GLUE-based uncertainty assessment of SURR model in Imjin River basin

Trinh, Ha Linh Sejong University 2016 국내석사

The uncertainty in hydrological model can be caused by various sources, such as the model input data, non-optimal model parameters and model structures. For regional studies, the model parameters and uncertainty of input data are acknowledged as the two important sources of uncertainty. Particularly the Imjin river basin has the unique geographical characteristics with river cross operation between North and South of Korea. The insufficient metro-hydrological information in the northern region may lead to significant errors on the flow simulations. In the calibration and verification processes applied on the SURR (Sejong University Rainfall - Runoff) for rainfall - runoff simulations, it was found that the SURR simulated flows in the northern region of Imjin basin (Gunnam station) had the least accuracy as compared to the other two stations located in the southern area (Jeonkok and Jeogseong station) for all events (2007, 2008, 2009, 2010). This was due to the less precise calibrated parameters and the insufficient information of weather data in this station. Therefore the objective of the dissertation is to quantify the uncertainty of flow simulation based on GLUE (Generalized Likelihood Uncertainty Estimation) method model for the model parameter and precipitation input data in Imjin basin. To examine the uncertainty on streamflow simulation, two indexes are used: (i) p-factor, the ratio of the number of observations fall inside the uncertainty interval, and (ii) r-factor, the width of uncertainty interval. The results showed that the uncertainties of the simulated flow in the northern station are always high for both parameter and input uncertainty estimations. This was caused by the variations of two parameters, ALPHABF and SURLAG in the Gunnam station were significant larger than those of the southern stations. Meanwhile, the interpolated rainfalls in this area were also less accuracy due to the extremely far distance to the observed rain gauges in southern area. This reason explains why the input uncertainty in the northern area of Imjin basin always higher than the southern area. In addition, the dissertation also stated that the parameter uncertainty in streamflow was also influenced by the different periods of the hydrograph. The peak flows had higher uncertainty than the flows at the beginning or the end of each event.

One-dimensional flux power spectrum and cosmology with the SDSS-IV eBOSS lyman-α forest

이영배 Sejong University 2019 국내석사

As a unique tracer of the high-redshift cosmic web complementary to lower-z probes, the Lyman-α (Lyα) forest represents a powerful cosmological tool especially sensitive to the dark sector, via significant attenuation effects on the matter and flux power spectra at small scales. Among all Lyα forest observables, the small-scale one-dimensional flux power spectrum (P 1D F ) plays a key role, being highly sensitive to a wide range of cosmological and astrophysical parameters, neutrino masses, and the nature of dark matter. Hence, there is considerable interest in obtaining estimates of P 1D F from highquality data. To this end, we present here novel detailed measurements of P 1D F spanning the redshift interval 2.2 < z < 4.4, from the Data Release 14 (DR14) of the Extended Baryon Oscillation Spectroscopic Survey (eBOSS), part of the Sloan Digital Sky Survey IV (SDSS-IV). In particular, we carefully quantify and subtract all the contaminations induced by metals and pipeline noise, along with additional systematics caused for instance by the presence of Damped Lyman-α Systems (DLAs). We then confront our P 1D F measurements with state-of-the-art hydrodynamical simulations from the ‘Sejong Suite’, and find excellent agreement. Our new estimates of the small-scale eBOSS DR14 line-of-sight flux power spectra open the door for obtaining competitive cosmological constraints on neutrino masses, warm dark matter, and dark radiation – exploiting high-redshift quasars. Our study also provides a more solid understanding of the sensitivity and constraining power of the Lyα forest to the dark sector, and on the high-z small-scale clustering.

Development of a Hybrid Model for the Prediction of Precipitation and Discharge using Radar Reflectivity and Gated Recurrent Unit Networks

DINH THI LINH Sejong University 2023 국내박사

This dissertation is my research result during the 4 years that I studied and researched at Sejong University, Korea. The topic of this dissertation is related to the application of radar data in hydrological areas, specifically in quantitative precipitation estimation and in predicting streamflow and water level. The objectives of this dissertation are first, to study and propose a new framework for the estimation of radar precipitation and second, to develop a novel hybrid model for the prediction of discharge from radar reflectivity. Recently, the application of deep learning in hydrological areas is becoming more and more popular. The advantage of deep learning is the ability to analyze data and process large amounts of information with high quality. Algorithms in deep learning methods are based on statistical data, allowing them to detect dependencies more accurately, thereby enabling accurate predictions to be established. Thus, it could be a way to replace the physical model in predicting hydrological factors. Therefore, this dissertation has applied the deep learning approach, specifically the Gated Recurrent Unit (GRU) network, to serve the 2 major research contents mentioned above. For the purpose of studying and proposing a new framework for radar quantitative precipitation estimation (QPE), this dissertation has introduced a “complete” three-dimensional Constant altitude Plan Position Indicator (“complete” 3D CAPPI) with a cubic grid of 4x4 km2 and has investigated different CAPPI grid structures for an urban area in Seoul, Korea using the GRU model. The results show that the proposed framework is superior to other models in radar QPE with improvements by 3.3~46.2%, 25~70% and 1.2~44.8% in the root mean square error (RMSE), bias and correlation coefficient (CC) for 2 testing strategies including traditional test strategy and leave-one-out cross-validation (LOOCV) strategy. This proposed framework has contributed an important foundation for radar QPE and then is applied for the second objective in the dissertation. For the second objective, in developing a new hybrid model, namely the SGGP model, for the predictions of streamflow and water level, the SGGP model has applied the new framework for radar QPE as a result of the first objective mentioned above to modify the Spatial-scale Decomposition method (SCDM) for radar QPF (namely, mSCDM). Here, the QPE in the SCDM model is modified by replacing the Marshall-Palmer approach with a framework similar to the first research above. Then, utilizing the modified SCDM model obtained from the previous step, mSCDM, in combination with observation data (rain-gauge station data) using the GRU model (mSCDM-GRU model) to calibrate and generate 3-hour mean areal precipitation (MAP) forecasts, which has been used as input data for the predictions of streamflow and water level. After that, a model to predict hourly streamflow and water level from 3-hour MAP forecasts data and hydrological stations data has been established using a combination of the GRU model and particle swarm optimization (PSO) algorithm. The results in calibrating and generating 3-hour MAP forecasts indicate that utilizing the mSCDM-GRU model helps to significantly improve the accuracy of MAP forecasts, in which the CC values increase to over 0.6 at the time step of 180-minute, and are acceptable performance indicators like Critical success index (CSI), Probability of detection (POD), RMSE and Forecast Bias (FB), and are used as input data for the GRU-PSO to predict streamflow and water level. The performance of the proposed SGGP model has been evaluated for multistep-ahead predictions and compared with four competitive models. The predicted results from the SGGP model also illustrate the ability of accurate prediction through the statistical indicators of the RMSE, CC, Nash-Sutcliffe efficiency (NSE), Mean absolute error (MAE) and Mean absolute percentage error (MAPE) and peak errors. This SGGP model is a meaningful contribution to improving the accuracy of prediction results based on its high performance. In conclusion, the major contributions of this dissertation include proposing a new framework for radar QPE using a “complete” 3D CAPPI and deep learning approach; modifying the SCDM QPF framework (mSCDM) by applying the new framework for radar QPE; applying the mSCDM model and deep learning approach for establishing the mSCDM-GRU model to calibrating and generating 3-hour MAP forecasts; proposing the GRU and PSO algorithm for forming up the GRU-PSO model to predicting the streamflow and water level.

Improving the short-range flood forecasting capability of a coupled meteorological and hydrological model

Nguyen, Hoang Minh Sejong University 2019 국내박사

Flooding is one of the most serious and frequently occurring natural disasters in many regions around the world. In particular, given climate change, the impacts of flooding increase over time. To mitigate the damages caused by floods, flood forecasting is required to support water resource managers. Unfortunately, the accuracy of such forecasts remains limited because of the uncertainties that arise from various sources. These uncertainties may result in ineffective mitigation of flood damages. The coupling of meteorological and hydrological models is one of the most common methods for predicting streamflow. In this method, the capability of the streamflow predictions depends strongly on the performance of numerical weather prediction (NWP) model. However, the outputs of NWP models, especially quantitative precipitation forecasts (QPFs), are usually only somewhat skillful. This leads to the requirements for increasing the skills of QPFs. The improvements in QPFs are not required only in term of qualitative and quantitative skills but they also must involve the increase in probabilistic skill that is very valuable for the warnings of heavy rainfall and flooding. Because deterministic forecast may not really useful in the probabilistic forecast, ensemble predictions, which could address this problem, are thus becoming increasingly common and widely applied. Ensemble predictions allow analyzing qualitative and quantitative skills by using ensemble mean and probabilistic skill by considering all of their members. Ensemble QPFs can be derived from multiple runs of NWP model and statistically post-processing deterministic forecasts. By dint of the quick operation in ensemble streamflow prediction and non-requirement of high computational resources, the latter has been a subject of interest in several previous studies. However, the methods proposed in previous studies often require a large amount of historical forecast data for estimating parameter set. Hence, they are not suitable for applying to deterministic NWP models that have been launched operationally for a short time. Besides, these methods were only applied to medium-range (3-15 days) forecasts with accumulated rainfall in several hours or seasonal forecasts with daily rainfall totals. The applications for short-range (12-72 hours) forecasts with hourly rainfall have not been considered because hourly rainfall has a more greatly skewed distribution and higher intermittency of precipitation. Even though a method that could address these issues was proposed in a study, its effectiveness was not high due to the strong dependence of generated ensemble predictions on deterministic forecasts. Therefore, this study was conducted to propose a method, which could address the limitations of previous studies, for improving the capability of short-range rainfall and flood forecasts using a coupled meteorological and hydrological model. Three individual approaches including ensemble generation, blending, hydrological correction and real-time correction (Kalman filter technique) are developed and applied. To generate the ensemble precipitation prediction from single-valued rainfall forecast, the qualitative and quantitative skills are firstly evaluated at previous time step window and are then used to define the perturbation weights of QPF issued at current time by looking up a table. Ensemble QPF, which is generated as a kind of Monte-Carlo simulation, is highly dependent on the performance of deterministic NWP model because its members oscillate around the deterministic rainfall forecast. A blending technique is applied to mitigate this dependence by taking the advantages of radar-based rainfall prediction which is the best forecast for very short-range (6 hours ahead). This technique only reduces the rainfall intensity errors without considering the spatial rainfall errors (mis-location and extent) that have considerable contributions in the accuracy of rainfall and flood forecasts. Thus, a rainfall correction method that may involve the spatial rainfall errors is proposed. Target basin is located into different zones to account for the errors inside the basin and contiguous areas. Forecast errors in each zone basin are evaluated at previous time step window and then displaced to the target basin by performing a hydrological correction. Next, the forecast errors are continuously predicted and updated in real-time by using a Kalman filter technique and are then used to verify the rainfall forecast of each ensemble member; this process is called Kalman correction. Finally, each ensemble member is forced in turn as input for a rainfall - runoff model to obtain ensemble streamflow prediction. To validate the performance of the proposed method, it was applied to improve the capability of short-range rainfall and flood forecasts for the two flood events that occurred in 2013 and 2016 on the Yeongwol watershed using a coupled Local Data Assimilation and Prediction system (LDAPS) and Sejong University Rainfall - Runoff (SURR) model. Ensemble precipitation predictions were generated from the rainfall forecasts of the deterministic LDAPS model and were then blended with the McGill algorithm for precipitation nowcasting by Lagrangian extrapolation (MAPLE) rainfall predictions to produce hybrid rainfall forecasts. These hybrids continued to be verified by applying the rainfall correction method that involves the forecast errors caused by mis-location and extent. Eventually, each ensemble member of the verified hybrids was used to driver the SURR model to obtain the ensemble streamflow predictions on the watershed. The performance of the method was evaluated in term of qualitative (Proportion Correct (PC), Critical Success Index (CSI), Bias ratio, and False Alarm Ratio(FAR)), quantitative (Root Mean Square Error (RMSE), Mean Error (ME), Mean Absolute Error (MAE), Correlation Coefficient (CC), Model Efficient (NSE), and Absolute Relative in Volume (AREV)), and probabilistic (Brier score) skills. The results showed that the capability of rainfall and flood forecasts using the coupled model was improved sustainably step by step and exhibited the best skills after applying the final step that is Kalman correction. For rainfall forecast, the values of PC, CSI, and Bias averaged over total lead time were improved dramatically by 59%, 60%, and 106%, respectively, for flood event in 2013; 39%, 40%, and 257%, respectively, for flood event in 2016, whereas the improvements in RMSE, ME, MAE, and CC were 32%, 74%, 16%, and 32%, respectively, for flood event in 2013; 15%, 97%, 8%, and 33%, respectively, for flood event in 2016. The probabilistic skill increased dramatically with values of Brier score decreasing by 48% and 37% for flood events in 2013 and 2016, respectively. For flood forecasting, the improvements in RMSE, NSE, AREV, and ME were 30%, 63%, 24%, and 103%, respectively, in 2013; 42%, 42%, 53%, and 134%, respectively, in 2016. The success of this case study proved the viability of the methodology proposed in this study. A notable point is the application of the linear Kalman filter technique for correcting the rainfall forecast of each ensemble member, which may not produce the highest effectiveness of the rainfall correction method. Further study would be conducted to investigate the performance of non-linear ensemble Kalman filter technique for improving the effectiveness of the method.

A framework for uncertainty quantification and parameter optimization of rainfall-runoff models focusing on transboundary river basins

NGUYEN THI DUYEN Graduate School of Sejong University 2023 국내박사

Rainfall-runoff models (RRMs) have been an effective tool for several purposes in hydrology during the past decades. Nevertheless, RRMs are often posed with uncertainty and parameter optimization, especially for transboundary river basins due to inconsistent hydrological data. The uncertainty of an RRM may be brought about by several sources. The parameter optimization of an RRM depends on the calibration strategy. This study proposed a novel framework to (1) evaluate the uncertainty and (2) optimize the parameters of RRMs focusing on transboundary river basins. The first study objective is implemented by a methodological comprehensive integration. They are the performance measures of streamflow simulation, the delayed rejection adaptive Metropolis (DRAM), and the three uncertainty assessment measures. The second study objective is acquired by combining the particle swarm optimization (PSO) algorithm with the objective functions. Two other optimization algorithms, grid search and random search, are used to confirm the optimization performance of the PSO algorithm. A case study was conducted to test the proposed framework. It is the Sejong university rainfall-runoff model (SURR) applied for the Imjin transboundary river basin. The Imjin River basin is one of the seven largest rivers in Korea. Five-year extreme flood events from 2008, 2009, 2010, 2013, and 2017 were used in the case study. The two events of 2008 and 2009 were utilized in the methods’ calibration. The three events of 2010, 2013, and 2017 were used in the methods’ verification. Some findings from the case study were obtained. The uncertainty of the SURR model was quantitatively and comprehensively assessed. Two sources of the parameters and the input data were selected to quantify the uncertainty of the SURR model. The sensitivity of the subjective model parameters was analyzed. The certain effects of the four performance measures in blending with the adaptive DRAM algorithm were confirmed by the three uncertainty evaluation criteria. In addition, the PSO algorithm outperformed grid search and random search under the four objective functions. The parameters of the SURR model were effectively optimized by the PSO algorithm. The findings from the case study demonstrated the practicality of the proposed novel framework. The contribution of the study is the application of the DRAM algorithm to estimate the uncertainty of an RRM. Moreover, the DRAM algorithm is combined with performance measures of simulation for uncertainty estimation in an RRM. In addition, this thesis proposes a comprehensive method of three uncertainty assessment measures to evaluate the uncertainty of an RRM. Therefore, the uniqueness of the study is that it provides a comprehensive quantification of the uncertainty and an effective optimization of parameters for RRMs focusing on transboundary river basins.

Size dependent characteristics and ginsenosides bioconversion efficacy of red ginseng powder

김영은 Sejong University 2018 국내박사

인삼 (Panax ginseng C.A. Meyer)은 우리나라 및 아시아에서 가장 일반적으로 사용되는 전통 한약재이다. 시중에서 판매되는 인삼은 수삼, 백삼 그리고 홍삼으로 분류된다. 홍삼은 수삼을 98 ~ 100 ℃에서 2 ~ 4 시간 동안 쪄서 건조시켜 제조한다. 홍삼은 고유한 진세노사이드 프로파일을 가지고 있으며, 진세노사이드는 온도, pH, 효소 및 미생물의 반응에 의해 탈글리콜화 되고, 이렇게 생성되어지는 진세노사이드의 약용 가치가 있다는 연구결과들을 다수 확인할 수 있다. 홍삼이 함유하고 있는 진세노사이드는 인체에서 대부분 흡수되지 않아, 다른 형태의 진세노사이드로 전환되어야 하는 진세노사이드 Rb1, Rb2, Rc 및 Rd를 함유한다. 체내 흡수가 용이하고, 생체이용율이 높은 진세노사이드 Rg3, Rg5, Rk1, Rh1, F2, compound K 및 PPD와 같은 진세노사이드는 Rb1, Rb2, Rc 및 Rd와 같은 기질 사포닌으로부터 위산, 효소 및 미생물에 의해 전환되어 생성된다. 최근 초미세 분쇄 기술에 대한 관심 및 연구가 많아지고 있다. 초미세분쇄 기술은 비표면적을 최대화하여 용해도 및 분산성을 향상시키는 것이특징이며, 재료의 물리적 손상과 화학적 변형을 최소화하면서 손상을 최소화 할 수 있다는 장점을 가졌다. 본 연구에서는 홍삼분말의 입자크기가 진세노사이드의 생체 및 효소전환에 미치는 영향을 조사하기 위해 먼저 홍삼분말의 사이즈를 다르게 제조하였다. 일반적인 분쇄공정을 거쳐 제조되어진 홍삼분말 (CRP, Coarse red ginseng powder), 그리고, 초미세분말 제조장비를 이용하여 1차 제조되어진 미세홍삼 분말 (FRP, Fine red ginseng powder) 이후 3단계의 초미세 공정을 거쳐 제조되어진 3가지 사이즈 (32 μm, 23 μm, 20 μm)의 초미세 홍삼분말 (URP, red ginseng powder)과 같은 5가지의 사이즈가 다른 홍삼 분말을 준비했다. 준비된 홍삼분말CRP, FRP 및 URPs의 평균 입자 크기는 각각 509.799 μm, 93.848 μm, 32.814 μm, 23.164 μm, and 20.814 μm 이었고, Uniformity 값은 0.5 ~ 0.8 범위 이내였다. 진세노사이드의 전환 등에 가장 크게 영향을 줄 것이라고 판단되는 비표면적 값은 각각 0.031 ± 0.001, 0.202 ± 0.001, 0.412 ± 0.003, 0.471 ± 0.002, and 0.513 ± 0.002 m2/g이었다. 이 결과를 토대로 입자가 작아질수록 비표면적이 커짐을 확인 할 수 있었다. 비표면적이 넓어질수록 체내흡수 및 기능성이 증가한다는 가설 확인을 위해 DPPH 라디컬 소거능을 확인해본 결과 사이즈가 비표면적이 넓어질수록 항산화 효능이 커짐을 확인할 수 있었다. 인삼 및 홍삼을 이용하여 제조되어진 초미세 분말은 대부분 추출물의 수율 향상 및 용해도 증진에 대한 연구에 이용되며, 체내흡수율 및 효능에 대한 연구는 추출물을 이용한 것이 대부분이다. 하여 본 연구에서는 제조되어진 5가지 홍삼분말을 이용하여, 체내흡수가 용이한 형태의 진세노사이드로의 전환율을 검증하는 연구를 수행하였다. 먼저 in vitro 위장관 조건에서의 진세노사이드 전환율 확인을 위해 사이즈가 다른 홍삼분말을 물에 넣은 후 37 ℃의 항온 배양기에서 펩신 용액과 리파아제, 담즙 추출물 및 팬 크레아틴 혼합물을 첨가 하였다. 반응시간 동안 Rg3, Rg5, Rk1 및 PPD와 같이 생성되는 진세노사이드는 시간에 따라 증가하였고, 기질 화합물 (ginsenoside Rg1, Re, Rb1, Rb2, Rc 및 Rd)은 감소하였다. 홍삼 분말의 사이즈가 작아질수록 증가 된 비표면적은 위장관 상태에서 빠르게 기질 화합물을 전환시켜 체내흡수가 용이한 형태의 진세노사이드를 생성하였다. URP-23과 URP-20 이 FRP 보다 많은 기질 ginsenosides를 전환시킨 이유는 2 단계 분쇄에 의해 얻어진 URPs가 작은 입자의 형성과 그에 따른 비표면적의 증가로 인해 FRP보다 빠른 방출력을 가졌기 때문이라고 생각된다. 그리고 종양의 형성을 차단하고 암세포의 침입을 억제함으로써 암의 발생 및 전이를 예방한다고 보고되어졌고, 진세노사이드 Rb1, Rb2, Rc 및 Rd의 주요 대사 산물인 진세노사이드 F2와CK의 생성에 관해 연구하였다. CK는 홍삼 뿌리에서는 발견되지 않지만, 대장에서 효소의 작용에 의해 진세노사이드 Rb1, Rb2, Rc 및 Rd가 CK로 전환되어 우수한 생체 이용률을 갖는다. 진세노사이드F2와 CK의 효소 전환은 pH 4.3과 55 ± 1 ℃에서 cytolase PCL 5를 사용하여 수행되었다. 홍삼 분말의 입자 크기가 작을수록 진세노사이드 F2와 CK의 전환율이 증가함을 확인할 수 있었다. 이처럼, 홍삼 분말의 입자 크기가 작아짐에 따라 체내대사 중에 항산화 효과가 증가함을 확인 할 수 있었으며, 생체 및 효소 처리를 통한 대사에서 체내흡수가 용이한 형태의 진세노사이드의 양이 증가 하는 등의 결과는 홍삼 분말의 입자 크기를 줄이는 것이 인체에서의 진세노사이드의 생체 이용률 향상에 크게 기여 함을 시사한다. Red ginseng contains ginsenoside Rg1, Re, Rb1, Rb2, Rc, and Rd which are hardly absorbed in the human body but are converted to other forms of ginsenosides. Bioavailable ginsenosides such as Rg2, Rg3, Rg5, Rk1, F2, compound K, PPT and PPD are produced through bioconversion from substrate ginsenosides such as Rg1, Re, Rb1, Rb2, Rc, and Rd by gastric juices, enzymes, and microorganisms. In this study, the effects of particle sizes of red ginseng powder on its ginsenoside bioconversion under simulated gastrointestinal conditions or enzymatic treatment were investigated in order to improve the bioavailability of red ginseng powders. Specially, the ultrafine milling technique that has the advantage of maintaining the physicochemical properties of plant materials and increasing the extraction yield and solubility, was applied to prepare red ginseng powders with five different particle sizes such as coarse red ginseng powder (CRP), fine red ginseng powder (FRP), and ultrafine red ginseng powders of three different sizes (URP-32, 23, 20). The average particle sizes of CRP, FRP, URP-32, URP-23 and URP-20 were determined to be 509.799, 93.848, 32.814, 23.164, and 20.814 μm, respectively. The uniformity value ranges of CRP, FRP, URP-32, URP-23 and URP-20 were in a range of 0.555-0.815. For bioconversion tests under simulated gastrointestinal conditions, red ginseng powders were dissolved in water, pepsin solution and a mixture of lipase, bile extract, and pancreatin was added to the solutions, followed by the incubation at 37°C. During the bioconversion tests, the levels of the substrate compounds (ginsenoside Rg1, Re, Rb1, Rb2, Rc, and Rd) decreased while those of the product compounds such as Rg3, Rg5, Rk1, and PPD increased over time. The total conversion rates of substrate ginsenosides under the simulated gastrointestinal conditions were 31.19%, 40.19%, and 45.79% for FRP, URP-23, and URP-20, respectively. The DPPH radical scavenging activities of CRP, FRP, URP-32, URP-23 and URP-20 were 0.32%, 0.90%, 1.43%, 1.72%, and 2.25%, respectively. The antioxidant effect of FRP, URP-23, and URP-20 increased as the particle sizes of the red ginseng powders decreased. Compound K, a major metabolite of ginsenosides Rb1, Rb2, Rc, and Rd, has been recognized to prevent the development and the metastasis of cancer by blocking the formation of tumors and suppressing the invasion of cancerous cells. Specific surface area determination supported this observation. When the specific surface areas of red ginseng powders were larger, substrate ginsenosides were more rapidly bioconverted under the gastric condition. The bioconversion of substrate ginsenosides into ginsenoside F2 and compound K was implemented using cytolase PCL 5 at pH 4.3 and 55 ± 1°C. The contents of ginsenoside F2 and compound K after enzymatic treatment were 11.22 mg/g, 10.25 mg/g, and 5.06 mg/g in the case of URP-23, URP-20, and FRP, respectively. The rates of enzymatic conversion into ginsenoside F2 and compound K increased as the particle sizes of the red ginseng powders decreased. Based on the results, two types of bioconversion, one under simulated gastrointestinal conditions and the other through enzymatic treatment of product ginsenosides increased as the particle sizes of the red ginseng powders decreased. These findings suggest that the reduced particle size of red ginseng powder can contribute significantly to the enhancement of the bioavailability of ginsenosides in the human body.

Identification and characterization of cell surface major vault protein (csMVP) in circulating tumor cells (CTCs) derived from liver cancer patients

이현민 Sejong University 2018 국내박사

Circulating tumor cells (CTCs) in peripheral blood play a major role in the initiation of hepatocellular carcinoma (HCC) metastasis and recurrence even after curative treatments. Major vault protein (MVP) is upregulated during malignant progression and drug resistance in various cancer cells. MVP has been known as a cytonuclear protein, but we found that MVP is expressed on the surface of HCC cells but not on the surface of normal hepatocytes. MVP knockdown decreases HCC cell growth and increases cell death, and treatment with anti-MVP antibodies (α-MVPs) recognizing cell surface MVP (csMVP) inhibits HCC cell proliferation, suggesting that csMVP contributes to HCC cell proliferation. The expression of csMVP is induced under stressful environments, and cell sorting revealed that csMVP (+) HCC cells have a higher clonogenic survival than csMVP (-) HCC cells. Treatment with α-MVP inhibits clonogenic survival and invasive and migratory potential of HCC cells, suggesting that csMVP contributes to cell survival, invasion and migration. csMVP-associated signaling also supports for the role of csMVP in the survival and metastasis of HCC cells. csMVP (+) CTCs are detected in HCC patients (approximately 88%) but not in healthy donors, and the number of csMVP (+) CTCs is further increased in cancer patients with liver metastases. csMVP is only detected in CTCs with mesenchymal phenotype or intermediate phenotype with neither epithelial nor mesenchymal markers. The results suggest that csMVP is a novel surface marker on CTCs with nonepithelial phenotype and promotes cancer progression through improving survival of CTCs in HCCs. 암 환자의 혈액에 존재하는 순환종양세포(Circulating tumor cells, CTCs)은 간암의 초기 전이나 재발에 중요한 역할을 한다. Major vault protein (MVP)는 다양한 암세포에서 악성으로 진행되고, 약물의 저항성 유발시 발현이 증가한다. MVP는 현재까지 세포질과 핵 단백질로 알려져 있었으나, 본 논문에서는 정상 간세포의 표면에는 발현하지 않으나, 다양한 간암세포 표면에서 발현하는 MVP를 동정 및 특성을 분석하였다. MVP를 결핍시킨 간암세포에서 세포 주기(Cell cycle)이 G2/M기에 정지되어 간암세포의 증식이 크게 감소되고, 세포 사멸도 크게 증가하였다. 세포 표면 MVP 항체를 처리해 본 결과 역시 간암세포의 증식이 억제되었다. 이러한 결과는 csMVP가 간암세포의 증식에 관여한다는 것을 알 수 있었다. csMVP(+) 간암세포와 csMVP(-) 간암세포를 분리하여 clonogenic survival을 분석한 결과 csMVP(+) 간암세포의 생존율이 csMVP(-) 간암세포의 생존율 보다 높게 나타났다. 동일 실험을 MVP항체를 처리한 것과 처리하지 않은 것으로 진행하였을 때 MVP항체를 처리한 군에서 생존율이 크게 떨어지는 것을 확인 하였으며, Migration과 Invasion 역시 크게 감소하는 것을 확인 할 수 있었다. 이러한 결과를 토대로 csMVP가 간암세포의 생존과 Migration 및 Invasion에 크게 관여하는 것을 알 수 있었다. 이러한 결과를 토대로, csMVP는 종양형성과 전이에 있어서 중요한 역할을 하고 있으며 FAK/mTOR 신호를 통해 간암의 성장과 전이를 조절 하는 것이라 생각된다. csMVP는 간암환자 혈액에 존재하는 CTC는 검출하지만, 정상인의 혈액세포에는 발현하지 않는다. 또한 타장기에서 간으로 전이된 암환자의 혈액에서 csMVP(+)-CTCs의 수가 크게 증가함을 알 수 있었다. 추가적으로, 이러한 csMVP(+)-CTCs가 중간엽성 (mesenchymal) 표현형을 보이는 것을 확인 할 수 있었다. 이러한 결과는 csMVP가 중요한 비상피성 표현형을 갖는 CTCs의 마커이며, 또한 csMVP(+)-CTCs는 암의 진행 및 생존에 있어서 중요한 역할을 할 것이라 제안 할 수 있겠다.

Redox behavior of some selected metals in room temperature ionic liquids

Dilasari, Bonita Sejong University 2016 국내박사

Room temperature ionic liquid (RTIL) is a salt, consisting of large asymmetric cation and complex anion, which is stable in liquid phase below 100°C. Properties of RTILs are solely dependent on their constituent cation and anion. In terms of electrochemistry, the most significant properties of RTILs that should be considered are viscosity, conductivity, and electrochemical stability window. Electrochemical stability window, which refers to a potential range where cation and anion are neither reduced nor oxidized, of RTILs is typically wider than aqueous electrolytes. This feature enables electrodeposition of metals with large negative reduction potential such as aluminum, zinc, magnesium, and lithium, which are otherwise unable to be recovered in aqueous solvents. In battery application, replacing conventional organic solvents with nonvolatile and nonflammable RTILs would help to improve battery safety. This dissertation focuses on the electrochemical reactions of metal electrodes in RTILs, including electrodeposition and corrosion. Three different topics regarding the oxidation and reduction reactions of some selected metals in RTILs are discussed in this dissertation. The effect of RTIL structures on the electrodeposition/dissolution behavior of lithium is investigated in the first topic. The second topic is about the oxidation and reduction reversibility of zinc anode and the effect of water addition in RTIL-based zinc-air battery application. In the third topic, the corrosion susceptibility of some selected metals is evaluated in a protic RTIL containing available proton on cation, and compared with aprotic one with a similar structure. RTIL structure is confirmed to affect the electrodeposition/dissolution potential of lithium. The largest portion of lithium metal in deposits is obtained from a sample electrodeposited in RTIL with the widest electrochemical stability window, 1-butyl-1-methylpyrrolidinium bis(trifluoromethylsulfonyl)imide ([BMPyr] [NTf2]). In regard to the application of RTILs as an electrolyte for zinc-air battery, zinc electrode shows reversible oxidation-reduction reactions in [BMPyr] [NTf2]. However, the presence of water gives a significant effect on the redox reactions of zinc electrode. Protic RTIL (1-butylpyrrolidinium bis(trifluoromethylsulfonyl)imide) is confirmed to show a narrower electrochemical stability window compared to that of aprotic RTIL ([BMPyr] [NTf2]) of similar structure and is more reactive towards metals. Lower corrosion potentials and higher corrosion current densities are observed in the protic RTIL.

(The) relationships among quality, emotions, satisfaction, desire and behavioral intentions in hotel spas

Pham Thi Mai Thuong Sejong University 2016 국내석사

The purpose of this paper is to clarify the relationships among quality, emotion, satisfaction, and behavioral intentions in a spa business context. This paper aims to find out the influence of spa quality to customer emotions, as well as how customer emotions influence customer satisfaction or the way emotions influence customer desire to stay, how satisfaction and desire to stay influences behavioral intentions. The data for this study was collected by an online survey. The sample in this study included 300 people who visited to the spa in the deluxe hotels for at least one time in the last three years (from the time of the interview). They were subjected to a series of multiple regression analyses. The Baron and Kenny approach was employed to test mediating impacts. The results of this study indicated that three dimensions of spa quality: tangibility, assurance and empathy were significant factors in explaining the pleasure emotions of customer. However, responsiveness and arousal emotions are not significant variables in all the relationships with other variables. Customer pleasure emotion was proven to have significant impact on satisfaction and desire to stay. Besides, customer satisfaction and desire to stay were also proven to have important roles in predicting customer behavioral intentions. Additionally, the outcomes of this study indicated that pleasure, satisfaction and desire to stay also play the mediating roles in the model. The results offer managerial insights into marketing strategies, design, and operational and human-resource management. The results of this study give opportunities for spa managers to have more effective planning, financial allocation and managerial decision-making with respect to spa service quality so that they can satisfy the needs of the consumers. Through the training of staffs to capture the emotions of customers, spa managers can set out strategies to promote customer emotions and customer satisfaction, so that they can achieve the goal of improving customer behavioral intentions. Key words: spa quality, customer satisfaction, emotion, pleasure, arousal, desire, behavioral intentions

Preparation of EGCG and piperine coloaded nanocarriers and evaluation of their stability, release property, and antioxidative activity

김혜원 Sejong University 2016 국내석사

에피갈로카테킨 갈레이트(EGCG)는 녹차 카테킨의 한 종류로 항산화능, 항염작용이 있는 물질로 노화방지에 큰 도움을 주며, 이와 관련된 많은 연구가 진행 중이다. 또 다른 물질인 피페린은 후추의 지표성분으로서 기관지염, 소화장애, 관절염 등에 효능이 있는 것으로 알려져 있다. 특이적으로 피페린은 글루콘산화(glucuronidation)를 저해하는 기능이 있어, 어떤 물질의 분해를 지연시킴으로써 생체 내에서 보다 머물러 생체이용률을 높일 수 있다. 하지만 EGCG와 피페린은 구조적 불안정성으로 인해 열, 산소, 빛에 노출되면 쉽게 산화될 가능성이 있다. 이러한 EGCG와 피페린의 제한을 극복하고자 두 물질을 나노전달체에 탑재하여 이용을 극대화 하고자 한다. 따라서 본 연구에서는 EGCG와 피페린을 나노전달체에 동시 탑재하고, 이의 이화학적 특성, 저장 안정성, 방출 특성 및 항산화능을 평가하였다. 나노전달체는 초고압균질법 (High pressure homogenization)을 이용하여 solid lipid nanoparticle (SLN)과 nanostructured lipid carriers (NLC) 두 가지 형태로 제조되었다. 이 두 가지의 나노전달체는 지질 성분으로 구성 된 나노전달체인 공통점을 갖지만, 나노전달체의 중심부에 존재하는 지질의 성상에 따라 구분할 수 있다. 나노전달체의 중심부에 고체 지질만을 포함하는 것이 SLN이며 고체지질과 액체지질을 둘 다 함유한 것이 NLC이다. SLN과 NLC 각각에 EGCG와 피페린을 동시에 탑재하고 이화학적 특성을 분석한 결과, SLN은 평균 150-200 nm 크기의 나노입자가 형성되었으며 NLC의 경우 평균 100 nm 크기의 나노입자가 형성되었다. 형성 된 입자를 4주간 저장했을 때 NLC가 SLN보다 초기형태를 더 유지했으므로 저장 안정성이 더 높았다. 또한 NLC가 SLN보다 효율적으로 캡슐화되는 경향을 보였다. 이 후, 제조 된 SLN과 NLC는 in vitro 소화 모방조건에 적용하여 EGCG와 피페린의 방출특성과 항산화능을 분석한 결과, EGCG와 피페린은 모방 위장조건에서 보호 및 유지되다가 모방 소장조건에 이르자 캡슐화되었던 두 코어물질이 방출되는 것을 통해 SLN과 NLC 구조로 방출을 조절할 수 있음을 확인하였다. 또한 ABTS와 DPPH 방법을 이용하여 제조 된 SLN과 NLC의 항산화능을 평가하였다. 캡슐화하지 않은 EGCG와 피페린은 in vitro 소화 모방조건에서 자유 라디칼 소거능을 급격히 감소했지만 SLN과 NLC에 탑재 된 EGCG와 피페린의 경우 in vitro 모방 위장조건에서 급증한 것을 확인하였다. 이는 SLN과 NLC구조가 EGCG와 피페린의 항산화능을 보다 오랫동안 유지하도록 했음을 뜻한다. 또한 EGCG를 나노전달체에 단일탑재한 경우보다 피페린과 동시 탑재했을 때 항산화능이 더 높게 나타났다. 이는 피페린의 특이적인 성질로 글루콘산화를 저해하여 EGCG의 분해를 지연시켜 생체이용률을 높인 것으로 보인다. 결론적으로 SLN과 NLC는 방출특성 및 항산화능 측면에서 유사한 기능을 하였지만, 이화학적 특성이나 캡슐화 효율 측면에서 NLC가 EGCG와 피페린을 동시탑재하기에 더 적합한 형태임을 확인하였다. 일반적으로 NLC를 제조할 때, poly-vinyl alcohol (PVA) 코팅을 하여, 보다 안정적인 NLC를 제조한다. 하지만 이는 비 식용물질이기 때문에 식품이나 음료에 적용하는데 한계가 있다. 따라서 비 식용물질인 PVA를 대체할 수 있는 식용 가능한 whey protein isolate (WPI)나 키토산을 이용해 NLC를 외부 코팅하여 코어물질의 손실을 최소화 하고자 했다. 따라서 코팅물질에 따라 NLC를 다르게 제조하고 이를 물, 0.02 M 인산완충용액, 가상음료에 적용하여 시간에 따라 EGCG와 피페린이 방출되는 경향을 분석하였다. PVA와 WPI를 이용한 NLC는 물과 0.02 M 인산완충용액에서 비슷한 경향을 보였으나, 산성을 띄는 가상음료에서는 WPI의 단백질 변성으로 인해 부적합함을 확인하였다. 결론적으로 본 연구를 통해 EGCG와 피페린을 동시 탑재한 나노전달체를 제조하였고 이화학적 특성, 저장안정성, 방출특성 및 항산화능을 평가하였다. 제조 된 나노전달체는 EGCG의 안정성과 생체이용률을 증가시키는데 도움을 주며, 다양한 건강기능성 성분에 적용 가능할 것으로 예상된다. 주요어: EGCG, 피페린, 나노전달체, 안정성, 방출특성, 항산화능 Epigallocatechin-3-gallate (EGCG) is the major component of catechin from green tea. It has been demonstrated to exert anti-inflammatory as phytotherapeutics and antioxidative properties as a polyphenol component (Singh, Shankar, & Srivastava, 2011). In addition, EGCG is effective component for adult disease like hyperlipidemia and artery hardening by increasing high density lipoprotein (HDL) cholesterol level and decreasing low density lipoprotein (LDL) cholesterol level and triglyceride (Raederstorff, Schlachter, Elste, & Weber, 2003). Additionally, it has been shown that piperine, an alkaloid-amine derived from black pepper and white pepper, is effective for bronchitis, gastrointestinal disorder, and arthritis. Also, it has reported that breast cancer, colon cancer, and lung cancer can be prevented by piperine (Bano, Raina, Zutshi, Bedi, Johri, & Sharma, 1991). Generally when some chemicals which has the health functionality come into the body, these are degraded in the liver or released outside the body via glucuronidation. So, these chemicals cannot be used effectively in the body for a long time. This glucuronidation can be overcome by piperine. Because piperine as bio-enhancer has an effect on inhibiting the glucuronidation, released time of some chemicals can be delayed. So it can be expected to overcome the low availability of core materials and piperine will be taken by a role of increasing the bioavailability of bioactive substances (Moorthi & Kathiresan, 2013; Patil, Singh, & Chakraborty, 2011). However, EGCG and piperine as core materials have limiting factors in structural instability because of oxidation, heat, pH or light. In order to protect EGCG and piperine continuously, these core materials were needed for loading in delivery system until they were absorbed in our body. There were a lot of drug delivery systems such as conventional emulsions, nanoemulsions, multilayer emulsions, hydrogel particles, liposomes and so on. Among these drug delivery systems solid lipid nanoparticles (SLN) and nanostructured lipid carriers (NLC) was selected by nanocarriers. It was because SLN and NLC as colloidal nanocarriers based on lipid have been reported that they were effective at the loading efficiency, encapsulation efficiency, and release controlling than any other conventional emulsion. Both of nanocarriers were colloidal system based on lipid in common. However, there was a difference about interior space of nanoparticles. In case of SLN, it consists of solid lipid as like glycerol monostearate (GMS). But NLC contained both solid lipid and oil as like oleic acid (OA). So it can lead to difference of physicochemical properties, stability, release property, and antioxidative activity. A lot of researches about comparison of single loaded SLN and NLC have studied recently. But there were few studies about coloading core materials in SLN and NLC. Therefore, the purpose of this study was prepare SLN and NLC coloaded with EGCG and piperine respectively, and to analyze their physicochemical properties and to evaluate stability, release property, and antioxidative activity. As a result, SLN coloaded with EGCG and piperine was developed with average size about 150 to 200 nm. The size of NLC coloaded with EGCG and piperine was approximately 100 nm. And both nanocarriers were moderate stable in aqueous phase. For storage period, the particle size and the zeta-potential of NLC were maintained better than SLN. In addition, the release property of EGCG and piperine could be protected until they were reached at small intestine by nanoencapsulation. SLN and NLC were similar at respect of encapsulation efficiency, release property, and the antioxidative acitivity. But NLC was more potential and outstanding results in all experiments. After selecting more suitable nanocarriers between SLN and NLC, EGCG and piperine coloaded NLC would be coated as different coating material in order to protect the activity of EGCG and piperine for a long time during preparation, storage, intake, and absorption in our body. So, after making nanoparticles as different coating materials, protection of core materials would be comparative analyzed in specific solution. In general, NLC was coated by poly-vinyl alcohol (PVA), which was inedible material. So it was limited in application of food or beverage. Therefore, NLC was coated by PVA, whey protein isolate (WPI), and chitosan. And then the release property was evaluated in specific solution. As a result, the absorbance of NLC was increased gradually because of released EGCG and piperine. In addition, there was no difference with PVA and WPI in water or 0.02 M PBS. Because PVA was inedible material, while WPI was edible material, PVA can be replaced into WPI. In this study, it confirmed that EGCG and piperine were protected by NLC structure. The prepared NLC coloaded with EGCG and piperine maintained stability and it could control the release time of core materials. In conclusions, this NLC system based on lipid was potential and outstanding structure at the respect of encapsulation efficiency, release property, and the antioxidative activity. Therefore, it can help increase aqueous stability and oral bioavailability of EGCG and piperine. Keywords: EGCG, piperine, nanocarriers, release property, stability, antioxidative activity