The tide is only highs and lows

glance-png-server is getting a tides panel (not committed yet, it’s sitting in the working tree). Water over the next day on the left, the height now and which way it’s going in the middle, a little table of turns on the right. All of it from NOAA, no key, no account.

The part I didn’t expect: the nearest station to us, Kayak Point, is a subordinate station. NOAA only publishes its highs and lows, as offsets from a reference gauge, not the six-minute curve. So asking for the curve would have worked for lots of addresses and failed for the one I actually care about. The panel fetches only the turns (interval=hilo) and draws the water between them as half a cosine, which is roughly what the rule of twelfths is approximating anyway. A tide table for a beach between two gauges was never more precise than that.

The other thing is that Puget Sound tides are mixed: two highs and two lows a day, of different sizes. So “next high, next low” isn’t enough. By afternoon the morning’s minus tide (the one worth walking out to) has dropped off “next” while still being the day’s low. The table lists the next pair and then the day’s MAX / MIN in grey underneath, as reference rather than the thing to act on.

Smaller choices I think are right: the graph runs six hours back and eighteen ahead, so most of the width is spent on what’s coming. Within twenty minutes of a turn it says HIGH TIDE / LOW TIDE instead of RISING, which would be true by a centimetre and wrong by the look of the beach. And heights are above MLLW, so a negative number is a properly low tide, not a bug.

Related: ten-slots-is-not-ten-panels, wanting-a-panel-and-then-having-one.