• Journal of Korea Water Resources Association
• /
• v.53 no.6
• /
• pp.395-408
• /
• 2020

This paper aims to develop and apply three different machine learning approaches (i.e., artificial neural networks (ANN), adaptive neuro-fuzzy inference systems (ANFIS), and wavelet-based neural networks (WNN)) combined with an evolutionary optimization algorithm and the k-fold cross validation for multi-step (days) streamflow forecasting at the catchment located in Algeria, North Africa. The ANN and ANFIS models yielded similar performances, based on four different statistical indices (i.e., root mean squared error (RMSE), Nash-Sutcliffe efficiency (NSE), correlation coefficient (R), and peak flow criteria (PFC)) for training and testing phases. The values of RMSE and PFC for the WNN model (e.g., RMSE = 8.590 ㎥/sec, PFC = 0.252 for (t+1) day, testing phase) were lower than those of ANN (e.g., RMSE = 19.120 ㎥/sec, PFC = 0.446 for (t+1) day, testing phase) and ANFIS (e.g., RMSE = 18.520 ㎥/sec, PFC = 0.444 for (t+1) day, testing phase) models, while the values of NSE and R for WNN model were higher than those of ANNs and ANFIS models. Therefore, the new approach can be a robust tool for multi-step (days) streamflow forecasting in the Seybous River, Algeria.

• Korean Journal of Agricultural and Forest Meteorology
• /
• v.9 no.2
• /
• pp.109-120
• /
• 2007

The information on flowpath, storage, residence time, and interactions of water and carbon transport in a catchment is the prerequisite to the understanding and predicting of water and carbon cycling in the mountainous landscapes of Korea. In this paper, along with some up-to-date results, we present the principal methods that are currently used in HydroKorea and CarboKorea research to obtain such information. Various catchment hydrological processes have been examined on the basis of the water table fluctuations, the end-member mixing model, the cross correlation analysis, and cosmogenic radioactive isotope activity. In the Gwangneung catchment, the contribution of surface discharge was relatively large, and the changes in the amount, intensity and patterns of precipitation affected both the flowpath and the mean residence time of water. Particularly during the summer monsoon, changes in precipitation patterns and hydrological processes in the catchment influenced the carbon cycle such that the persistent precipitation increased the discharge of dissolved organic carbon (DOC) concentrated in the surface soil layer. The improved understanding of the hydrological processes presented in this report will enable a more realistic assessment of the effects of climate changes on the water resource management and on the carbon cycling in forest catchments.

• Journal of Korean Society of Environmental Engineers
• /
• v.34 no.7
• /
• pp.504-515
• /
• 2012

Groundwater is widely used as drinking water supplies around the world. However, microbial contamination of groundwater is a serious environmental problem that degrades drinking water quality and poses a great threat to human health. Among the pathogenic microorganisms such as viruses, bacteria, and protozoa, viruses are not readily removed during transport through soils, having high mobility in groundwater environment due to their smaller size compared to bacteria and protozoa. Studies regarding the fate and transport of viruses in soils and aquifers are necessary to determine the vulnerability of groundwater to microbial contamination and to secure safe drinking water sources. Also, these studies provide important information to establish the regulations and policies related to public health. This review paper presented the field and laboratory studies conducted for the fate and transport of viruses in subsurface environments. Also, the paper provided the factors affecting the fate and transport of viruses, the characteristics of bacteriophages used for virus studies, and virus transport model/colloid filtration theory. Based on this review work, future researches should be performed actively to set up the viral protection zone for the protection of groundwater from viral contamination sources. Especially, the researches should be focused on the development of mathematical models to calculate the setback distance and travel time for the viral protection zone along with the accumulation of information related to the model parameters.

• Journal of the Korean Association of Geographic Information Studies
• /
• v.15 no.1
• /
• pp.64-75
• /
• 2012

To predict the future distribution of forest vegetation, the present forest stand distributions of South Korea were represented by multinomial logit model with the following environmental variables: summer average precipitation, the coldest month average temperature, elevation, degree of base saturation, and soil organic matter. The future forest community was predicted by applying the MIROC3.2 hires A1B scenario. The future climate data were downscaled by statistically method. The coldest month average temperature increased $4.4^{\circ}C$, $6.0^{\circ}C$, and $9.4^{\circ}C$, and 3 months average precipitation changed -1.2%, 5.7%, and 5.3% for 2020s, 2050s, and 2080s respectively. For the projected summer precipitation and the coldest temperature, the future deciduous and mixed forests in the study area increased 56.9% and 8.3% and the coniferous forest decreased 11.2% in 2080s based on present.

• Journal of Korean Society of Forest Science
• /
• v.88 no.1
• /
• pp.38-46
• /
• 1999

Canopy structure conductances of a Norway spruce forest in the Solling Hills(Central Germany) and Central Forest Biosphere Reserve(320km to the north-west from Moscow) were derived from LE(latent heat flux) and H(sensible heat flux) fluxes measured(by Eddy correlation technique and energy balance method) and modelled(by one dimensional non-steady-state) SVAT(soil-vegetation-atmosphere-transfer) model(SLODSVAT) using a rearranged Penman-Monteith equation("Big-leaf" approximation) during June 1996. They were compared with canopy stomatal conductances estimated by consecutive intergrating the stomatal conductance of individual needles over the whole canopy("bottom-up" approach) using SLODSVAT model. The result indicate a significant difference between the canopy surface conductances derived from measured and modelled fluxes("top-down" approach) and the stomatal conductances modelled by the SLODSVAT("bottom-up" approach). This difference was influenced by some nonphysiological factors within the forest canopy(e.g. aerodynamic and boundary layer resistances, radiation budget, evapotranspiration from the forest understorey). In general, canopy surface conductances derived from measured and modelled fluxes exceeded canopy stomatal conductance during the whole modelled period, The contribution of the understorey's evapotranspiration to the total forest evapotranspiration was small (up to 5-9% of the total LE flux) and was not depended on total radiation balance of forest canopy. Ignoring contribution of the understorey's evapotranspiration resulted in an overestimation of the canopy surface conductance for a spruce forest up to 2mm/s(about 10-15%).

• Journal of the Korean Geosynthetics Society
• /
• v.13 no.2
• /
• pp.59-67
• /
• 2014

The amount of the contaminants that can be adsorbed on the drain was evaluated for the effective remediation of the contaminated soil, and the contaminants adsorptivity of the drain was evaluated by comparing the isothermal adsorption model after carrying out the contaminants adsorption test of the reactants coated on the surface of the drain. The reactant used in the experiment is a natural zeolite, and the contaminants are copper, lead and cadmium. The results that Freundlich and Langmuir adsorption isotherm model are compared to the adsorption amount according to the change of the initial concentration by the contaminants. As a result of the component analysis, because Si, Al and O are contained approximately 28%, 11% and 48%, respectively, it is identified that the material coated on the surface of the drain is the component of the zeolite which is the reactant for the adsorption of the heavy-metal (Cu, Pb, Cd) contaminants. The heavy-metal adsorption kinetic of the zeolite which is the reactant was decreased in order of lead, copper and cadmium. The important factor of the performance evaluation of the adsorbent is the reaction rate, and if zeolite is used as the reactant in the relationship between the maximum amount of adsorption and reaction rate, it can be utilized as the design factor that determine the removal order of the complex heavy-metal. In other words, because the maximum adsorption quantity of lead is smaller compared to copper but the reaction rate is relatively fast, it can be primarily removed, and copper can be removed after removing the lead. It was analyzed that Cadmium can be finally removed after that other heavy-metal is removed.

• Journal of the Korean Geotechnical Society
• /
• v.21 no.7
• /
• pp.55-62
• /
• 2005

The use of the bearing capacity equation is subjected to several uncertainties. In this study, estimation of the bearing capacity of footings based on the cone resistance q$_{c}$ is investigated. Non-linear finite element analyses based on a state-dependent stress-strain model were performed to obtain the load-settlement responses of axially loaded circular footings. Various soil and footing conditions, including different relative densities, depths of embedment, and footing diameters were considered in the analyses. Based on the finite element results, load-settlement curves were obtained and used to determine the unit limit bearing capacity in terms of the cone resistance q$_{c}$ for footings subjected to surcharge. Values of the unit bearing capacity for different embedment depths were in a narrow range, while considerable variation was observed with relative density D$_{R}$. It was observed that the unit limit bearing capacity normalized with respect to q$_{c}$ decreases as D$_{R}$ increases for a given surcharge.

• Journal of the Korean Geotechnical Society
• /
• v.29 no.7
• /
• pp.57-74
• /
• 2013

In this study, the virtual fixed point analysis and 3D full-modeling analysis for pile bent structures are conducted by considering various influencing factors and the applicability of the virtual fixed point theory is discussed. Also, a discrete analysis calculating separately both the superstructure and substructure of pile bent structures is performed on the basis of an equivalent base spring model by taking into account the major influencing parameters such as soil conditions, combined loading and pile diameter. The results show that the settlement and lateral deflection of the virtual fixed point theory are smaller than those of 3D full-modeling analysis. On the other hand, the virtual fixed point analysis overestimates the axial force and bending moment compared with 3D full-modeling analysis. It is shown that the virtual fixed point analysis cannot adequately predict the real behavior of pile bent structures. It is also found that discrete analysis gives similar results of lateral deflection and bending moment to those of unified analysis. Based on this study, it is found that discrete analysis considering column-pile interaction conditions is capable of predicting reasonably well the behavior of pile bent structures. It can be effectively used to perform a more economical design of pile bent structures.

• Journal of the Korean Association of Geographic Information Studies
• /
• v.15 no.3
• /
• pp.52-65
• /
• 2012

The net biomass accumulation (or net primary production, NPP) and gross primary production (GPP) have closely related with carbon accumulations(or carbon exchange) in vegetation. There are many approaches to estimate biomass using remote sensing techniques. The vegetation indices (VIs) can be a methodology to estimate biomass which assumes total chlorophyll contents. Various VIs were characterized with difference development conditions as vegetation species, input datasets. The hyperspectral data have also different spatial/spectral resolutions for aerial surveying. Additionally they need particular spectral bands selection difficulty to calculate the VIs. The objective of this study is to evaluate the correlations with airborne hyperspectral data (compact airborne spectrographic imager, CASI) and spectral unmixing model (or spectral mixture analysis, SMA) to characterize vegetation indices in forest area. The spectral mixture analysis was used to model the spectral purity of each pixel as an endmember. The endmembers are the fraction components derived from hyperspectral data through the SMA. In this study, we choose three endmembers represented vegetation pixels in the hyperspectral data. These endmembers were compared with 9 VIs by the Pearson's correlation coefficient. The results show MTVI1 and TVI have same correlation coefficient with 0.877. The MCARI, especially has very high relationship with vegetation endmembers as 0.9061 at less vegetation and soil distributed site. The MTVI1 and TVI have high correlations with the vegetation endmembers as 0.757 in whole test sites.

• Journal of Korea Water Resources Association
• /
• v.50 no.11
• /
• pp.759-767
• /
• 2017

Two main parameters of NRCS-CN method are curve numbers and intial loss ratio. They are generally selected according to the guideline of US National Engineering Handbook, however, they might cause errors on estimated runoff in Korea because there are differences between soil types and hydrological characteristics of Korean watersheds and those of United States. In this study, applying asymptotic CN regression method, we suggested eight modified NRCS-CN models to decide optimum runoff estimation model for Korean watersheds. RSR (RMSE-observations standard deviation ratio) and NSE (Nash-Sutcliffe efficiency) were used to evaluate model performance, consequently M6 for gauged basins (Avg. RSR was 0.76, Avg. NSE was 0.39) and M7 for ungauged basins (Avg. RSR was 0.82, Avg. NSE was 0.31) were selected. Furthermore it was observed that initial loss ratios ranging from 0.01 to 0.10 were more adequate than the fixed ${\lambda}=0.20$ in most of basins.