the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 06 Mar 2026
UK-Flow15 Part 1: Development of a coherent national-scale 15-min flow dataset
Abstract. High-resolution river flow data is essential for modelling flood dynamics and assessing hydrological change. In the UK, there are a wealth of sub-daily flow records that have been collected for more than 70 years. However, they remain scattered across multiple agencies and lack consistent quality assurance, limiting their use for large-sample and national-scale analysis. This paper presents UK-Flow15, a quality-controlled, 15-min, national scale, flow dataset for the UK, based on records from over 1,300 gauging stations and more than 1.8 billion observations. Data were collected via APIs and in conjunction with UK measuring authorities, then compiled into a single national dataset. Duplicate timestamps and inconsistencies in temporal resolution were systematically identified and resolved using a combination of automated filters and manual review. A comprehensive quality-control framework, specifically tailored for the dataset, was then applied to identify and document data anomalies; the full methodological development and evaluation of this framework are presented in the companion paper of this series. The final dataset is accompanied by transparent documentation of flagged issues and a suite of metadata files detailing data resolution, QC outcomes, and all processing decisions to support traceability and user interpretation. We demonstrate the dataset’s utility through a practical case study by systematic removal of unreliable data, interpolation of anomalous spikes, and manual verification of significant high-flow and truncation events. These steps illustrate how the dataset can be effectively curated for detailed hydrological research and operational applications. This publicly accessible, robust, and transparent dataset significantly enhances capabilities for sub-daily hydrological research in the UK, offering essential resources for improved flood prediction, management strategies, and policymaking.
- Preprint
(1423 KB) - Metadata XML
- BibTeX
- EndNote
Status: final response (author comments only)
- RC1: 'Comment on essd-2026-152', Alexander Dolich, 01 Apr 2026
-
RC2: 'Comment on essd-2026-152', Anonymous Referee #2, 24 Jun 2026
The article “UK-Flow15 Part 1: Development of a coherent national-scale 15-minute flow dataset”, submitted by F. Fileni et al. for publication in ESSD, is an original and very valuable contribution for the readers of the journal. This represents a major data curation effort aimed at making available to the scientific community a dataset of river flow time series at a standardised 15-minute time step, derived from the network of operational hydrometric stations operated by various UK agencies. The most significant added value worth highlighting is the high temporal resolution (15 minutes), which was not systematically available in open access, the standardisation of the data quality assessment protocol, the standardisation of the data format, and the ease with which this nationwide dataset can be accessed. This brings several advantages: (i) it is now possible to study flash flood-type hydrological events, which tend to be more frequent and occur at higher latitudes as a result of climate change; (ii) the released dataset is an attempt to comply with the FAIR principles; (iii) this enhances the reliability and robustness of future analyses and results produced from this dataset. In my view, this article merits publication after taking into account some modifications.
General comments
In the introduction, I think that references to other initiatives elsewhere in the world are missing. I am thinking in particular of the Consortium of Universities for the Advancement of Hydrologic Science (CUAHSI), the TERENO infrastructure (https://agupubs.onlinelibrary.wiley.com/doi/10.1029/2024EF004510), the eLTER infrastructure (https://agupubs.onlinelibrary.wiley.com/doi/10.1029/2025EF006743), the Theia-OZCAR portal (https://www.tandfonline.com/doi/full/10.1080/02626667.2020.1764568), and the BDOH (https://www.tandfonline.com/doi/abs/10.1051/lhb/2014005).
Regarding the scope of the dataset, I think it would have been interesting to also include the hydrometric stations managed by academic research laboratories in the UK. I can only see the CEH stations in this study. I am not very familiar with the UK research community, but I am thinking of centres such as the University of Exeter and the University of Aberdeen, for example.
As for the data itself, I feel that information is missing regarding the data curation process carried out by the data producers themselves. Would it be possible to add some further details here? Any information referring to documents that explain how to control the data would be useful. For example, the authors do not specify the type of hydrometric station. I found some of this information in the “supporting document” (Table 10), but this type of information could have been included in the metadata files (00_station_id_meta.csv). Ideally, the genealogy of the versions of the stage-discharge rating curves used to generate the discharge time series would have been included. I recognise that this represents a considerable additional step, but it would have been ideal in terms of reproducibility and for the use of the data in other fields, such as hydrometry. In this regard, information on the station’s sensitivity to changes in the stage-discharge curves would be very useful.
Another issue that is not addressed in the article is the question of intermittence or river flow interruption. Are there any stations where flow cease during certain periods? How were these periods identified, as this remains a sensitive issue?
There is one point that is unclear when reading the article. At one point, the authors state (lines 206–208): “Instead of removing or altering observations, we adopted a flagging approach designed to maximise transparency, minimise subjective decisions, and allow users to select the level of curation appropriate for their needs.” Later in the article, it is stated: “In the dataset processing, to ensure that the data was both comprehensible and suitable for analysis, it was necessary to apply several modifications; however, to retain full transparency, these were documented in dedicated metadata files. These allow users to trace each processing step, enabling informed use of the data and facilitating scrutiny of the resulting dataset.” In the end, I am in the doubt, based on these two statements, whether the dataset has been corrected or not, and whether the erroneous data has been archived somewhere.
Overall, “NA” values are frequently found in the data files. There can be several reasons for these “NA” values: for example, lack of data (gaps) or removal of erroneous data. This type of information could have been tracked in greater detail using codes.
Similarly, in the metadata, “NA” values often appear. Are these solely due to missing data, or are there other possible explanations? For example, the “NA” values in the ‘Split’ column of the file ‘01_common_sense_anomalies_meta.csv’ are not straightforward to interpret.
Overall, there would be significant hydrological benefit in making catchment rainfall data available in the same repository or a supplementary repository. Normally, this data has been used to validate discharge data for the identification of anomalous spikes (see section 5.3). It would be a great added value to include them in the dataset (if they are available) or give an access link.
Specific comments
l.70: It is not the Turia river which was so much flooded but rather the Poyo river. See https://www.investigacionesgeograficas.com/article/view/30056 and https://www.mdpi.com/2073-4441/17/21/3177
l.83-85: you can add other references such as https://doi.org/10.1016/j.ejrh.2026.103491 and https://doi.org/10.1016/j.jhydrol.2024.131267
l.138-140: you can add this reference https://www.nature.com/articles/s44221-024-00351-1
Section 4.1: the format of the data files is not specified
Table 2: the issue related to data gaps do not appear in this table. The absence of flow due to intermittency could also be included in this table.
Table 4: “Station identification metadata”: it would have been useful to include in the description the type of hydrometric station (Weir Flume, Velocity Area, Others, Weir/Flume + VA, Other combined measurement methods), whether the station is subject to changes in the stage-discharge rating curve, and whether the station is subject to intermittency. It would also be useful to provide the topographic catchment area in order to calculate specific discharges and to assess the consistency of discharge from upstream to downstream or between geographically neighbouring stations
l.323–325: are there any bibliographical references to support this assertion?
l.352: what does FOI mean?
l.374–376: I do not understand this sentence
l.396: I do not understand properly this sentence “Spurious flows during timesteps unlikely to affect high-flow analyses”
Figure 6: a), b), c) and d) are missing from the figure
l.463: the term FAIR is never mentioned in the article.
It would be interesting to know the time spent on this work and the programme that funded it. This would benefit from being clarified in the ‘Acknowledgements’ section. Indeed, an analysis covering more than 50,000 years of data spread across more than 1,300 stations, requiring expert visual checks in addition to automated rules, is a very substantial work, not easy to replicate. It seems important to realise the scale of the work involved if others wish to carry out similar work in other countries.
Technical corrections
l.40-41 : the authors wrote “for a specific measurement point in time”. I would rather say “at a specific moment”
l.297: “they follow” instead of “they follows”
l.380: “differentfrom”
Citation: https://doi.org/10.5194/essd-2026-152-RC2
Data sets
Sub-hourly river flow data observations from 1369 river gauges in the UK, 1948-2023 (UK-Flow15) F. Fileni, H. J. Fowler, E. Lewis, M. Fry, H. Cooper, O. Swain, G. Coxon, F. McLay, E. Bruce, L. Yang, D. Archer https://doi.org/10.5285/211710ac-f01b-4b52-807f-373babb1c368
Model code and software
UK-Flow15-Usage_Notes F. Fileni https://github.com/felipef93/UK-Flow15-Usage_Notes
Viewed
| HTML | XML | Total | BibTeX | EndNote | |
|---|---|---|---|---|---|
| 318 | 174 | 30 | 522 | 26 | 30 |
- HTML: 318
- PDF: 174
- XML: 30
- Total: 522
- BibTeX: 26
- EndNote: 30
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1
The UK-Flow15 dataset is a valuable addition to the large-sample hydrological datasets, where sub-daily resolutions are especially scarce. Thank you to the authors for the tedious work of collecting and controlling the 15 minute data for this large amount of gauging stations.
I recommend acceptance of the manuscript after minor revisions, see the attached PDF file for my comments.