Estimation of Nutrient Export from Different Land-Use Patterns & Uncertainty Assessment in a Watershed Scale Based on Bayesian Approach.

Estimation of nutrient export from different land-use patterns is vital to soil fertility management and non-point source pollution control. All export coefficient models and many complex mechanistic models for estimating nutrient export are relied on the nutrient Export Coefficients (ECs) to quantify and identify nutrient source. However, determining the ECs of each nutrient source remains a challenge because of their spatiotemporal variability and mixture of multiple nutrient sources in a watershed or catchment scale. Based on ECs’ prior knowledge and local conventional water quality monitoring data, we introduced a Bayesian method to estimate the ECs of nitrogen and phosphorus by combining the export coefficient model and nutrient in-stream removal model in six catchments with mixed land-use patterns (including dry land, paddy field, woodland, and residential land) of Changle River basin, eastern China in 2004 to 2009. Modeling validation results showed that the mean relative errors between the observed and modeled export loads of TN and TP were 20.8% and 11.3%, respectively. The average annual TN ECs were 70.3, 41.7, 26.1, and 8.90 kg ha^-1 yr^-1 for dry land, paddy field, residential land and woodland, respectively; and for TP, the average annual ECs were 5.43, 1.98, 1.39, and 0.699 kg ha^-1 yr^-1, respectively. In-stream retention processes were considered in estimating TN and TP export coefficients. For the whole basin, the average in-stream retention ratios of TN and TP were account for 21.3% and 23.8% of their total loading, respectively. The variability of mean ECs of TN and TP among different land-use patterns was much higher than that among different catchments. Fully aware of the variability and uncertainty in Estimation of nutrient export has important practical significance for the development of a quantitative non-point source pollution control planning.