APPLICATION OF GREY MODEL AND ARTIFICIAL NEURAL NETWORKS TO FLOOD FORECASTING1

The main focus of this study was to compare the Grey model and several artificial neural network (ANN) models for real time flood forecasting, including a comparison of the models for various lead times (ranging from one to six hours). For hydrological applications, the Grey model has the advantage that it can easily be used in forecasting without assuming that forecast storm events exhibit the same stochastic characteristics as the storm events themselves. The major advantage of an ANN in rainfall‐runoff modeling is that there is no requirement for any prior assumptions regarding the processes involved. The Grey model and three ANN models were applied to a 2,509 km² watershed in the Republic of Korea to compare the results for real time flood forecasting with from one to six hours of lead time. The fifth‐order Grey model and the ANN models with the optimal network architectures, represented by ANN1004 (34 input nodes, 21 hidden nodes, and 1 output node), ANN1010 (40 input nodes, 25 hidden nodes, and 1 output node), and ANN1004T (14 input nodes, 21 hidden nodes, and 1 output node), were adopted to evaluate the effects of time lags and differences between area mean and point rainfall. The Grey model and the ANN models, which provided reliable forecasts with one to six hours of lead time, were calibrated and their datasets validated. The results showed that the Grey model and the ANN1010 model achieved the highest level of performance in forecasting runoff for one to six lead hours. The ANN model architectures (ANN1004 and ANN1010) that used point rainfall data performed better than the model that used mean rainfall data (ANN1004T) in the real time forecasting. The selected models thus appear to be a useful tool for flood forecasting in Korea.     

Removal of organic matter and nitrogen from river water in a model floodplain

A significant improvement in river water quality cannot be expected unless nonpoint-source contaminants are treated in addition to the further treatment of point-source contaminants. If river water is sprayed over a floodplain, the consequent water filtration through the sediment profile can simultaneously remove organic matter and nitrogen in the water through aerobic and denitrifying reactions. This hypothesis was tested using lysimeters constructed from polyvinyl chloride (PVC) pipe (150 cm long, 15 cm in diameter) packed with loamy sand floodplain sediment. Water was applied to the top of the lysimeters at three different flow rates (48, 54, and 68 mm d(-1)). Concentrations of NO3 and dissolved oxygen (DO), chemical oxygen demand (COD), and redox potential (Eh) in the water were measured as functions of depth after the system reached steady states for both water flow and reactions. At the rate of 68.0 mm d(-1), a reducing condition for denitrification developed below the 5-cm depth due to the depletion of O2 by organic matter degradation in the surface oxidizing layer; Eh and DO were below 205 mV and 0.4 mg L(-1), respectively. At a depth of 70 cm, COD and NO3-N concentration decreased to 5.2 and 3.8 mg L(-1) from the respective influent concentrations of 17.1 and 6.2 mg L(-1). Most biodegradable organic matter was removed during flow and further removal of NO3 was limited by the lack of an electron donor (i.e., organic matter). These results indicate that the floodplain filtration technique has great promise for treatment of contaminated river water.     

Polychlorinated dibenzo-p-dioxins and dibenzofurans from food salt samples processed by heat treatment in Korea

Eight samples of processed food salt collected from five plants in Korea were analyzed for 2,3,7,8-substituted polychlorinated dibenzo-p-dioxins (PCDDs) and polychlorinated dibenzofurans (PCDFs) using liquid-liquid extraction, clean-up procedures, and high resolution gas chromatography-high resolution mass spectrometry. The study included the analyses of two kinds of salt product sample: bamboo-salt and parched salt. The levels of toxic PCDD/Fs found in the salt product samples were extremely low: the results revealed TEQ levels ranging between the sub pg TEQ/g and sub fg TEQ/g. The differences in the TEQ values of toxic PCDD/F were observed between the salt product samples, which were treated with different frequency of baking using four different fuels (firewood, pine wood, pine resin, and indirect heating by gas) at temperatures from 300 to 2000 degrees C. In bamboo-salt samples, the concentrations and TEQ values of toxic PCDD/Fs ranged between 0.57-66 pg/g and 5.7x10(-5)-0.64 pg TEQ/g, respectively. PCDD/Fs levels in bamboo-salts baked by firewood were found to be higher than those baked by pinewood or pine resin. In parched salt samples, the concentrations and TEQ values of toxic PCDD/Fs ranged between 0.97-3.7 pg/g and 0.0097-0.017 pg TEQ/g, respectively. The data was discussed regarding the concentration and the distribution pattern of congeners.     

Abundance and sources of hydrophilic and hydrophobic water-soluble organic carbon at an urban site in Korea in summer

In this study, the characteristics of total water-soluble organic carbon (WSOC) and isolated WSOC fractions were examined to gain a better understanding of the pathway of organic aerosol production. 24 h PM(2.5) samples were collected during the summer (July 28-August 28, 2009) at an urban site in Korea. A glass column filled with XAD7HP resin was used to separate the filtered extracts into hydrophilic (WSOC(HPI)) and hydrophobic (WSOC(HPO)) fractions. The origins of air mass pathways arriving at the sampling site were mostly classified into three types, those originating over the East Sea of Korea that passed over the eastern inland urban and industrial regions (type I); those from the marine (western/southwestern/southern marine) and passed over the national industrial complex regions (type II); and those from northeastern China that passed through North Korea and metropolitan areas of South Korea (type III). Measurements showed an increase in the average WSOC fraction of total OC from the type II to III air mass (53 to 64%) periods. Also, higher SO(4)(2-)/SO(x) (=SO(2) + SO(4)(2-)) was observed in the type III air mass (0.70) than those in the types I (0.49) and II (0.43). According to the average values of WSOC/OC and SO(4)(2-)/SO(x), measurements suggest that the aerosols collected during the type III air mass period were more aged or photo-chemically processed than those during the types I and II air mass periods. The relationship between the SO(4)(2-)/SO(x) and WSOC/OC (R(2) = 0.64) suggests that a significant fraction of the observed WSOC at the site could be formed by an oxidation process similar to SO(4)(2-) aerosols, probably the oxidation process using OH radicals, or in-cloud processing. The photochemical production of WSOC(HPO) was also observed to significantly contribute to the total OC.     

Removal of gaseous elemental mercury by dielectric barrier discharge

The removal of elemental mercury (Hg(0)) with the reactive species produced from dielectric barrier discharge (DBD) was studied. The effects of the operating parameters, such as the applied voltage, residence time, initial concentration and co-existence of other pollutants, were investigated. The removal of Hg(0) was significantly promoted by an increase in the applied voltage of the DBD reactor system. The presence of NO gas decreased the Hg(0) removal efficiency within the range of input powers tested compared to the case of Hg(0)-only due to the competition for ozone between Hg(0) and NO gas in the DBD reactor.     

Development of the Spatial Rainfall Generator (SRGEN) for the Agricultural Policy/Environmental Extender Model

Accurate spatial representation of climatic patterns is often a challenge in modeling biophysical processes at the watershed scale, especially where the representation of a spatial gradient in rainfall is not sufficiently captured by the number of weather stations. The spatial rainfall generator (SRGEN) is developed as an extension of the “weather generator” (WXGEN), a component of the Agricultural Policy/Environmental eXtender (APEX) model. SRGEN generates spatially distributed daily rainfall using monthly weather statistics available at multiple locations in a watershed. The spatial rainfall generator as incorporated in APEX is tested on the Cowhouse watershed (1,178 km²) in central Texas. The watershed presented a significant spatial rainfall gradient of 2.9 mm/km in the lateral (north‐south) directions based on four rainfall gages. A comparative analysis between SRGEN and WXGEN indicates that SRGEN performs well (PBIAS = 2.40%). Good results were obtained from APEX for streamflow (NSE = 0.99, PBIAS = 8.34%) and NO₃‐N and soluble P loads (PBIAS ≈ 6.00% for each, respectively). However, APEX underpredicted sediment yield and organic N and P loads (PBIAS: 24.75‐27.90%) with SRGEN, although its uncertainty in output was lower than WXGEN results (PBIAS: ?13.02 to ?46.13%). The overall improvement achieved in rainfall generation by SRGEN is demonstrated to be effective in the improving model performance on flow and water quality output.     

Prediction of effluent concentration in a wastewater treatment plant using machine learning models

Of growing amount of food waste, the integrated food waste and waste water treatment was regarded as one of the efficient modeling method. However, the load of food waste to the conventional waste treatment process might lead to the high concentration of total nitrogen(T-N) impact on the effluent water quality. The objective of this study is to establish two machine learning models—artificial neural networks(ANNs) and support vector machines(SVMs), in order to predict 1-day interval T-N concentration of effluent from a wastewater treatment plant in Ulsan, Korea. Daily water quality data and meteorological data were used and the performance of both models was evaluated in terms of the coefficient of determination(R~2), Nash–Sutcliff efficiency(NSE), relative efficiency criteria(d rel). Additionally, Latin-Hypercube one-factor-at-a-time(LH-OAT) and a pattern search algorithm were applied to sensitivity analysis and model parameter optimization, respectively. Results showed that both models could be effectively applied to the 1-day interval prediction of T-N concentration of effluent. SVM model showed a higher prediction accuracy in the training stage and similar result in the validation stage.However, the sensitivity analysis demonstrated that the ANN model was a superior model for 1-day interval T-N concentration prediction in terms of the cause-and-effect relationship between T-N concentration and modeling input values to integrated food waste and waste water treatment. This study suggested the efficient and robust nonlinear time-series modeling method for an early prediction of the water quality of integrated food waste and waste water treatment process.     

Enrichment of the metallic components from waste printed circuit boards by a mechanical separation process using a stamp mill

Printed circuit boards incorporated in most electrical and electronic equipment contain valuable metals such as Cu, Ni, Au, Ag, Pd, Fe, Sn, and Pb. In order to employ a hydrometallurgical route for the recycling of valuable metals from printed circuit boards, a mechanical pre-treatment step is needed. In this study, the metallic components from waste printed circuit boards have been enriched using a mechanical separation process. Waste printed circuit boards shredded to <10mm were milled using a stamp mill to liberate the various metallic components, and then the milled printed circuit boards were classified into fractions of <0.6, 0.6-1.2, 1.2-2.5, 2.5-5.0, and >5.0mm. The fractions of milled printed circuit boards of size <5.0mm were separated into a light fraction of mostly non-metallic components and a heavy fraction of the metallic components by gravity separation using a zig-zag classifier. The >5.0mm fraction and the heavy fraction were subjected to two-step magnetic separation. Through the first magnetic separation at 700 Gauss, 83% of the nickel and iron, based on the whole printed circuit boards, was recovered in the magnetic fraction, and 92% of the copper was recovered in the non-magnetic fraction. The cumulative recovery of nickel-iron concentrate was increased by a second magnetic separation at 3000 Gauss, but the grade of the concentrate decreased remarkably from 76% to 56%. The cumulative recovery of copper concentrate decreased, but the grade increased slightly from 71.6% to 75.4%. This study has demonstrated the feasibility of the mechanical separation process consisting of milling/size classification/gravity separation/two-step magnetic separation for enriching metallic components such as Cu, Ni, Al, and Fe from waste printed circuit boards.