Köhn et al. (2026) Evaluation of SWOT HR PIXC version D water level time series of small lakes

Identification

• Journal: Hydrology and earth system sciences
• Year: 2026
• Date: 2026-10-22
• Authors: Simon Jakob Köhn, Karina Nielsen
• DOI: 10.5194/hess-30-6443-2026

Research Groups

• DTU Space, The Technical University of Denmark, Kongens Lyngby, Denmark

Short Summary

This study evaluates the Surface Water and Ocean Topography (SWOT) satellite's ability to monitor water levels in 37 small Danish lakes (0.25 to 40 km²) using a customized processing workflow of low-level PIXC data, finding it doubles temporal resolution compared to standard products while revealing residual roll errors.

Objective

• Evaluate a customized processing workflow for SWOT HR PIXC version D data to track relative water surface elevation (WSE) anomalies in small Danish lakes.
• Compare the performance of this customized workflow against the standard SWOT LakeSP product, particularly regarding temporal resolution and accuracy.
• Investigate the presence and characteristics of residual time-varying roll errors in inland SWOT observations.

Study Configuration

• Spatial Scale: 37 small Danish lakes, ranging from 0.25 km² to 40 km² in surface area, located within 54.7–57.2° N and 8.1–12.6° E.
• Temporal Scale: July 2023 to May 2026 (approximately 3 years). In-situ gauge data has daily resolution, while SWOT provides an effective temporal resolution of approximately 9.5 days for the study area.

Methodology and Data

• Models used: A customized processing workflow for SWOT L2 HR PIXC vD data, incorporating spatial masking, outlier exclusion (3-sigma rule), and specific data selection criteria (e.g., minimum 200 observations, geo-flag filtering). Statistical evaluation employed Root Mean Square Error (RMSE) and Pearson Correlation Coefficient (PCC).
• Data sources:
  • SWOT L2 HR PIXC version D data (KaRIn water mask point cloud).
  • SWOT L2 HR LakeSP product (for comparative baseline).
  • Daily in-situ gauge data from Vandportalen, referenced to the Danish height system DVR90.
  • Lake geometries/masks from GeoDanmark (vector layer product).
  • EGM2008 geoid model.

Main Results

• The customized PIXC processing workflow achieved a median Root Mean Square Error (RMSE) of 5.57 cm and a Pearson Correlation Coefficient (PCC) of 0.91 across the 37 gauged lakes.
• This workflow retained a median of 125 valid timestamps per lake, effectively doubling the temporal resolution compared to the filtered LakeSP product (median of 60 timestamps).
• The filtered LakeSP product exhibited slightly better metrics (median RMSE of 4.92 cm, median PCC of 0.94) but with significantly fewer valid observations.
• Residual time-varying roll errors were detected in SWOT's PIXC data over Danish lakes, affecting approximately 18% of all passes with more than 10 observed lakes. This error degrades spatial precision by expanding the 1-sigma spread of the data distribution.
• The workflow demonstrated relative insensitivity to moderate boundary adjustments of lake masks (e.g., ±20 m buffer), with minor changes in RMSE and PCC.

Contributions

• Demonstrated the successful utilization of lower-level SWOT HR PIXC data to generate highly accurate relative water surface elevation (WSE) time series for small lakes, significantly improving temporal resolution compared to standard operational products.
• Provided a comprehensive validation of SWOT's capabilities for small inland water bodies using a dense network of in-situ gauges in Denmark.
• Identified and characterized the presence of a residual time-varying roll error in inland SWOT data, highlighting its potential impact on WSE accuracy for lakes with low hydrological amplitudes and on water surface slope measurements.
• Offered insights into the critical trade-offs between operational simplicity (LakeSP) and maximizing temporal frequency of water level observations (custom PIXC processing) for future SWOT data users.

Funding

• Klimadatastyrelsen (KDS)
• Miljøstyrelsen, HOFOR, and various Danish municipalities (Aarhus, Haderslev, Halsnæs, Lolland, Næstved, Ringkøbing–Skjern, Ringsted, Roskilde, Skanderborg, Silkeborg, Thisted, and Vejle–Kommune), as well as Aage V. Jensen Naturfond, Nationalpark Kongernes Nordsjælland, Vestforsyning A/S, and WSP for providing open access to gauge data via Vandportalen.

Citation

@article{Köhn2026Evaluation,
  author = {Köhn, Simon Jakob and Nielsen, Karina},
  title = {Evaluation of SWOT HR PIXC version D water level time series of small lakes},
  journal = {Hydrology and earth system sciences},
  year = {2026},
  doi = {10.5194/hess-30-6443-2026},
  url = {https://doi.org/10.5194/hess-30-6443-2026}
}