Uncertainty analysis of total phosphorus spatial-temporal variations in the Yangtze River Estuary using different interpolation methods

Interpolation processes and results are generally accompanied by uncertainty which affects the spatial and temporal properties of pollutants. Based on the 4 period sample data of total phosphorus (TP) collected from the Yangtze River Estuary (YRE) in 2010 and 2011, the uncertainty of spatial-temporal variation was analyzed with interpolation methods of inverse distance weighted (IDW), local polynomial interpolation (LPI), ordinary kriging (OK) and disjunctive kriging (DK). The root mean square errors (RMSE) and the mean relative errors (MRE) were used to analyze the accuracy of different interpolation methods. The results showed that the uncertainty of DK was the lowest and the uncertainty of LPI was the highest among the 4 methods. The subtraction results between different interpolation methods showed that there was some distinct area of value in the disparate interval (not in [-0.05, 0.05) (mg/ L)) in the 4 seasonal results, which was mainly distributed in the boundary region and around some sample sites. Both standard deviation (SD) and coefficient of variance (CV) in August 2010 were the highest in the 4 seasons and annual mean. The uncertainty may be caused by choice of interpolation methods, spatial data discrepancy and the lack of sample data.