Combining neural network models to predict spatial patterns of airborne pollutant accumulation in soils around an industrial point emission source

Neural networks (NNs) have the ability to model a wide range of complex nonlinearities. A major disadvantage of NNs, however, is their instability, especially under conditions of sparse, noisy, and limited data sets. In this paper, different combining network methods are used to benefit from the existence of local minima and from the instabilities of NNs. A nonlinear k-fold cross-validation method is used to test the performance of the various networks and also to develop and select a set of networks that exhibits a low correlation of errors. The various NN models are applied to estimate the spatial patterns of atmospherically transported and deposited lead (Pb) in soils around an historical industrial air emission point source. It is shown that the resulting ensemble networks consistently give superior predictions compared with the individual networks because, for the ensemble networks, R2 values were found to be higher than 0.9 while, for the contributing individual networks, values for R2 ranged between 0.35 and 0.85. It is concluded that combining networks can be adopted as an important component in the application of artificial NN techniques in applied air quality studies.     

Are savannas patch-dynamic systems? A landscape model

Savannas are ecosystems characterized by the coexistence of woody species (trees and bushes) and grasses. Given that savanna characteristics are mainly formed from competition, herbivory, fire, woodcutting, and patchy soil and precipitation characteristics, we propose a spatially explicit model to examine the effects of the above-mentioned parameters on savanna vegetation dynamics in space and time. Furthermore, we investigate the effects of the above-mentioned parameters on tree–bush–grass ratios, as well as the degrees of aggregation of tree–bush–grass biomass. We parameterized our model for an arid savanna with shallow soil depth as well as a mesic one with generally deeper and more variable soil depths. Our model was able to reproduce savanna vegetation characteristics for periods of time over 2000 years with daily updated time steps. According to our results, tree biomass was higher than bush biomass in the arid savanna but bush biomass exceeded tree and grass biomass in the simulated mesic savanna. Woody biomass increased in our simulations when the soil's porosity values were increased (mesic savanna), in combination with higher precipitation. Savanna vegetation varied from open savanna to woodland and back to open savanna again. Vegetation cycles varied over ∼300-year cycles in the arid and ∼220-year cycles in the mesic-simulated savanna. Autocorrelation values indicated that there are both temporal and spatial vegetation cycles. Our model indicated cycling savanna vegetation at the landscape scale, cycles in cells, and patchiness, i.e. patch dynamics.     

Co-occurrence profiles of trace elements in potable water systems: a case study

Potable water samples (N?=?74) from 19 zip code locations in a region of Greece were profiled for 13 trace elements composition using inductively coupled plasma mass spectrometry. The primary objective was to monitor the drinking water quality, while the primary focus was to find novel associations in trace elements occurrence that may further shed light on common links in their occurrence and fate in the pipe scales and corrosion products observed in urban drinking water distribution systems. Except for arsenic at two locations and in six samples, rest of the analyzed elements was below maximum contaminant levels, for which regulatory values are available. Further, we attempted to hierarchically cluster trace elements based on their covariances resulting in two groups; one with arsenic, antimony, zinc, cadmium, and copper and the second with the rest of the elements. The grouping trends were partially explained by elements’ similar chemical activities in water, underscoring their potential for co-accumulation and co-mobilization phenomena from pipe scales into finished water. Profiling patterns of trace elements in finished water could be indicative of their load on pipe scales and corrosion products, with a corresponding risk of episodic contaminant release. Speculation was made on the role of disinfectants and disinfection byproducts in mobilizing chemically similar trace elements of human health interest from pipe scales to tap water. It is warranted that further studies may eventually prove useful to water regulators from incorporating the acquired knowledge in the drinking water safety plans.     

Linking Hydrogeological and Ecological Tools for an Integrated River Catchment Assessment

Ecological (biological and hydrochemical assessment) and hydrogeological (vulnerability and pollution risk mapping) tools have been combined to assess the ecological quality and hydrogeological vulnerability of an agricultural river basin. In addition, the applicability of the recently developed vulnerability assessment approach (COP method) in the particular environmental conditions was tested by comparing its results with hydroecological assessment tools (i.e., pollution metrics). Five sampling sites were selected and sampled for benthic macroinvertebrates and physicochemical variables during summer and spring. Overall, sites ranged from moderate to poor ecological quality. The results illustrated that 26% of the study area was of moderate pollution risk, while 65% was classified as of low and very low risk zones. However, the higher elevation zones where calcareous rock formations are encountered presented moderate to high pollution risk that was accredited by the ecological quality assessment. Pollution metrics facilitated from hydrochemical analysis indicated a significant association with groundwater vulnerability, thus validating vulnerability and risk estimations. This study indicated that the particular groundwater pollution risk mapping methodology and the water quality assessment indices can be well combined to provide an integrated evaluation tool at a catchment scale.     

Determining the ventilation and aerosol deposition rates from routine indoor-air measurements

Measurement of air exchange rate provides critical information in energy and indoor-air quality studies. Continuous measurement of ventilation rates is a rather costly exercise and requires specific instrumentation. In this work, an alternative methodology is proposed and tested, where the air exchange rate is calculated by utilizing indoor and outdoor routine measurements of a common pollutant such as SO₂, whereas the uncertainties induced in the calculations are analytically determined. The application of this methodology is demonstrated, for three residential microenvironments in Athens, Greece, and the results are also compared against ventilation rates calculated from differential pressure measurements. The calculated time resolved ventilation rates were applied to the mass balance equation to estimate the particle loss rate which was found to agree with literature values at an average of 0.50 h^?1. The proposed method was further evaluated by applying a mass balance numerical model for the calculation of the indoor aerosol number concentrations, using the previously calculated ventilation rate, the outdoor measured number concentrations and the particle loss rates as input values. The model results for the indoors’ concentrations were found to be compared well with the experimentally measured values.     

The role of materials selection in the urban heat island effect in dry mid-latitude climates

This work investigates the role of materials selected for different urban surfaces (e.g. on building walls, roofs and pavements) in the intensity of the urban heat island (UHI) phenomenon. Three archetypal street-canyon geometries are considered, reflecting two-dimensional canyon arrays with frontal packing densities (λ_f) of 0.5, 0.25 and 0.125 under direct solar radiation and ground heating. The impact of radiative heat transfer in the urban environment is examined for each of the different built packing densities. A number of extreme heat scenarios were modelled in order to mimic conditions often found at low- to mid-latitudes dry climates. The investigation involved a suite of different computational fluid dynamics (CFD) simulations using the Reynolds-Averaged Navier–Stokes equations for mass and momentum coupled with the energy equation as well as using the standard k-ε turbulence model. Results indicate that a higher rate of ventilation within the street canyon is observed in areas with sparser built packing density. However, such higher ventilation rates were not necessarily found to be linked with lower temperatures within the canyon; this is because such sparser geometries are associated with higher heat transfer from the wider surfaces of road material under the condition of direct solar radiation and ground heating. Sparser canyon arrays corresponding to wider asphalt street roads in particular, have been found to yield substantially higher air temperatures. Additional simulations indicated that replacing asphalt road surfaces in streets with concrete roads (of different albedo or emissivity characteristics) can lead up to a ~5 °C reduction in the canyon air temperature in dry climates. It is finally concluded that an optimized selection of materials in the urban infrastructure design can lead to a more effective mitigation of the UHI phenomenon than the optimisation of the built packing density.     

Groundwater risk assessment at a heavily industrialised catchment and the associated impacts on a peri-urban wetland

Industrial and agricultural activities often impose significant pressures to the groundwater quality and consequently degrade wetland ecosystems that depend mostly on subsurface water flow. Groundwater vulnerability and risk mapping is a widely used approach to assess the natural protection of aquifers and the associated pollution potential from human activities. In the particular study, the relatively new Pan-European methodology (COP method) has been applied in a highly industrialized peri-urban wetland catchment, located close to Athens city, to map the intrinsic vulnerability of the aquifer and evaluate the risk potential originating from local land uses. Groundwater analysis results for various parameters, including Phenols, PCBs and nutrients, have been used to validate the vulnerability and risk estimations while a biological assessment occurred to associate the mapping results with the wetland's ecological status. The results indicated that even though the natural protection of the aquifer is relatively high due to the dominant hydrogeologic and geomorphologic conditions, the groundwater pollution risk is considerable, mainly because of the existing hazardous land uses. The water quality of the groundwater accredited these findings and the ecological status of this peri-urban wetland also indicated significant impacts from industrial effluents.     

Combining Neural Network Models to Predict Spatial Patterns of Airborne Pollutant Accumulation in Soils around an Industrial Point Emission Source

Abstract

Neural networks (NNs) have the ability to model a wide range of complex nonlinearities. A major disadvantage of NNs, however, is their instability, especially under conditions of sparse, noisy, and limited data sets. In this paper, different combining network methods are used to benefit from the existence of local minima and from the instabilities of NNs. A nonlinear k-fold cross-validation method is used to test the performance of the various networks and also to develop and select a set of networks that exhibits a low correlation of errors. The various NN models are applied to estimate the spatial patterns of atmospherically transported and deposited lead (Pb) in soils around an historical industrial air emission point source. It is shown that the resulting ensemble networks consistently give superior predictions compared with the individual networks because, for the ensemble networks, R² values were found to be higher than 0.9 while, for the contributing individual networks, values for R² ranged between 0.35 and 0.85. It is concluded that combining networks can be adopted as an important component in the application of artificial NN techniques in applied air quality studies.     

Soil column experiments used as a means to assess transport,sorption, and biodegradation of pesticides in groundwater

Soil column experiments are used to investigate the fate of three pesticides of high, intermediate, and low solubility in groundwater: N- phosphonomethyl glycine (glyphosate); O,O-diethyl-S-[(ethylthio)methyl]phosphorodithioate (phorate); (2,4-dichlorophenoxy)acetic acid (2,4-D). Feed solutions are prepared by adding each pesticide (100 mg/L glyphosate, 50 μ g/L phorate, 50 mg/L 2,4-D) along with conservative tracer, KBr, in synthetic groundwater. The concentration of the pesticides in effluents is detected by ion chromatography (glyphosate, 2,4-D) and GC-FID (phorate). The Br^? breakthrough curves are employed to estimate the dispersion coefficient and mean pore velocity in each column. Solute transport and reactive models accounting for equilibrium/non-equilibrium sorption and biodegradation are coupled with inverse modeling numerical codes to estimate the kinetic parameters for all pesticides.     

Immobilized-cell-augmented activated sludge process for treating wastewater containing hazardous compounds.

A novel bioaugmentation scheme called immobilized-cell-augmented activated sludge (ICAAS) was developed. Offline enricher reactors were used to maintain immobilized acclimated cells applied to augment completely mixed activated sludge (CMAS) treating a pentachlorophenol (PCP) pulse loading. Cellulose triacetate (CA) and powder activated carbon (PAC) combined with CA (PAC + CA) were the two media types used for entrapping the PCP-degrading culture. With ICAAS at 5% by volume augmentation, PCP removal of 73.1 and 75.1% via biodegradation, volatilization, and adsorption onto suspended cells, entrapped cells, and media was achieved for the systems with CA and PAC + CA media, respectively, while PCP removal in a control CMAS, which had a comparable level of combined PCP adsorption onto suspended cells and volatilization as the ICAAS, was 48.7%. Results further showed that the immobilized cells retained their PCP-degrading ability when they were fed with the inducer (PCP) once every 20 days.