Investigating the potential of Morris algorithm for improving the computational constraints of global sensitivity...

Area of Science:

• Environmental science
• Hydrology
• Computational modeling

Background:

• Sensitivity analysis (SA) is crucial for calibrating and optimizing complex hydrological models.
• Sobol global SA is effective but computationally expensive, limiting its application.
• Morris global SA offers a potential alternative with lower computational demands.

Purpose of the Study:

• To evaluate the efficacy of Morris global SA compared to Sobol global SA for hydrological models.
• To explore a new approach of increasing Morris algorithm replications for improved results.
• To assess the computational cost-effectiveness of Morris SA.

Main Methods:

• Performed sensitivity analysis using both Morris and Sobol algorithms on a coupled hydrological model across two catchments.
• Employed two target functions for each algorithm.
• Increased the number of replications for the Morris algorithm (from 1000 to 3000).

Main Results:

• Morris SA with increased replications produced results comparable to Sobol SA but with significantly reduced computational cost (63,000 runs vs. 660,000 runs).
• Morris SA achieved a 50% improvement in parameter ranking compared to Sobol SA with a threefold increase in replications and minimal additional computational time.
• Parameter sensitivity was influenced by target functions and catchment characteristics.

Conclusions:

• The enhanced Morris SA method is a promising, computationally efficient alternative for highly parameterized hydrological models, especially under time constraints.
• This approach makes SA more accessible for complex model applications, overcoming computational barriers.
• The study advocates for wider adoption of SA in hydrological modeling.