Plaster Creek Daily Mean Streamflow

“Water Years” 2017–2019. Measured at USGS site 04119055, under 28th Street

During the pandemic months, I’ve turned to tactile crafts. I've learned to knit with my arms as needles and print with cyanotype chemicals. I even looked into these iron-together, mosaic beads meant for kids. Given a grid and physical pixels in ordered spectrums of colors, I thought it’d be natural to render some data.

This is a sort of companion to the hourly streamflow visualizer—low-tech vs. high, manual depiction vs. automated, and a three-year daily history vs. hourly updates.

Data Process

In addition to hourly measurements, the USGS Plaster Creek site calculates an average discharge for each day, and puts those in a table to summarize each water year. (A “water year” begins each October to account for early snowfall that doesn’t melt until spring.)

I scraped these into a Google Sheet and used conditional formatting to shade each cell blue by value.

This is really the heatmap I was after, but I needed another step to help me make it a physical object. I took a screenshot, and in Photoshop, used the posterize effect to quantize or “round” the continuous color scale up or down to five shades of blue available in Perler’s Tropical bead tray line: Robin's Egg, Mist, Lagoon, Teal, and Midnight!

This became my key as I lay beads. It probably took three hours total. A really soothing, hypnotic passtime!

Note on units

I actually skipped a step above: in the spreadsheet, I first converted each measurement to its value log2. I created this heatmap actually using that value instead, so that color variation communicates a truer seasonal arc.

Bear with me: so, the streamflow gauge is capable of answering the question: every second, how many cubes of water, a foot in each dimension, are fully displaced from their position here? When streamflow increases basically at all, the answer to this question surges exponentially. The USGS site uses a logarithmic scale by default for this ft3/sec unit. So I straightened out that curve, rescaling the high variation in large readings to the low variation in small readings. The scale in the bottom corner makes this plain.

(I log’d the discharge reading on my hourly update page as well, in calculating the height of the blue wave. Little fluctuations on dry days then have as much visible range as big flood fluctuations do.)

((The USGS station started collecting water height in April 2020. This unit increases more linearly with rainfall and probably my subjective sensation of streamflow, though it doesn’t capture speed.))

Anyway, this is all to say: logarithmic scales can be really hard to make sense of! They can require some reinterpretation, which sometimes feels like data manipulation. Numbers feel so objective; attaching a unit to them grounds them in measurement to the real world, though still abstractly enough to require interpretation. I think this communication is a main concern of data visualization practice.