Spatial Variability And Distribution Pattern Of Groundwater Nitrate Pollution In Farming Regions Of Shandong Province

In recent years, groundwater nitrate pollution in some regions of China is very serious. Especially, nitrate pollution in intensive cultivation areas is more serious because of the application of a large number of nitrogen fertilizer. To control pollution, spatial variability and distribution pattern of groundwater nitrate pollution should be made clear.Detailed field data on groundwater NO3--N concentration have been collected at 502 farming sites in 2009. Based on the data, the pattern of groundwater nitrate pollution has been analyzed in different regions, different sampling depth and uses of wells, different landuse types. And related data has been collected and analyzed, such as the soil data, data of groundwater resource modulus, nitrogen fertilizer data, precipitation data, elevation gradient and the slope data. The correlation analysis has been implemented between groundwater NO3--N concentration and these data. On that basis, geostatistical method and neural network were used to do spatial variability and distribution pattern analysis on the data of groundwater NO3--N concentration in growing areas in Shandong Province. From the geostatistical method we can come to the conclusion that there was a certain difference of the nitrate content of groundwater in different regions; furthermore, there was a clear trend effect and variability. Through the correlation analysis, two factors (Coarse sand percentage and total nitrogen percentage in soil) were chosen and were using as synergy factors of Co-Kriging to interpolate the groundwater nitrate pollution in Shandong Province. By means of comparative analysis, Co-Kriging got high accuracy compared with Ordinary Kriging. On the other hand, a Back Propagation Neural Network (BPNN) was developed for modeling groundwater NO3--N concentration. By reference to the results of correlation analysis, groundwater resource modulus (M), Coarse sand percentage in soil (C), total nitrogen percentage in soil (Q), Organic matter percentage in soil (Y) were chosen as input features of the BPNN for having the best correlation with groundwater NO3--N concentration. Groundwater NO3--N concentration was as the output feature.Overall, using geostatistical method and neural network to simulate groundwater NO3--N concentration in Shandong Province obtained similar spatial distribution. Results of the two methods showed that the spatial distribution reflecting in the directional effect that gradually increasing from the southwest to the northeast, while areas with higher nitrate content in groundwater mainly locating in Weifang, Qingdao, Yantai growing areas, such as Pingdu, Laixi of Qingdao and Shouguang of Weifang and other developed growing areas.Precision analysis showed that the BPNN obtained higher space extrapolating precision. Both methods provide a basis for the study of the spatial distribution of Shandong Province. Based on the results, the present situations and reasons of groundwater nitrate pollution in Shandong Province were discussed. Combined with the analysis of relevant factors, related development tendency and the protection measures were put forward.