I’ve been delaying my first post a bit, mostly because I’ve been perplexed re: how to begin writing about this project. Initially, I thought I might start out by explaining the history of code within the context of the history of the Jacquard loom, and situating this project within that context (note: this project is situated within that context). But then I spent some hours scouring the library website and Google Scholar for suitable histories of weaving and the loom, and realized this explanation was going to be more involved than I thought. I currently have a folder entitled “Loom Readings,” that is, well, looming over me. Anyway, this is just about how every project I embark on begins: some trepidation about where to begin, a lot of scouring, more trepidation, then an epiphany about how involved it will be, and then an acceptance that I have to begin somewhere and nothing is ever perfect. And with weaving, quilting, knitting, etc., it is tradition to slip a mistake into every piece, because only deities are perfect. So, here we are.
Save the looming Loom Readings, I’ve been progressing on this project in other ways. So today, I’ll tell you a little bit about my first Landsat knitting pattern, which will inevitably involve a short discussion of the electromagnetic spectrum, infrared light, postphenomenology, and Fair Isle knitting techniques. Inevitably. And if you don’t know what any of those things are, that’s perfectly ok: at one point in time, I didn’t either. And for that reason, I’ll try to explain everything as clearly as I can.
I started with this Landsat 8 image of Houston, TX, taken on May 13th, 2013.
It’s an infrared image, which means that it uses Landsat 8 bands 5, 4, and 3 (typically used for viewing vegetation. Let’s put a pin in the concept of infrared light — I’ll explain that in a moment. First, though, you might be asking: what are band combinations? Heck, what are bands? To understand what bands and band combinations are, you must first have a basic understanding of the electromagnetic spectrum. You might not think about the electromagnetic spectrum very much, but it’s in everything you see and do. Every color you perceive is really just light reflecting off of a material in a certain way. For instance, it’s technically incorrect to say that healthy trees are green. Rather, their leaves reflect light in a particular way that makes them appear green, and they absorb all other kinds of wavelengths. Similarly, ocean water sometimes appears blue or turquoise to us because it reflects electromagnetic wavelengths in the form of blue light, and absorbs all other kinds of wavelengths. Color is an illusion. Note that I am using the terms light and wavelength interchangeably. However, I should also note that there is probably some nuance that could disrupt my interchangeable use, much of which is beyond the scope of this blog and the scope of my expertise.
However, all of the light I’ve talked about up until this point is from the visible portion of the electromagnetic spectrum. UV, or ultraviolet, light — the kinds of wavelengths that damage our eyes and skin — are actually invisible to us. They are comprised of waves that are closer together than those within the visible part of the electromagnetic spectrum. Remote sensing satellites don’t just capture wavelengths that we can see with our eyes: they can “see” larger portions of the electromagnetic spectrum. Hence, the image of Houston pictured above.
However, satellites don’t just capture any and all wavelengths at all times. Rather, they’re equipped with a variety of sensing instruments that capture light in portions of the spectrum that we deem most valuable in a given instance. We refer to these chosen portions of the electromagnetic spectrum as a satellite’s bands. Think of it this way: the electromagnetic spectrum is a sliding scale, and teams of scientists and engineers that design satellites select particular points on the spectrum they’d like to see reflected from a target material (in this instance, a particular feature on earth, like trees or the ocean). Landsat 8, for example, measures wavelengths by way of 11 different spectral bands. There are all kinds of band combinations that allow you to view a single scene within many different contexts. For instance, on Landsat 8, bands 4, 3, and 2 allow you to view an image as a true color approximation. Here’s a complete list of those bands and band combinations and their technical specs, if you’d like to see them. Now, let’s contextualize this within the image of Houston: it shows us light from bands 5, 4, and 3. Band 3 measures green light in the visible spectrum; band 4 measures red light within the visible spectrum; band 5 measures near-infrared (NIR) light.
So, what is infrared light, and why do we care about it? Healthy vegetation reflects light in the near-infrared portion of the spectrum, wavelengths that are slightly too long for us to observe with our eyes. Different disciplines define the infrared portion of the spectrum differently, and the boundaries between NIR, short-wave infrared wavelengths (or SWIR), mid-range infrared wavelengths (MWIR), and far-range infrared wavelengths (FWIR) are fuzzier than one might think. But here’s what I feel confident explaining to you: near-infrared (NIR) and short-wave infrared (SWIR) come to us in the form of reflected light. We detect mid-range (MWIR) and far-range (FWIR) in the form of heat: they are thermal. While NIR and SWIR wavelengths are longer than the ones we can observe with our corporeal vessels, we can view these wavelengths with the same kinds of instruments we would use to detect visible light (i.e. the light we can see with our eyes). NIR and WIR wavelengths, however, are not as long as MWIR and FWIR wavelengths. And because MWIR and FWIR wavelengths are quite long, we use different instruments to detect them (i.e. radiometers).
But back to the image. Check out the slider on this page to view this image through infrared color bands and as a true color approximation. You’ll see that everything that looks red through infrared color bands is something that looks green and in a true color approximation (and that would look green to us): so, trees, grass, etc. The light blue and silver portions are infrastructure: roads, buildings, etc. And the giant turquoise blob is Trinity Bay, reaching out to Houston by way of smaller bays and bayous.
I love Landsat imagery because it opens up space for us to think about how even our own bodies mediate and confine the ways in which we see and experience life — and that there are many other possible ways to see and experience a scene (Houston, in this case). The idea that experience is always mediated through some kind of instrument — even if that instrument is our own eyesight — comprises a subfield of philosophy called postphenomenology. I’ll talk more about postphenomenological theory in future posts. But for now, I wanted to point out that we can think through Landsat imagery in both physical and philosophical terms. In fact, I’d be willing to bet that philosophers reading this likely conceptualized my explanations of the electromagnetic spectrum and color bands as both a philosophical project and a scientific exploration.
For now, though, I’d like to think through this Landsat 8 image in terms of fiber art. Recently, I discovered a fun and free online software called Chart Minder. Chart Minder lets you create your own Fair Isle knitting patterns. Fair Isle is a knitting technique that I’ll sum up for you as knitting pretty and colorful patterns. Fair Isle patterns typically involve five or less colors of yarn, and only knitting and purling stitches (the most basic of stitches). Fair Isle also involves some techniques that help you carry the strands of yarn that are dormant in your pattern for a given stitch, series of stitches, or rows. Knitters call this catching floats. Catching floats is like the knitting equivalent of packing a lunch in the morning and carrying it to school with you, so it’s there for you a couple of hours later. We carry strands of yarn that aren’t presenting in the pattern along with us at the back of the project for a similar reason: so they are present when we need them.
You can make almost any pattern into a knitting pattern. If it’s really loopy and detailed, it may appear a bit pixellated (much like an old digital rendering of PacMan, who we know is all cute and round, but has some jagged edges because of pixellation). Chart Minder takes this idea and runs with it, inviting you to scan in an image of your choosing. It uses an algorithm (likely some kind of image recognition algorithm) to turn your images into discrete, knit-able squares. Each square represents a single stitch. And so, this infrared band image of Houston…
…becomes this infrared band image of Houston:
where each square represents a single stitch.
As you can see, it’s far from a perfect rendering. Unfortunately, Chart Minder did not want to make my image 300 x 300 stitches, which would have been about the size of a very large blanket. This would have allowed for much more detail. However, Chart Minder kept crashing on me whenever I tried to adjust the image settings to allow for this kind of detail. So I settled for a pattern that’s about 140 stitches wide and 90 stitches tall. This will make a small blanket (maybe a blanket for a cat).
I also discovered that, limited by traditional Fair Isle specifications, Chart Minder only lets you select five colors for the algorithm to automatically match to the image. I later found out that you can go in after the rendering, choose additional colors manually, and paint them in with a paintbrush tool. I’m currently working on another image where I’m employing this manual technique heavily. I’ll reveal this image in time. 🙂