Wow, Java is like one huge city. They find one man there and look where it leads.
Well. OK, but on, that current map, nearly all the fires seem to be in Queensland and the Northern Territory, and nearly all of those that are in WA are close to the northern and north-western coast, not in the middle of WA as we are seeing on those satellite pictures.
Well, it is the world’s most populous island, and one of the most densely populated regions on Earth (unless you count cities as regions, I guess).
On the first picture in the OP, there’s a bright light on the north coast of Alaska. The north coast of Norway also looks brighter than I would have imagined.
Any thoughts?
NASA provide a definitive answer.
That’s just amazing if slightly confusing!
If the maps are composite of many orbits then a single ‘light’ moved each time is shown as bigger than a stationary light. The instinctive way to interpret the map is that if you could look at the earth at one moment in time, at night, you would see all these lights … but that is not the actuality of the map.
That’s just a function of history and climate. High rainfall seasons in a location produce more fuel. Fires reduce the fuel load. Rainfall and fire in Australia are both highly unpredictable. Added to that, northern Australia tends to burn in late spring early summer with the onset of the storms. Western Australia tends to burn any time it damn well likes.
So the areas that burn at any given time are a total crapshoot. You can’t do any sort of comparison from one year to the next, or even one month to the next.
I suspect that the maps are false colour composites composed from a broad spectrum. So you aren’t actually seeing light at all, you are seeing a computer’s interpretation that is based on a combination of visible light, UV and heat. corrected for apparent cloud cover, angle of incidence, relative albedo and so forth.
If you could look at the Earth in IR, UV and visible simultaneously, you probably would see something that looked much like that map because your eye automatically adjusts for shadow, contrast and so forth. So the image is trying to replicate what your eye might see on a cloudless night, based on certain assumptions. But it kind of fails because the assumptions about what generates heat were wrong.
The NASA info I cited above suggests the spectral range is limited to green through near IR. The final image seems to just map some aggregate intensity to a yellow to pure white colour range - which is probably a good trick to try to represent intensity visually. They also imply that there is a lot of compression of the dynamic range. So you could be standing in the middle of the outback with a flashlight pointed upwards and just about register, and in the final image you be represented with an apparent intensity that vastly over represents your actual brightness relative to a big city. What NASA do point out is that the images are composites from many overpasses. So the final image is like a time exposure over many days - and thus every fire that was burning, and not obscured by cloud in the period the overpasses, may be added to the image. There was never a time that all these fires were burning together.
The original NASA info did however say they explicitly removed fires, and given fires could reasonably have a definitive IR signature, this seemed reasonable, but it now seems this was a mistake on the part of whoever wrote up the images, and that fires had not been removed from the images. The new news release suggests that they do intend to work on this, and maybe release further images where fires are not included.
Or that any one fire was burning as represented by the map. Compare the fires in Colibri’s post.
What I want to see is the fires over time.
This is a really interesting issue. The question is whether the image of any fire is an instant or whether it has been integrated over time. Looking at the information on the satellite the slightly surprising answer is that these are not integrated over time pictures. The satellite is in a sun synchronous earth orbit, 845 km high, which takes it over every point on the earth once a day, and the image width for any pass is about 3000km - which means pretty much all of Oz, or the continental USA in one pass.
But wildfires burn as a front not a solid area - so why do the fires look so big? If you look at the very high res images the fire look like an illuminated cloud - which is why I was confused by them at first. The answer appears to be that the satellite’s imaging system is so sensitive it can see the smoke cloud illuminated by the fire, even when the fire is actually many miles away. Indeed the satellite is so sensitive it can see airglow. So the images are really very deceiving. They represent smoke clouds illuminated from many miles away by a fire with almost the same intensity as the lights in a big city.
There is one particular fire over Western Australia that looks perfect - with a clear edge to a fire front, pure black behind it, and streaming smoke clouds being driven in the same direction as the advancing front. Which is pretty impressive.
I’m not giving up on the fire theory however I stumbled across another forum post suggesting the illuminated clouds are lightning flashes which does make some sense.
April and October are the wrong time of year for lightning in Australia.
The images can’t be lightning. The satellite’s imager - the VIIRS sensor uses a rotating telescope, and only has what amounts to a single scanning pixel. As the telescope orbits the rotating telescope scans a line across the path, and the forward motion of the satellite moves the field of view forward ready for the next scan line. (This is how most low earth orbit Earth imaging systems work.) In systems like this you get linear streaks at right angles to the orbital path from lightning strikes, not a fully illuminated cloud. The length of the streak depends upon how long the lightning strike(s) last for as the imager scan over the cloud.
The only way to get fully lightning illuminated clouds would be to have a conventional camera that takes the entire scene in one hit. Since the imager doesn’t work this way, there is no way it can be lightning.
This paper (PDF) describes the capability of the VIIRS sensor, and discusses both fire and lightning detection. Figure 7B depicts lightning strikes as seen by a similar satellite imaging system.
The definitive description of the sensor would seem to be NPOESS VIIRS sensor design overview. PDF, and $25. (My university library subscribes, so I have access.)
October is the *ideal *time of year for lightning in northern Australia. That’s when the pre-monsoon thunderstorms arrive.
No, not really. My experience living in Darwin is that October has a small amount of activity but not much compared to the true wet season. The graph (A) on page 97 of this PDF reflects my memories of the climate. Yes there is some thunderstorm activity but it is light compared to the summer months. April is similar. I’ll grant you that I was incorrect to say those months are the “wrong” time of year, they are on the cusp, but I certainly wouldn’t expect large numbers of storms and certainly not in the more inland areas of the country where the lights on the maps are. When living in Darwin, October was the month that you wished the blasted storms would hurry up and break but they rarely did.
October and April are not months where you’d expect large numbers of storms, and the distribution in the NASA maps is completely wrong for wet season thunderstorm activity.
“Fires and other lighting could have been detected on any one day and integrated into the composite picture even though they were temporary. That seems to be the case in central and western Australia, where many lights appear in this map. Different areas burned with wildfire at different times that the satellite passed over, giving the impression (in the composite view) that the entire area was lit up at once.”
The image isn’t just one picture, its made from data collected over a longer period of time.
You know, from a lot of these answers I really get the feeling many dopers don’t seem to realise how big Australia is. You know that its land area is basically the same as the lower 48, right?
No, you don’t understand us Americans. For us, all other countries are much smaller than us even if geographically larger. 
Fires are most likely to be caused by the *first *lightning strike after the grass has dried off. It’s not like it requires 12.7 lightning strikes to ignite grass. As such, the *earliest *storms will be when you expect to get the majority of your fires.
Added to that, once the wet season starts the chances of fire are greatly reduced because the grass becomes green and wet. This is one factor that shows up in every single study of fire in Northern Australia: fire drops off to almost zero once the wet season starts.
The fact that October has only 1/3 the number of strikes as January would seem to be utterly irrelevant. If you want to see lightning starting grass fires, October is the *ideal *time of year.
Ok. What has that got to do with what I was talking about? Poster A said he thinks they may be lightning flashes, I said it is not the right time of year for lightning flashes. Now you are talking about starting fires? Sure the fires may have been started by lightning, but the lights are not lightning flashes, too many for that time of year and in the wrong part of the country.
Edit: Just to be clear, I was responding to someone who raised the theory that the lights themselves were lightning flashes rather than fires.