Solution of export coefficients of nitrogen from different land-use patterns based on Bayesian analysis

Area of Science:

• Environmental Science
• Water Quality Management
• Hydrology

Background:

• Non-point source (NPS) pollution estimation relies on export coefficients (ECs), but uncertainty exists regarding nitrogen (N) ECs across different land-use patterns.
• Traditional regression models struggle with inherent uncertainties in ECs, complicating accurate pollution assessment.

Purpose of the Study:

• To estimate total nitrogen (TN) export coefficients (ECs) and stream loss rates (K) for five distinct land-use patterns.
• To apply a Bayesian approach to quantify uncertainty in ECs and improve NPS pollution management.

Main Methods:

• Utilized Bayesian theory, combining published data with monthly monitoring data from the ChangLe River system (2004-2008).
• Employed Markov chain Monte Carlo (MCMC) simulations for parameter estimation, achieving convergence with minimal errors.
• Estimated TN ECs and stream loss rates (K) for paddy fields, dry lands, residential areas, woodlands, and barren lands.

Main Results:

• Average TN ECs varied significantly across land uses, with residential land showing higher values (41.7 ± 6.9 kg ha⁻¹ yr⁻¹).
• The average stream loss rate (K) was 0.33 d⁻¹, with a coefficient of variation (CV) of 11.3%.
• Model predictions for 2008-2009 validated the approach, demonstrating its efficacy.

Conclusions:

• The Bayesian model successfully determined TN ECs and stream loss rates, effectively addressing uncertainty issues inherent in regression models.
• This approach enhances NPS pollution management by providing a robust method for EC estimation using prior knowledge and monitored data.
• The model facilitates explicit consideration of uncertainty, leading to more reliable water quality predictions and management strategies.