How to set a shoreline setback line from historical erosion rates

A setback line only works if the erosion rate behind it holds up to scrutiny. Get the rate wrong, or base it on three shoreline positions instead of thirty years of them, and the line either overprotects land that was never really at risk or lets a structure get sited where the shoreline will be in twenty years. Here's the method most coastal programs use, and where it tends to break down.
Start with the erosion rate, not the setback distance
A setback distance is just an erosion rate multiplied by a planning horizon, plus whatever buffer your jurisdiction adds for dune or bluff instability. So the real work is the rate.
The standard approach is end-point rate or linear regression across a series of historical shoreline positions, usually drawn from aerial photography, LIDAR, or satellite imagery going back as far as the record allows. More shoreline positions and a longer time span give you a more stable number. A rate built from two surveys taken a decade apart can swing wildly if one of those years happened to follow a storm that cut the beach back 15 meters that didn't come back. A rate built from fifteen or twenty annual positions averages that noise out.
This is the part that trips up a lot of setback studies: the historical record is usually patchy. Aerial photo flights happen on whatever schedule a county could afford, so you might have good coverage for 1985 to 1995, a nine-year gap, then two more flights before the record goes quiet again. Digitizing a waterline from a 1988 aerial photo scanned at who-knows-what resolution introduces its own error, and that error compounds when you're trying to extract a trend from six or eight data points spread unevenly over forty years.
Build the setback from three numbers
Once you have a defensible long-term rate, most codes combine three things into the final setback distance:
The erosion rate itself, expressed as meters or feet per year, averaged over the longest reliable period you have.
The planning horizon, which is a policy choice, not a measurement. Jurisdictions that use a 30-year horizon are accepting more risk at the shoreline than ones using 60 or 100 years. This number usually comes from your enabling statute or local ordinance, not from the data.
A stability or safety factor layered on top, often a fixed addition for dune escarpment risk, bluff retreat, or storm-induced erosion that the long-term trend doesn't capture on its own.
Multiply the rate by the horizon, add the stability factor, and you've got your setback distance from the reference line (typically the most landward position of the toe of dune or line of vegetation, not the current waterline). That distance then gets applied against a current baseline to draw the actual setback line on a map, segment by segment, because erosion rate is never uniform along a stretch of coast. A shoreline next to a tidal inlet can retreat ten times faster than one two kilometers down the beach.
Why consistency matters more than precision
Most setback disputes come down to one question: was the shoreline position in year 1 measured the same way as the shoreline position in year 20. If one came from a photo-interpreted waterline and another from a different analyst's digitizing judgment call on a different tide stage, the "erosion rate" partly reflects measurement drift, not actual land loss. That detail is what holds up in a hearing when a property owner challenges the setback: can you show the waterline was pulled the same way, at the same tidal datum, every single year.
That's the gap an annual, consistently extracted waterline layer closes. Shoreline Change pulls the waterline from a high-resolution satellite pass every year, segment by segment, so the retreat or accretion rate behind your setback line comes from a record built the same way each time instead of a stitched-together archive of whatever imagery happened to exist.
If your setback study is due for an update and the historical record has more gaps than data points, it's worth seeing what a clean annual baseline would do for the rate.