Employing K-means clustering to reconstruct missing water surface elevations in LiDAR digital elevation models for...

Area of Science:

• Hydrology
• Geographic Information Systems (GIS)
• Environmental Modeling

Background:

• Digital Elevation Models (DEM) are crucial for hydrodynamic modeling, but contain uncertainties over water bodies and riverbanks.
• Inaccurate DEM data can lead to errors in estimating runoff depth and river storage.

Purpose of the Study:

• To develop and validate a DEM reconstruction approach for accurate water surface elevation estimation.
• To improve runoff simulations and urban flood risk prediction using enhanced DEM data.

Main Methods:

• Utilized k-means clustering to identify river and riverbank areas.
• Applied inverse distance weighting for riverbank elevation reconstruction.
• Replaced river and lake elevations with minimum riverbank elevation to estimate water surface elevations.

Main Results:

• Estimated water surface elevations showed high agreement with observed values (R² = 0.98).
• Reconstructed DEM data improved hydrodynamic simulations, correcting overestimated river storage and underestimated runoff depths.
• The method provides a reference for areas with DEM data scarcity or uncertainty.

Conclusions:

• The proposed DEM reconstruction method effectively estimates water surface elevations.
• This approach enhances the accuracy of runoff simulations and urban flood risk assessments.
• The findings offer valuable insights for hydrological studies in data-limited regions.