What would sunset look like if you were on the planet HD209458b, a gas giant orbiting a star 150 light years away? According to exoplanetary scientist Frédéric Pont, it looks like this:
Isn't that pretty? And there's quite a bit of science in that, too.
First things first: HD209458 is a star pretty similar to our Sun. It was one of the first stars determined to have a planet orbiting it (way back in 1999) -- the aforementioned HD209458b, nicknamed Osiris -- and it turns out the planet's orbit is so close to edge-on as seen from Earth that we see that planet passing directly in between us and that star once per orbit. When the planet transits that star the amount of light we see dips a little bit. From that we can get the period of the orbit and the size of the planet (a bigger planet blocks more light).
But we can get more, too. There's a camera on board Hubble called the Space Telescope Imaging Spectrograph, or STIS. It can take the light from an object and break it up into thousands of separate narrowly sliced colors, called a spectrum. By analyzing that spectrum we can find out an astonishing amount of things about astronomical objects: their temperature, rotation, even their composition!
Shortly after HD209458b was discovered to be a transiting exoplanet, STIS was pointed at the star. The camera took hundreds of very short exposures during a transit in the hope of being able to detect the atmosphere of the planet. Osiris was known to be massive, about 70% as massive as Jupiter, so it most likely has a thick atmosphere. It also orbits so close to its parent star -- 6.7 million km (4 million miles), much closer than Mercury orbits the Sun -- that the heat from the star puffs the atmosphere up, making it easier to see.
In fact, the spectra did reveal the presence of an atmosphere; the first time the atmosphere of an alien planet was ever observed. Different elements and molecules absorb light at different colors, so in the spectrum there are dark spots where the planet's air absorbs the light from the star behind it during a transit, and how dark that spot gets tells you how much light is absorbed.
It's this information Prof. Pont used to create the image above (inspired by investigation and an animation done by Alain Lecavelier des Etangs). By knowing the color of the star itself, and using the way the planet's atmosphere absorbs light, he created this image of the star using sophisticated computer modeling. The star itself is white, like the Sun, and so you might expect the sunset to look red like it does on Earth. But there are different processes involved with Osiris's air! Sodium absorbs red light, and there's enough of it floating around in the atmosphere of Osiris that the sunset takes on a bluish cast, but then as the star sets further, the blue light gets scattered away, much as it does here on Earth. The result is a green star -- something not usually seen!
Pont also modeled the layering in the atmosphere too, and how each layer would affect the star's color, producing the image seen. Even the glowing layers around the star are real (at least, real in the computer model); the reddish one is from those same sodium atoms re-emitting the red light they absorb, and the bluer layers from the light scattered away. By the way, he did this same analysis for the planet orbiting HD 189733, and got a far more terrestrial looking sunset.
Two things to note: the planet has no surface, so Pont put his imaginary sunset-watcher 10,000 km above the planet's surface, observing as the star sinks beneath the planet's limb. Imagine being in a space station (with the AC set to ultra-super-duper maximum), floating in front of a window, and seeing such a thing unfold! I would never have guessed the sunset would be green. Which brings me to the second thing: this model of the sunset is an average over the entire atmosphere. Where I live in Boulder, for example, the sunsets are different than they are in the eastern US, or in, say, Greenland, due to local conditions. What Pont did was take a planetary average for Osiris, since the STIS observations look at the whole planet all at once.
And a personal note, too. I was still working on STIS when these observations came in. Those were heady times; the idea of exoplanets was still pretty new, and being able to detect one this way was very new. I remember my boss, Don Lindler, very excitedly telling me he had the data from the observations and was going to do the basic processing of them for the scientists involved. He let me see them, and it was funny: to the eye, all the spectra (684 of them!) looked the same. But when you analyzed them carefully, subtle changes could be seen as the planet moved across the face of its sun. It was by far the best observations of a transiting planet ever seen. I remember Don and I were freaking out... well, I freaked more than Don did; I'm more of a dork. As a scientist working on Hubble I knew I couldn't tell a soul about any of this -- the investigators involved did the work, so they get the glory -- and it was really tough. But oh, those few minutes of seeing that.
And at the time, of course, it didn't occur to me this data could be used to model what a sunset would look like. I wish I had thought of that! But I'm glad someone else did, and made this dream-like vista. It's nice to know -- to see -- the fruits of those earlier labors paying off so colorfully.
Image credits: Prof. Frédéric Pont at the University of Exeter; NASA/JPL-Caltech.