Student or educator? Sign up with your school email and get 6 months of Pro free.
CoChat

The DOI matches its Crossref record; 39 of the 40 references it cites were found in a registry or an index, and 1 are not checked yet; none of the 25 sources its claims cite is checked for support yet.

Global mapping of river sediment bars

Open access
Carbonneau, Patrice E., Bizzi, Simone · 2024 · Earth Surface Processes and Landforms

Abstract

Recently, deep learning has been increasingly applied to global mapping of land‐use and land‐cover classes. However, very few studies have addressed the problem of separating lakes from rivers, and to our knowledge, none have addressed the issue of mapping fluvial sediment bars. We present the first global scale inventory of fluvial gravel bars. Our workflow is based on a state‐of‐the‐art fully convolutional neural network which is applied to Sentinel‐2 imagery at a resolution of 10 m. We use Google Earth Engine to access these data for a study site that covers 89% of the Earth's surface. We count 8.9 million gravel bars with an estimated area of 41 000 km 2 . Crucially, the workflow we present can be executed within a month of highly automated processing and thus allows for global scale, monthly, monitoring of gravel bars and associated rivers.

Checked 2026-11-19 · 1 reference not checked yet

Overview

CoChat Summary

TL;DR

The paper presents the first global-scale inventory of fluvial gravel bars using a fully convolutional neural network applied to Sentinel-2 imagery at 10 m resolution. It maps 8.9 million gravel bars covering about 41,000 km² and argues the workflow can enable monthly global monitoring.

Key findings

  • The global inventory counts 8.9 million gravel bars with an estimated area of 41,165 km² with cloud correction and 33,665 km² without cloud correction; the abstract rounds this to 41,000 km².
  • Validation on manually labelled data gives river producer’s/user’s accuracies of 95%/92%, lake 94%/84%, and bar 61%/84%; a merged water-body class reaches 87% user’s and 95% producer’s accuracy. A global random check of 10,000 samples per class gives lower user’s accuracies of 85% for rivers, 78% for lakes, and 77% for bars.
  • Total river area is 433,544 km², rising to an estimated 562,925 km² after cloud correction. Global bar sizes follow a near log-linear power-law decay with a slope of −1.59.
  • High bar densities occur in Siberia/Northern Russia, Alaska, Nunavut, south-west of the Andes, around the Licungo and Zambezi rivers in Africa, and in rivers draining the Himalayas; apparent hotspots around the Nile valley and north of the Arabian Peninsula are false positives from arid lands adjacent to canals.

Methods

  • The workflow uses a fully convolutional DenseNet architecture (“Tiramisu”) for pixel-level semantic classification of Sentinel-2 Bands 8, 4, and 3 at 10 m resolution, with 224 × 224 pixel chips, accessed through Google Earth Engine. Classes are background, rivers, lakes, and fluvial bars, plus inherited classes for ocean, glaciated terrain, snow, clouds, and no-data.
  • The study area covers 468 MGRS zones constituting the non-polar world, or 89% of Earth’s landmass, using July 2021 imagery composited from available images with less than 20% cloud cover and an s2cloudless mask. Training first uses 740,000 automatically labelled samples from ESRI 2020 land-cover data, then fine-tunes with manually labelled data from 293 randomly selected locations; 50 sites are held for validation, and augmentation yields 59,000 high-quality samples of 2.24 × 2.24 km.
  • Post-processing filters use the second-highest class probability plus contextual corrections from HydroSHEDS and the Randolph Glacier Inventory. Validation uses confusion matrices from 50 manual sites and a global random check of 10,000 locations per river, lake, and bar class (30,000 total) assessed visually.

Limitations

  • The paper states that using only July 2021 leaves data gaps in the tropics and the Indian Monsoon, and that a one-month window cannot deliver globally cloud-free imagery. It also states that bar producer’s accuracy is only 61%, so a significant portion of global bar area is missed and the bar results should be seen as a lower bound accounting for a little over half the global population of bars.
  • The paper states that false positives occur at medium scales and that very large bars in excess of 10,000 pixels with a single count suggest large patches of false positives; arid areas adjacent to canals, such as around the Nile valley and north of the Arabian Peninsula, are falsely classified as bars. It further notes that validation outcomes are sensitive to method, with random global validation producing lower accuracies than set-aside labelled sites, and that no common benchmark datasets exist for Earth Observation.
  • Evident from the text: the first training phase relies on “more error-prone” automated labels from ESRI 2020 land-cover data, manual labelling covers only 293 locations, and the global random validation user’s accuracies (85%, 78%, 77%) are lower than the manual-site validation figures.

Model-written summary. Check it against the paper before relying on it.

Integrity signals

Registry

Crossref · 10.1002/esp.5739

Cites

39 of 40 references verified0 citation errors · 0 not checkable

Statistics

Nothing to recompute.

Figures

Image 1
CC-BY · From the PDF
Image 2
CC-BY · From the PDF
F I G U R E 3 Result examples of our data from July 2021 with comparison to ESA WorldCover 2021, HydroRIVERS and SWORD v2 data. (a) River McKenzie, Canada; (b) Jacaré River (Amazon Basin), Brazil; (c) Po River, Italy; and (d) Popigay River, Russia. Greyscale background is Band 8 (NIR) of Sentinel 2. [Color figure can be viewed at wileyonlinelibrary.com]
F I G U R E 3 Result examples of our data from July 2021 with comparison to ESA WorldCover 2021, HydroRIVERS and SWORD v2 data. (a) River McKenzie, Canada; (b) Jacaré River (Amazon Basin), Brazil; (c) Po River, Italy; and (d) Popigay River, Russia. Greyscale background is Band 8 (NIR) of Sentinel 2. [Color figure can be viewed at wileyonlinelibrary.com]CC-BY · From the PDF
F I G U R E 5 Global scaling relationships for bar sizes. (a) Magnitude frequency plot of bar objects. (b) Scatter plot in log scale of River and Stream Surface Area versus Exposed Sediment Bar Area for all 2  -2  squares with non-zero bar densities.
F I G U R E 5 Global scaling relationships for bar sizes. (a) Magnitude frequency plot of bar objects. (b) Scatter plot in log scale of River and Stream Surface Area versus Exposed Sediment Bar Area for all 2  -2  squares with non-zero bar densities.CC-BY · From the PDF

Verification

Claims the paper makes

0 Valid · 1 Not valid · 15 Not checked · 71 Own results and interpretations

50 more claims

Create a free account to see all 87 claims and to verify the 20 references still open.

See all claims — free account
  • It may state two separate claims as one. 17
  • Probably not a substantial claim. 19

Create a free account to see these claims.

Claims other papers make about this one

Each is checked against this paper’s own text: does it say what the citing paper says it does?

Self-check

Percentages, sums and reported differences re-computed from the paper’s own numbers.

Each contradiction between two passages listed here was confirmed by an independent check.

1 of 1

Citations

Cites

Show all 39 references

Research works printed but not linked: 1

  1. Carbonneau P.E., 2023, Semantic_Classification_of_Rivers_and_Bars (v1.3)

Cited by

See all 27

Metadata

Authors
Carbonneau, P. E., & Bizzi, S.
Affiliations
  • Durham University (GB)
  • University of Padua (IT)
Corresponding author
Carbonneau, Patrice E.
Venue
Earth Surface Processes and Landforms
Publisher
Wiley
Published
2023-11
Volume
49
Issue
1
Pages
15-23
ISSN
  • 0197-9337 (ISSN-L)
  • 1096-9837
OpenAlex
W4388111963
Cited by
27
References
39
Citations by year
2026: 10 · 2025: 9 · 2024: 7 · 2023: 1
FWCI
1.72
Type
journal article
Language
en
Open access
Locations
  • Earth Surface Processes and Landforms (publishedVersion, cc-by)
  • Durham Research Online (Durham University) (publishedVersion, other-oa)
  • Durham Research Online (Durham University) (submittedVersion, cc-by)
  • Padua Research Archive (University of Padova) (submittedVersion, cc-by)
Topics
  • Physical Sciences › Computer Science › Computer Vision and Pattern Recognition › Remote-Sensing Image Classification
  • Physical Sciences › Earth and Planetary Sciences › Earth-Surface Processes › Hydrology and Sediment Transport Processes
  • Physical Sciences › Environmental Science › Global and Planetary Change › Flood Risk Assessment and Management
Keywords
gravel bars, river mapping, global mapping, fully convolutional network, Sentinel-2 imagery, automated image processing, river monitoring, land use and land cover mapping

About these checks

Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.

A language model drew the claims from the paper on 2026-11-19, under version 3 of the claim definition (model deepseek-v4.1-flash, prompt extract-v3).

A second model labelled each claim on 2026-11-19 (model jev-1.13-20260917, definition revision gold-guideline-v0.2).

Statistics and arithmetic are recomputed, not judged.

Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.

What counts as a claim ›How we measure confidence ›How we check this ›

Are you an author, or do you see a problem with this page? Request a review

How to cite

Cite CoChat for the checks on this page, and the paper for its own findings.

This page
CoChat. (2026). Citation check: Global mapping of river sediment bars. https://app.cochat.ai/papers/10.1002/esp.5739
The paper
Carbonneau, P. E., & Bizzi, S. (2024). Global mapping of river sediment bars. Earth Surface Processes and Landforms. https://doi.org/10.1002/esp.5739

Data sources

Bibliographic data: Semantic Scholar, OpenAlex, Crossref.

Contains information from the Semantic Scholar Academic Graph Datasets, made available under the ODC Attribution License (ODC-BY 1.0).

Semantic Scholar’s data is described in: Kinney, R., et al. (2023). The Semantic Scholar Open Data Platform. arXiv:2301.10140.