Monday, April 3, 2017

Understanding Written Science

Forbes and Nature recently published these two articles on the complexity of science writing:

These articles is related to one of my "deep thoughts" about science. Scientists have a need for precise communication to other scientists in order to accurately convey the details of our analysis. As a result, we have developed what are effectively entirely new languages full of technical jargon. Reading a biology paper is completely foreign to me. Even sub-fields within astronomy can read like different "dialects." For example, we can use different words between fields for the same mathematical analysis technique. This article is more about the quality of writing (sentence and word complexity), but the point is related. I'm glad to see that at least some people are attempting to understand this phenomenon and its implications. The need for science communicators (aka "translators") for the general public speaks to my fear of the (growing?) disconnect between scientific progress and the public perception of science. I fear that many of our failures - e.g. global warming, evolution - stem from a failure, or at least an underfunded effort, to communicate (translate) effectively. I'd even go so far as to change the typical formatting of a scientific papers in journals to require as a new standard the inclusion of the equivalent of an "executive summary" for the public - not just an abstract, but something free of jargon and comprehensible to a high school graduate. 

To summarize: The output of science shouldn't be accessible to only a select few who speak a specific "language" that requires years of specialized training to understand. Years of specialized training to undertake said research, sure. But not to understand the gist.

Monday, January 2, 2017

The Superbowl of Astronomy

Over 2400 students, professors, scientists and journalists will gather at the Gaylord Convention center in Grapevine, Texas this week to listen to over 1600 talks, explore exhibit booths, catch up with colleagues, and network.  We're here for the 229th meeting of the American Astronomical Society, referred by as the Superbowl of Astronomy.

I'm bringing four of my students to the conference, and my department colleague, Dr. Mike Reed, is also bringing three of his students.  Springfield, MO is close to the center of the country, so we decided to make the trip to Texas by car - a 7 hour road trip! (for reference, the east coast is a ~16 hr drive; west coast ~24 hr drive, and Chicago ~8 hrs).
I'm reminded of this scene from the PhD movie sequel:

 I've apparently reached the PI career stage

To entice you to attend our posters and talks, below is a list of what Dr. Reed, myself, and our research students will be presenting.  Four students (marked below with an *) are applying to PhD programs, so if you are on a grad school admissions committee be sure to talk to them!   Many more of my former students are also going to the AAS meeting, and I look forward to catching up with them, listening to their talks and seeing their posters.

First year masters student, Bryson Cale*, will be giving a talk on Saturday morning (11:10-11:20am, Texas D, Session 403. Extrasolar Planets Detection: Radial Velocity II).  Bryson will present the motivation and first light observations with an isotopic methane gas cell on the iSHELL spectrometer at the NASA Infrared Telescope facility.  Our goal is to measure relative radial velocities of stars to better than 3 m/s precision to search for habitable zone exoplanets around M dwarfs, and to confirm planets discovered by the TESS mission.
Bryson Cale

First year masters student Patrick Newman*, will be presenting a poster on Wednesday (146.10 Extrasolar Planets: Detection Poster Session).  Patrick is working on radial velocity survey yield simulations for input into a future flagship direct imaging mission like WFIRST, HabEx, and LUVOIR. WFIRST will be launching in the 2020s, and HabEx and LUVOIR are under study right now for the 2020 Astrophysics Decadal survey for launch possibly in the 2030s.

Patrick Newman

Senior undergraduate Ryan Hall* will be presenting a poster on Wednesday (146.14 Extrasolar Planets: Detection Poster Session).   Ryan is the current lead of MICRONERVA, an array of small eight-inch telescopes to work robotically and autonomously together to synthesize the light gathering power of a larger telescope.  Ryan will present on the current capabilities and development of the prototype array of four telescopes.  Ryan and I will also be presenting at a splinter meeting for the MINERVA team on Thursday.
Ryan Hall

Senior undergraduate Shannon Dulz* will be presenting a poster on Thursday (245.13 Extrasolar Planets: Characterization & Theory Poster Session) on measuring Transit Timing Variations of two Kepler planetary systems. Shannon developed her own custom code, using a custom transit shape model to speed up the transit fitting process in determining the time of transit midpoint.
Shannon Dulz presenting at a conference in April 2016

Junior undergraduate and computer science major Frank Giddens will be demoing (at Ryan Halls poster) his node.js web server interface to MICRONERVA, complete with Python bridges to the hardware. Watch with us on a webcam as we remotely control the telescope and accessories back in our lab 400 miles away, as long as we don't have any live-demo technical difficulties!
Frank Giddens

Undergraduates John Crooke and Ryan Roessler will be presenting a poster (433.17 Stars of Many Stripes Late Poster Session) on their work with Dr. Reed on the asteroseismology of the subdwarf B star PG 1219+534.

I (Dr. Peter Plavchan, Assistant Professor, Missouri State University) will be presenting a talk on Friday (2:40-2:50pm Texas D Session 320. Extrasolar Planets Detection: Radial Velocity I).  I will be presenting the discovery of a candidate Jovian exoplanet around AU Mic!  The paper is drafted and circulated to co-authors, and will be submitted soon!

My colleague Dr. Reed will be presenting a talk on Wednesday (3-3:10pm Texas 4 Session 130. Variable Stars, Asteroseismology) on his latest research on the asteroseismology of subdwarf B stars.

I'm ready for the kickoff for the astronomy Superbowl!

Sunday, October 30, 2016

MICRONERVA - my "moon shot"

I finally updated my research group web page, admittedly more than halfway though a busy semester!   On my web page you can meet the new students in my research group, along with new articles and podcast appearances, and a description of our new masters degree emphasis in astronomyNow it is time for a new blog post too!

This past summer MICRONERVA started to take up residence at Baker Observatory.  In this post I will highlight the work of Central Methodist University senior physics major Denise Weigand, and how we went from a drawing on a Powerpoint slide to the built frame for the MICRONERVA prototype enclosure.

Denise was funded as a Missouri SpaceGrant Consortium intern to come work with me at Missouri State University.  I spent most of my summer traveling on the east coast (two immediate family weddings, a couple conferences, etc.), so we had to work remotely.  For that, we used Google+ Hangouts to share screens, audio and video to communicate effectively from a distance.

First, a bit about MINERVA, the "parent" observatory concept behind MICRONERVA.  For a number of years I've been helping with the more complex, larger version called MINERVA.  It's designed to be completely robotic and autonomous after initial setup and regular maintenance:

The basic principle behind MINERVA is to synthesize a larger telescope aperture by combining the light from multiple, smaller telescopes at a lower cost.  The cost of a telescope has a steeper than linear dependence on the telescope primary mirror area:

The cost of single aperture telescopes (red and blue data points) compared to multiple aperture telescopes (green data points).  The solid lines in red and blue indicate approximate power law relationships.  The jump at a diameter of ~1.2 meters marks the transition from the amateur market to the lower demand market that is partially dictated by military expenditures.  From Figure 1 of Swift et al. 2015.

For seeing-limited telescopes there is no loss in angular resolution, which is true at visible wavelengths for any primary mirror larger than ~8-inches in diameter without adaptive optics.  Additionally, the spectrograph shrinks in volume as N3/2, where N is the number of telescopes, making it a lot cheaper and more stable too. In the case of MINERVA, four 0.7 m telescopes need a spectrograph 1/8th the volume of a spectrograph for a single 1.4 m telescope.  Thus, one saves costs on both the telescopes and the spectrograph.  And because the spectrograph is smaller, it makes it easier to measure precise radial velocities to hunt for exoplanets.  It's a win-win situation all around.  As far as I know, MINERVA is the first successful on-sky demonstration of this multi-telescope approach for precise spectroscopy.  One of the enabling technologies are fiber optics, which allow us to bring together the light from several telescopes to a single spectrograph entrance slit.

Anyway, back to MICRONERVA.

Upon my arrival at Missouri State University in the fall of 2014, I saw that we had a number of computer controlled CPC800 Celestron Telescopes lying around for educational use.  I also noted that with a simple tip-tilt adaptive optics, light could be effectively coupled from these eight-inch telescopes into special optical fibers called single mode fibers.  In other words, we could build a mini-version of MINERVA - aka MICRONERVA - if we could effectively turn these telescopes into computer controlled autonomous robots.  Additionally single mode fibers provide the ultimate limit in the stable illumination of spectrographs, critical for measuring precise radial velocities.

We decided to start with a prototype of array of four telescopes, but this concept is extremely scalable and even cheaper than MINERVA.  I envision a future array of many hundreds of these telescopes, similar to the HATPI concept by Gaspar Bakos for exoplanet transit searches, but this time for spectroscopy.  I find this is a potentially more cost effective approach for large aperture spectroscopy compared to other concepts such as the Mauna Kea Spectroscopic Explorer:

HATPI concept design

Fast forward to the spring of 2016.  My former student Claire Geneser, now a graduate student at Mississippi State University, managed to get the telescopes to point to a series of targets entirely under the control of a laptop computer:

We still had a number of issues to solve - could we guide at <1 arc-second precision?  Could we control all of these telescope autonomously from a web page?  I'll post about our progress on these challenges at a later date.  We also needed an enclosure to hold the telescopes, something that could open and close on its own.  That is the problem Denise chose to solve.

Denise joined my research group in mid-June of 2016. We had looked into buying a pre-built "dome" like the excellent domes from Astrohaven.  However, at costs of >$25,000, these were beyond the limits of our modest budget.  We were going to have to build one ourselves. Our prototype telescope array will be located at Baker Observatory, shown below:

Baker Observatory, north of Marshfield, MO, the birthplace of Edwin Hubble

In the photo above you can see two traditional telescope domes that are the usual landmark for an observatory.  However, these domes are far more complex than necessary to build an autonomous robotic facility.  Between the lower shutter, dome slit, and dome rotation, three separate motors are needed to open, close and point the dome.  A far simpler approach for autonomy is to design an enclosure that needs only one motor to open and close.  One motor is much easier to program and control than three, even if the end result isn't as exciting as looking at the domes above.

Below is the first sketch Denise and I made of what would eventually be our prototype enclosure in mid-June. This drawing was made with Powerpoint while we talked on Google+ hangouts more than 1000 miles apart:

The first MICRONERVA enclosure sketch.
The design involved the smallest number of exterior surfaces - five - and a single motor to make the design as simple as possible (one of the guiding principles in our design process).  It wasn't our first design, but it was the first design to make sense.  The four shapes inside the triangular room represent the envelops of the four telescopes - e.g., the space they will fill after going through all possible directions they point.  The red surfaces for the roof, the triangles the side walls, and the rectangles some legs to hold the structure off the ground, to prevent moisture wicking and contact with snow in the winter.   The whole thing would be oriented east-west, to enable the largest declination range viewable, at the expense of some east-west horizon obscuration by the triangular walls.

Denise set about her summer turning this sketch into reality.  First up was getting the dimensions right, and here were some sketches that passed back and forth between us.  We started with Powerpoint because it had an easier learning curve than AutoCAD, and let us quickly iterate on ideas.

At the end of June, it was time to graduate to three dimensions, which we used Google Sketchup to get started:

We also started thinking about how the roof would roll of and on.  We had already settled on this excellent motor drive from MVO Controls:

Denise made a parts list, and readied to go to Lowe's and Home depot to acquire the wood we needed.  We decided to start with wood rather than aluminum or metal, since it was more forgiving of us to make mistakes, and it was cheaper too.  We'd learn our mistakes on this prototype, which would make things easier the second time around if we went for more permanent and expensive materials.

We next enlisted the help of our campus mechanical engineer, Brian Grindstaff, to help with some of the engineering logistics - e.g. building the floor like a deck with joists and slatted cross-beams and such.  For the first three weeks in the July heat, Denise and Brian build the decking floor and roof frame in the loading dock outside our departments building.  Having it on campus made it easier to work on every day, but then we'd have to transport it out to Baker Observatory, a 40 minute drive!

Brian helped Denise put her design into AutoCAD, and pick the roofing material, wheels, wheel channel. and many other practical aspects of our design under Denise's guidance.

By August 8th, Denise and I used Google maps to plot a final location for the enclosure at Baker Observatory, next to my colleague Dr Mike Reed's robotic 16-inch telescopes, BORAT.  It was ready to ship! The night prior Denise and I went to Baker Observatory with wooden stakes, string and a level, and staked out the position of the enclosure and made sure it'd be on ground that was level enough to be fixed with a shovel and some sweat.

To get the enclosure to Baker Observatory, we made the enclosure relatively easy to disassemble into three pieces - the roof, the deck floor, and the extension for the rails when the roof is "open".  We transported it on one of my students 16'x8' trailers.

The morning of August 9th, we loaded the enclosure onto the trailer.  We left as early as possible to beat the Southwest Missouri summer heat and humidity.  I also recruited my entire research team to help with the lifting and moving.  We had gotten good at this teamwork, because back in June we built a fence at the observatory.  We used reclaimed wood from a long fence in my subdivision that had been torn down and replaced after a pickup truck crashed through it:

Missouri State physics majors Joe Huber and Ryan Hall

Fence building at Baker Observatory in June of 2016 to help with car headlights during our public viewing nights.  Featuring Missouri State CS majors Frank Giddens and America Nishimoto, physics major Laura Ketzer, and Dr. Mike Reed.

Unloading the enclosure pieces on August 9th took a lot of work, and we were all very dehydrated by midday:

Lifting the roof onto the platform was the hardest part.

Denise Weigand next to her roll-off roof enclosure design built and installed!  August 9th, 2016

Two days later we came back and added cross-beam support to the legs, and the roof panels:

We'd add the walls by the end of that week in August.  Denise returned to Central Methodist University for her senior year. Today, the enclosure is buttoned up, complete with tornado tie-downs to keep it from getting blown over:

It's not quite ready for the MICRONERVA telescopes yet. We're waiting on funding to add some interior environment control. The motor has been tested and rolls the roof on and off.  I'll update on the other aspects of MICRONERVA in a future blog post.

Saturday, June 25, 2016

Guest blog post - Weighted Trend Filtering Algorithm and Machine-Learning Template Selection for Time-Series Analysis

This months exoplanet research blog post comes from a former undergraduate and current colleague of mine, Giri Gopalan.  Giri and I have known each other since 2009 when he was a Summer Undergraduate Research Fellow at Caltech.  I had already identified a student to work with me that summer, but then Giri showed up at my door.  He was so convincing in his confidence, so exuberant in his interest, and so knowledgeable with applied mathematics.  I had to take him on as a second summer student, and I'm glad I did.  The Kepler telescope hadn't launched yet, and discovering transiting exoplanets with ground-based surveys were de rigueur.  The result of our summer working together, which we revisited to finish last year, was recently published in the Publications of the Astronomical Society of the Pacific.  Giri took a successful algorithm for filtering ground-based time-series data, and improved upon it with mixed results. Giri is on his way to a three year PhD program at the University of Iceland in the fall of 2016.


Photometric time series data have provided a fruitful resource for astronomers in recent times; the curation and analysis of photometric time series have allowed for the detection of transients, perhaps most notably exoplanets. A planet which revolves around a star with respect to our line of sight results in a characteristic dip or transit  in the star’s light curve and by detecting such dips, astronomers gain evidence for the existence of an exoplanet.

Unfortunately photometric time series are subject to noise of two major varieties: white and systematic noise. White noise is of the standard independent and identical structure from measurement to measurement, whereas systematics correlate heavily between different stars and time scales (e.g., instrumental measurement errors or seeing conditions that vary on a nightly basis). Such noise complicates the process of detecting transients,  (e.g. the characteristic depressions or transits from exoplanets), and so it is prudent to decontaminate light curves of this noise before analyzing them to look for transients. It turns out that in practice, systematics tend to dominate noise patterns in comparison to white noise. Our work concerns the implementation and application of a well known systematic trend filtering methodology by Bakos and Kovacs, the Trend Filtering Algorithm (TFA), and the investigation into ways to improve its performance.   

The methodology (in the non-reconstructive mode used for our analysis) essentially consists of two major components: the creation of “template light curves” which are meant to encapsulate typical systematic noise patterns and the filtering of systematic noise from a particular light curve given this set of templates. TFA assumes that systematic noise is a linear combination of these “basis” vectors and the residual of the projection of a light curve onto this “basis” is the filtered signal. Hence we wrote MATLAB code to perform this least squares projection in addition to weighted least squares (where the weights are the inverse of measurement uncertainties of the light curve to be filtered as noted in the thesis of A.Pal) leveraging matrix algebra. An illustration below demonstrates the idea:

For the template selection procedure we implemented and tested a version of hierarchical clustering, introduced by Kim et al. in 2009.

Our overall results were mixed; the analysis on real data (select PTF and 2MASS data) seemed to indicate that overfiltering occurs if a template set is not chosen carefully. On the other hand simulation studies indicate that the modifications we investigated did not improve exoplanet detection, but potentially variable detection (“potentially” because the variable detection numbers were very small for all combinations of factors tried in our simulation study).

Nonetheless, it was an extremely gratifying experience to tie up work begun during the summer of 2009 as a Caltech summer undergraduate research fellow. Peter’s guidance and willingness to see this work through was pivotal and has set an example of mentorship that I hope to fulfill [Editor's note: Thanks Giri!]. Moreover, I am very grateful for the feedback I received from the remainder of my collaborators. Looking forward I still think there is much that can be done with this work: the first being the translation of the code into other languages (e.g., R or Python) and a systematic noise removal methodology which leverages a fully probabilistic (e.g., Bayesian hierarchical) model.

Sunday, May 29, 2016

The Goldilocks Trap: Guest post from Andrew Vanderburg, PhD student at Harvard

This guest post is from Harvard astronomy graduate student Andrew Vanderburg.  Andrew is better known for his extensive work on the NASA K2 mission, including the discovery of a disintegrating minor planet orbiting a white dwarf!   Andrew and I collaborated on a paper that is published in the Monthly Notices of the Royal Astronomical Society, and is available in pre-print form here:  I remember when I had the genesis of the idea for this paper -  I was at the American Astronomical Society meeting in Washington DC in January of 2014; I excitedly rushed around the meeting to other exoplanet scientists bouncing the idea off of them.  Several of them would end up being co-authors on our paper.  No one had thought of this angle before. However, soon thereafter an excellent series of observational papers would come out from Paul Robertson demonstrating that we were on to something.  Read on for more from Andrew!

A major goal for astronomers studying exoplanets is to learn how common is it to find small, rocky planets orbiting in their stars’ habitable zones -- that is, orbiting just far enough away from their host star for the planet to have liquid surface water (and possibly alien life). One of the most promising ways to address this question is by closely monitoring the star’s radial velocity, or the speed at which stars are traveling towards or away from us here on Earth. We can detect the presence of planets around these stars because they tug on their host stars very slightly as they orbit. For the past 20 years, radial velocity measurements have been one of the most successful methods for discovering exoplanets.

Radial velocity measurements of the first exoplanet discovered around a sun-like star. The planet tugging on the star as it orbits causes a wobble in the radial velocity.

As our technology has improved and we learned to measure radial velocities (or RVs for short) more and more precisely, the situation has become a bit more complicated. It turns out that other processes on stars can cause apparent changes in its radial velocity, which can either mask (or more disturbingly) mimic the signatures of small exoplanets. One of the main culprits are starspots, which are the same as sunspots on our own Sun. When starspots rotate in and out of view as the star spins, they can cause RV variations which look a lot like the wobble of exoplanets that orbit their stars every time the star rotates.

Recently, there have been several high-profile cases where starspots have likely fooled astronomers into believing a planet orbited a star (see for example For astronomers trying to find small planets orbiting in their host stars’ habitable zones, this is a scary lesson! For one thing, two planets which we thought might be at the right temperature for liquid surface water to exist actually never existed. And for another, how will we ever be able to confidently say that a particular signal is a planet instead of starspots?

For these reasons, Peter and I decided to investigate the phenomenon of stellar activity (like starspots) mimicking and masking habitable exoplanets. We wanted to figure out under what circumstances could this type of planetary deception take place and how we might avoid it when searching for habitable exoplanets in the future.

We performed an analysis using a mixture of theoretical stellar models, observational data from NASA’s Kepler space telescope, and empirically determined relations describing the rotation of stars. First, we figured out the periods at which starspots can mimic exoplanets. We thought that these periods should be related the period of the star’s rotation (because spots will rotate in and out of view every time the star rotates), and our analysis confirmed that intuition. We were a bit surprised, however, to see how important the “harmonics” of the star’s rotation period are for stellar activity. It turns out that starspots can mimic planets orbiting at one half the star’s rotation period, or one third the star’s rotation period, just as easily as at the rotation period itself. We also found that even if there is a real planet orbiting near the star’s rotation period, it’s considerably more difficult than usual to detect it with radial velocity measurements.

Once we knew the importance of the star’s rotation period (and its harmonics), we used gyrochronology relations to predict the rotation periods for stars of different masses and ages. As stars age, they start to spin more and more slowly, and gyrochronology is the study of how that deceleration happens. Generally, more massive stars spin faster than lower mass stars. So cool, low-mass M-dwarfs have longer rotation periods, which means that any spurious planet detections would happen at similarly long periods.

Finally, we compared the periods where we expect spurious planet detections from our analysis to orbital periods of exoplanets in the habitable zone. The habitable zone also depends on the mass of the star -- for stars like the sun, the habitable zone is at periods around 1 year (like the Earth). But for smaller stars, the orbital periods of habitable zone exoplanets are shorter because these stars are intrinsically fainter, and to stay warm enough for liquid water to form, you need to be closer to the star.

It turns out that stars about half the mass of the sun have rotation periods close enough to the orbital period of habitable zone exoplanets, and therefore, we expect to see stellar rotation mimicking habitable-zone exoplanets around this type of star. This is the exact type of star where starspots have previously been found to mimic habitable zone exoplanets, and we could see more of the same until better ways have been found to separate out the signatures of starspots and genuine exoplanets. Until then, our take-away message is that if you’re looking for habitable-zone exoplanets around stars about half the size of the sun, buyer beware! You risk being fooled by stellar activity.

The time-scales of stellar activity from starspots that confuse radial velocity measurements.  For M dwarfs, habitable zone exoplanets have orbital periods (green stripe) that are in the same range as rotation periods (grey region).  We've nicknamed this overlap the "Goldilocks trap."  Similarly, the amplitudes of the apparent and real radial velocity changes are in the same ranges too.

Friday, May 13, 2016

Undergraduate Student Research Highlights

I'd like to take a time to acknowledge all the hard work my undergraduate students put into their research projects this semester.

Claire Geneser is graduating (today!) and lead my research group project on MICRONERVA.  She is on her way to earn her PhD at Mississippi State University in the fall.  Here she is giving a poster presentation at our college Undergraduate Research Day on April 21st:

Claire also decided to photobomb Ryan Hall standing with his poster at the Undergraduate Research Day.  Ryan is finishing his junior year and will be applying to graduate schools in the fall.  He taking over the MICRONERVA project from Claire.

Claire and Ryan would go on to give back-to-back talks the next day(!) at the NASA SpaceGrant Missouri meeting in Rolla, MO:

Two additional students joined my research group this year, and they also presented their posters at both the Undergrad Research Day and the Missouri SpaceGrant meeting.  Frank Giddens is a CS major working on an altitude-azimuth sensor made from an arduino board:

Joe Huber is working with me on separating subgiants and giants using photometry only for improving the TESS mission target selection:

Finally, I wanted to highlight two students working with Dr. Reed in our department, on Transit Timing Variations of Kepler planets (Shannon Dulz) and the asteroseismology of sub-dwarf B stars (Laura Ketzer), seen in the photos below.  Both Shannon and Laura will be seniors in the fall:

Tuesday, April 26, 2016

Guest post from University of Arizona postdoctoral scholar Dr. Huan Meng

Today's post comes from my collaborator and current University of Arizona postdoctoral scholar Dr Huan Meng.  Our work is featured in a NASA/JPL press release today that is getting coverage in the news.   

I (Peter Plavchan) planned and proposed these observations back in 2009 with the Spitzer Space Telescope.  The Spitzer Space Telescope team accepted our proposal, and made the observations in April 2010 simultaneously with a coordinated ground-based effort with four large telescopes in the Northern and Southern Hemispheres.   I remember the observing runs vividly, because I was using a telescope at Kitt Peak in Arizona along with Dr. Kevin Covey, and after collecting some data we were beset by a "major" snow fall atop the mountain:

The ground-based effort was monumental - one telescope in South America was a queue based telescope, but the other two were used by a colleague in Mexico, and colleagues who had traveled to a second telescope in South America just to get this data.  We were fortunate that on one night when one telescope went down we had another telescope covering the gap in data, and we all communicated online between the various telescopes to check in and see how things were going.

The project generated so much data for the Spitzer Space Telescope that it filled up the storage space onboard the spacecraft.  The Spitzer Space Telescope scheduling team, for which I am forever grateful, had to space out the campaigns every-other-night instead of every night.  On the days inbetween our observations, the Deep Space Network had to communicate with the Spitzer Space Telescope and download our data to make room for the next night of observations.

In 2013, Huan Meng joined me from the University of Arizona for a six month visiting graduate student fellowship at the NASA Exoplanet Science Institute where we finished analyzing the data.  After a pause to work on other projects, in 2015 Huan wrote it up and submitted it for publication.  While the publication of our paper this year ended one 7 year long journey in science, it opened a whole new method for studying the inner accretion environments of young stars!

Read on for Huan's story:

Most, if not all, young stars are born with disks of gas (mostly hydrogen and helium) and dust (small solid smoke-like particles mostly made of silicates and carbon). In astronomy, this disk is called a "proto-planetary disk" because at later stages of evolution, material in the disk will aggregate, accrete, and give rise to a planetary system like the Solar System.

However, the architecture of our own Solar System may not be representative of all planetary systems. The more we learned about exoplanetary systems, the more diverse we found they are. For example, a well-known population of planets that is absent in the Solar Sytems is "hot Jupiters." The Solar System has two gas giant planets, Jupiter and Saturn, both fairly far from the Sun. Jupiter's orbit is 5.2 Astronomical Units in radius (1 AU is the average distance between the Sun and the Earth), and Saturn's is 9.5 AU. By contrast, Jupiter's hot cousins in many exoplanetary systems are merely a fraction of an AU from their central stars and are very hot in temperature. Did these "hot Jupiters" form in-situ close to the stars? Or did they form further out and migrate inward ever since?  And did that happen during or after the protoplanetary disk existed?  Because the structure and evolution of the protoplanetary disk sets the initial conditions for planet formation, a crucial piece of evidence in this debate is whether there is planet-building material so close to the star.

It is long known from theories and spectroscopic observations that a protoplanetary disk cannot reach the photosphere, or "surface," of its star — The disk always has a hole in the center. Two major mechanisms can make the hole. On the one hand, gas in the inner region of protoplanetary disk is ionized and interacts with the stellar magnetosphere. If getting too close, it will be diverted off the disk plane along the magnetic field lines and accreted onto the star near the stellar magnetic poles. This is also how young stars accrete mass. An important reference for this so-called "magnetospheric truncation" of a disk is the co-rotation radius, at which the orbital period of circumstellar material matches the rotation period of the star. For mass accretion onto the star to proceed, the disk inner edge has to stay interior to the co-rotation radius and the magnetospheric truncation distance also has to be smaller. On the other hand, if solid dust particles are placed inside of a distance from the star at which their temperatures would surpass the sublimation limit of the material, the solid particles will get too hot and will be vaporized. This mechanism can also truncate the disk at the "sublimation radius," which is typically outside of the co-rotation radius. These different disk truncation radii provide a diagnostic: by measuring where protoplanetary disks get truncated and comparing with the theoretical expectations, we may tell which mechanism is at work in which disk.

So, we should just measure the sizes of the inner disk holes, right?  Unfortunately, this is not as simple as laying a ruler on top of a photograph. The sizes of the inner disk holes are expected to be small. For a solar-mass young star (called a "T Tauri" star) in the nearest star-forming regions, the expected disk truncation radii are hundreds of millions of times smaller than their distances to us. They are too small to be directly resolved with the current astronomical technology. Over the past decade, the only technique that can systematically explore the inner regions of protoplanetary disks is near-infrared interferometry, for which an array of designated telescopes observes the same object, combining signals to reconstruct a partial image of the object with higher spatial resolution. To obtain a measurement from the interferometric data, people have to introduce some assumptions about the disk geometry that are not necessarily justified. Such interferometric measurements have suggested that some protoplanetary disks around the most massive young stars, called "Herbig Be" stars, are truncated by magnetospheric accretion; disks around intermediate-mass "Herbig Ae" stars can be well described by a directly heated, "puffed-up" inner rim truncated at distances that corresponds to temperatures between 1500 and 2000 Kelvin (2200 to 3100 degree Fahrenheit), a typical range of silicate dust grain sublimation temperatures. However, the trend does not extrapolate down to the regime of solar-mass T Tauri stars, many of which appear to have larger-than-expected inner disk cavities. Oversized inner disk holes around T Tauri stars have raised questions about the roles of unrecognized physical processes in addition to the two major mechanisms considered above.  And there are also possible problems with the model assumptions upon which the interferometric measurements are made.

Our work is a different and novel approach to this issue. Since T Tauri stars are known to be variable stars, we can simultaneously monitor the changing stellar emission at a shorter wavelength and the disk response at a longer wavelength in the infrared. Given the constant and limited speed of light, it takes time for the variable stellar emission to travel to the disk and trigger a response, just like "light echoes." Therefore, the long-wavelength disk light curve (echoes) should lag behind the short-wavelength stellar counterpart (direct light) by the amount of the additional light-travel time1. If such a time lag is detected, we can compute the corresponding light-travel distance between the central star and its inner disk rim. Compared with interferometry, such measurements are relatively independent of model assumptions and should be more robust.

The basic idea of the method, called "reverberation mapping", has been used in extragalactic astronomy for over 20 years to measure the distance between supermassive blackholes in active galactic nuclei and their surrounding moleculuar clouds ("broad line regions"). To carry out the experiment for the first time around stars, we pre-selected a field in the rho Ophiuchi cloud complex, one of the nearest star-forming regions to the Solar System, and coordinated four ground-based telescopes to observe the area simultaneously with NASA's Spitzer Space Telescope on three nights. The ground-based telescopes, in Arizona, Chile, and Mexico, were used to monitor the stars in the near-infrared H and K wavebands (1.6 and 2.2 micron), while Spitzer worked at 4.5 micron wavelength to keep an eye on the disks. To validate any time comparison, we had to first correct the light-travel time on our receiving end, especially between the Spitzer Space Telescope and the Earth. As a result, 27 young stars were observed in the common field of view.  One of the T Tauri stars, called YLW 16B, was found to vary significantly in brightness rapidly and  have a time lag. This is the first detection of light echoes on the stellar scale!

Detailed analysis of the data revealed that the variable signals of YLW 16B in H and K bands were synchronized all the time, consistent with both being from the accreting gas right above the stellar photosphere. The signals at 4.5 micron lagged behind both H and K by 74.5 +/- 3.2 seconds. Interestingly, YLW 16B is a known edge-on system because of its mid-infrared molecular spectrum. Taking into account the viewing geometry, our reverberation measurement of the radius of its inner disk hole was 0.084 +/- 0.004 AU. Considering the simplifications we had used to convert the time lag to a single radius, we estimated that the total error is likely larger than the nominal one by a factor of several, on the order of 0.01 AU.

We can place our measurement in the context of previous interferometric results.  See our data point in red in the figure above compared to previous work. The reverberation inner radius of YLW 16B, a solar-mass T Tauri star, is consistent with a "puffed-up" inner disk rim, in the presence of backwarming, truncated at 1500 Kelvin, a typical dust sublimation temperature. This is in line with the interferometric disk sizes measured around intermediate-mass Herbig Ae stars. But unlike the old interferometric measurements of most other T Tauri stars, YLW 16B does not have an oversized inner disk hole and does not require any additional mechanism. For the planet formation question asked earlier in the article, now we have settled one more piece of the puzzle.

Footnote 1. It may take time for the disk to respond to changing starlight. Reflection is instantaneous once allowing for light-travel time; thermal response time of isolated and exposed dust particles should be well under a second given their tiny thermal capacities, much faster than the time resolution we can achieve in real observations; radiative transfer becomes important in the bulk disk and should take days to take effect, but it has no influence on the short timescale of our interest. Hence, the disk response time can be safely neglected for our purpose.