Comparison Of Different Methods For Assessing Regional Soil Heavy Metal Contamination

With the rapid development of industrialization, urbanization and agricultural intensification process, the soil environment has been negatively influenced,among which soil heavy metal contamination is the most serious one.It has become increasingly obvious that restrict social sustainable development and become a hot issue. Therefore, scientific, accurate and comprehensive assessment about regional soil heavy metal contamination is the foundation of grasping soil heavy metal contamination and further remediation for it. Currently, quite a few methods of assessment about soil heavy metal contamination (include contamination level evaluation and spatial distribution estimation) have been proposed, and scholars carried out a great amount of practical work. But few studies have made comparison of different methods for assessing regional soil heavy metal contamination, therefore the accuracy, comprehensiveness of assessment results is difficult to grasp.In view of this, the booming Kunshan city was selected as the study area. We obtained heavy metal content by field sampling and indoor test, and choosed As,Hg as typical element. Geo-accumulation index was regarded as pollution index. At first simple statistics, normal fuzzy number, kernel density estimation were used for evaluating regional soil heavy metal contamination level, and evaluation results of these methods were compared; And then Kriging interpolation, sequential Gaussian simulation, BP network were used for estimating the spatial distribution of regional soil heavy metal contamination. On the basis of further grasping soil heavy metal contamination of local region, estimation results of these methods were compared; Finally, this paper led to a discussion of improvement of assessment methods of regional soil heavy metal contamination. The result was showed as follows:(1)The evaluation results of regional soil heavy metal contamination level were different, and convenience during evaluation process, accuracy and comprehe-nsiveness of evaluation result was also something different. On convenience, three methods were in the order of simple statistics>>ormal fuzzy number≥kernel density estimation; On accuracy, the order was kernel density estimation>normal fuzzy number=simple statistics; On comprehensiveness evaluation result contained, the order was normal fuzzy number>>ernel density estimation=simple statistics.(2)The spatial distribution of soil heavy metal contamination by functional area and grid samples layout was somewhat different. Judging by spatial variability, both presented characteristic that general distribution patterns were similar, but local region of them were much different. Judging by estimation error, the MSE based on grid layout was much lower than functional area layout, and this illustrated the general estimation result based on grid layout was more accurate than that based on functional area.(3) Spatial distribution estimation result of regional soil heavy metal contamination could further grasp contamination information of local region. The difference about distribution characteristic, spatial variability, estimation error of estimation results based on different methods was significant. All methods had their advantages and disadvantages at accuracy and comprehensiveness estimation result contained. Sequential Gaussian simulation result could preserve distribution characteristic of original data well and presented spatial variability obviously. Therefore, the information result contained was most comprehensive. But uncertainty estimation characteristic of it made accuracy lowest; BP network result was highest at accuracy, but it presented right skewed and aroused spatial autocorrelation significantly, therefore, the result contained less information; Kriging interpolation result was relatively poor at accuracy and comprehensiveness.(4) Using kernel density estimation model optimized by iterative algorithm could further improve accuracy of evaluation results of regional soil heavy metal contamination; After running multiple sequential Gaussian simulations of regional soil heavy metal geo-accumulation index, and then we may evaluate contamination level at specific location by normal fuzzy number. This could realize fuzzy evaluation at any location of study area and improve the comprehensiveness of assessment result.