Hacker News Reader

Comment by tedunangst | Hacker News Reader

I'm a little confused. The image that's captioned "Land surface temperature, Philadelphia" (top label: "Landsat 8 derived land temperature") is that estimated or measured? Because it sounds like the hot spots were detected by taking a satellite image and looking for places that "looked" hot, but then it's presented like those places really were hotter?

A little later, he says the black warehouses were located under the really hot spots. But weren't the really hot spots determined by finding all the black spots in the image?

View on HN

Landsat 8 has 11 different imaging bands, and 2 of them (bands 10 and 11) are thermal infrared bands just for measuring heat.

For more info, see this Landsat 8 cheatsheet (written by my awesome coworker Charlie Loyd): http://landsat.gsfc.nasa.gov/?page_id=5377

Ah, thanks all.

> Our calibration team has found that with current processing these surface brightness temperatures are accurate to within ~±1 K for many 15 – 35° C targets, e.g., growing season vegetated targets.

Wow, I would have never expected that!

Dump European question: ±1 K is the same as ±1° C, right?

Yes. 0K is -273.15C, and 273.15K is 0C.

Yes, 1 K is the same temperature difference as 1 °C.

See of course Wikipedia for more: https://en.wikipedia.org/wiki/Kelvin.

Oh, and the degree symbol goes with the C, not with the value. :)

You can see some other uses of the satellite data on its website: http://landsat.usgs.gov/gallery.php

> Dump European question: ±1 K is the same as ±1° C, right?

Yes the intervals are identical, the only difference is the scale's starting point. In fact although that's discouraged papers often use Celsius absolutes and Kelvin intervals, as is the case here.

The temperature was almost certainly determined by fitting the spectral data to the black body radiator equation, the reason that red-hot metal glows red. Basically, objects emit light when they are warm, and the hotter they are, the higher the energy of their emissions (more blue). At normal Earth temperatures, these emissions will be in the infrared, not visible. Remember, the Landsat data includes not just visible data, but also registered images at a bunch of other wavelengths (near and far IR, at least). The actual color of the object is not really relevant.

Another analogy -- its like those thermometers that can measure temperature from across the room by pointing at an object!

Reply on news.ycombinator.com