On agency bridge work there is usually a scour evaluation, a model, and a survey. On most other erosion work there is a photograph and an owner who wants a number.
That second case is normal and it is workable. What makes it workable is being explicit about what is estimated.
The substitutes, in descending order of confidence
A calibrated hydraulic model. The best available. See taking a design velocity from a model.
An uncalibrated model. Still much better than nothing; the geometry and hydrology are real even if the roughness is assumed.
A gauged record. Where a gauge exists on the reach, discharge is known and velocity can be derived from the section geometry.
Regional methods. Regression relationships for discharge from catchment characteristics, where they exist for your region. Wide uncertainty bands, and a legitimate starting point.
Documented professional judgement. An engineer’s estimate from experience of comparable sites, written down as such.
Observation. Debris lines and high-water marks establish the stage a past event reached, which with the section geometry gives a discharge and a velocity. Free, and frequently the most site-specific information available. See post-flood scour inspection.
The bank covers what information can be used when measured current velocities are unavailable, and the practical position is that all of the above are usable bases.
What is not a basis
A number with no stated origin.
The failure mode is not that an estimate was used. It is that an estimate got written into a drawing, then quoted in a specification, then relied on by a reviewer, with the word “estimated” dropped somewhere along the way.
That is the same failure this site’s no-invented-numbers policy exists to prevent on the product side — see which test standards apply — and it applies equally to site data.
What is observable without instruments
More than people expect, and it is cheap:
Debris lines and high-water marks, immediately after an event. Perishable and specific to your site.
Historical aerial imagery. Migration rate over decades, free, and the single most informative input on a bend. See outer bend erosion.
The maintenance record. How often this reach has been repaired, and what it cost. The most useful data an owner already has, and it informs both design and the lifecycle case. See lifecycle cost vs first cost.
Photographs over time from a fixed point, which convert an impression of rapid loss into a measured rate.
Bed material, by looking and by a sample.
Probing, which distinguishes consolidated bed from loose post-event infill and is worth more than looking at the water.
Bank signatures, which identify the mechanism. Tension cracks mean stability rather than erosion, and that changes everything. See how to read a failing bank.
That last one is the highest-value observation on the list. Establishing the mechanism correctly matters more than getting the velocity right, because armor on a mass stability failure is money spent on the wrong problem.
The margin that goes with an estimate
More uncertainty means more margin, and it should be declared rather than buried. See safety factors in armor sizing.
Spend it where it addresses the dominant failure modes: extent first, toe second, unit size last. An estimated design with generous extent and a well-keyed toe is more robust than one with a larger unit on a footprint that stops at the visible damage.
And prefer the arrangement that tolerates being wrong: units that can be added at a boundary turn an under-estimate into a maintenance item. See repair and unit replacement.
The record that keeps it honest
Whatever the basis, write down:
- Each input, its value, and where it came from
- Which are measured, which are modelled, which are estimated
- The margin applied, and to what
- What would change the design
- What monitoring will confirm or correct the assumption
That record is what lets the design be revised later without it looking like a failure, and it is what a reviewer needs. See assembling the submittal package and budgeting at concept stage.
Monitoring as the correction
An estimated design is a hypothesis, and the baseline survey is how you test it.
Establish the datum at completion, inspect after the first significant event, and compare. If the bed went deeper than predicted, you find out while it is still a maintenance question. See post-installation inspection.
What we can do at this stage
Send photographs, rough dimensions, water levels, and an access description through the quote form and you get a size, a quantity, and a layout approach back within one business day, at no charge, including at feasibility stage with no funding in place.
The output is a shortlist, not a design. Final selection belongs to the engineer of record, and where our data does not support a case, when not to use rock bags says so.