A Tool for Classification and Regression Using Random Forest Methodology: Applications to Landslide Susceptibility Mapping and Soil Thickness Modeling

Classification and regression problems are a central issue in geosciences. In this paper, we present Classification and Regression Treebagger (ClaReT), a tool for classification and regression based on the random forest (RF) technique. ClaReT is developed in Matlab and has a simple graphic user interface (GUI) that simplifies the model implementation process, allows the standardization of the method, and makes the classification and regression process reproducible. This tool performs automatically the feature selection based on a quantitative criterion and allows testing a large number of explanatory variables. First, it ranks and displays the parameter importance; then, it selects the optimal configuration of explanatory variables; finally, it performs the classification or regression for an entire dataset. It can also provide an evaluation of the results in terms of misclassification error or root mean squared error. We tested the applicability of ClaReT in two case studies. In the first one, we used ClaReT in classification mode to identify the better subset of landslide conditioning variables (LCVs) and to obtain a landslide susceptibility map (LSM) of the Arno river basin (Italy). In the second case study, we used ClaReT in regression mode to produce a soil thickness map of the Terzona catchment, a small sub-basin of the Arno river basin. In both cases, we performed a validation of the results and a comparison with other state-of-the-art techniques. We found that ClaReT produced better results, with a more straightforward and easy application and could be used as a valuable tool to assess the importance of the variables involved in the modeling.     

Risk assessment and distribution of soil Pb in Guangdong, China

In this study, the spatial distributions of soil lead (Pb) concentration in three horizontal soils in Guangdong, China, were surveyed and analyzed using geostatistics and geography information systems (GIS). Findings indicated that the Pb geometric mean concentration of 23.3 mg/kg in surface soils was found to be higher than those in global soils, which ranged from 2.3–235 mg/kg. In addition, the Pb geometric mean concentrations from A- to C-horizon were found to be 23.3, 27.2, and 28.6 mg/kg, respectively. The classification of a soil Pb environmental risk in an area was likewise presented based on the different levels of environmental quality of Pb and was done by GIS technology. Accordingly, there is a higher local concentration of Pb in the surrounding areas of Guangzhou where there is higher population density and in the north of Guangdong, which is a historic mining area. The results obtained by the environmental risk assessment reveal that about 46% of total surveyed area was above the background value, that is, 2.7% of the total area was at risk of pollution.     

Study of the relationships between bone lead levels and its variation with time and the cumulative blood lead index, in a repeated bone lead survey

The study aims were to: (i) investigate long term human lead metabolism by measuring the change of lead concentration in the tibia and calcaneus; and (ii) assess whether improved industrial hygiene was resulting in a slow accumulation of lead in an exposed workforce. 109Cd excited K X-ray fluorescence was used to measure tibia lead and calcaneus lead concentrations in 101 workers in a secondary lead smelter. 51 subjects had had similar bone lead measurements 5 years previously. Most of the other subjects had been hired since the first survey. Measurements of whole blood lead were available for the large majority of subjects. Tibia lead concentrations fell significantly (p<0.001) in the 51 subjects with repeated bone lead measurements, from a mean of 39 microg Pb (g bone mineral)(-1) to 33 microg Pb (g bone mineral)(-1). The change correlated negatively with the initial tibia lead concentration, producing an estimate for an overall half-life of 15 years, with a 95% confidence interval of 9 to 55 years. Adding continuing lead exposure and recirculation of bone lead stores to the regression models produced half-life estimates of 12 and 9 years, respectively, for release of lead from the tibia. The repeat subjects showed no net change in calcaneus lead (64 microg Pb (g bone mineral)(-1) initially, 65 microg Pb (g bone mineral)(-1) 5 years later). Subjects not measured previously had average lead concentrations of 15 microg Pb (g bone mincral)(-1) in the tibia and 13 microg Pb (g bone mineral)(-1) in the calcaneus. The rate of clearance of lead from the tibia (9 to 15 years) is towards the more rapid end of previous estimates. The lack of a significant fall in the calcaneus lead was surprising. Attempts should be made to repeat this observation. If confirmed, it would have implications for models of lead metabolism. The relatively low lead concentrations in the non-repeat subjects are reassuring. However, observation after a longer period of employment would be desirable.     

Fast simultaneous determination of traces of Cu(II) and Co(II) in soils and sediments with the luminol/perborate chemiluminescent system

A flow injection analysis method based on ion chromatography and luminol chemiluminescence detection was used for the simultaneous determination of copper (II) and cobalt (II) trace levels in soils and sediments following microwave-assisted acid digestion. Detection was based on chemiluminescence (CL) of the luminol–perborate system in an alkaline medium, which is catalyzed by both transition metals. The concentration and pH of the eluent (oxalic acid) was found to affect CL intensities and retention times, both of which were inversely proportional to the oxalic acid concentration. The calibration curves for both metal ions were linear and allowed a limit of detection of 0.003 μg l^?1 for cobalt (II) and 0.014 μg l^?1 for cooper (II) to be calculated. The proposed method was successfully used to determine both metal ions in certified reference materials of stream and river sediments and soil samples. Based on the results, the determination is free of interferences from the usual concomitant ions.     

Organic interference of solidified/stabilized hazardous wastes

Liquid hazardous waste disposal in landfills is usually allowed only after solidification/stabilization. Although various procedures are commonly practiced, little is known about the mechanism(s) of the processes. A particular problem is the interference of organics. Small amounts of organics can interfere with the reaction between inorganic sludges and cementitous matrices. The present communication reports studies of the interaction between selected organic hazardous wastes and Type I Portland cement. Microscopic studies of the structural differences between cements set with water and those set with water plus organic liquids are discussed. In these studies the scanning electron microscope is used to observe samples fractured at 78K. The results provide technical background data on the ultimate stability of critical waste constituents solidified by various binding agents.     

Seasonal Variation of Toxic Benzene Emissions in Petroleum Refinery

Petroleum refineries are largest chemical industries that are responsible for the emission of several pollutants into the atmosphere. Benzene is among the most important air pollutants that are emitted by petroleum refineries, since they are involved in almost every refinery process. Volatile organic compounds (VOCs) are a major group of air pollutants, which play a critical role in atmospheric chemistry. These contribute to toxic oxidants, which are harmful to ecosystem, human health and atmosphere. The variability of pollutants is an important factor in determining human exposure to these chemicals. The ambient air concentrations of benzene were measured in several sites around the Digboi petroleum refinery, near the city of Gowahati in northeast India, during winter and summer 2004. The seasonal and spatial variations of the ambient air concentrations of this benzene were investigated and analyzed. An estimation of the contribution of the refinery to the measured atmospheric levels of benzene was also performed. The ambient air mixing ratios of benzene in a large area outside the refinery was generally low, in ppbv range, much lower than the ambient air quality standards. This article presents the temporal and spatial variation of air pollution in and around petroleum refinery and showed that no health risk due to benzene is present in the areas adjacent to the refinery.     

An evaluation of selenium concentrations in water, sediment, invertebrates, and fish from the Solomon River Basin

The Solomon River Basin is located in north-central Kansas in an area underlain by marine geologic shales. Selenium is an indigenous constituent of these shales and is readily leached into the surrounding groundwater. Portions of the Basin are irrigated primarily through the pumping of selenium-contaminated groundwater from wells onto fields in agricultural production. Water, sediment, macroinvertebrates, and fish were collected from various sites in the Basin in 1998 and analyzed for selenium. Selenium concentrations were analyzed spatially and temporally and compared to reported selenium toxic effect thresholds for specific ecosystem components: water, sediments, food-chain organisms, and wholebody fish. A selenium aquatic hazard assessment for the Basin was determined based on protocol established by Lemly. Throughout the Basin, water, macroinvertebrate, and whole fish samples exceeded levels suspected of causing reproductive impairment in fish. Population structures of several fish species implied that successful reproduction was occurring; however, the influence of immigration of fish from low-selenium habitats could not be discounted. Site-specific fish reproduction studies are needed to determine the true impact of selenium on fishery resources in the Basin. The U.S. Government’s right to retain a non-exclusive, royalty free license in and to any copyright is acknowledged.