Setting elevation band thresholds for basin snowmelt timing

Elevation banding sounds like a modeling exercise until you're staring at a hydrograph three days early because the 2,500-3,000 m band cleared faster than your rule assumed. Getting the band width and the depletion threshold right is less about statistics and more about matching the choice to how your basin sheds snow.

Why band width matters more than people think

Most forecast offices start with 100 m or 200 m bands because that's what the DEM resolution and the old model setup handed down. Fine for a first pass. But a steep, narrow basin with a tight hypsometric curve (most of the area packed into 500 vertical meters) behaves very differently from a broad, gently sloped basin where elevation and area are spread evenly.

In a compressed basin, 100 m bands give you real separation: each band melts out at a distinguishable date, and the sequence tells you something about aspect and radiation load. In a flat, high-plateau basin, 100 m bands often clear within a day or two of each other, so you're carrying extra bands that add noise without adding lead time. For those, 200-300 m bands usually track the melt front just as well and cut your QC workload roughly in half.

Rule of thumb: plot your basin's hypsometric curve first. Where the curve is steep (lots of elevation change over little area), narrow the bands. Where it flattens out, widen them. Don't pick one band width for every basin on your desk.

Setting the depletion threshold

The threshold is the percent-snow-covered-area value you call "melted out" for a given band. Pick it wrong and you either flag melt-out weeks before the band is functionally bare (too high a threshold) or miss the window entirely because a few shaded pockets or north-facing microsites never quite clear (too low).

A threshold in the 10-20% SCA range is a reasonable starting point for most mid-latitude basins. It covers persistent shade patches and old avalanche debris without waiting for every last pixel to go bare, which in some bands never fully happens before the next snowfall resets the clock anyway.

Test it against last season's melt-out dates if you have them. If your 15% threshold call lines up within a day or two of when streamflow actually started responding to that band's melt, keep it. If it's consistently early or late by more than two or three days, nudge the threshold, not the band width. Those are two different knobs and conflating them is the most common mistake I see on this.

Watching the whole stack move

Once band width and threshold are set, the real value is in watching every band deplete in sequence rather than checking one elevation zone at a time. A basin that melts out low-to-high in a tidy progression behaves very differently in your runoff timing than one where a warm spell takes out three bands simultaneously, mid-elevation and high together, and dumps a compressed pulse into the channel network.

That simultaneous-depletion pattern is the one that catches forecast desks off guard, because it doesn't show up if you're only reviewing bands weekly or relying on a single cloud-free satellite pass per month. Daily classification, even coarse 10-30 m snow/no-snow imagery, is enough to catch that two bands crossed threshold on the same day instead of three days apart, which is often the difference between a manageable rise and a flashy one.

This is the part of the workflow Snow Cover Monitoring was built around: a daily snow/no-snow call per basin, tracked by elevation band over the season, so you're watching the depletion curve move in near real time instead of reconstructing it after the fact from a handful of scattered scenes.

Once the band widths and thresholds are dialed in for your basins, feeding them a daily series instead of a monthly snapshot is the next step worth taking.