Wednesday, February 26, 2014

A UFO case study.

Last Thursday night we saw a UFO fly by at Poker Flat. Here's a clip of the all-sky video, you'll see it as a white cloud coming up from the west (left) horizon then disappearing.

Here's a 100% crop on the relevant part of the image:

It's definitely a cloud of some sort, you can watch it spread out and disperse. And if there was any doubt in your mind about that, here's a shot from a photographer on Pedro Dome (~10 miles south of Poker Flat) through a telephoto lens:

A wide angle shot from the same photographer, showing it's position in the sky:

And here's another shot from all the way over in Unalakleet, AK on the edge of the Bering Sea, ~650km west of here:

The one thing that jumps out to me instantly is they all have the object near the Andromeda Galaxy. Here's an annotated version of my cropped AASAP shot:

Okay, the Andromeda Galaxy in these photos is, um, maybe not all that obvious to most people even after it's been pointed out, but it's noticeable if you spend enough time looking at the sky and know the patterns. It's in all the photos except the telephoto shot.

Why is this relevant? Well, if even the camera 650km away is seeing it in roughly the same part of the sky, the object must be very far away. In orbit.

So I discussed this with a few rocket scientist colleagues and we decided the most likely answer is it's a fuel dump from a satellite or rocket. it's so high up that the dumped fuel is still in sunlight, that's why we can see it so well. And that's why it seems to follow such a constant path - no winds up there. We're just watching the cloud of fuel spread out and disperse.

Two days later, after the image was posted on, a satellite tracker wrote in about a Delta 4 rocket carrying a new GPS satellite which was in that exact location in the sky at that time. And, further, the rocket had just 12 minutes earlier separated a stage, which entails dumping excess fuel.

Consider that a UFO turned into an IFO.

Sunday, February 16, 2014

Auroral Forecasting: A Study Guide

There's two main things I look at when I'm trying to figure out what the aurora's about to do: Solar wind and Earth's magnetosphere. I guess that much is obvious since the whole show is caused by the interaction of those two things.

I'm sort of assuming you know the basic idea of how auroras work here: The Sun emits plasma which makes up the solar wind, when the solar wind collides with Earth's magnetosphere (if the solar wind is 'south' or 'negative') it strips magnetic field lines off the dayside of Earth and drapes them back behind the planet, where they 'reconnect' and accelerate plasma down the field lines to make aurorae. It still happens when the solar wind is north, BTW, just the geometry is different so where the plasma comes down to Earth is different.

So what you mostly care about is how much stuff is coming in on the solar wind, if the magnetic field geometry is right (southwards), and how much Earth's magnetosphere is being squished and stretched in response. I look at the readouts from two satellites to get this: ACE which sits about an hour upstream of Earth measuring the solar wind, and GOES-15 which is a geosynchronous satellite sitting at 135W longitude measuring (among other things) the shape of Earth's magnetosphere.

Here's a readout of the last two hours at ACE. I like this two hour display but everyone else seems to like the longer ones. Whatever. Either way, you look for the speed or density to be high (the amount of stuff coming in) and for Bz to be negative (which means the field is southward, and the more southward the better). Since ACE is about an hour upstream, whatever you see happening at ACE will reach Earth an hour later.

And here's the readout from GOES. This is measuring the component of Earth's magnetic field parallel to the dipole axis (at the location of the satellite... you'll see a daily up/down cycle which is the satellite going around the planet once a day). In simple terms, when the satellite is near midnight (marked 'M' on the plot), a high value means the field is not very compressed, and a low value means the field is compressed and storing magnetic energy. When that energy releases you see the value suddenly jump upwards, and that probably means some plasma has been accelerated down the field lines and you should expect to see the aurora brighten in roughly 10 or 15 minutes. The higher the jump up, the more energy released, and the more auroral activity you should expect. When the value drops below about 50 you should expect a release soon.

And oh, but they give it as a three day plot? How awkward is that when you're looking for features on the scale of minutes? So this 3 hour plot is better.

With that in mind, I put together something of a 'study guide' display. It's one thing to hear this stuff as theory, but it's much better to see it in practice. So I took the timelapse movie of the night's activity from AASAP and overlaid the last hour of ACE Bz data and the last 30 minutes of GOES Hp data. This is useful I think for anyone from a beginner to a seasoned vet because it lets you watch in a minute or two how these factors work together over a whole night. Here's the video from Feb 08, when we had a fairly short but quite spectacular display:

So if you watch, Bz was around zero or weakly south from the beginning, leading to some slow but steady buildup of magnetic energy in Earth's magnetosphere, reflected in the Hp number being down around 40nT. Around 06:00UT Bz turned more solidly south, and one hour later that led to a pretty strong buildup of magnetic energy as Hp went down to around 20nT. Bz jumped back sharply to the north around 07:00UT, and when that jump reached Earth an hour later it triggered the release of all that stored magnetic energy - GOES Hp jumped nearly 50nT in 10 minutes! That's a really large jump, and as expected 15 minutes later we see that westward traveling surge come in from the east (right) side of the image and the aurora gets very bright and active. Once the energy is used up the bright arcs fade away, leaving a large mass of drifting, pulsating diffuse aurora, which is pretty typical - this is the recovery phase of the substorm. Bz remains north for a few hours, and so isn't 'driving' the aurora very much and we just get some very weak diffuse stuff for a while, but around 12:00UT it turns south again. This late in the evening the GOES satellite is no longer well positioned, but an hour later we do see the aurora brighten up and get more active. This late in the evening Poker Flat is coming out of the 'auroral electrojet' where the brightest, most active displays happen, so instead of super bright arcs like earlier the display is dominated by large patches that drift around and pulsate. This continues for the remainder of the night as Bz remains southward.

So yeah. I'll be making more of these overlays when I get good example nights, and I hope people can learn by watching them. I chose this night as the debut because we had that really impressive 50nT jump in Hp. Here's what the GOES 3-day plot of that looked like:

And here's a 5 hour plot:

And, in case you were wondering what that looked like in person, here's a video from an intensified CCD video camera. It's black and white, and kind of noisy, but it can record the aurora in real time video, so that's something. I sped it up by about 4x to better see the dynamics of the westward traveling surge come in, and the arc movement. This should give a good idea as to how much it was actually moving around.

Sorry about the big black thing blocking part of the sky. The camera is sensitive enough that the Moon will ruin it, so that's the lawnmower battery we set next to the camera to block the moonlight. It's very high-tech.

Saturday, February 15, 2014

Rocket Science, Part 1

Well I mentioned in the last post that I would be on a rocket campaign. I've been waiting until we launched the rocket to make an update, but turns out we made it through the whole launch window without favorable conditions to launch. We actually had pretty good science conditions on several nights but the rocket was grounded due to high winds which left a significant probability of blowing out of our intended landing zone. So we're waiting for the Moon to not be so bright then we'll open another window.

The image above is one I took during a night waiting to launch, the three red streaks are weather balloons released to measure the winds. It's a composite of about 40 two minute exposures. I've been thinking about a better way to process star trails for a while, that would preserve the single image 'background' so that the aurora would be detailed and pretty (and, in other situations, would give you stationary foreground foliage even if it's windy) while still allowing the stars to trail over many hours. So you're looking at something like 1.5 hours (~40 exposures) of stars but only 2 minutes (1 exposure) of aurora. What I did was loaded all the images into IDL and calculated the standard deviation on each pixel. I picked a single image to use as a background, then I went through and picked the pixels with a large positive standard deviation, meaning something temporary and bright happened in that pixel (e.g. a star moved through), and replaced that pixel from the background image with the pixel from the brightest single image. The tricky part is I can do the statistics on a single color channel, so if I ignore the green channel the process is blind to the aurora movement. I used the red channel for this example to get the red flares on the weather balloons, but normally I'd probably use the blue channel since there can be significant auroral emissions in red. It still needs some work but I think it's a good start. Eventually I plan to make it into a simple GUI where people can use it without needing to know how to script.

Other than that, we did actually launch a rocket, a small experimental prototype designed by some of the folks here at UAF. Since it was a test it didn't need to launch into an aurora like our typical rockets, so it actually launched in daytime. Here's a shot I got of it going up with the 150mm Sigma lens and the D7000:

And here's a video I took with the new NEX F3 and a 50mm Nikon lens:

I don't know why Flickr isn't offering me the video embed code that it offers for every other video.

Here's a shot of it leaving the launch pad from a camera that was very close, taken by Jeff Rothman:

And click here for an album of it taking off then descending under parachute in the sunset, taken by Mark Conde.

Our next launch window opens on the 24th of this month, so hopefully we'll get the big rocket off the ground without too much trouble this time.

In other news, I got rear-ended by a big truck while waiting for the car in front of me to turn, sandwiched between the two, and my car is probably totaled. So I'm shopping around and I think I'll get a 2014 Subaru Forester. I am disappointed with Subaru for not offering the ridiculously oversized moonroof on manual transmission models.