Optimization of rain gauge sampling density for river discharge prediction using Bayesian calibration.

Area of Science:

• Hydrology
• Geostatistics
• Environmental Modeling

Background:

• River discharge predictions rely on calibrated rainfall-runoff models.
• Quantifying uncertainty from input, parameters, and model structure is crucial for accurate predictions.
• Bayesian calibration is a key method for uncertainty quantification in hydrological models.

Purpose of the Study:

• To integrate geostatistics and Bayesian calibration for analyzing rain gauge density's effect on river discharge prediction accuracy.
• To assess the influence of varying rain gauge network densities on hydrological model performance.
• To provide recommendations for handling input uncertainty in Bayesian calibration for river discharge prediction.

Main Methods:

• Calibrated the HBV hydrological model using Bayesian methods.
• Accounted for input, initial state, model parameter, and model structural uncertainty.
• Incorporated discharge measurement uncertainties and geostatistical methods for rainfall prior distribution.

Main Results:

• Model parameter uncertainty was the primary contributor to overall uncertainty.
• A low rain gauge density proved adequate for Bayesian calibration.
• Increasing rain gauge density improved predictions up to one gauge per 340 km².

Conclusions:

• Bayesian calibration is robust even with limited rain gauge data.
• Optimal rain gauge density is case-study specific but provides a benchmark.
• The study presents a methodology for assessing input uncertainty's impact on discharge prediction accuracy.