CSAT: The Dredging Record Inside the Forecast

Maintenance changes the measured trend

An owner forecasting the next dredging campaign needs a history of the bed and a history of the work done to it. A deeper sounding after maintenance contains the contractor's excavation. Treating that change as natural sediment movement can reduce the forecast rate of shoaling and leave the next campaign short of volume, time or funding.

USACE's Corps Shoaling Analysis Tool, CSAT, uses historical surveys to forecast channel maintenance requirements. The May 2024 programme storyboard describes its use in budget justification and district planning, with applications in Matagorda, Corpus Christi, Houston, Calcasieu and Savannah. The operating question is how much sediment will occupy a particular channel reach when the dredge arrives. ^[2]

A 2024 study by USACE researchers tested the effect of adding dredge-position records to maintenance-event identification. Across 684 reaches with maintained depths of at least 20 feet in seven districts, 358 contained an additional GPS-indicated dredging event missed by the legacy method. The analysed surveys fell within January 2012-November 2023. The current product page lists version 2.7.0 and an October 2026 software update. [1,3]

Four surveys with one maintenance event

The following IMWO arithmetic example isolates the effect of recognising excavation. It assumes a uniformly changing 100,000-square-foot area, four annual surveys on one fixed vertical datum, and maintenance between Years 1 and 2. Bed elevations are positive upwards. Natural deposition raises the bed by one foot in each undisturbed year.

The calculation uses the duration-weighted survey-pair averaging described in the paper. For each admitted pair, elevation change is divided by elapsed years. Its weight is that elapsed period. Summing each rate multiplied by its weight and dividing by the sum of weights gives the forecast rate. This example explicitly sets the maintenance flag; the arithmetic omits the software's automatic event-detection tests. ^[1]

With every possible pair admitted, six pairs contribute. Their elevation changes sum to minus six feet and their observation durations sum to ten years. The resulting rate is minus 0.6 feet per year. Recognising maintenance removes the four pairs spanning the excavation. The remaining Year 0-Year 1 and Year 2-Year 3 pairs each show one foot of deposition over one year, producing a rate of plus 1.0 foot per year.

The forecast changes the quantity

Starting at the final survey elevation of minus 41 feet, a two-year projection gives minus 42.2 feet with the unfiltered rate and minus 39 feet with maintenance recognised. For a required bed elevation of minus 40 feet, the first calculation forecasts zero material above grade. The second forecasts a one-foot layer over 100,000 square feet: 100,000 cubic feet, or 3,704 cubic yards rounded. These are illustrative inputs and results, with no allowance for side slopes, overdepth, mobilisation or production losses.

The numerical difference enters an estimate before a unit rate is applied. A planner using the zero-volume result may leave the reach out of a campaign or allocate too few dredging days. Recognising maintenance restores a positive requirement in this example. A real estimate then needs current survey coverage, the target prism, the material route and achievable production. Each changes the vessel time and expenditure associated with the forecast.

A single average also conceals the location of a shoal. Two reaches with the same total forecast quantity can require different campaigns if one contains a short controlling high spot and the other has a thin deposit over a long corridor. The first may justify targeted early removal; the second may favour a continuous production run. The forecast should retain the spatial distribution that supports that choice.

Retain the observations outside the cut

The 2024 paper identifies a further complication: treating any dredging within a reach as an event for every grid cell can discard valid observations in untouched areas. Its two-cell Whitehall Reach example compares excavation effects at different locations within one reach. The authors call for wider validation of cell-level methods. They also report that 40% of GPS-identified events missed by the legacy method coincided with a net sediment gain across the reach. ^[1]

That combination has a straightforward physical explanation. Excavation can deepen one part of the channel while deposition raises another part by a greater volume. A whole-reach net change then hides the cut. Reducing a single net-volume trigger can capture more excavation events, but it also admits more natural scour as apparent maintenance. A dated work footprint offers a more specific way to distinguish the two.

For an owner commissioning a forecast, the deliverable should show which survey pairs were retained at the controlling shoals, which were removed, and the work record supporting each removal. That small audit trail lets an engineer investigate an unexpectedly low rate. It also exposes a reach where sparse surveys leave the answer dependent on one short observation period.

Turn the estimate into a campaign plan

USACE's public directory provides district input-file links. Those datasets need to be matched to the project area and survey dates used in the estimate. The Detroit District training record documents an emphasis on irregular observation periods and project-specific configurations, illustrating the engineering work required between obtaining data and using a forecast. [4,5]

IMWO proposes a forecast handover with three connected records. The survey record carries acquisition date, coverage, vertical datum and uncertainty. The maintenance record carries work dates and the excavated footprint. The campaign estimate carries forecast horizon, target grade, volume and the production assumptions that convert it to vessel days. The owner's engineer should reconcile all three before freezing the dredging allowance.

Test the estimate against a later survey withheld from the rate calculation, concentrating on the controlling parts of the channel. Record the forecast error in elevation and in material above the intended grade. If local scour, a storm or a revised channel geometry dominates the interval, use that condition in the next forecast review. A model calibrated to a different geometry or sediment regime needs a fresh engineering basis.

The immediate procurement benefit is a quantity allowance tied to a stated date and a traceable maintenance history. Bidders can then test campaign timing, plant availability and material handling against the same physical starting point. Updating the history before pricing reduces the chance that a missing work record becomes an unexplained quantity change after mobilisation.