Methodology

Why this page is the point

The Alaska salmon fishery is a small community that will notice any misuse of ADF&G's data. So this page spells out exactly where every number comes from, how the scoreboard is computed, and what we leave off. The honesty is the moat.

The three sources

Everything comes from ADF&G, all public. The live harvest is the Blue Sheet, ADF&G's preliminary in-season commercial salmon catch report, the cumulative harvest of the season in thousands of fish, by region, area, district and species, refreshed twice daily from mid-May through September. The benchmark is the April forecast, ADF&G's Run Forecasts and Harvest Projections report, Table 1, the projected commercial harvest by area and species. The value data is ADF&G's preliminary season-summary tables, the annual ex-vessel price and value by area and species.

The scoreboard math

The scoreboard is one calculation: the cumulative harvest divided by the preseason forecast, as a percent, per area and species. Both sides are in thousands of fish, so the percent is a clean pace read. It is descriptive, not a good-or-bad judgment, and not a forecast: a run at 60 percent of forecast in early July is on pace, not weak, because the run is still coming. We show the percent only where both a harvest number and a forecast number exist.

Reading the harvest against the forecast

The Blue Sheet reports harvest down to the district, while the April forecast reports by management area, so we roll the harvest up to the area the forecast uses and compare the two at that grain. Bristol Bay's five districts become one Bristol Bay line, matched to the Bristol Bay forecast; Prince William Sound's natural and hatchery projections are summed to one forecast; and so on. An area's all-species total forecast is ADF&G's own printed total where the report prints one, otherwise the sum of the species projections, which can differ from a printed total by a rounding unit, as ADF&G notes rows may not total exactly.

Confidential cells and the null rule

When too few permit holders fish an area to publish a number without revealing an individual business, ADF&G suppresses the cell, and that catch is left out of the cumulative totals at source. On this site a blank is a blank: a fishery that has not reported, has no harvest, or has a suppressed cell. We never fill a blank with a zero, and we never sum around a suppression to reconstruct a confidential number.

The effective dates

Each number is dated to what it reports. The Blue Sheet is a live cumulative snapshot with no printed report date, refreshed twice daily, so a harvest reading is dated to the day it was pulled, the report date. The forecast is dated to its April publication. Each ex-vessel price is dated to the season it reports. A backfill of an old season therefore reads as that season, never as today's news.

The price lag, stated honestly

ADF&G does not publish ex-vessel prices in-season. It publishes preliminary annual finals months after the season, and postseason adjustments and bonuses can land later still. So the price on any page is the last published season's number, labeled as a look back, never presented as a current in-season price. Where ADF&G has published no price, we show none. We never invent a price.

Areas and districts

ADF&G calls the levels the scoreboard is built at management AREAS: Bristol Bay, Prince William Sound, Southeast and the rest. The level below is the DISTRICT. In Bristol Bay the five districts, Naknek-Kvichak, Egegik, Nushagak, Ugashik and Togiak, are what the fleet watches every morning at the peak, so the Bristol Bay page carries a by-district drill-down of cumulative harvest, read straight off the Blue Sheet's district rows. A district that has not reported shows a blank, never a zero.

Escapement, and why harvest is not the whole run

This site tracks the commercial HARVEST against the harvest forecast. It does not yet track ESCAPEMENT, the salmon that pass upriver to spawn. That matters, because ADF&G manages fisheries to escapement goals, and the total run is harvest plus escapement, so a harvest-only percent of forecast can understate a run where escapement is front-loaded, a lot of fish let up the rivers early before the fishery takes its share. ADF&G publishes daily escapement counts by river in season; folding them into a Bristol Bay total-run read is a planned addition. Until then, read the percent of forecast as a harvest pace, not a run-strength verdict, and see the glossary entry for escapement.

Run-timing versus normal, and why we do not show a pace-vs-history number yet

We tried to build the most useful in-season read, this year's pace against a normal year at the same point in the run, and came up honestly short. Building it needs several prior seasons of cumulative harvest at the same date. ADF&G's own multi-year timing charts sit behind a dashboard we cannot cleanly automate, so we reconstructed what we could from the Internet Archive's Wayback Machine, which has captured this same ADF&G Blue Sheet page across past seasons. The result: coverage is too sparse and badly timed to support a reliable typical-pace baseline. The captures cluster in August and September, after most of the fishing, and the late-June-into-July Bristol Bay peak, the exact window that matters, is almost never captured. No prior season has two captures close enough on either side of early July to pin the pace there without drawing a straight line across the whole peak. So we do not show a pace-versus-typical number, and we will not interpolate one across the peak. Why a straight line is the wrong tool: a salmon run is pulsed, not linear. It sits near zero for weeks, explodes over a handful of days at the peak, then tails off, so the fraction of the season that is in by a given date is a steep S-curve, not a ramp. Reading pace by calendar date against a real historical S-curve is meaningful; interpolating a straight line across a fifty-day gap that spans the entire peak is not, and it would invent the false precision this site exists to avoid. We keep building our own dated history each season, and a versus-normal read will appear once the coverage genuinely supports one.

What is not here yet

CFEC permit-value data, the price of a Bristol Bay or a Southeast permit, sits behind the same barrier as ADF&G's other dashboards, so it is not yet wired in. Escapement and a Bristol Bay total-run read are the other planned additions. When any of these can be automated cleanly it will get its own place. We would rather leave a thing out than scrape it unreliably.

What we do not compute

We do not forecast runs, we carry no price model, and we never tell a reader where to fish, when to sell, or what a permit is worth. Where a public number does not exist we leave it blank rather than estimate it. Charts and figures on this site are free to republish with visible attribution and a link back.

For the plain-English version of how to read a page, see how to read this.