Papadimos et al. (2025) Performance Evaluation of a Distributed Hydrological Model Using Satellite Data over the Lake Kastoria Catchment, Greece

Identification

• Journal: Hydrology
• Year: 2025
• Date: 2025-12-20
• Authors: Dimitris Papadimos, D. Papamichail
• DOI: 10.3390/hydrology13010002

Research Groups

• The National Museum of Natural History Goulandris—Greek Biotope/Wetland Centre, Thessaloniki, Greece
• Department of Hydraulics, Soil Science and Agricultural Engineering, Faculty of Agriculture, Forestry and Natural Environment, Aristotle University of Thessaloniki, Thessaloniki, Greece

Short Summary

This study evaluates the effectiveness of a coupled distributed hydrological model (MIKE SHE/MIKE HYDRO River) in simulating the long-term Lake Surface Elevation (LSE) and water balance of the Lake Kastoria catchment, Greece, using satellite precipitation and Leaf Area Index (LAI) data. The research concludes that satellite products (GPM_3IMERGDF for precipitation and GEOV3 for LAI) can adequately replace ground station data for LSE prediction and water balance quantification, providing valuable information for water resource management.

Objective

• To investigate the ability of using long-term satellite precipitation and LAI data, in combination with an integrated surface water—groundwater and fully distributed modeling system (MIKE SHE/MIKE HYDRO River), to simulate the hydrological regime and surface elevation of Lake Kastoria in Northern Greece.
• To demonstrate the effectiveness of fully distributed hydrological models combined with satellite data for estimating the long-term water level fluctuation of natural lakes under specific dry-thermal conditions prevailing in Greece.

Study Configuration

• Spatial Scale: Lake Kastoria catchment, Northwest Greece, covering an area of 262.72 km². Elevation ranges from 595 m to 2049 m above mean sea level. The model domain was discretized with a cell size of 200 m × 200 m.
• Temporal Scale: The study period for model calibration and validation was from November 2012 to December 2022 (10 years). Calibration was performed for November 2012–December 2019, and validation for January 2020–December 2022. Satellite data were utilized for the period 1 January 2013–31 December 2022.

Methodology and Data

• Models used:
  • MIKE SHE: A fully distributed, physically-based hydrological simulation system.
  • MIKE HYDRO River (MHR): A deterministic one-dimensional, finite-difference system for simulating water flow in open conduits, coupled dynamically with MIKE SHE.
• Data sources:
  • Satellite Data:
    • Precipitation: GPM Level 3 IMERG Final Daily 10 × 10 km, V06 (GPM_3IMERGDF) from NASA.
    • Leaf Area Index (LAI): VPROVA-V LAI, 300 m Version 1.0 (GEOV3) from Copernicus Global Land Service (CGLS).
  • Ground-based Observations:
    • Daily precipitation and temperature from the Kastoria meteorological station (Unit METEO of the National Observatory of Athens).
    • Mean monthly climatic data and lapse rates from the Hellenic National Meteorological Service (HNMS) climatic atlas.
    • Monthly Lake Surface Elevation (LSE) measurements from the National Water Monitoring Network of Lakes.
    • Groundwater table measurements from three observation wells (YKS020, YKS053, YKS010) of the National Monitoring Water Network for Water Framework Directive (WFD).
    • Field visits and Google Earth images for stream cross-sections and geometry.
  • Geospatial and Derived Data:
    • Digital Terrain Model (DTM): SRTM data (90 m spatial resolution).
    • Hydrogeological data from previous studies [51–53].
    • Soil types and spatial distribution from ESDB—ESDAC [22,23], with water retention characteristics from [54,17].
    • Land use data from Corine Land Cover (CLC) database (2012, 2018) [20,21].
    • Reference evapotranspiration (ETo) calculated using [61–63] with Greek climatic adjustments [64].
    • Manning coefficient for surface runoff from [67].
    • Crop coefficients (Kc) estimated following [63].

Main Results

• Base Model Performance (using ground station data):
  • The model showed very satisfactory simulation of long-term LSE during calibration (2012–2019), with statistical metrics: ME = -0.002 m, MAE = 0.027 m, RMSE = 0.035 m, CC = 0.979, NSE-R2 = 0.958.
  • During validation (2020–2022), the model slightly underestimated LSE in consecutive dry years, with metrics: ME = 0.071 m, MAE = 0.080 m, RMSE = 0.106 m, CC = 0.926, NSE-R2 = 0.701.
  • Groundwater table simulation was also very satisfactory, generally within observed ranges, with minor underestimation in the validation period.
• Satellite Data Scenarios (SatR, SatL, SatRL):
  • All scenarios using satellite data (SatR: satellite precipitation; SatL: satellite LAI; SatRL: both) provided very satisfactory simulations of long-term LSE.
  • Overall statistical performance (2013–2022) for satellite scenarios was comparable to the Base model, with mean absolute errors (MAE) ranging from 0.041 m to 0.044 m and RMSE from 0.062 m to 0.069 m. R2 values ranged from 0.885 to 0.944, and NSE-R2 from 0.858 to 0.883.
  • Catchment Water Balance:
    • SatR scenario showed a decrease in rainfall (-7.88%) and snowfall (-18.79%), leading to a 33.79% reduction in stream outflow to the lake and a 60.71% increase in groundwater storage change. The runoff coefficient (ns) decreased from 0.16 (Base) to 0.12.
    • SatL scenario had minor effects on total inflows (-0.32%) and outflows (-0.54%), with ns at 0.17.
    • SatRL scenario combined effects, with Sat_R's influence prevailing, resulting in decreased total inflows (-7.13%) and outflows (-7.36%), and ns at 0.13.
    • All scenarios indicated a negative water balance for the catchment, implying water resource depletion.
  • Lake Water Balance:
    • SatR and SatRL scenarios showed decreases in both total inflows (up to -10.00%) and outflows (up to -11.13%) for the lake. The total runoff coefficient (nt) for the lake's catchment decreased from 0.1 (Base) to 0.07–0.08.
    • SatL scenario showed marginal increases in total inflows and outflows (+1.47%) for the lake, with nt at 0.11.
    • The lake's water balance was negative across all scenarios, indicating an average annual LSE drop of 0.04 m to 0.05 m.

Contributions

• Demonstrates the robust capability of coupled MIKE SHE/MIKE HYDRO River models to integrate complex physiographic features, water resource management practices, and dynamic hydrological interactions within a catchment.
• Validates the use of GPM_3IMERGDF (precipitation) and GEOV3 (LAI) satellite products as reliable alternatives to conventional ground-based data for long-term hydrological simulations, particularly in data-scarce regions.
• Provides a comprehensive, quantified assessment of the Lake Kastoria catchment's water balance under different data input scenarios, offering valuable information for decision-making in water resource management.
• Establishes a high-performance modeling platform suitable for future research, including evaluating climate change impacts on LSE and comparing the effectiveness of various satellite products for hydrological parameters.

Funding

• Act MIS 5001204, financed by the European Union Cohesion Fund (National Strategic Reference Framework 2014–2020).
• Acts MIS 371010, 371138, 371140, 371144, 371145, financed by the European Regional Development Fund (National Strategic Reference Framework 2007–2013).

Citation

@article{Papadimos2025Performance,
  author = {Papadimos, Dimitris and Papamichail, D.},
  title = {Performance Evaluation of a Distributed Hydrological Model Using Satellite Data over the Lake Kastoria Catchment, Greece},
  journal = {Hydrology},
  year = {2025},
  doi = {10.3390/hydrology13010002},
  url = {https://doi.org/10.3390/hydrology13010002}
}