Ecohydrology, remote sensing, and environmental physics — building the sensors first, then arguing with the data they send back. 2026-07-22T20:14:54+00:00 https://shadmanamin.github.io/ Md Shadman Aminmdshadman_amin@ucsb.edu 2026-05-18T00:00:00+00:00 https://shadmanamin.github.io/blog/2026/05/18/what-the-wildlife-taught-me/

A field note on the gap between a system that works and a system that keeps working, written mostly while replacing cable.

<p>There is a version of fieldwork that exists in the methods section,where instruments are <em>deployed</em> and data is <em>collected</em>. Then there is theversion where you drive two hours, hike the last stretch because the road hasbecome a creek, and find that something with teeth has spent the winterredesigning your wiring harness.</p><p>I have now replaced enough cable at the Dangermond Preserve to have opinions.</p><h2 id="the-site-will-throw-everything-it-has-against-you">The site will throw everything it has against you</h2><p>In three field seasons, I do not think a single sapflow probe has failed on itsown terms. What fails is everything around it: the cable run, the connector,the enclosure seal, the strain relief, the mounting that was fine until thebranch it was attached to moved four centimetres.</p><p>While the design of the sensors could surely use an upgrade, it is just realizing thatthe thing you spend your time designing and the thing that determines youruptime are usually different objects.</p><figure class="figure-wide"> <img src="/assets/img/cards/chewedcable.jpg" alt="Cable insulation stripped away by an animal" loading="lazy" /> <figcaption>Exhibit A.</figcaption></figure><h2 id="what-i-do-differently-now">What I do differently now</h2><p>A few things that have survived contact with reality:</p><ul> <li><strong>Assume the cable is the weak point, and make it replaceable in the field.</strong>A connectorised run that a cold, tired person can swap in ten minutes beatsan elegant sealed one that requires a bench.</li> <li><strong>Go wireless wherever possible, it is surprising people don’t appreciate LoRA more</strong></li> <li><strong>Armour the runs that touch the ground.</strong> Conduit is heavy to carry in andcheaper than a lost month.</li> <li><strong>Log locally, always.</strong> A node that keeps writing to its own storage whenthe uplink drops turns a data loss into a data delay. This one change hasrecovered more of my dataset than any other.</li> <li><strong>Design the enclosure as habitat.</strong> Not to keep animals out — you will losethat fight — but so that when something moves in, it cannot reach anythingload-bearing.</li></ul><h2 id="the-part-i-did-not-expect">The part I did not expect</h2><p>The maintenance cycle changed how I read my own data. When you personallyreplaced the connector on node four in March, you look at the Marchdiscontinuity in node four differently than if you had inherited the timeseries as a CSV.</p><p>I would not describe hiking a battery bank up a hill as a <em>research method</em>.But knowing exactly which of my measurements I trust, and why, comes almostentirely from having been the one standing there when it broke.</p> 2026-03-02T00:00:00+00:00 https://shadmanamin.github.io/blog/2026/03/02/tesla-oatmeal/

Nikola Tesla was punk before it was cool.

<iframe src="https://theoatmeal.com/comics/tesla" width="100%" height="500px"></iframe>