2.4. Diffuse N modeling and interaction analysis
The yearly loading of organic-N (Org-N), nitrate (NO3–N), and
total nitrogen (TN) from 2008 to 2011 was simulated. The average
yearly loading during this period was applied in the spatial
distribution and the subsequent correlation analysis, which can
neutralize the impact of seasonal or annual weather variance. The
diffuse N loading of different land use conditions was then
delineated with the spatial distribution of land use and diffuse N
loading. Based on the analysis of eight soil parameters for the 1.5-
km-grid sampling sites, the Kriging interpolation method was used
to develop the spatial distribution of the soil properties of each
sub-basin at two depths (Robinson and Metternicht, 2006). After
the interpolation, the average value in each sub-basin was
calculated. After the diffuse N loading and soil properties were
determined on the same spatial scale, the relationships at two
depths were analyzed. Based on the regional land use distribution,
it was found that the paddy rice, upland, forest and wetlands
covered most of the area. Their spatial correlations were
categorized
firstly and the impacts of land use were determined.
The upland and rice paddy sub-basins were the major diffuse N
contractors, so the
fitting lines for their sub-basins were
determined. Finally, a partial least squares (PLS) regression method
was applied to assess the spatial correlations of the eight soil
parameters with diffuse N loading under two kinds of farmlands
(Chighladze et al., 2012). During the PLS analysis, the diffuse N
loading was the dependent variable and the soil parameters were
independent variables, which aimed to select the suitable
indicators to assess the diffuse N loading (Fox and Metla, 2005).