Thursday, August 14, 2008

When in doubt, blame the instrument: It wasn't the lightning.

The WGN Weather blog shows a video from Bucktown of the intense thunderstorm of August 6th here:, but they claim that the nearby lightning strike at the end of the video actually produced arcing close to the camera. That wasn't the case. The "arcing" is actually an artifact of the CCD sensor in the video camera. To understand what's going on, you'll have to deal a little bit with the physics of CCDs. In silicon, incoming photons will excite electrons out of a lower energy state and into the "conduction" band where it can then migrate through the material. You can call this liberating the electron. In a CCD control voltages create zones where these freed electrons are trapped in the silicon until they are moved out and measured. Those zones are best known as pixels. Depending on the type of CCD, when the exposure is over, the electrons are moved pixel by pixel in columns to be read.

Intensely bright sources of light will produce so many electrons that they will overwhelm the control voltage and flood out of the pixel and into the surrounding pixels and circuitry, producing spurious effects. You've probably seen these effects -- it starts showing up at 1:57 in the Yeah Yeah Yeahs video for Maps for example, or in the SOHO image above (it's Venus doing the blooming).
In this case the electrons flow out and down the columns that the electrons would normally be read. The Sun is a great source of column bleeding in a lot of videos online: see this one, for instance. Or bright stars--here is a weak version (it's the faint vertical column, not the diagonal streaks):

Since it's difficult to control all the sources of light in any possible photo scene, the CCD manufacturers have ways of trying to mitigate the overflowing electrons. One technique is to put drainage canals around the pixels and dump the electrons. This is good, but the extra space for drainage costs you some light sensitivity and light measuring accuracy.

Another problem can develop while you are moving the electrons off the CCD to be measured--if you have a shutterless camera, then light is still hitting all the pixels and can still cause overflow problems. One technique (used a lot for video cameras, at least in the old days) is to make the CCD twice as big with half of the chip covered up. At the end of the exposure you quickly move the electrons in the lit part over to the dark part and then leisurely read them out. This helps, but you can still have those overflowing electrons come down into your dark area.

So, in the lightning video, you can see that the extremely bright strike produces too many electrons in the CCD of the camera, and they flow 1. into the dark frame-transfer area and 2. down the columns (the vertical bleeds).
You can see at least one of these effects in some of the other strikes in the video.

No comments: