Skip to content

Norway: NVE

Documentation index · Usage
Provider no_nve
Country Norway
Published by Norges vassdrags- og energidirektorat (NVE), the Norwegian Water Resources and Energy Directorate
Quantities Discharge, stage, water temperature
Stations in the catalogue 4,902 locations; 3,804 have catalogued series for the supported quantities and resolutions
Station metadata Station name, water-body name, drainage area and elevation. See Fields by provider
Credentials Required for retrieval: NVE_API_KEY; see Credentials and requests
Licence stated by NVE Norwegian Licence for Open Government Data (NLOD)
Agency documentation HydAPI documentation

Set NVE_API_KEY in the process environment or a private .env file in the directory from which Python runs, using a key obtained from NVE. Catalogue browsing works without it. Retrieve one week of published daily mean discharge at station 2.605.0:

import rivretrieve as rr

selection = rr.find(
    provider="no_nve",
    station="2.605.0",
    quantity="discharge",
    frequency="daily",
    statistic="mean",
)

result = rr.fetch(selection, start="2024-01-01", end="2024-01-07", cache="bypass")

preview = result.data.select("time", "time_zone", "value", "unit").head(3)
print(preview.write_csv(float_precision=3), end="")

print(result.data.height)
for issue in result.issues:
    print(f"{issue.severity}: {issue.message}")

Output:

time,time_zone,value,unit
2024-01-01T11:00:00.000000,+00:00,401.750,m3/s
2024-01-02T11:00:00.000000,+00:00,415.396,m3/s
2024-01-03T11:00:00.000000,+00:00,444.500,m3/s
7
info: no_nve published source quality code 2 without RivRetrieve interpretation
info: no_nve published source correction code 0 without RivRetrieve interpretation

The request returned seven daily means in m³/s. The preview shows three values rounded to three decimal places. Both endpoint dates are included. RivRetrieve reads the published means; it does not calculate them from more frequent observations. The two informational messages report source metadata, not retrieval problems. NVE calls quality code 2 PrimaryControlled and correction code 0 No changes. These are NVE’s descriptions of the observations, not a RivRetrieve quality judgement. They are separate from the series version, which is 1 in this example.

cache="bypass" requests HydAPI rather than a local cache. This output was checked on 2026-09-23. NVE can revise historical observations, so later retrievals may return different values. Run the examples in order in the same Python session. See Usage for general selection and result handling.

Who measures, and who publishes

NVE is responsible for Norway's national hydrological observation networks and databases. It operates monitoring stations and publishes historical and current observations through HydAPI, its programming interface for hydrological data. RivRetrieve reads HydAPI.

NVE does not own or produce every measurement in that archive. The network also includes stations owned by river-regulation operators. Owners of regulation facilities carry out and fund hydrological investigations required by NVE; NVE checks and archives the mandated data. Keep that distinction when describing the origin of a station's record.

NVE may withhold recent observations from public access, including in regulated catchments and at hydropower installations. Its publication guidelines specify a 14-day withholding period for stage and discharge where those restrictions apply. The station owner may permit earlier publication. Owners of privately collected data that are not required by a public authority can decide what to publish. HydAPI access does not promise an unrestricted real-time record at every station.

Credentials and requests

NVE describes HydAPI as a free service. Request a key at hydapi.nve.no/Users, entering an email address and accepting the terms. Store the key securely: NVE says it cannot be retrieved later. Set NVE_API_KEY in the process environment or a private .env file in the directory from which Python runs. Process variables take precedence, including blank values. Keep credentials out of code, version control, logs and shared files. See credential configuration for details. rr.providers() reports ready when the credential is present locally; it does not authenticate with NVE or prove that the key is accepted.

NVE limits requests per key and the number of observations in a response. Its user documentation does not give numeric limits. RivRetrieve sends separate observation requests for each station, quantity, resolution and version. It requests a slightly wider period than selected, then clips the returned rows. For Norway, it does not automatically subdivide that period to meet the source's observation-count limit. If NVE reports that a response is too large, request shorter periods, as its documentation advises. Check the returned issues rather than treating a failed request as missing observations.

What you can retrieve

Quantity filter Source field / variant Frequency Statistic Source unit Returned unit
discharge Raw resolution Unknown Mean or instantaneous, depending on series m³/s m³/s
discharge Hourly resolution Hourly Mean or instantaneous, depending on series m³/s m³/s
discharge Daily resolution Daily Mean or instantaneous, depending on series m³/s m³/s
stage Raw resolution Unknown Mean or instantaneous, depending on series m m
stage Hourly resolution Hourly Mean or instantaneous, depending on series m m
stage Daily resolution Daily Mean or instantaneous, depending on series m m
temperature Raw resolution Unknown Mean or instantaneous, depending on series °C °C
temperature Hourly resolution Hourly Mean or instantaneous, depending on series °C °C
temperature Daily resolution Daily Mean or instantaneous, depending on series °C °C

These units require no numerical scaling. Stage is water level; RivRetrieve has not established its vertical reference or datum.

HydAPI publishes raw, hourly and daily resolutions. RivRetrieve establishes frequency="hourly" or frequency="daily" from the published resolution. For raw resolution, frequency remains unknown. The source method establishes statistic="mean" or statistic="instantaneous" separately. An hourly or daily series can contain instantaneous values; its resolution alone does not establish an averaging operation. Use both frequency and statistic filters for daily means, as in the example. RivRetrieve does not derive additional statistics.

Of the 4,902 catalogue locations, 3,804 list series for the supported quantities and resolutions. The other 1,098 list other parameters, such as groundwater level and temperature, air temperature, precipitation or snow. Their presence in the catalogue does not make those parameters retrievable through RivRetrieve.

The catalogue contains daily mean discharge at 1,620 stations, daily mean stage at 2,790, and daily mean temperature at 1,153. These counts can overlap and describe the packaged snapshot, not an exhaustive current inventory. A station being listed does not guarantee observations for every quantity or requested period. A successful short request does not establish continuous historical coverage.

Source versions

HydAPI can publish several versions of a station's series. variant is the published version number, represented as a string. Version numbers identify records at a particular station and quantity; they are not global quality codes. RivRetrieve keeps those records separate. A larger number is not a quality ranking, and no universal meaning for each version number has been established. Without an explicit version choice, retrieval checks current source metadata and requests all versions matching the selection's physical filters. Each version can return values, an empty record, or a separate failure.

The example returned version 1 at station 2.605.0. Another station, 109.42.0, has three catalogued versions of daily mean discharge:

Station Quantity variant Frequency Statistic
109.42.0 Discharge 1 Daily Mean
109.42.0 Discharge 2 Daily Mean
109.42.0 Discharge 3 Daily Mean

These numbers identify versions at this station, not quality levels. Inspect them before making a version-specific request:

print(rr.series(result).select("station_id", "variant").rows())

other_station = rr.find(
    provider="no_nve", station="109.42.0", quantity="discharge",
    frequency="daily", statistic="mean",
)
print(rr.series(other_station).select("station_id", "variant").sort("variant").rows())

version_two = rr.pick(other_station, variant="2")
print(rr.series(version_two).select("station_id", "variant").rows())

Output:

[('2.605.0', '1')]
[('109.42.0', '1'), ('109.42.0', '2'), ('109.42.0', '3')]
[('109.42.0', '2')]

The second and third lines describe catalogue candidates, not newly retrieved observations. pick narrows the selection; it does not make an observation request. Passing version_two to fetch requests only that station's version 2, without substituting another version when observations are unavailable. This choice does not select quality code 2 or assert that this version is preferable.

Time and data status

HydAPI timestamps carry explicit UTC labels. RivRetrieve returns their wall-clock values in time, paired with time_zone="+00:00". Read both columns together: the datetime column itself is timezone-naive. The daily values above are labelled at 11:00 UTC. That clock label does not establish the averaging interval or its start and end. NVE's documentation gives conflicting wording for the daily time basis; RivRetrieve leaves exact daily boundaries and interval anchors unknown.

HydAPI observations carry quality and correction codes. RivRetrieve does not filter observations by those codes, interpret them as a quality ranking, or add per-row quality columns to result.data. Informational issues summarize the published codes, their counts and first and last source timestamps. Those summaries can include the extra dates requested around the selected period, so their counts need not equal the returned row count. Optional receipts retain the observation response bytes when the original per-row codes are needed.

A published null remains a row with value=null. An absent observation supplies no row. A failed request remains an issue and an identified outcome alongside independent successful records. Check issues before interpreting gaps or empty results. Neither a returned number nor the absence of retrieval errors establishes the source's approval of its quality.

Terms and citation

NVE's HydAPI documentation licenses the data under the Norwegian Licence for Open Government Data (NLOD). It states that the data are provided "as is" and may contain errors or omissions. The verified observation response includes the licence URL in its JSON metadata.

NVE asks:

When using data from this service, if possible, please refer to this service as origin of data.

The linked NLOD 2.0 terms permit reuse, including commercial reuse, subject to their conditions. They require attribution and a licence reference, links to the source and licence where practical, and a clear indication of changes. Where the licensor has not specified attribution wording, the licence gives:

Contains data under the Norwegian licence for Open Government data (NLOD) distributed by [name of licensor].

Identify NVE HydAPI as the publication service, and retain the station, quantity, version, requested period and access date to make the retrieved record traceable. These record details are practical citation guidance, not an official NVE citation template. NLOD's compatibility provisions distinguish databases from other material; do not assume an unqualified equivalence with CC BY 3.0.

Sources

Publisher pages checked on 2026-09-23:

The station counts describe the packaged catalogue. Live public retrieval and the recording path were checked on 2026-09-23. The private verification archive retains commands, source checks, exact output and the limits of these checks.