Great Lakes Runoff Intercomparison Project Phase 3: Lake Erie (GRIP-E).

Area of Science:

• Environmental science
• Hydrology
• Water resource management

Background:

• The Lake Erie watershed faces challenges like flooding, erosion, and eutrophication due to nutrient loads.
• Understanding water flow sources and pathways is crucial for addressing these complex issues.
• This study is part of the Great Lakes Runoff Intercomparison Projects, focusing on the Lake Erie watershed.

Purpose of the Study:

• To compare the performance of seventeen diverse hydrologic and land-surface models.
• To evaluate model agility in simulating streamflow, evaporation, and soil moisture.
• To assess model performance in both calibration and independent validation scenarios.

Main Methods:

• Seventeen hydrologic and land-surface models were set up using identical meteorological forcings.
• Simulated streamflows were compared at 46 calibration and 7 independent validation stations.
• Model performance was evaluated based on their ability to replicate observed hydrological variables.

Main Results:

• Machine learning models showed decreased performance during validation due to limited training data.
• Models calibrated at individual stations maintained good performance in validation.
• Distributed models, despite challenges in urban areas, outperformed others during validation when calibrated across the entire domain.

Conclusions:

• Model calibration strategies significantly impact validation performance, particularly for data-driven approaches like machine learning.
• Distributed hydrologic models demonstrate robust performance in watershed-scale simulations, even with complexities like urban areas.
• Effective watershed management requires accurate hydrologic modeling, highlighting the need for careful model selection and calibration.