How we count

01Sea lice and the legal limit

Source. Farms report lice counts weekly to the Norwegian Food Safety Authority (Mattilsynet). We read its open API (akvakultur-offentlig-api), the same data BarentsWatch shows. The counts are self-reported by the farms and not independently verified.

Coverage. Complete from ISO week 40 of 2023. Earlier weeks need access to BarentsWatch, which we don't have yet.

Corrections. A farm can submit several reports for the same week. We keep the latest. Weeks stay open for corrections, so we re-read the last weeks every night.

The legal limit (Regulation on combating salmon lice § 8, unchanged since 6 March 2017) depends on the county the farm is in, not latitude or production area:

• Before 6 March 2017 the limit was 0.5 all year.
• If we know neither a farm's county nor its position, its spring limit is unknown and the week is not counted as over or under. Farms deleted from the register are placed using the Directorate of Fisheries' layer of deleted sites.
• The rule says “fewer than” the limit, so a count equal to the limit is over it.
• Slaughter holding cages are exempt. Broodstock sites may have an individual, higher limit that is not published. We flag them and say that the limit may differ.

02Two measures of exceedance

• Above the legal limit: the weekly average is at or above that week's limit.
• Enforcement practice: in the 0.2 spring weeks, the Food Safety Authority treats one count of 0.3 or more, or three successive counts between 0.2 and 0.3, as a real exceedance. Outside spring it equals the legal measure.

Since July 2026 the Food Safety Authority follows up repeated or prolonged exceedances rather than single ones, so we also count long exceedances (three or more consecutive weeks over) and the longest streak.

03When we write “unknown”

We never fill in missing numbers. A week without a report is shown as unknown, never as “below the limit” and never as 0. Farms must count every week at 4 °C or above, but only every 14 days below 4 °C, and not at all if the site will be emptied within 14 days, so a missing week is not necessarily a breach either. Unknown weeks count neither over nor under.

04Company and group rankings

• Who gets the week. Each weekly report carries the organisation number of the company that submitted it. A farm-week is credited to that company, so ownership changes are handled correctly.
• Groups (Mowi, SalMar, Lerøy …): no open register publishes ownership trees. We assign groups by name rules plus a hand-kept list of overrides. Only well-known, unambiguous brands are included.
• Fairness. Large companies have more farms, so we always show the share of weeks over the limit next to the count, and the number of counted weeks. Companies with fewer than 52 counted weeks in a year get no rank (too little data for a fair share) but are still listed.
• Like with like. Rankings include only farms holding a commercial (or development) licence for salmon and trout food fish. Research, teaching, display and broodstock sites (e.g. the Institute of Marine Research, Aquagen) are left out by default: they have other purposes and may have individual lice limits. Farms missing from the current register count as commercial.
• Rankings start in 2024, the first full year of open data with organisation numbers. Lice history before autumn 2023 has no organisation numbers and is shown on farm pages only.
• Reports without a lice count are counted separately. Weeks with no report at all are not in the company table (see the map and farm pages).

05Treatments

From the same weekly reports: medicinal (bath and in-feed, with active substance), non-medicinal (mechanical, thermal, freshwater, other) and combinations. A combination treatment (e.g. freshwater plus mechanical delousing) counts as one treatment; its steps still count towards each method. One entry can cover several pens.

06Disease

Source. The Food Safety Authority's register of listed diseases (ISA, PD, BKD, francisellosis and systemic Flavobacterium psychrophilum). Complete from 26 November 2024, when the register went live; older cases appear only if they were still open then. Suspicions that were ruled out are kept but marked and left out of all counts. Cases start on the quality-assured suspicion date, else the diagnosis date, else the notification date. Dates are in Norwegian local time.

Which company. A case is credited to the company that reported lice at the farm in the 26 weeks before the case started. If there is none, we leave it unattributed rather than guess.

07Escapes

The Directorate of Fisheries' escape register, published with a delay of about two months. Numbers are the companies' own estimates; many are given as bands (“1–10”), which we show as text next to exact counts.

08Wellboat visits (derived by us)

There is no open list of wellboat visits, so we derive them from the Coastal Administration's open AIS track archive:

1. Vessels. Every vessel registered as “Fish Carrier” in Kystdatahuset, refreshed weekly. Vessels under 25 m are flagged as not wellboats. The category also includes some harvest and transport carriers, and there is no open flag that separates them, so we call them “wellboats (fish carriers)” and keep a manual list of known harvest boats.
2. Stops. AIS track segments (minutes to hours long) with an average speed under 0.4 knots. Speed rather than distance, because a long stop accumulates drift.
3. At a farm. The stop's centre lies within 500 m of the farm's registered position (nearest sea farm wins). Farms can span several hundred metres, so a smaller radius misses visits; a larger one risks attributing a stop to a neighbour.
4. Visit. Consecutive stops at the same farm with gaps under 60 minutes, lasting at least 30 minutes. The archive cuts tracks at midnight UTC, often leaving a 1–3 hour hole, so across midnight we allow gaps of up to 3 hours.

Precision. Arrival and departure times are accurate to roughly an hour. The AIS archive lags about 17 days and has occasional gaps (e.g. early June 2025). BarentsWatch runs its own visit analysis; once we have access, we will compare the two.

09Wellboats and disease outbreaks

This shows association, not transmission. It answers: how often was a wellboat that had recently been at an infected site seen at a farm shortly before that farm's outbreak, compared with ordinary farm-weeks?

• A visit is from an infected site if, in the 14 days before, the same vessel had a visit at another farm with an active ISA or PD case, or inside an active ISA or PD protection zone. Surveillance zones are left out: they are so large that a third of all visits would qualify.
• An outbreak is exposed if its farm had such a visit (same disease) in the 28 days before the case started.
• The baseline is the same measure for every farm-week at salmon and trout farms without an active case of that disease, leaving out the 28 days before each outbreak. Cases are also limited to those farms, so both groups come from the same population.
• We report both shares with 95% Wilson confidence intervals, and their ratio. Case counts are small, so intervals are wide: say “roughly N times as common” only with the interval next to it.
• Only the period where both datasets are complete is used: from December 2024, when disease data starts, plus 28 days.

Confounders we do not control for: farms near other infected farms get more visits from boats working in the area, larger farms get more visits, and treatment trips increase when a farm already has problems.

10Municipalities

Aquaculture Fund payouts come from the Directorate of Fisheries' yearly spreadsheets. Payouts made to municipalities that later merged are summed into today's municipality (following Statistics Norway's code changes). The few that later split cannot be divided and are left out. River catches are from Statistics Norway table 08991, placed in the municipality of the river mouth.

11Sources

12Licence and attribution

Our figures and tables can be reused freely under CC BY 4.0. Credit “Source: The Salmon Ledger (Lakseregnskapet)” and the original source. The raw data is licensed by the agencies as shown below; the attribution also applies when you use our downloads.

Downloads and feeds are on the data page.