Unfortunately, for many hours of the day the sun seems to refuse to provide enough light for the webcam to operate properly - in fact, rumor says at such times the light is blocked by a rather large chunk of rock called "the earth"...
However, although nighttime images of the webcam seem to contain little more than random color noise when viewed with the naked eye, they still contain a lot of hidden, useful information. With a few image processing techniques, this information can be extracted and enhanced in such a way that the resulting images can provide a nice supplement to the daytime images, and give an impression of the actual volcano condition during nighttime (...well, at least when the moon is out - see below).
Here is an example of an unprocessed webcam image at night:
No image details except some strange bright spot above the center can be discerned.
After greyscaling the image, the color noise vanishes:
(Well, duuh!, of course it vanishes - it's a grey scaling, remember?)
True, but the overall noise level is still highly reduced by this step. At low light conditions, the internal noise caused by the CCD chip and the electronics becomes much stronger compared to the signal, and is visible as random color fluctuations in individual pixels or image areas. Since these fluctuations apparently have about the same intensity for all pixels, but are randomly assigned to a color channel (red, green or blue), simply ignoring the color information greatly increases the image quality.
The remaining grid-like noise pattern is mostly due to the JPG compression of the original image.
Already at this state, with a little imagination the dark crater region, as well as the snow covered flank on the right can be recognized. In addition, the bright spot mentioned above now resolves into two distinct blobs where we would expect the active crater dome.
Contrast enhancing the image reveals even more details:
(I had to crop the text in order for the histogram levelling to be effective)
Now the shape of the mountain can be clearly recognized, despite the strong JPG artifacts.
This is about all you can get from a single image.
However, when combining several consecutive images and averaging them into one (which is equivalent to a very long exposure time), the noise level - especially from the image compression - drops dramatically:
This picture is the result of processing 18 individual images taken on that day (well, night) within the 3 1/2 hours where Mount St. Helens was most active. Many details on the crater rim and around the mountain can be easily identified - the only unusual sight being the two intense spots on top of the new crater dome.
Under extreme light conditions, the camera switches to a black-and-white mode, where it becomes much more sensitive to (near) infrared light, i.e. the heat from sources such as the superheated steam and smoke emerging from the crater, but also red magma, or even part of the rock that is not hot enough to create a red glow, but still radiates in the IR spectrum.
That night, on Feb 23rd 2005, the volcano was active in two distinct regions of its dome. The first eruption occured at or before 02:13 PST, in the right part of the dome. The second one at approx. 04:08 PST in the left part, and again in the right part at approx. 04:28 PST:
 02:13 |
 02:18 |
 04:08 |
 04:13 |
 04:18 |
 04:23 |
 04:28 |
 04:33 |
 04:38 |
 04:43 |
 04:53 |
 05:03 |
 |
After the actual events, the (presumably) erupted lava took quite some time to cool down and was still glowing in the infrared for several frames.
Averaging all images of the whole night (in this case, 112) yields an even smoother image. However, the influence of the (relatively short) eruption is diminished:
Unfortunately, the camera iris is fully open during the night, causing it to unfocus. Otherwise, the overall quality of these nighttime images would be quite high.
(Note the bright pixel to the left of the center: this is a "hot pixel", a pixel within the CCD chip that always returns a brightness value, even when no light is received by it. It is however only a very small value which becomes noticeable only when many images are averaged.)
The mountain looks different than in the averaged image above. The reason is that the Feb 23rd was close to a full moon night, and the volcano is therefore illuminated by the moon.
While in the first image the moon is positioned somewhere to the right actually casting a shadow over the crater, in this image the moon is completing almost half a circle over the mountain, illuminating it from all sides in the process. The reason that the left rim is still dark is simply because there is no spoon.
...uh, snow.
Now compare this image to the original picture taken by the camera:
 |
 |
| before | after |
-------
Here are a few more impressions of various nighttime situations:
Moonless nights, however, are too dark even for this camera:
This was the night of Oct 13th 2004, where the volcano was also active, this time on the backside of the right part of the dome. Granted - at this time there was no snow on the mountain, which makes it even more difficult for the camera to see anything.
The "hot pixel" is visible once again, and the grey shape in the lower left corner is a minor camera artefact, probably from the optics.
-------
Thanks to the guys at the Mount St. Helens National Volcanic Monument for continuously providing the webcam images.
I hope you enjoyed this page. Take care!
-------
2005 by Wolfram Kresse.