Spatial Analysis Of Water/soil Pollutions And Source Identification

Soil and wate are the most important resources for humanbeing. They have direct impact on the quality of human life, while human activities directly affect soil and water environment. With great social economic development and land use/land cover changes in recent decades, soil and water environment have gradually deteriorated. Qiantang River Basin and Haining City locate in the regions with fast-growing economy, rich industrial type, and frequently changing of land use/land cover, which all affect soil and water environment, and eventually lead to various pollution patterns. This dissertation characterized the spatial patterns of pollutions and identified the potential sources of pollution types.Spatial analysis and source apportionment of water pollution in Qiantang River Basin were conducted. In this work, we considered data for 13 water quality variables collected during the year 2004 at 46 monitoring sites along the Qiantang River (China). Fuzzy comprehensive analysis categorized the data into three major pollution zones (low, moderate, and high) based on national quality standards for surface waters, China. Most sites classified as "low pollution zones" (LP) occurred in the main river channel, whereas those classified as "moderate and high pollution zones" (MP and HP, respectively) occurred in the tributaries. Factor analysis identified two potential pollution sources that explained 67% of the total variance in LP, two potential pollution sources that explained 73% of the total variance in MP, and three potential pollution sources that explained 80% of the total variance in HP, respectively. UNMIX was used to estimate contributions from identified pollution sources to each water quality variable and each monitoring site. Most water quality variables were influenced primarily by pollutants from industrial wastewater, agricultural activities and urban runoff. In LP, non-point source pollution such as agricultural runoff and urban runoff dominated; in MP and HP, mixed source pollution dominated. The pollution in the small tributaries was more serious than that in the main channel. These results provide important information for developing better pollution control strategies of the Qiantang River.Spatial analysis and source identification of heavy metal pollution in soils of Haining City were performed. A total of 309 topsoil samples were collected in 2005 from agricultural land. Each sample was analyzed for 10 elements:arsenic (As), cadmium (Cd), cobalt (Co), chromium (Cr), copper (Cu), mercury (Hg), nickel (Ni), lead (Pb), tin (Sn) and zinc (Zn). With the combination of enrichment factor and multivariate statistics, the variables were classified into the high contaminating (Pb, Hg and Sn) and low contaminating elements (As, Co, Cr, Cu, Cd, Ni and Zn). The former were predominantly influenced by anthropic inputs, and the latter derived from natural sources or marginal human influence. Spatial analysis showed that high-value areas of Pb, Hg, Cd and Sn were isolated and mainly situated around the urban areas. The influence of soil type on uncontaminated or low contaminating elements was greater than on high contaminating elements, except for Hg. Land use had no distinct impact on most elements. Pb, Sn, and Hg were greatly influenced by point or linear pollution sources. Along the traffic lines, Pb was mainly from car exhausts, while Cu, Ni and Zn weremainly from wear and tear of mechanical parts and tires. For As, Cr, Ni, Pb, Cu and Zn, their ecological hazard was minor; The area with medium ecological hazards of Cd and Hg accounted for about 98.4% and 63.1%, respectively. The strongly hazardous area of Hg occupied 33.8%. The rest area was at slight risk. For biological types, risk probability of As with medium risk was 43.7%. For ecological process, risk probability of As and Cd was 25.2% and 25.3%, respectively. Risk level of the rest pollutants was low.Fuzzy clustering analysis was used to delineate combined pollution zones. The ten elements were divided into three groups:Group I (As, Cd, Co, Cr and Ni), Groupâ…¡(Cu, Zn), Groupâ…¢(Hg, Pb, Sn). Each group was divided into three zones with different pollution levels. One way ANOVA showed that the significant differences of mean concentration existed among different pollution zones. The overall combined pollution levels in groupâ… and groupâ…¡were low, and membership was relatively explicit in the two groups. In groupâ…¢, severe pollution zone located in the south of Haining and along the Qiantang River; moderate pollution zone mainly located in the west and near the urban region; low pollution zone was mainly in central and southeastern regions. The impact of soil type on the delineation of pollution zones was related with the design of the sample points and soil properties. Land-use types were not important in the delineation of pollution zones. The differences of pH between different zones were not obvious. Organic matter contents were important factors in pollution delineation of groupâ… andâ…¡. In Groupâ…¢, organic matter contents of low-pollution area were higher than those in high-contaminated area.Local Moran's I index was used to identify pollution hot spots. Weight function, data transformation methods and extreme values all affected the number and spatial location of hot spots. The results of removing the extreme values were similar to those in data transformation. High-high zones were increased after removing the extreme values and the area of high-high will also increase.Uncertainty assessment of contaminated areas delineation was conducted. Ordinary kriging was compared with Sequential Gaussian Simulation in pollution mapping. Indicator Kriging and Sequential Gaussian Simulation were used to produce the probability distribution map exceeding certain probability level. Overlaying simulated interpolation maps and simulated probability maps produced reliable contaminated areas. The results showed that Kriging interpolation maps tended to be smooth, with less local details. However, the maps produced by simulation had rich spatial structure information. Contaminated areas of Hg were mainly distributed in several towns in the southwest along the Qiantang River; Pb contamination areas scattered throughout the region, and the proportion of contaminated areas was small; Contaminated areas with high confidence of Sn were mainly in the southwest along the river and the two towns in the southeast.Spatial analysis was conducted based on Bayesian maximum entropy(BME). Integrating soft data and hard data in BME (BME_HS) had less bias than ordinary Kriging did in interpolating spatial data with highly skewed values. But, on prediction of variation (or fluctuation), BME_HS did not always perform better than ordinary Kriging. BME only with hard data also performeds better than ordinary Kriging. Predictions of extremely high values based on BME were closer to the real values than those based on ordinary Kriging.In sumary, this dissertation made the following improvements:1) Fuzzy method, multivariate statistics and GIS were used to explore the levels and spatial patterns of water pollution. The integrated uses of multiple techniques provided better understanding of water pollution, and the findings help developing better pollution control strategies of Qiantang River.2) This study used a variety of methods (multivariate statistics, enrichment factor method, geostatistics and GIS) to charaterize the status of soil pollution, quantify the relationships between pollution levels and soil type, land use/cover.3) The complexity of soil systems and human activity determined that the delineation of the contaminated area was difficult. In his study, fuzzy theory was used to analyze ambiguity of pollution levels, which could better reflect the reality.4) Comparing with kriging alone, the combination of simulation probability and simulations values produced better prediction to delineate the contaminated area.5) Mean error and mean squared error showed that prediction results of Hg based on BME were better than those based on Kriging. Mean error showed that prediction results of Pb based on BME were better than those based on Kriging, while mean squared error of Pb based on BME was greater than that based on kriging.