November 10, 2019 0

John Bochanski: “Data-Driven Discovery: Astronomy in the Era of Large Surveys” | Talks at Google


JUSTIN: I wanted
to introduce John. We have a great astronomy
focused talk today. John Bochanski, a professor
of physics at Rider University is joining us. He’ll discuss how large digital
surveys of the night sky have revolutionized
how astronomy is done. He will also explore
the motivation behind large surveys,
detail some of his results, and discuss the
data rich feature of large astronomical surveys. John most recently discovered
the two most distant stars in our galaxy, over one
million light-years away. So in addition to
being a professor, John is a blogger at
“Sky and Telescope”, a prominent magazine
for astronomers, and expecting his first
child in March, as well. So congratulations. He’s also a very
close friend of mine and an all-around amazing guy. So just please give a
warm welcome to Professor John Bochanski. JOHN BOCHANKSI: Thank you. [APPLAUSE] Thank you, Justin. And thanks to all of
you for coming out to listen a little bit about
some of the surveys that are really shaping the
future of astronomy. And for those of you who
have been in the room for a few minutes,
you’ve seen the video that’s playing in the
background for my title slide. That was put together by folks
at Johns Hopkins University and the Adler
Planetarium in Chicago. It uses data on galaxies. This is a map of the
universe at the scale of about a billion light-years–
one of the more detailed maps of the universe that
has ever been produced. And that’s really what I’m
going to focus on a lot today, is one of the essential
jobs of astronomers, for almost as long as
astronomers have been around, they’ve been trying
to make maps. So they’ve looked at
images like this one, seen the Milky Way galaxy
splayed across the sky, and tried to map out the
stars and the dust that are shown in this
image, and tried to make sense of the
physical structure that makes up our galaxy. And that’s what I’m
going to focus on today. Most of my research is focused
on mapping our own Milky Way galaxy. And so we’re going to
be talking about scales of about a million
light-years or so. One of the first astronomically
correct maps of the Milky Way was produced by Thomas Wright. This was in the mid-1700s. He was one of the
first to really begin to suss out the details in
terms of the physical structure of our galaxy. It was no longer
just a uniform sphere that was perfect in every
dimension and direction. Instead, he was
able to figure out that it was a flattened system. And William Herschel,
who actually usually gets the credit for first coming up
with a map of the Milky Way, used one of the largest
telescopes at the time to conduct a survey
of the night sky and inferred this distribution,
with the sun somewhere in the center here, and
found that the Milky Way was a flattened
disk structure. And this kind of work has
gone on from the 1700s, when the first modern
maps of the Milky Way were really produced,
to the present day. And so I want to tell you
the story behind my own work and how I produced one of
these maps along the way. My story starts not far from
where this image was taken. So this is Ocean
City, New Jersey. I grew up about an
hour away from where this image was taken. And that’s kind of what
the night sky looked like. You could see some bright
stars, but you really couldn’t see a
whole lot of detail. And it wasn’t until I traveled
to this place, Kitt Peak National Observatory, just
outside Tucson, Arizona, as part of a summer internship,
and used this telescope right here. We were in the middle
of a long observation. And so I was able to go
outside, take a long look at the night sky,
and saw an image that looks a lot like this. And the other thing I
want to point out, too, the telescope in
the upper right, we’ll be naming that later. Is there anybody
in the room that thinks they know what
telescope that is? AUDIENCE: Dark Energy Survey? No, unfortunately. So that, I would
argue, is probably one of the most
important telescopes that has ever been around, but
no one knows its name. But you will by the
end of the talk. AUDIENCE: Sloan
Digital Sky Survey? JOHN BOCHANKSI: Oh, yes. Thank you. So I looked up at the night
sky, saw the Milky Way for the first time, and
was immediately enthralled. I wanted to know how
the galaxy came to be, how it’s evolved over the last
tens of billions of years, and really why it was
shaped the way that it was. And so fast forward
10 years later, and I was taking data
that actually mapped out and produced one of the more
precise maps of our Milky Way that’s been produced to date. And so I know many
of you in the room probably are not up on
your astronomical jargon. So I’m going to try and keep
it to an absolute minimum. But there’s a few things
that you need to know. So the first thing
is that the parsec is the basic unit of distance
that most astronomers use. And the reason that
it’s named a parsec is that it is the
angular shift that you see for something that
is a distance away. So an object that
is one parsec away will have a parallactic angle
shift of one arc second. The closest star to us is almost
right around a parsec away. And so the way
this angular shift works– everyone in this
room is familiar with it– if you hold your arm out,
or your finger out at arm’s length, and blink your
eyes back and forth, you’ll see your finger move
across distant background objects. You bring your finger
closer to your nose, and that shift becomes
greater and greater. The same thing
happens with stars. So a nearby star, shown in
blue here, if one eyeball is in July– you’d take an image
of that field of stars in July, and that’s like blinking
your eye back and forth– you take the same image again
six months later in January, and that nearby star
will appear to shift its position with respect
to more distant stars. And so that shift is known
as the parallactic angle. It is a key distance indicator
for all of astronomy. The reason that this is so
important and so fundamental– and I’ll come back
to this a few times– is that we can
measure the distance to nearby stars
with no assumptions. This is a purely
geometric measurement. It’s very hard to do. So an arc second is an
extremely small angle. Most stars have shifts that
are nearly 100 times smaller than that. And so the limit that
we can get out to is something like 50 to
100 parsecs, currently. And so that means that
there’s only a few hundred, maybe about 1,000 stars, where
we can measure their distances directly using this technique. But they’re very, very valuable. They’re what’s known
as a fundamental wrong on the distance ladder. Because if we can get
the distance directly, then we can calculate
something known as the absolute magnitude. So this is the only equation
that will appear in the talk. If you have the apparent
magnitude of a star, which is just what you would see
if you just measured it. So it’s saying that star is
bright, that star is faint. If you have the
distance here, then you can calculate the absolute
magnitude of the star. And why that’s important,
the absolute magnitude is a tracer of the star’s
fundamental properties. So if you know the
absolute magnitude, you know something about
the size of the star, you know something
about its mass, potentially about
its composition. And so it becomes
very, very valuable to use these nearby stars
to calibrate relations between absolute magnitude
and other observables. And for those of you who are
up on your logarithm scales, the absolute
magnitude is formally defined as what the star’s
apparent magnitude would be at a distance of 10 parsecs. You plug in 10 here,
this side is zero. And so it places all stars on
the same fundamental scale. And so if you don’t
know the distance, if the star is beyond
our current capabilities of measuring a
parallactic angle shift, then we need to use
the star’s color to infer its absolute magnitude. And the way this is
done is with images like the one that’s
being shown here. This is an image of
a cluster of stars and it essentially– I’ll
step through and show you how astronomers sort
this kind of image based on color and brightness,
or their intrinsic brightness, their absolute magnitude. So this image is actually
a little bit of a fake. So the images that we
get from the telescope don’t actually appear in color. We use filters. And so the combination
of different filters give you the nice color
image that you see here. And this was taken with
the Hubble Space Telescope. So astronomers will
use a blue filter and recover only the blue
light from a particular star, a green filter that looks
at only the green part of the electromagnetic
spectrum, and a red filter to produce the full color
image that you see here. And so you can see that,
depending on whether or not a star is intrinsically
red, like the one in the upper right hand part of
the diagram that contains quite a lot of red light, or a
blue star like these two here on the bottom– let’s
skip back just a little bit– those intrinsically are
producing more blue light then the red stars that we see. So the color of stars
actually gives us a lot information
about the stars. And my wife may actually
disagree about this, but astronomers really like
to be orderly and organized. And so they always try
to sort things out. And so we like to put red
stars on the right, blue stars on the left. So we’ll do that. And we also like to characterize
things by their brightness. And typically, the
intrinsically bright stars are placed on the
top of diagrams, and the intrinsically faint
stars are placed on the bottom. And so what this
diagram really is, it’s known as a color
magnitude diagram. So on the x-axis,
we talk about color. We put blue stars on the
left, red stars on the right. And the magnitude, short
for absolute magnitude, we take stars that are
intrinsically bright, put them on the top. Intrinsically faint
stars are on the bottom. Most of the stars
fall along what’s known as the Main Sequence. Our sun is on the Main Sequence. Our sun would be kind of towards
the top of the sequence here. The majority of stars are what’s
known as Main Sequence Red Dwarfs. They would fall in this region. And then there’s pockets of
smaller, more intrinsic– or, excuse me, larger, more
intrinsically bright stars. Those are the stars
that you would see at the top of the diagram. They’re known as red giants. And on the left, we have
blue horizontal branch stars. And this sequence really
led to the development of stellar evolution
theory in general. Stars don’t evenly
populate this phase space. They fall along sequences. And so understanding
those sequences became an important part,
or very important subject, in the early 1900s. So this is another version
of a color magnitude diagram. I want to show this just
to play some numbers on the kind of stars that
we’ll be dealing with. And my own work
focuses on red stars. So we’re talking
about things that are anywhere from about
4,000 Kelvin or cooler. Our own sun is
about 6,000 Kelvin. This is the sun here. And on the y-axis, we
have absolute magnitude and luminosity. And so the stars that
are on the bottom part of the main sequence,
the cool red dwarfs, are putting out something
like 1/100th to 1/1000th of the light that
the sun produces. And so they’re extremely
difficult to see, even though they make
up 70% of all stars. So you need a telescope
to see these things. There’s one red dwarf
that you can actually see with your naked eye, but
you can’t see it from this area. You have to go way
into the desert and have extremely
dark sky conditions. The flip side, on the bright
upper part of the diagram, we’re talking about giant stars. So they produce anywhere from
10,000 to 1 million times more light than the sun. And so what this means
for us is that if we want to map the nearby galaxy,
we’re going to take the most numerous tracers, the red dwarfs
that make up 70% of all stars. There’s hundreds of billions
of these in our own galaxy. And if we can determine accurate
distances for all these stars, then we have a very nice measure
of the density distribution of the galaxy. And using the same
survey data, we can look for the giants
that are producing millions of times more
light than our own sun. They can be seen at much,
much larger distances. And so they give
us the best shot at finding stars at very,
very large distances. So I’m going to walk you
through a little bit of how we use these two types of stars
to map out the nearby parts of the galaxy and the really
distant parts, as well. So a little more Astronomy
101, the Milky Way galaxy is a spiral galaxy. It’s a flattened disk system. We are about eight
kiloparsecs or so from the center of our galaxy. And we’re essentially in
the disk of the galaxy. So that’s why as you look
out in a dark night sky you see that band of stars. That’s because
we’re in that disk. The Galactic Center
itself is composed of older, redder stars. It appears yellow to the eye. There’s also a lot
of gas and dust that’s settled into the
disk of the Milky Way. As gas settles
down and cools, it begins to spark
new star formation. And so one of the best examples
of this is the Orion Nebula. It’s a site of recent
star formation, and you can go out
tonight and see it. And so using survey data, we
can get a slightly nicer model of the Milky Way. So the sun would be somewhere
around 8,000 to 8,500 parsecs away from the center
of our galaxy. And it’s composed
of two disks known as the thin and thick disk. They have different
structural properties. But essentially,
they make up the bulk of the stars that are along
the disk of our galaxy. This is not the
Milky Way, it’s just an image of a galaxy that
looks similar to the Milky Way. And as you get either
below or above the disk, you see fewer and fewer stars. But mapping out
exactly that fall off, either as you get away
from the Galactic Center, or as you go above
and below, tells you a lot about the composition and
the structure of our galaxy. And there’s a sparse
halo of stars. So you’re talking
about maybe 1% or so of the total number of
stars in our galaxy. But as you get further
and further away, it’s the remnants of
Galactic formation. So our best picture
of Galactic formation is that smaller protogalaxies
were pulled together by mutual gravitation,
produced the Milky Way, and the leftover is what is
in a quasi-spherical shroud around our galaxy. And that’s actually one
of the more exciting areas of active research today. But we can zoom in. And so for those of you
that have very good vision, and because we have fabulous
projectors here at Google, you may see a
tiny, tiny red dot. So that tiny red
dot is essentially the limit for the
nearby stars that we can measure good parallaxes to. It’s known as the
solar neighborhood. It’s the stars that we used
to calibrate all our distance relations, and it’s great for
studying very faint objects. Because if it’s a
faint object, you’re not going to see it
at large distances. The downside to studying objects
in the solar neighborhood is that it’s not actually
good for telling you a lot about the structure
of the Milky Way. We’re only probing a very,
very, very tiny volume. And so it’s difficult
to really recover a lot of information
about our galaxy. Fortunately, there are surveys. And I’ll tell you about the
Sloan Digital Sky Survey today, as well as two
that are on the horizon. They probe somewhere around
a kiloparsec or so out into our Milky Way. And so, to date,
they’ve really given us the best picture that we have
of the structure of our galaxy at relatively large distances. The surveys that are
planned for the future are going to blow
Sloan out of the water. So we have surveys
that are about to start in the next decade
that will produce precise maps of our Milky Way
out to distances of around 10 kiloparsecs or so. The survey that I want to spend
a little bit of time on today is known as the Sloan
Digital Sky Survey. It started in 1998, actually the
same year that Google started. It’s housed in the Apache Point
Observatory in New Mexico. And each clear night,
this entire structure would slide down the rail
and unveil this telescope. So this is arguably one of the
most important astronomical telescopes that’s
ever been constructed. You can see two
people here to scale. The Sloan Digital
Sky Survey worked in a scanning mode
that made it very, very efficient to image large
parts of the sky continuously. And so essentially every
clear night since 1998, it’s been imaging
the night sky using a relatively wide field of view. At the time of its construction,
the SDSS, the Sloan Digital Sky Survey, was home to the largest
digital camera ever made. It was about 120
megapixels and covered about 1 and 1/2 square degrees. So it’s about eight times
the area of the full moon. So it was a really
large field of view. And so it made it
great at covering wide areas of the sky
each and every night. And so you can see a person
behind the camera there, too. That’s Jim Gunn, a
professor from Princeton. And Jim was really the
founder of the entire survey. So you can see, you
may be able to pick out some of the color differences
in the cameras there. And there were
multiple CCDs that made up the entire camera array. Well, those different
colors correspond to different filters. So the red, green, and blue that
I showed earlier in the talk is now split up into
five different filters, and that covers the
visual spectrum. So those five filters image
the night sky at the same time, and it gave us a large amount
of information about the stars and galaxies that it
was imaging at the time. Sloan also recovered
something known as a spectrum. So you can think of a
spectrum as a detailed look into the light output
of an individual object. And so what I have
here are two spectra. One, shown in red,
corresponds to a red star. The other one shown in blue
corresponds to a blue star. Sloan not only imaged
the night sky– so it’s covered now over half
of the entire night sky– but it also took spectra
of millions of objects. Why this is important, this
spectrum, essentially you’re taking the light, spreading
it out through a prism, and looking at the
detailed abundances of what makes up these
stars and galaxies. So this contains a
lot more information, but it’s painstakingly tough
to make these observations. Sloan was actually pretty
efficient at doing this. And so the way they operated was
that they would image the night sky and then take
algorithms to target the most interesting
objects, the most interesting stars and galaxies. And at the University
of Washington, and the machine shop
down in the basement, plates would be
drilled that covered the entire field of view. And each of those
holes correspond to either an interesting
star, or a galaxy, or a distant quasar that
Sloan wanted to use to try and map out the distribution
of galaxies in the universe, or the distribution of
stars in the Milky Way. And it was able to go from
hundreds of millions of objects down to millions of
very interesting objects that we could get detailed
information about. And so the way this was done
was through fiber optics. There are 640 plugs, or
640 holes, on each plate. All of those needed
to be plugged by hand. And so there are some very
dedicated telescope operators down in New Mexico
that now all have carpal tunnel because
they’ve done this for a very long
time at this point. There’s millions of
spectra that were recovered during the course of the survey. But really, what
it’s led to, Sloan is arguably the most
efficient discovery engine that astronomy
has ever produced. The latest data release,
which actually just came out last week,
catalogs nearly 470 million different photometric
objects that have been imaged by the
Sloan Digital Sky Survey. 60 million of those, and
probably more at this point, are these cool red dwarfs. So this is really where I
try and focus my research on, and try to use these
cool red dwarfs to map out the structure of the galaxy. And just over 4 million
spectra– again, all those holes
have been plugged by hand– have been
obtained mostly on galaxies. But there are at
least 70,000 stars that we’ve identified
using the spectra. And probably, the
most exciting thing is that, since 1998, over 6,000
papers and 250,000 citations have been produced from this
small telescope in New Mexico. So that’s really the hallmark
of the legacy of Sloan. And in the public
domain, Ann Finkbeiner wrote a very nice book titled
“A Grand and Bold Thing”, detailing the history
and the politics that went into getting
Sloan up and running. And in 2008, Jim Gunn received
the Presidential Medal of Science. And so it is really a
testament to his work. And so now I want to
give you a little flavor. I want to show you some of
the data, some raw images from the Sloan survey. And you’re going to
see a few maps here. I just want to orient
your field of view. Astronomers love
Aitoff projection maps. So this is an Aitoff
projection of the earth. This is what the galaxy looks
like in Aitoff projection. So you can see the plane
of the Milky Way here, the large and small
magellanic clouds, and you can see that the
Galactic Center really is composed of yellow,
older stars as well as a lot of dust that’s blocking
the light from the stars behind it. The Sloan sky coverage
looks something like this. This map should be updated, but
I haven’t gotten around to it with the latest data release. Most of the data is contained
in the northern hemisphere of our galaxy. There are three stripes in
the southern hemisphere. And actually, a lot
of this area has been filled in in the
last five years or so. But it’s imaged roughly about
half of the observable sky from New Mexico. And so to give you a sense
of just how many stars, you may be able to– if you
squint your eyes a little bit, you can convince
yourself that, as you go from the pole of the Milky
Way down towards the equator, the spatial density
of stars increases. There’s more stars
at lower latitudes. So that’s the broad
brush structure that we’re trying to measure. But we’re going to do that in
a much more detailed fashion. And fortunately, there’s
quite a large number of stars available to us when we use
the Sloan Digital Sky Survey. So this is a movie
that I put together of one square degree of the sky. So the entire sky is 40,000
or so square degrees. This is a very tiny
footprint on the sky. And each successive
frame is highlighting the cool red dwarfs that are
in that small square degree. And so we use the colors
of all these stars to estimate their absolute
magnitudes, and then their distances. And each and every
one of these stars becomes a valuable
probe for measuring the density and the
structure of our Milky Way. And so we can take
that two degree image, associate a distance
with all those stars, and then put it into a
model of our Milky Way, and it looks
something like this. So this is a visualization put
together by some collaborators at the American Museum
of Natural History. This is the center
of the Galaxy, this is the sun’s location. And what you can see is
that most of the stars are above the plane
of the Milky Way, and that as you get further
away from the plane, there’s fewer stars. We’re going to rotate through
the center of the Milky Way now, and you’ll see
those three stripes that I pointed out a few
slides ago come into focus. So the first stripe is coming
into focus right around now. There’s the second,
and then the third. And the exact distribution
of these stars, how they fall away– how
the density of these stars falls as you get above
or below the plane, and how the density
falls as you get further away from the Galactic
Center, tells us something about the
structure of the Milky Way. So if we take one
slice of these stars, we can paste it onto a
cartoon of the Milky Way. What we’re really
measuring is the structure of the density of stars
as you get further away from the Galactic Center,
and as you get further away from the Galactic Plane. So the sun would be here. And we see that there’s
a little bit of a fall as you go from here to here. And there’s a fall as
you get above the plane. So this is one of the most
precise maps of the Milky Way that’s ever been produced
for the nearby Milky Way. So within about 1
and 1/2 kiloparsecs, we have a great idea as to
exactly how our galaxy appears. What I’m working
on now is actually a much different scale. So we’re also using
the same survey data, but now trying to pull out the
most intrinsically bright stars to go after the shape of
the halo of our Milky Way. So we’re looking at scales
of 100 kiloparsecs or so. Prior to the work
that we’ve been doing, the most distant star was only
about 120 kiloparsecs away. We’ve identified two that are
over 200 kiloparsecs and likely some that are probably
closer to 300. So we’ve been using
telescopes all over the world. Some of these, the top row, are
mostly telescopes in Hawaii. The middle row are telescopes
in the continental US, and the lower two are
telescopes in Chile. There’s a wide team
of collaborators that we’ve called
upon to try and get detailed spectral observations
to tell the difference. It’s actually
relatively difficult to use images alone to tell the
difference between a giant star and a dwarf star. And so we rely on
spectra to nail down the difference between
these two stars. What’s exciting
about this project is that these stars are really
at unexplored distances. So using a combination
of two different surveys, we’ve been able to identify
about 500 stars that are likely anywhere from 200 to potentially
600 kiloparsecs away. Now, the three
different histograms correspond to
different assumptions about the exact chemical
composition of the star, and that actually tweaks the
color, absolute magnitude relation. But the punchline is
that the Milky Way would appear much,
much different as viewed from the stars. So rather than seeing a
wide band across the sky, the Milky Way
would appear almost like just another small
galaxy in your night sky. So we published this
work during the summer, and it made a little
bit of a splash. So “Gizmodo” and
“Huffington Post”– which put me next to Britney
Spears for some reason– “Discovery Channel,” “The
Philadelphia Inquirer” back home, and NBC News
all ran with this story. And so it’s really been
an exciting result. And we’re actually–
this is ongoing work. So we just got five
nights of observing time with an eight meter telescope
in Chile to look at about 10% of our samples. We’re very excited
about this work. And in the last
15 minutes or so, I want to tell you a little bit
about where this kind of work is going. So all the results
that I’ve used so far have relied on survey data. Surveys are a very
different mode of operation for astronomers. We’re no longer concerned about
our own individual projects and going to the
telescope to look at the small number of stars
that we’re interested in. Instead, schools and large
groups of astronomers are working together
to produce surveys that produce enough data that
everyone can share and use. There’s two in particular that I
want to draw your attention to. The first is a European
mission known as Gaia. Gaia launched last
year and is currently starting its observations. It is going to produce
a parallax survey of a billion stars. So this is going to produce the
cleanest fundamental distance maps that astronomy and
humanity has ever done. All the stars that
it’s going to observe will get parallax distances. And so it’s essentially
going to be a high definition look at the nearby galaxy. It will also measure kinematics,
so velocity information, for about 150 million stars. And it can measure
a parallactic shift of about 10 microarc seconds. So that’s the angle
that a quarter would subtend if you
put it on the moon. So it’s a ridiculously
precise measurement, and like I said, it successfully
launched in December, and observations are ongoing. And because it’s 2015,
Gaia has a Twitter account. And so every once
in a while, there are these types of
images posted to Twitter. So you can follow along
if you’re interested. But the granddaddy of
them all, and the one that I and most
astronomers hopefully are very excited about, is
a project known as LSST. It stands for the Large
Synoptic Survey Telescope. This telescope is
going to be in Chile. And its slogan is
“wide fast, deep.” What it’s going to do is
image the entire night sky every three days for 10 years. So it’s essentially going to
reproduce the entire Sloan survey every three days
and run for 10 years. It’s an 8.4 meter telescope. It will produce some of the
finest images ever seen. This is the filter
assembly being shown here. The focal plane is
shown in blue there, and you can see that,
as one of these filters comes into view,
that will help get us detailed observations
about the colors of stars and galaxies. The camera itself
weighs about three tons. It’s going to hang off
of the secondary mirror. So this is a three
mirror design. I’ll show you a little bit more
about that in just a second. This is not how the
construction is going to go, but construction is under way. So they’ve begun to level
off the site down in Chile that it’s going to be at. And some of the hardware is
actually already in place. The camera is going to be built
by the Department of Energy, and the mirror– I have a
few pictures of the mirror. The primary mirror has
reached a milestone. It’s essentially completed
its polishing process. So the mode of
operation for LSST is to tile the night sky
every three days for 10 years. So it’s going to use six
different filters, and every 30 seconds, take an
image of the sky and move to the next
position, take another image, and continue to tile the night
sky each observable night. So this is one of the
most ambitious projects that’s ever been undertaken. It’s the highest
ranked– this is what happens when
you loop backwards. The data is disappearing! This is the highest ranked
ground-based astronomy project that’s underway. The imaging camera is going
to be the largest camera ever constructed, about
3.2 gigapixels. And this is the
actual size to scale. And you can see the angular
size would cover an area much, much larger than the full moon. So here’s another detailed look
at the camera with a person to scale. It weighs about three tons. And for each ton, you get about
a gigapixel worth of imaging. The images, obviously
we don’t have images yet from the LSST camera. But the survey
design specifications are looking for
image quality that is slightly better
than Sloan, than SDSS. And this is kind
of the difference that you would see
between the two surveys. It’s also going to
push deeper than Sloan. So it’s going to image the sky
to fainter apparent magnitudes. And each field is
going to contain millions of stars and galaxies. So before, the movie
that I showed you, it had a few thousand
cool red dwarfs. We’re now going to be
looking at millions of stars in each image. So it’s really going to be an
exquisite map of our universe. Mirror construction started
at the University of Arizona in 2008. Just last year– and you
can see a lot of the VIPs there for scale– last year,
I, along with a professor at Haverford College,
Beth Willman, took a few undergrads
to the mirror lab. And so that’s the primary
and tertiary mirrors of LSST sitting behind us. You may be able to see
a slight change in pitch between the outer ring
and the inner ring. So the outer ring is
the primary mirror, the inner ring is
the third mirror that the photons will bounce off
of on their way to the camera. And actually, just
this past weekend, the polishing of this
mirror was completed. So there was a little
bit of a celebration. People were popping
champagne bottles. And it’s really a testament
to the work that’s done at University of Arizona. It’s an 8.4 meter
mirror, and it’s polished to within 10
nanometers of focus. So it’s a painstaking process. But it’s finished, and
things are looking good. First light, the
first observation, should be taken
somewhere around 2019, with full science observation
starting around 2021. When that happens, we’re
going to need your guys’ help, because there’s going
to be a lot of data that’s going to be produced
by this telescope. Google’s actually already
a partner in LSST, and it’s the type of challenges
that come out of this much data is something that we
really– astronomers need help with optimizing
our analysis of this data. There will be gigabytes of
public data available nightly. So because we’re imaging
the night sky over and over and over again, one
of the questions that astronomers– and
potentially the public in general– will be
interested in is things like, did a star explode in this
part of the sky last night? We’re working on
developing alerts that can be configurable to any
type of scientific question. And obviously, if you’re imaging
2 million objects per field, and potentially hundreds
of fields per night, you don’t want every
alert being triggered. There is going to be all
sorts of variability that’ll be observed. Stars rotate, and their
brightness changes with time. Nearby asteroids would move
across the field of view. Distant stars may explode
in other galaxies. And so all those questions
are important for astronomy, but they’re important to
different astronomers. And so for someone like myself
that’s mapping out the Milky Way, I don’t want
to be getting alerts on my phone saying that there’s
an asteroid buzzing around. And so that’s an ongoing
problem that we’re trying to work on in
preparation for the survey. In addition to that type
of question, what’s great about these large
surveys is that we also get the public involved. So one of the more exciting
avenues for public involvement that’s come out of
Sloan is a project known as Galaxy
Zoo, where anyone with an internet connection can
go online and look at images of galaxies and begin
to classify them. And it’s actually
led to a number of different academic papers. But one really cool
discovery, there was a schoolteacher
in Denmark that discovered an object
that’s really, really rare– to the point where, when
it was actually discovered, no one really quite knew
what we were looking at. It’s since been sussed out,
but that type of work really only comes out of large
surveys, where you’re not biasing your view of
the night sky at all. In terms of the imaging
that will come out of LSST, some of the
data rates, you’re looking at somewhere around
100 petabytes of imaging data by the time the survey
is completely over. The catalogs– so
most astronomers don’t deal with raw images. Instead, we look at
catalogs of objects. So whether or not you may
be interested in the star catalog, or the galaxy
catalog, all that information at the end of the day is
somewhere around 50 petabytes. And catalogs will be produced
throughout the duration of the survey. And for the final
catalog alone, we’re looking at something
like 15 petabytes or so. So the bottom line
is that LSST is one of the most exciting surveys
that’s ever been planned. It’s going to result in the
best sky image ever taken. With 60 petabytes of
astronomical image data, you would need something
like 3 million HDTV sets to view all the pixels
that will be coming out of this telescope. If you stacked all
those images up, it would take about 11 months
to view the entire movie. So this is really the best movie
that has ever been constructed. And for the first
time, we’re going to produce an
astronomical catalog. Somewhere around 40
billion stars and galaxies will be in this catalog. And that, for the first time,
will dwarf the number of people here on earth. So it’s an exciting time
for surveys and astronomy, and I thank you for
your time and attention. And I’ll take any questions. [APPLAUSE] AUDIENCE: So you mentioned
that the Sloan telescope sends the data to the
University of Washington to be printed out
on those sheets. What was the purpose of that? I don’t understand. JOHN BOCHANKSI: Yeah. So you take an image
of the night sky, and you would
really– ideally, you would want a spectrum
of everything, right? But you can’t do that, because
you don’t have the telescope time in order to make
all those observations. And so there were pipelines. There was essentially
code that was written that said, all right, if I’m
interested in distant galaxies, look at this imaging
data, tell me where we think all those
distant galaxies are, and record their location. And then there’s another group
that’s interested in stars, and they run a
different code, and that produces the most
interesting stars that they want to get those
detailed observations on. And so it was kind of a
democratic process that was set up, with the goal being
that, at the end of the survey, the big scientific questions
that we were interested in were answered. And so for each
position on the sky, 640 objects were selected to
get those detailed observations. And so that’s when
the positions were sent to the University
of Washington. The holes were drilled,
and then those plates were shipped back
down to New Mexico. AUDIENCE: I see. So what was the advantage
of actually drilling them versus just
creating computer images and viewing them on a monitor? JOHN BOCHANKSI: Because it’s
a different observation, all right? So one, you’re taking
an image, and you get essentially position on the
sky, some color information, and the shape of either
your star or your galaxy. But if you take a
spectrum of an object, that tells you things that you
can’t get out of the image. So that tells you more about
the composition of the object, like what elements are in the
atmospheres of these stars. It also tells you something
about the velocity, because you use the
Doppler shift to measure the relative velocity
between you and the object that you’re interested in. So essentially, it gives
you more information than just an image alone. AUDIENCE: The star that’s
1 million light-years away, how do you know it
belongs to our Milky Way? JOHN BOCHANKSI: That’s
a good question. So we know the
velocity of the star. And we know its distance. And what we do is compute
a potential orbit. We assume a model for the
gravitational potential of the Milky Way, and
we then– we essentially turn the clock backwards and
say, if we look at the orbit and see where it started,
did it have enough velocity, or enough energy associated
with it, to leave the Milky Way? So if it was above
or below, it’s known as the escape velocity. So for the Earth, you have
to hit about 11 kilometers per second or so to
escape the Earth. For the Milky Way,
it’s somewhere between 500 and 600
kilometers per second. And the velocities
of these stars are much smaller than that. AUDIENCE: One that’s
been in the press a lot that you haven’t mentioned was
the James Webb Space Telescope. Could you talk a
little bit about that? And how does relate to what
you’ve been talking about? JOHN BOCHANKSI: Sure. So the James Webb
Space Telescope, which is the
successor to Hubble, it’s essentially
a different mode of operation from the surveys
that I’ve been discussing. So if you wanted to do
something like Sloan with the Hubble
Space Telescope, it would take you over a century
to just image the sky. What’s great about
these space telescopes, though, is that if you find
a very interesting object, then you can get very detailed
and precise measurements on the object. And what’s great
about James Webb– so Hubble works in the visual
part of the spectrum, which is what our eyes see. James Webb is
going to be focused on the infrared part
of the spectrum. And why that’s
important, primarily, the thing that I’m kind
of most excited about, is that that lets you
get a better handle on the properties of
planets around other stars. And so we actually talked about
this a little bit at lunch, but James Webb will
essentially give us the first definitive measurement
of atmospheres around planets, around other stars. We’re starting to
do this right now, but the signal-to-noise of those
measurements is fairly small. With James Webb, it won’t
be small at all anymore. So that’s probably one of
the most exciting things that will come
out of James Webb. AUDIENCE: You mentioned imaging
the same part of the sky again and again every few nights. So I could see one thing
you’re looking for is changes. But is another thing that
you’re looking for building up better statistics
of the same area? JOHN BOCHANKSI: Yes. Yeah, and actually thanks for– AUDIENCE: I was wondering
how many times you do that before you’re really
not getting much more, you know? JOHN BOCHANKSI: And thanks
for mentioning that, because I failed to bring
that up in the talk. Each part of the
sky is essentially going to get about
1,000 pictures taken of it over the
course of the survey. And so one of the
things that you can do is essentially add all
those images together, and you’re going to be able to
push a few magnitudes deeper than a single image. And so that, at the
end of the survey, you’re going to have a slightly
deeper view of the Milky Way, primarily of its
halo, than you would from just a single image alone. And so 1,000 is about the
number that we’re shooting for. And one of the nice
modes of– or one of the nice things that comes
out of surveys like Sloan or LSST, and the
reason that “wide” is the first thing
that’s mentioned in the slogan for the
survey, is that by going wide you recover more stars and
galaxies than if you just push deeper and deeper. So that’s the first
thing that you try and do with a survey is cover as
much of the sky as possible. AUDIENCE: It would seem, taking
the full picture multiple times like that, you would have
a really great opportunity to catch things like nearby
novae or faraway supernovae as they’re just starting up. But that also– the
amount of processing to do that very quickly
in real time, or at least over the course of a day or two
so you can catch them when it’s still useful, is
that something that seems feasible with
what we have now? Or do we need to develop some– JOHN BOCHANKSI: That
will be hopefully one of the main science products
that comes out of the survey. So the actual cadence
of observations is something that’s
still being worked on. But essentially, there’s
going to be two short visits to catch things that vary on the
scale of about anywhere from 20 to 40 seconds, and then a
return a few days later. Supernovae tend to brighten on
the scale of about a few days. But they’re one of the primary
science drivers for the survey. They should– within the
first year of observations, we’ll get somewhere
around 200,000 supernovae. And the reason that
supernovae are important, the Nobel Prize was just
awarded for observations of about 40 or so
supernovae that led to discovering
the characterization of dark energy. In the first year
alone, we’re going to be able to do
very similar work, but with 200,000 supernovae. And so that will
really help us pin down the exact composition of
regular matter, dark matter, and dark energy, and
hopefully illuminate some of the questions
that we still have. AUDIENCE: Hi. Have you ever worked–
have you ever heard about SIM, Space
Interferometer Mission? JOHN BOCHANKSI: Yes,
I did hear about SIM. AUDIENCE: So I worked at
JPL for many years on SIM. And so the purpose was to
measure the motion of stars to the microarcsecond level. So interferometers, there’s been
also adaptive optics programs. So you know, these
surveys are really getting really
impressive in terms of the resolution, the
megapixels, gigapixels. So the question is,
so the aim of SIM was mostly to actually
detect exoplanets, to see planets around stars. So from the data you
measure, to that level, can you actually
measure– can you tell if there is an
exoplanet around a star, you know, by lateral motions,
or perhaps with microeclipses, you know by seeing
the magnitude change? The microeclipse
technique, or side motions. JOHN BOCHANKSI: Yeah. So you hit the nail on the head. So Gaia is going to be a
little better than LSST. AUDIENCE: Gaia was
actually basically the European version of SIM. So SIM was canceled,
unfortunately, and they decided
to go with Gaia. JOHN BOCHANKSI: Yeah. So Gaia will do the best
job of detecting planets, and the European
astronomers have worked on making estimates
of exactly how many planets may come out of measuring
the wobble of stars as they get tugged by
their planetary companions. And so there’s– I’ve seen
estimates of hundreds to I think 1,000 or so. But I looked at that paper
a long, long time ago. And LSST will also be somewhat
good at detecting some planets. It’s not its primary mission. But certainly, there’s going
to be planetary companions that pass between us and
their host stars. So the best
telescope that’s ever been made at finding
planets is Kepler. It has confirmed about
1,000 exoplanets to date, and it has something
like almost 5,000 more. That uses a transit method. AUDIENCE: Kepler is the
real tool to do that. JOHN BOCHANKSI: Kepler
is the real tool. And Kepler is still going, but
in a slightly different mode. LSST will pick some
of these up, but it’ll be mostly serendipitous. AUDIENCE: What’s the model
for working with this data once you’ve got it? I mean, can Joe Shmo sitting
at home submit a query and say, send me an alert
every time there’s a star whose
magnitude is 3.1415? JOHN BOCHANKSI: Yeah–
well, so all right. So there’s not too many stars
that have that magnitude. But yes, yes, you can. The data will be public. So if you have an
internet connection, you can get this data. And that’s really one
of the primary– that’s been in the description of
the survey from day one. And how that interaction
actually happens, typically what we use now are large SQL
queries and that kind of thing. So some of that
will be available. There will be nicer
UIs that are developed. I think there’s
already an iOS app. So it’s kind of a
work in progress, exactly how you
interact with the data. But anyone with an
internet connection will have access to it. MALE SPEAKER: Thank you, John. JOHN BOCHANKSI: Thanks.

Leave a Reply

Your email address will not be published. Required fields are marked *

Recent Posts
Recent Comments
Tags
© Copyright 2019. Tehai. All rights reserved. .