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

The DOI matches its Crossref record; 49 of the 49 references it cites were found in a registry or an index; none of the 54 sources its claims cite is checked for support yet.

Improved Water Level Retrieval in Complex Riverine Environments: Sentinel‐3 and Sentinel‐6 Altimetry Over China's Rivers

Open access

Abstract

The decline in in situ water level measurements since the 1980s has impeded our ability to fully understand hydrological and hydrodynamic processes, particularly in ungauged river reaches, and how global and regional water cycles respond to climate change. Satellite altimetry offers a valuable means of supplementing these gaps in river water level data, both temporally and spatially. However, existing radar waveform retracking techniques often struggle to accommodate rivers with varying morphologies and surrounding environments. This study presents an Improved Multiple Subwaveform Analysis (IMSA) algorithm based on the 50% Threshold and Ice‐1 Combined (TIC) algorithm, incorporating noise filtering into the subwave search module and refining the retracking strategy for multiple subwaves, independent of coarse digital elevation models (DEMs). We validated the IMSA algorithm using in situ data from 23 gauging stations and applied it to Sentinel‐3 and Sentinel‐6 altimetry across 57 virtual stations (VSs) in China, covering rivers with widths ranging from 20 to 1,500 m, generating 79 validation results (each representing an RMSE value comparing altimetry with in situ measurements). The IMSA algorithm demonstrated significant enhancements at over 48 VSs with more than 64 validation results compared to the original TIC, achieving the lowest median RMSE of 0.61 m (0.13–0.50 m lower than the OCOG, Threshold, MWaPP, and TIC algorithms), with strong resilience to environmental noise. Error analysis revealed that the altimetric accuracy is primarily influenced by the underlying surface characteristics of VSs, with built‐up areas exerting significant interference. Additional disturbances stem from surrounding waters, large slopes, river channels running parallel to the satellite's ground track, and unique features such as sandbars, braided and ice‐covered rivers, and hydroelectric stations. The synthetic aperture radar (SAR) mode was found to mitigate some of these land cover impacts, further improving water level retrieval accuracy. Finally, the results show that river width and topography (whether mountainous or flat) do not inherently affect altimetric accuracy, provided that the on‐board tracking system is supported by accurate prior DEMs and minimal slope interference.

Overview

CoChat Summary

TL;DR

The paper develops an Improved Multiple Subwaveform Analysis (IMSA) retracking algorithm for Sentinel-3 and Sentinel-6 altimetry and validates it over 57 virtual stations on China’s rivers. IMSA achieves the lowest median RMSE (0.61 m) among eight retrackers and shows accuracy is driven mainly by surrounding land cover, not river width or mountain/plain terrain.

Key findings

  • IMSA achieved the lowest median RMSE of 0.61 m, 0.13–0.50 m lower than OCOG, Threshold, MWaPP, and TIC algorithms; improvements occurred at over 48 VSs with more than 64 validation results, with minimum RMSE 0.09 m at VS56 and maximum 4.75 m at VS11.
  • SAR mode outperformed LRM: S6_LR median RMSE was 0.63 m versus 0.53 m for S3 (SAR) and 0.58 m for S6_HR (SAR); SAR performed better at 15 S6 VSs (68%), and at VS46 SAR reduced RMSE by 1.07 m compared with LRM.
  • Built-up area significantly degraded accuracy (r > 0.37, p < 0.10), vegetation had a positive but less significant influence (r < 0.22), and cropland/bare land had minimal influence (|r| < 0.17, p > 0.52). River width and mountain/plain terrain did not inherently affect accuracy, while slope and orientation did: orientations 10–40° had the highest RMSEs, and orientations > 40° had median RMSEs below 0.50 m.
  • Waveform complexity affected accuracy: single-subwave proportion correlated positively with accuracy (r = 0.27), while more than two subwaves correlated negatively (r = 0.42); IMSA identified single-subwave over-water waveforms in 90%, 64%, and 73% of cases for S3, S6_LR, and S6_HR, respectively.

Methods

  • IMSA builds on the TIC algorithm with three modules: buffer generation and over-water waveform selection, waveform retracking, and post-processing (water level retrieval, outlier identification, hooking-effect correction). It uses Pulse Peakiness > 1.5 and maximum power > 10,000 for over-water waveform selection and refined subwave thresholds.
  • The algorithm was validated using in situ data from 23 gauging stations and applied to Sentinel-3A/B and Sentinel-6A altimetry across 57 virtual stations in China, covering river widths from 20 to 1,500 m and generating 79 validation results.
  • Performance was compared against OCOG, 30%/50%/80% Threshold, MWaPP, TIC_30, and TIC_80 using RMSE between satellite and gauge stage anomalies; Pearson correlation was used to relate RMSEs to land cover areas (ESA WorldCover 2020) and river slope (SWORD), with land cover analyzed in 1.5 km radius buffers.

Limitations

  • The paper has no dedicated limitations section. Stated limitations include: ice-covered river measurements retain significant uncertainty due to ice thickness and wave propagation effects within ice; earlier Sentinel-6 cycles used unvalidated corrections; and the primary focus remains China, with global validation only in supporting information.
  • Evident from the text: validation is limited to 57 VSs and 23 gauging stations in China with 79 validation results; some VS retrieval periods were constrained by available gauge records (VS22 ∼27 months, VS57 ∼12 months); and in situ data are available only upon request to the corresponding author.

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

Integrity signals

Registry

Crossref · 10.1029/2024wr039705

Cites

49 of 49 references verified0 citation errors · 0 not checkable

Statistics

Nothing to recompute.

Figures

Figure 1 Geographic distribution of virtual stations (VSs), with circle sizes representing river widths. Subfigures provide zoomed-in views of the distribution map to illustrate river morphology near each gauging station. The ground tracks of various satellites contribute to the formation of VSs, indicated by S3A (green), S3B (blue), and S6A (red). Black numbers correspond to the serial indices of gauging stations, as detailed in the distribution map.
Figure 1 Geographic distribution of virtual stations (VSs), with circle sizes representing river widths. Subfigures provide zoomed-in views of the distribution map to illustrate river morphology near each gauging station. The ground tracks of various satellites contribute to the formation of VSs, indicated by S3A (green), S3B (blue), and S6A (red). Black numbers correspond to the serial indices of gauging stations, as detailed in the distribution map.CC-BY · From the PDF
Image 2
CC-BY · From the PDF
Figure 2 Flowchart of the IMSA algorithm. The algorithm is structured into three core modules: (1) buffer generation and over-water waveform selection, (2) waveform retracking, and (3) post-processing. The waveform retracking module includes a subwave search to extract critical water surface information, while the post-processing module handles water level retrieval, outlier identification, and correction of the hooking effect. These enhancements ensure more precise water level measurements by improving the selection of over-water waveforms and refining retracking strategies. MAD refers to the median absolute deviation.
Figure 2 Flowchart of the IMSA algorithm. The algorithm is structured into three core modules: (1) buffer generation and over-water waveform selection, (2) waveform retracking, and (3) post-processing. The waveform retracking module includes a subwave search to extract critical water surface information, while the post-processing module handles water level retrieval, outlier identification, and correction of the hooking effect. These enhancements ensure more precise water level measurements by improving the selection of over-water waveforms and refining retracking strategies. MAD refers to the median absolute deviation.CC-BY · From the PDF
Image 4
CC-BY · From the PDF
Image 5
CC-BY · From the PDF
Figure 4 Water level retrievals at four virtual stations VS15 (a), VS49 (b), VS22 (c), and VS57 (d) using the IMSA and other baseline algorithms. The in situ water levels were adjusted to the altimeter's reference using the average vertical offset. The retrieval period at VS22 ( ∼ 27 months) and VS57 ( ∼ 12 months) was constrained by available stage records from gauging stations. The color scheme for each retracking method matches that used in Figure 3.
Figure 4 Water level retrievals at four virtual stations VS15 (a), VS49 (b), VS22 (c), and VS57 (d) using the IMSA and other baseline algorithms. The in situ water levels were adjusted to the altimeter's reference using the average vertical offset. The retrieval period at VS22 ( ∼ 27 months) and VS57 ( ∼ 12 months) was constrained by available stage records from gauging stations. The color scheme for each retracking method matches that used in Figure 3.CC-BY · From the PDF
Image 7
CC-BY · From the PDF
Figure 5 Reflection gate identification using the IMSA and seven baseline algorithms: (a) single-peak waveform during the dry season; (b) single-peak waveform during the wet season; (c)-(f) double-peak waveforms; and (g)-(h) complex waveforms with multiple subwaves. The 'real bin' is derived from in situ water level measurements and serves as a reference with errors less than 4 cm. The colors representing different retracking methods correspond to those in Figure 3, highlighting performance variations across algorithms.
Figure 5 Reflection gate identification using the IMSA and seven baseline algorithms: (a) single-peak waveform during the dry season; (b) single-peak waveform during the wet season; (c)-(f) double-peak waveforms; and (g)-(h) complex waveforms with multiple subwaves. The 'real bin' is derived from in situ water level measurements and serves as a reference with errors less than 4 cm. The colors representing different retracking methods correspond to those in Figure 3, highlighting performance variations across algorithms.CC-BY · From the PDF
Image 9
CC-BY · From the PDF
Figure 6 Relationship between RMSEs and percentages of over-water waveforms with (a) single subwave, (b) double subwaves, (c) more than two subwaves, and (d) more than one subwave. Subwaves were identified using the IMSA algorithm. The Pearson correlation coefficient ( r ) is shown in each plot to indicate the strength of the relationship.
Figure 6 Relationship between RMSEs and percentages of over-water waveforms with (a) single subwave, (b) double subwaves, (c) more than two subwaves, and (d) more than one subwave. Subwaves were identified using the IMSA algorithm. The Pearson correlation coefficient ( r ) is shown in each plot to indicate the strength of the relationship.CC-BY · From the PDF
Figure 7 Comparison of LRM and SAR for S6 at VS25. (a) Over-water waveforms captured in LRM and SAR modes. (b) Water level retrievals in LRM and SAR modes using the IMSA algorithm. In situ water levels were adjusted to the altimeter's reference using the average vertical difference.
Figure 7 Comparison of LRM and SAR for S6 at VS25. (a) Over-water waveforms captured in LRM and SAR modes. (b) Water level retrievals in LRM and SAR modes using the IMSA algorithm. In situ water levels were adjusted to the altimeter's reference using the average vertical difference.CC-BY · From the PDF
Image 12
CC-BY · From the PDF
Image 13
CC-BY · From the PDF
Figure 9 Proportions of the four major land cover types vegetation, cropland, built-up area, and bare land within the buffer zone of each VS. The right panel illustrates the optimal RMSE values for each validation result, with different colors and shapes representing various gauging stations and satellite configurations (i.e., working modes). The VSs indices correspond to those shown in Figure 1 and Tables S1-S4 in Supporting Information S1.
Figure 9 Proportions of the four major land cover types vegetation, cropland, built-up area, and bare land within the buffer zone of each VS. The right panel illustrates the optimal RMSE values for each validation result, with different colors and shapes representing various gauging stations and satellite configurations (i.e., working modes). The VSs indices correspond to those shown in Figure 1 and Tables S1-S4 in Supporting Information S1.CC-BY · From the PDF
Image 15
CC-BY · From the PDF
Figure 10 Waveform examples from selected VSs, including VS24-25 for S6_LR, and VS22, VS16 for S3. (a) VS24, primarily surrounded by built-up areas, also includes city park lakes; (b) VS25, dominated by cropland, with a sandbar present in the scene; (c) VS22, mainly characterized by cropland; (d) VS16, situated in bare land, with solar panels along the satellite's ground track contributing to waveform complexity.
Figure 10 Waveform examples from selected VSs, including VS24-25 for S6_LR, and VS22, VS16 for S3. (a) VS24, primarily surrounded by built-up areas, also includes city park lakes; (b) VS25, dominated by cropland, with a sandbar present in the scene; (c) VS22, mainly characterized by cropland; (d) VS16, situated in bare land, with solar panels along the satellite's ground track contributing to waveform complexity.CC-BY · From the PDF
Image 17
CC-BY · From the PDF
Figure 11 Water level monitoring results using the best retrackers for various VSs with distinctive underlying surfaces: (a) braided river, (b)-(c) hydroelectric power station, (d) ice-covered river, (e)-(f) sandbar. In situ water levels have been adjusted to match the altimetric stages using the average vertical difference, highlighting the specific challenges posed by each unique surface feature for altimetric accuracy. WSE refers to water surface elevation.
Figure 11 Water level monitoring results using the best retrackers for various VSs with distinctive underlying surfaces: (a) braided river, (b)-(c) hydroelectric power station, (d) ice-covered river, (e)-(f) sandbar. In situ water levels have been adjusted to match the altimetric stages using the average vertical difference, highlighting the specific challenges posed by each unique surface feature for altimetric accuracy. WSE refers to water surface elevation.CC-BY · From the PDF
Image 19
CC-BY · From the PDF

Verification

Claims the paper makes

0 Valid · 0 Not valid · 37 Not checked · 149 Own results and interpretations

100 more claims

Create a free account to see all 186 claims and to verify the 39 references still open.

See all claims — free account
  • It may state two separate claims as one. 69
  • The quote only partly states this claim. 1
  • Probably not a substantial claim. 16

Create a free account to see these claims.

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.

2 of 2

Citations

Cites

Research works printed but not linked: 2

  1. 10.17613/pr06‐fy12 · doi:10.17613/pr06-fy12
  2. Wingham D. J., 1986, Paper Presented at the In Proceedings of the 1986 International Geoscience and Remote Sensing Symposium, Zurich, Switzerland

Cited by

See all 17

Metadata

Authors
Fang, C., Long, D., Huang, Q., Zhao, F., Liu, H., Duan, X., & Hou, A.
Affiliations
  • Tsinghua University (CN)
  • China University of Geosciences (Beijing) (CN)
  • Yunnan University (CN)
  • Ministry of Water Resources of the People's Republic of China (CN)
Corresponding author
  • Long, Di
  • Huang, Qi
Venue
Water Resources Research
Published
2025-04
Volume
61
Issue
4
ISSN
  • 0043-1397 (ISSN-L)
  • 1944-7973
OpenAlex
W4409378075
Cited by
17
References
49
Citations by year
2026: 12 · 2025: 5
FWCI
3.76
Type
journal article
Language
en
Open access
Locations
  • Water Resources Research (publishedVersion, cc-by)
  • DOAJ (DOAJ: Directory of Open Access Journals) (submittedVersion, cc-by-sa)
Funders
National Natural Science Foundation of China (52325901)
Topics
  • Physical Sciences › Environmental Science › Water Science and Technology › Hydrology and Watershed Management Studies
  • Physical Sciences › Environmental Science › Global and Planetary Change › Flood Risk Assessment and Management
  • Physical Sciences › Environmental Science › Environmental Engineering › Soil Moisture and Remote Sensing
Keywords
Sentinel-3 altimetry, virtual stations, noise reduction, river width, digital elevation model, China rivers

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

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.