Every overflow on a sewer network is reported to Miljødirektoratet each year, in Altinn, by 1 March: how many times it ran, for how many hours, how much water went out and how much phosphorus with it. The directorate publishes an import sheet with exactly those columns (Miljødirektoratet). In August 2026 the Statsforvalter for Østfold, Buskerud, Oslo and Akershus wrote, after reading 111 reports, that overflow reporting is still deficient in several municipalities, and that many still have no flow meter at their overflows (Statsforvalteren).
Many overflows without a meter still have a level sensor or an on/off signal, and the municipality knows the height and length of the overflow edge from a drawing or its network model. From those, Auge counts each overflow’s events and hours, works out the volume over the edge, and writes the directorate’s sheet, ready to import. Each figure says how it was found: measured, or an estimate from a level and an edge. Where something cannot be known, the cell is left empty with the reason: a level with no edge height, an on/off signal with no capacity, a year with less than half its readings. The phosphorus uses the directorate’s own typical value for a diluted overflow, with its range beside it. The same tool writes the yearly report’s import file with each sample day’s water volumes taken from the readings, so nobody types them across.
We tested the estimate before trusting it, on an overflow in Graz where a meter in the overflow channel measured what went out, over 844 days from 2008 to 2011. From the chamber level and the edge in the utility’s own model alone, it found 112 of 118 overflow events, with 2 it should not have. The hours came to 155 against 166 measured. A typical event’s volume was off by a factor of 1.5, against 2.2 for a fixed rate times the hours.
But the volume was low: 20% in all, and low in each of the 4 years, by up to 35%. We did not correct it afterwards, because a factor fitted to this overflow would only fit this overflow. So the register calls such a volume an estimate, and says how it did here. It is one overflow in one city; more are needed before anyone leans on the volume.
P59, written down before the run: SUPPORTED.