine this with a 3-star fit on the 2023 Keck data. Microlensing events with a binary source star are not uncommon, e.g., D. P. Bennett et al. (2018b) and C. Han et al. (2023a, 2023b). Effects on the light curve due to a binary source can be seen in the form of xallarap (source acceleration due to due to a binary companion) (M. Domi- nik 1998) or magnification of the second source (K. Griest & W. Hu 1992). We find that OB161195 is not an ultracool dwarf orbited by an Earth-mass planet, instead we find that it is a cold sub- Neptune (or super-Earth) orbiting an M-dwarf, supporting the conclusion of B17 that was reached using a Bayesian analysis with a standard Galactic model and a prior that assumes that for a mass ratio, q, the probability of hosting a planet is independent of the host mass. This prior assumption that is Figure 10. Left panel: Three different models of the Spitzer data. The flat-line model from the 2019 baseline Spitzer data has a χ2 = 120.059. The t0 model inferred from S17 has a χ2 = 121.263. The t0 model inferred from the Keck follow-up results has a χ2 = 102.117. Right panel: S17 ’s original magnification curve from Spitzer photometry as a comparison, with magnification ranging from −1 < y < 7. 12 The Astronomical Journal, 169:319 (15pp), 2025 June Vandorou et al. commonly used in microlensing has to be updated as described in Section 4. New research (N. Koshimoto et al. 2021; D. P. Bennett et al. 2024) indicates that planet detection frequency scales linearly with host mass, for a fixed mass ratio. Although the two priors differ (M∧0 and M∧1), the final results for OB161195 are completely independent of the prior because we have enough measurements to constrain the masses. Cases that only have θE measurements (such as B17) are sensitive to the prior assumption. Although OB161195 is not an ultracool dwarf as reported in Y. Shvartzvald et al. (2017), false detections of these substellar objects could have negative implications on the understanding of their formation and evolution. Ultracool dwarfs are predicted to make up 15%–30% of stars in the Galaxy (P. Kroupa 2001; K. Muzic et al. 2017); however, their formation and evolution are currently only incompletely understood, since direct mass measurements are difficult to achieve for these faint objects. Techniques such as transits (e.g., T. W. Carmichael et al. 2020) and direct imaging (e.g., A. J. Burgasser et al. 2006) have had some success with ultracool dwarf binary systems; however, these methods are only sensitive to close binaries with mass ratios q  0.4. Additionally, they can only measure the total mass of the system, not the individual masses. Microlensing complements these techniques because it has the ability to detect wide-orbit binaries with low-mass ratios. The individual masses can also be measured if the microlensing light curve exhibits both finite source effects and a parallax effect. With OB161195 as an example, however, parallax effects measured by Spitzer are often contaminated by systematics and/or have low signal-to-noise (N. Koshimoto & D. P. Bennett 2020). Therefore, high-resolution follow-up observations are the optimum solution to determine the physical properties of these low-mass ratio systems. Parallaxes in general are not easy to measure and typically result in degenerate light curves. In the case of Spitzer parallaxes, N. Koshimoto & D. P. Bennett (2020) found that >75% of parallaxes obtained by the Spitzer campaign were higher than the median averages predicted by Galactic models, from a study of 50 microlens parallax measurements. They concluded that the main source of this discrepancy came from systematic errors in the Spitzer photometry. In particular N. Koshimoto & D. P. Bennett (2020) draw attention to this event, noting that the Spitzer light curve does not show a clear peak feature and attempts to measure a faint source star, both of which contribute to unusually high systematic errors. A point of interest for OB161195 is its very low-mass ratio, q = (4.94 ± 0.49) × 10−5, because it puts it just below the estimated power-law break seen in the statistical analyses of ∼30 microlensing planetary events by D. Suzuki et al. (2016, 2018). In their work, the authors studied the measured mass ratios of events detected from 2007 to 2012 by the MOA survey, and compared them to the planet distributions from core accretion theory population synthesis models. Although the core accretion models predict planets with mass ratios q < 10−5, measurements indicate a break at q ∼ 1.7 × 10−4, indicating that we do not know the occurrence rates of planets beyond this. The precise location of the break is not well constrained, with its location more likely to be at = ´- + -q 6.7 101.8 9.0 5. Therefore, OB161195 with q ≈ 4.84 × 10−5, is just below the mass ratio function break. High-resolution follow-up observations are crucial for accurate mass measurements of microlensing systems. In this paper we have presented the event OB161195 which had contradictory conclusions from the two detections papers, B17 and S17. Through multiepoch high-resolution Keck AO data, we have shown that Spitzer parallax measurements are not always a reliable mass measurement method because they are prone to systematic errors in the photometric data when the target is faint. Low cadence observations from the space telescope can also impact the credibility of the parallax measurement. The types of planets and host stars that microlensing can detect complement the other techniques, offering a unique window into the population and distribution of these objects in our Galaxy. Acknowledgments This work is supported by NASA through grant NASA- 80NSSC18K0274 and by the University of Tasmania through the UTAS Foundation and the endowed Warren Chair in Astronomy and the ANR COLD-WORLDS (ANR-18-CE31- 0002). This research was also supported in part by the Australian Government through the Australian Research Council Discovery Program (project number 200101909) grant awarded to A.C. and J.P.B. The Keck Telescope observations and analysis were supported by a NASA Keck PI Data Award, administered by the NASA Exoplanet Science Institute. Data presented herein were obtained at the W. M. Keck Observatory from telescope time allocated to NASA through the agency’s scientific partnership with the California Institute of Technol- ogy and the University of California. The Keck I and Keck II real-time controller upgrades are funded by the National Science Foundation’s Mid-Scale Innovations Program award AST-1836016 and Major Research for Instrumentation Pro- gram award AST-1727071, respectively. The Observatory was made possible by the generous financial support of the W. M. Keck Foundation. The authors wish to recognize and acknowl- edge the very significant cultural role and reverence that the summit of Maunakea has always had within the indigenous Hawaiian community. We are most fortunate to have the opportunity to conduct observations from this mountain. D.P. B. and A.B. were also supported by NASA through grant NASA-80NSSC18K0274. Work by C.R. was supported by the Alexander von Humboldt Foundation. L.D. acknowledges support in part through the Technologies for Exo-Planetary Science (TEPS) PhD Fellowship, the Natural Sciences and Engineering Research Council of Canada (NSERC)’s Canada Graduate Scholarship and Banting Postdoctoral Fellowship. L. D., T.B.K., and N.B.C. acknowledge camaraderie and intellectual support of the Institute for Research on Exoplanets and the McGill Space Institute. Appendix Misidentified Target Star As mentioned in the paper, the initial analysis conducted with the 2020 data was wrong since the target star had been misidentified. After submitting the paper with the wrong target star, the problem was brought to the author’s attention by A. Gould et al. (2023). We correct this misidentification and present the results in this paper. In this Appendix, we show the 2020 analysis conducted on the wrong target, along with new data obtained with Keck in 2023. Image reduction and the PSF fitting method is the same as what is described in Section 3. The only difference is the target star that is being analyzed. 13 The Astronomical Journal, 169:319 (15pp), 2025 June Vandorou et al. For both the 2020 and 2023 epochs we see a visible residual in the 1-star fits (when the target star has been subtracted). We name the visible bright target “Star A” and the fainter residual star “Star B.” Star B is northwest of Star A, as seen in Figure 11. In the 2020 Keck data, we find a separation between Star A and Star B of S = 54.49 ± 2.36 mas, and a flux ratio of fB/fA = 0.065 ± 0.007. If Star A and Star B were in fact source and lens, then 4.12 yr would have passed since the peak of the microlensing event and the relative proper motion between these two stars could be calculated to be μrel = 13.22 ± 0.94 mas yr−1. I. A. Bond et al. (2017) predicted a relative proper motion of ∼9.38 ± 0.75 mas yr−1 from the light-curve model. Although this is a difference of ∼3σ, light-curve models without constraints from high-resolution follow-up data or robust higher order effect measurements, such as parallax or xallarap, cannot always predict physical properties of the system accurately. In addition, with the flux ratio measured and the total K magnitude of the target from the Keck 2018 wide image (KA+B = 16.92 ± 0.02), we calculate KStarA,2020 = 16.98 ± 0.05 and KStarB,2020 = 19.96 ± 0.15. Star A is indicated in red and Star B in cyan in Figure 11. I. A. Bond et al. (2017) predict a source K magnitude of ∼17.3, which agrees with our Star A magnitude value if we were to consider this the source star. Since the target’s magnitudes agreed so well with that predicted in the light-curve model, we felt confident that we were working on the correct target star. However, this agreement was just a coincidence, and the true source star for OB161195 was ∼1 mag brighter than predicted. In the 2023 data, we measure a separation of S = 79.71 ± 0.49 mas and a flux ratio of fB/fA = 0.132 ± 0.004 between the two stars. Using these two epochs of data we can calculate the star’s relative position to each other in 2016 using the relative proper motion seen in the 3 yr between 2020 and 2023. At the peak of the microlensing event, the separation should be 0 mas. Therefore, we find a relative proper motion of μrel = 8.41 ± 0.81 mas yr−1, which places Star B at a separation of ∼16–20 mas from Star A in 2016. Using the flux ratio found in 2023, and the calibrated total magnitude from the 2018 Keck wide data, we find KStarA,2023 = 17.06 ± 0.02 and KStarB,2023 = 19.25 ± 0.03. From the OGLE identification of OB161195 that we present in this paper, and the A. Gould et al. (2023) identification, we know that these stars (Star A and Star B) are unrelated nearby stars to the microlensing system OB161195. ORCID iDs Aikaterini Vandorou https://orcid.org/0000-0002- 9881-4760 Lisa Dang https://orcid.org/0000-0003-4987-6591 Figure 11. The 2020 and 2023 Keck narrow image data of the misidentified target star, and the 1-star fit residual image. The best-fit MCMC contours of the two stars’ pixel positions (X, Y) are shown in cyan (Star B) and in red (Star A). North is up and east is to the left. 14 The Astronomical Journal, 169:319 (15pp), 2025 June Vandorou et al. https://orcid.org/0000-0002-9881-4760 https://orcid.org/0000-0002-9881-4760 https://orcid.org/0000-0002-9881-4760 https://orcid.org/0000-0002-9881-4760 https://orcid.org/0000-0002-9881-4760 https://orcid.org/0000-0003-4987-6591 https://orcid.org/0000-0003-4987-6591 https://orcid.org/0000-0003-4987-6591 https://orcid.org/0000-0003-4987-6591 David P. Bennett https://orcid.org/0000-0001-8043-8413 Naoki Koshimoto https://orcid.org/0000-0003-2302-9562 Sean K. Terry https://orcid.org/0000-0002-5029-3257 Andrzej Udalski https://orcid.org/0000-0001-5207-5619 Jean-Philippe Beaulieu https://orcid.org/0000-0003- 0014-3354 Joshua W. Blackman https://orcid.org/0000-0001- 5860-1157 Ian A. Bond https://orcid.org/0000-0002-8131-8891 Andrew A. Cole https://orcid.org/0000-0003-0303-3855 Nicolas B. Cowan https://orcid.org/0000-0001-6129-5699 Jean-Baptiste Marquette https://orcid.org/0000-0002- 7901-7213 Clément Ranc https://orcid.org/0000-0003-2388-4534 Natalia E. Rektsini https://orcid.org/0000-0002-1530-4870 Peter Wizinowich https://orcid.org/0000-0002-1646-442X References Batista, V., Beaulieu, J.-P., Gould, A., et al. 2014, ApJ, 780, 54 Beaulieu, J.-P. 2018, Univ, 4, 61 Beaulieu, J. P., Bennett, D. P., Batista, V., et al. 2016, ApJ, 824, 83 Beaulieu, J. P., Bennett, D. P., Fouqué, P., et al. 2006, Natur, 439, 437 Bennett, D. P. 2010, ApJ, 716, 1408 Bennett, D. P., Batista, V., Bond, I. A., et al. 2014, ApJ, 785, 155 Bennett, D. P., Bhattacharya, A., Anderson, J., et al. 2015, ApJ, 808, 169 Bennett, D. P., Bhattacharya, A., Beaulieu, J.-P., et al. 2020, AJ, 159, 68 Bennett, D. P., Bhattacharya, A., Beaulieu, J.-P., et al. 2024, AJ, 168, 15 Bennett, D. P., Bond, I. A., Udalski, A., et al. 2008, ApJ, 684, 663 Bennett, D. P., & Rhie, S. H. 1996, ApJ, 472, 660 Bennett, D. P., Rhie, S. H., Nikolaev, S., et al. 2010, ApJ, 713, 837 Bennett, D. P., Rhie, S. H., Udalski, A., et al. 2016, AJ, 152, 125 Bennett, D. P., Udalski, A., Bond, I. A., et al. 2018a, AJ, 156, 113 Bennett, D. P., Udalski, A., Han, C., et al. 2018b, AJ, 155, 141 Bertin, E., 2010 SWarp: Resampling and Co-adding FITS Images Together, Astrophysics Source Code Library, ascl:1010.068 Bertin, E., & Arnouts, S. 1996, A&AS, 117, 393 Bessell, M. S., & Brett, J. M. 1988, PASP, 100, 1134 Bhattacharya, A., Beaulieu, J. P., Bennett, D. P., et al. 2018, AJ, 156, 289 Blackman, J. W., Beaulieu, J. P., Bennett, D. P., et al. 2021, Natur, 598, 272 Bond, I. A., Abe, F., Dodd, R. J., et al. 2001, MNRAS, 327, 868 Bond, I. A., Bennett, D. P., Sumi, T., et al. 2017, MNRAS, 469, 2434 Boyajian, T. S., van Belle, G., & von Braun, K. 2014, ApJ, 147, 47 Burgasser, A. J., Kirkpatrick, J. D., Cruz, K. L., et al. 2006, AJS, 166, 585 Calchi Novati, S., Gould, A., Yee, J. C., et al. 2015, ApJ, 814, 92 Carmichael, T. W., Quinn, S. N., Mustill, A. J., et al. 2020, AJ, 160, 53 Chin, J. C. Y., Cetre, S., Wizinowich, P., et al. 2022, Proc. SPIE, 12185, 121850V Chung, S.-J., Gould, A., Skowron, J., et al. 2019, ApJ, 871, 179 Cumming, A., Butler, R. P., Marcy, G. W., et al. 2008, PASP, 120, 531 Dang, L., Calchi-Novati, N. B., Carey, S., & Cowan, N. B. 2020, MNRAS, 497, 5309 Delfosse, X., Forveille, T., Ségransan, D., et al. 2000, AAP, 364, 217 Dominik, M. 1998, A&A, 329, 361 Drimmel, R., & Spergel, D. N. 2001, ApJ, 556, 181 Fazio, G. G., Hora, J. L., Allen, L. E., et al. 2004, ApJS, 154, 10 Furusawa, K., Udalski, A., Sumi, T., et al. 2013, ApJ, 779, 91 Gaudi, B. S. 1998, ApJ, 506, 533 Gaudi, B. S., Bennett, D. P., Udalski, A., et al. 2008, Sci, 319, 927 Gould, A. 1992, ApJ, 392, 442 Gould, A., Dong, S., Gaudi, B. S., et al. 2010, ApJ, 720, 1073 Gould, A., & Loeb, A. 1992, ApJ, 396, 104 Gould, A., Shvartzvald, Y., Zhang, J., et al. 2023, AJ, 166, 145 Griest, K., & Hu, W. 1992, ApJ, 397, 362 Han, C., & Gould, A. 1995, ApJ, 449, 521 Han, C., Udalski, A., Jung, Y. K., et al. 2023b, A&A, 670, A172 Han, C., Zang, W., Jung, Y. K., et al. 2023a, A&A, 678, A101 Henry, T. J., Franz, O. G., Wasserman, L. H., et al. 1999, ApJ, 512, 864 Henry, T. J., & McCarthy, D. W., Jr. 1993, AJ, 106, 773 Ida, S., & Lin, D. N. C. 2004, ApJ, 604, 388 Kennedy, G. M., Kenyon, S. J., & Bromley, B. C. 2006, ApJ, 650, L139 Kim, S.-L., Lee, C.-U., Park, B.-G., et al. 2016, JKAS, 49, 37 Koshimoto, N., Baba, J., & Bennett, D. P. 2021, ApJ, 917, 78 Koshimoto, N., & Bennett, D. P. 2020, AJ, 160, 177 Kroupa, P. 2001, MNRAS, 322, 231 Lecavelier des Etangs, A., & Lissauer, J. J. 2022, NAR, 94, 101641 Lissauer, J. J. 1993, ARA&A, 31, 129 Lu, J. R., Gautam, A. K., Chu, D., Terry, S. K., & Do, T., 2021 Keck- DataReductionPipelines/KAI: v1.0.0, Zenodo, doi:10.5281/zenodo. 6677744 Mao, S., & Paczynski, B. 1991, ApJL, 374, L37 Minniti, D., Lucas, P. W., Emerson, J. P., et al. 2010, NewA, 15, 433 Mulders, G. D., Pascucci, I., & Apai, D. 2015, ApJ, 798, 112 Muzic, K., Schodel, R., Scholz, A., et al. 2017, MNRAS, 471, 3699 Nishiyama, S., Nagata, T., Kusakabe, N., et al. 2006, ApJ, 638, 839 Pascucci, I., Mulders, G. D., Gould, A., & Fernandes, R. 2018, ApJL, 856, L28 Quenouille, M. H. 1949, Ann. Math. Stat., 20, 355 Quenouille, M. H. 1956, Biometrika, 43, 353 Reid, I. N., Gizis, J. E., & Hawley, S. L. 2002, AJ, 124, 2721 Shvartzvald, Y., Yee, J. C., Calchi Novati, S., et al. 2017, ApJL, 840, L3 Shvartzvald, Y., Yee, J. C., Skowron, J., et al. 2019, AJ, 157, 106 Stanek, K. Z., Mateo, M., Udalski, A., et al. 1994, ApJL, 429, L73 Stetson, P. B. 1987, PASP, 99, 191 Surot, F., Valenti, E., Hidalgo, S. L., et al. 2019, AAP, 623, A168 Suzuki, D., Bennett, D. P., Sumi, T., et al. 2016, ApJ, 833, 145 Suzuki, D., Bennett, D. P., Udalski, A., et al. 2018, AJ, 155, 263 Terry, S. K., Barry, R. K., Bennett, D. P., et al. 2020, ApJ, 889, 126 Terry, S. K., Bennett, D. P., Bhattacharya, A., et al. 2022, AJ, 164, 217 Terry, S. K., Bhattacharya, A., Bennett, D. P., et al. 2021, AJ, 161, 54 Tukey, J. W. 1958, Ann. Math. Stat., 29, 581 Udalski, A. 2003, AcA, 53, 291 Udalski, A., Szymanski, M., Kaluzny, J., et al. 1994, AcA, 44, 227 Udalski, A., Ryu, Y. H., Sajadian, S., et al. 2018, ACTAA, 68, 1 Vandorou, A., Bennett, D. P., Beaulieu, J.-P., et al. 2020, AJ, 160, 121 Wizinowich, P., Lu, J., Cetre, S., et al. 2022, Proc SPIE, 12185, 121850Q Wizinowich, P. L., Le Mignant, D., Bouchez, A. H., et al. 2006, PASP, 118, 297 15 The Astronomical Journal, 169:319 (15pp), 2025 June Vandorou et al. https://orcid.org/0000-0001-8043-8413 https://orcid.org/0000-0001-8043-8413 https://orcid.org/0000-0001-8043-8413 https://orcid.org/0000-0001-8043-8413 https://orcid.org/0000-0003-2302-9562 https://orcid.org/0000-0003-2302-9562 https://orcid.org/0000-0003-2302-9562 https://orcid.org/0000-0003-2302-9562 https://orcid.org/0000-0002-5029-3257 https://orcid.org/0000-0002-5029-3257 https://orcid.org/0000-0002-5029-3257 https://orcid.org/0000-0002-5029-3257 https://orcid.org/0000-0001-5207-5619 https://orcid.org/0000-0001-5207-5619 https://orcid.org/0000-0001-5207-5619 https://orcid.org/0000-0001-5207-5619 https://orcid.org/0000-0003-0014-3354 https://orcid.org/0000-0003-0014-3354 https://orcid.org/0000-0003-0014-3354 https://orcid.org/0000-0003-0014-3354 https://orcid.org/0000-0003-0014-3354 https://orcid.org/0000-0001-5860-1157 https://orcid.org/0000-0001-5860-1157 https://orcid.org/0000-0001-5860-1157 https://orcid.org/0000-0001-5860-1157 https://orcid.org/0000-0001-5860-1157 https://orcid.org/0000-0002-8131-8891 https://orcid.org/0000-0002-8131-8891 https://orcid.org/0000-0002-8131-8891 https://orcid.org/0000-0002-8131-8891 https://orcid.org/0000-0003-0303-3855 https://orcid.org/0000-0003-0303-3855 https://orcid.org/0000-0003-0303-3855 https://orcid.org/0000-0003-0303-3855 https://orcid.org/0000-0001-6129-5699 https://orcid.org/0000-0001-6129-5699 https://orcid.org/0000-0001-6129-5699 https://orcid.org/0000-0001-6129-5699 https://orcid.org/0000-0002-7901-7213 https://orcid.org/0000-0002-7901-7213 https://orcid.org/0000-0002-7901-7213 https://orcid.org/0000-0002-7901-7213 https://orcid.org/0000-0002-7901-7213 https://orcid.org/0000-0003-2388-4534 https://orcid.org/0000-0003-2388-4534 https://orcid.org/0000-0003-2388-4534 https://orcid.org/0000-0003-2388-4534 https://orcid.org/0000-0002-1530-4870 https://orcid.org/0000-0002-1530-4870 https://orcid.org/0000-0002-1530-4870 https://orcid.org/0000-0002-1530-4870 https://orcid.org/0000-0002-1646-442X https://orcid.org/0000-0002-1646-442X https://orcid.org/0000-0002-1646-442X https://orcid.org/0000-0002-1646-442X https://doi.org/10.1088/0004-637X/780/1/54 https://ui.adsabs.harvard.edu/abs/2014ApJ...780...54B/abstract https://doi.org/10.3390/universe4040061 https://ui.adsabs.harvard.edu/abs/2018Univ....4...61B/abstract https://doi.org/10.3847/0004-637X/824/2/83 https://ui.adsabs.harvard.edu/abs/2016ApJ...824...83B/abstract https://doi.org/10.1038/nature04441 https://ui.adsabs.harvard.edu/abs/2006Natur.439..437B/abstract https://doi.org/10.1088/0004-637X/716/2/1408 https://ui.adsabs.harvard.edu/abs/2010ApJ...716.1408B/abstract https://doi.org/10.1088/0004-637X/785/2/155 https://ui.adsabs.harvard.edu/abs/2014ApJ...785..155B/abstract https://doi.org/10.1088/0004-637X/808/2/169 https://ui.adsabs.harvard.edu/abs/2015ApJ...808..169B/abstract https://doi.org/10.3847/1538-3881/ab6212 https://ui.adsabs.harvard.edu/abs/2020AJ....159...68B/abstract https://doi.org/10.3847/1538-3881/ad4880 https://ui.adsabs.harvard.edu/abs/2024AJ....168...15B/abstract https://doi.org/10.1086/589940 https://ui.adsabs.harvard.edu/abs/2008ApJ...684..663B/abstract https://doi.org/10.1086/178096 https://ui.adsabs.harvard.edu/abs/1996ApJ...472..660B/abstract https://doi.org/10.1088/0004-637X/713/2/837 https://ui.adsabs.harvard.edu/abs/2010ApJ...713..837B/abstract https://doi.org/10.3847/0004-6256/152/5/125 https://ui.adsabs.harvard.edu/abs/2016AJ....152..125B/abstract https://doi.org/10.3847/1538-3881/aad59c https://ui.adsabs.harvard.edu/abs/2018AJ....156..113B/abstract https://doi.org/10.3847/1538-3881/aaadfa https://ui.adsabs.harvard.edu/abs/2018AJ....155..141B/abstract http://www.ascl.net/1010.068 https://doi.org/10.1051/aas:1996164 https://ui.adsabs.harvard.edu/abs/1996A&AS..117..393B/abstract https://doi.org/10.1086/132281 https://ui.adsabs.harvard.edu/abs/1988PASP..100.1134B/abstract https://doi.org/10.3847/1538-3881/aaed46 https://ui.adsabs.harvard.edu/abs/2018AJ....156..289B/abstract https://doi.org/10.1038/s41586-021-03869-6 https://ui.adsabs.harvard.edu/abs/2021Natur.598..272B/abstract https://doi.org/10.1046/j.1365-8711.2001.04776.x https://ui.adsabs.harvard.edu/abs/2001MNRAS.327..868B/abstract https://doi.org/10.1093/mnras/stx1049 https://ui.adsabs.harvard.edu/abs/2017MNRAS.469.2434B/abstract https://doi.org/10.1088/0004-6256/147/3/47 https://ui.adsabs.harvard.edu/abs/2014AJ....147...47B/abstract https://doi.org/10.1086/506327 https://ui.adsabs.harvard.edu/abs/2006ApJS..166..585B/abstract https://doi.org/10.1088/0004-637X/814/2/92 https://ui.adsabs.harvard.edu/abs/2015ApJ...814...92C/abstract https://doi.org/10.3847/1538-3881/ab9b84 https://ui.adsabs.harvard.edu/abs/2020AJ....160...53C/abstract https://doi.org/10.1117/12.2629614 https://ui.adsabs.harvard.edu/abs/2022SPIE12185E..0VC/abstract https://ui.adsabs.harvard.edu/abs/2022SPIE12185E..0VC/abstract https://doi.org/10.3847/1538-4357/aaf861 https://ui.adsabs.harvard.edu/abs/2019ApJ...871..179C/abstract https://doi.org/10.1086/588487 https://ui.adsabs.harvard.edu/abs/2008PASP..120..531C/abstract https://doi.org/10.1093/mnras/staa2245 https://ui.adsabs.harvard.edu/abs/2020MNRAS.497.5309D/abstract https://ui.adsabs.harvard.edu/abs/2020MNRAS.497.5309D/abstract https://ui.adsabs.harvard.edu/abs/2000A&A...364..217D/abstract https://doi.org/10.48550/arXiv.astro-ph/9702039 https://ui.adsabs.harvard.edu/abs/1998A&A...329..361D/abstract https://doi.org/10.1086/321556 https://ui.adsabs.harvard.edu/abs/2001ApJ...556..181D/abstract https://doi.org/10.1086/422843 https://ui.adsabs.harvard.edu/abs/2004ApJS..154...10F/abstract https://doi.org/10.1088/0004-637X/779/2/91 https://ui.adsabs.harvard.edu/abs/2013ApJ...779...91F/abstract https://doi.org/10.1086/306256 https://ui.adsabs.harvard.edu/abs/1998ApJ...506..533G/abstract https://doi.org/10.1126/science.1151947 https://ui.adsabs.harvard.edu/abs/2008Sci...319..927G/abstract https://doi.org/10.1086/171443 https://ui.adsabs.harvard.edu/abs/1992ApJ...392..442G/abstract https://doi.org/10.1088/0004-637X/720/2/1073 https://ui.adsabs.harvard.edu/abs/2010ApJ...720.1073G/abstract https://doi.org/10.1086/171700 https://ui.adsabs.harvard.edu/abs/1992ApJ...396..104G/abstract https://doi.org/10.3847/1538-3881/aced3c https://ui.adsabs.harvard.edu/abs/2023AJ....166..145G/abstract https://doi.org/10.1086/171793 https://ui.adsabs.harvard.edu/abs/1992ApJ...397..362G/abstract https://doi.org/10.1086/176076 https://ui.adsabs.harvard.edu/abs/1995ApJ...449..521H/abstract https://doi.org/10.1051/0004-6361/202245525 https://ui.adsabs.harvard.edu/abs/2023A&A...670A.172H/abstract https://doi.org/10.1051/0004-6361/202347366 https://ui.adsabs.harvard.edu/abs/2023A&A...678A.101H/abstract https://doi.org/10.1086/306793 https://ui.adsabs.harvard.edu/abs/1999ApJ...512..864H/abstract https://doi.org/10.1086/116685 https://ui.adsabs.harvard.edu/abs/1993AJ....106..773H/abstract https://doi.org/10.1086/381724 https://ui.adsabs.harvard.edu/abs/2004ApJ...604..388I/abstract https://doi.org/10.1086/508882 https://ui.adsabs.harvard.edu/abs/2006ApJ...650L.139K/abstract https://doi.org/10.5303/JKAS.2016.49.1.37 https://ui.adsabs.harvard.edu/abs/2016JKAS...49...37K/abstract https://doi.org/10.3847/1538-4357/ac07a8 https://ui.adsabs.harvard.edu/abs/2021ApJ...917...78K/abstract https://doi.org/10.3847/1538-3881/abaf4e https://ui.adsabs.harvard.edu/abs/2020AJ....160..177K/abstract https://doi.org/10.1046/j.1365-8711.2001.04022.x https://ui.adsabs.harvard.edu/abs/2001MNRAS.322..231K/abstract https://doi.org/10.1016/j.newar.2022.101641 https://ui.adsabs.harvard.edu/abs/2022NewAR..9401641L/abstract https://doi.org/10.1146/annurev.aa.31.090193.001021 https://ui.adsabs.harvard.edu/abs/1993ARA&A..31..129L/abstract http://www.ascl.net/10.5281/zenodo.6677744 http://www.ascl.net/10.5281/zenodo.6677744 https://doi.org/10.1086/186066 https://ui.adsabs.harvard.edu/abs/1991ApJ...374L..37M/abstract https://doi.org/10.1016/j.newast.2009.12.002 https://ui.adsabs.harvard.edu/abs/2010NewA...15..433M/abstract https://doi.org/10.1088/0004-637X/798/2/112 https://ui.adsabs.harvard.edu/abs/2015ApJ...798..112M/abstract https://doi.org/10.1093/mnras/stx1906 https://ui.adsabs.harvard.edu/abs/2017MNRAS.471.3699M/abstract https://doi.org/10.1086/499038 https://ui.adsabs.harvard.edu/abs/2006ApJ...638..839N/abstract https://doi.org/10.3847/2041-8213/aab6ac https://ui.adsabs.harvard.edu/abs/2018ApJ...856L..28P/abstract https://ui.adsabs.harvard.edu/abs/2018ApJ...856L..28P/abstract https://doi.org/10.1214/aoms/1177729989 https://doi.org/10.1093/biomet/43.3-4.353 https://doi.org/10.1086/343777 https://ui.adsabs.harvard.edu/abs/2002AJ....124.2721R/abstract https://doi.org/10.3847/2041-8213/aa6d09 https://ui.adsabs.harvard.edu/abs/2017ApJ...840L...3S/abstract https://doi.org/10.3847/1538-3881/aafe12 https://ui.adsabs.harvard.edu/abs/2019AJ....157..106S/abstract https://doi.org/10.1086/187416 https://ui.adsabs.harvard.edu/abs/1994ApJ...429L..73S/abstract https://doi.org/10.1086/131977 https://ui.adsabs.harvard.edu/abs/1987PASP...99..191S/abstract https://doi.org/10.1051/0004-6361/201833550 https://ui.adsabs.harvard.edu/abs/2019A&A...623A.168S/abstract https://doi.org/10.3847/1538-4357/833/2/145 https://ui.adsabs.harvard.edu/abs/2016ApJ...833..145S/abstract https://doi.org/10.3847/1538-3881/aabd7a https://ui.adsabs.harvard.edu/abs/2018AJ....155..263S/abstract https://doi.org/10.3847/1538-4357/ab629b https://ui.adsabs.harvard.edu/abs/2020ApJ...889..126T/abstract https://doi.org/10.3847/1538-3881/ac9518 https://ui.adsabs.harvard.edu/abs/2022AJ....164..217T/abstract https://doi.org/10.3847/1538-3881/abcc60 https://ui.adsabs.harvard.edu/abs/2021AJ....161...54T/abstract https://doi.org/10.1214/aoms/1177706635 https://doi.org/10.48550/arXiv.astro-ph/0401123 https://ui.adsabs.harvard.edu/abs/2003AcA....53..291U/abstract https://doi.org/10.48550/arXiv.astro-ph/9408026 https://ui.adsabs.harvard.edu/abs/1994AcA....44..227U/abstract https://doi.org/10.32023/0001-5237/68.1.1 https://doi.org/10.3847/1538-3881/aba2d3 https://ui.adsabs.harvard.edu/abs/2020AJ....160..121V/abstract https://doi.org/10.1117/12.2628275 https://ui.adsabs.harvard.edu/abs/2022SPIE12185E..0QW/abstract https://doi.org/10.1086/499290 https://ui.adsabs.harvard.edu/abs/2006PASP..118..297W/abstract https://ui.adsabs.harvard.edu/abs/2006PASP..118..297W/abstract 1. Introduction 2. Microlensing Event OGLE-2016-BLG-1195 3. Keck Follow-up of OGLE-2016-BLG-1195 3.1. Target Identification 3.2. PSF Fitting on 2023 Data 3.2.1.2-star Scenario 3.2.2.3-star Scenario 3.2.3. Errors with MCMC + Jackknife 3.3. Additional Keck Epochs 4. Image-constrained Light-curve Models 4.1. Source Colors and CMD 5. Planetary System Parameters 5.1. Reanalysis of the Spitzer Data 6. Discussion and Conclusion AppendixMisidentified Target Star References