• Proceedings of the Korea Water Resources Association Conference
• /
• 2012.05a
• /
• pp.71-71
• /
• 2012

현재 전 세계적으로 극한강우의 발생빈도가 점차 높아지고 있으며 홍수량 또한 강도가 커지고 있는 것이 현실이다. 하지만 과거의 홍수발생 빈도에 따라 설계된 홍수방어시설들이 점차 한계를 보이고 있으므로 이를 대비하기위한 구조적 대책뿐만 아니라 홍수피해 발생 가능지역에 사전 예경보를 시행하는 비구조적 대책마련 또한 필요하다. 기존의 홍수예측은 확정적인 하나의 유량예측값만을 제공함으로써 신속하고 편리하였지만 이에 대한 불확실성이 큰 경우 예상치 못한 큰 인적 물적 피해를 가져올 수 있다. 이처럼 확률론적 홍수예측의 필요성이 대두되어 지면서 유럽이나 미국등 선진국에서는 EFFS(European Flood Forecasting System)과 NWSRFS(National Water Service River Forecast System)같이 이미 확률론적 홍수예측에 대한 연구 및 기술개발이 활발하게 진행되어지고 있다. 하지만 홍수예측의 확률론적 접근에 있어서는 많은 불확실성들이 내포되어 있으므로 예측시스템에서 생성된 앙상블 유량예측 결과의 신뢰도 분석과 올바른 불확실성 정보의 제공이 필요하다. 본 연구는 확률론적 홍수예측 방법을 국내에 적용시켜서 기상청의 예측시스템 KLAPS(Korea Local Analysis and Prediction System), MAPLE(McGill Algorithm for Precipitation Nowcasting by Lagrangian Extrapolation), UM(Unified Model) 그리고 MOGREPS(Met Office Global Regional Ensemble Prediction System)으로부터 생성된 기상앙상블을 현재 국토해양부 홍수통제소에서 사용하고 있는 강우-유출모형인 저류함수모형(Storage Function Method)의 입력 자료로 사용한다. 확률론적 홍수예측에서 오는 불확실성을 분석하기 위해서 첫 번째로 제공되는 기상예측 시스템의 시 공간적 스케일 및 대상유역의 공간특성에 따라 어떠한 형태로 전파되어지는지를 분석하였다. 두 번째는 각각의 예측시스템들이 선행기간(Lead time)에 따라 불확실성의 특성이 어떻게 나타나게 되는지를 확인하였다. 이러한 불확실성의 특성을 정확하게 파악하게 된다면 예측에 있어서 현재 갖고 있는 문제점들로부터 개선해 나가야 할 방향을 제시해주어 향후연구에 유용하게 활용될 수 있을 것이다.

• Journal of the Korean Society of Hazard Mitigation
• /
• v.9 no.5
• /
• pp.93-102
• /
• 2009

A prediction of streamflow based on multi-objective function is presented to check the performance of Probability Distributed Model(PDM) in Miho stream basin, Chungcheongbuk-do, Korea. PDM is a lumped conceptual rainfall runoff model which has been widely used for flood prevention activities in UK Environmental Agency. The Monte Carlo Analysis Toolkit(MCAT) is a numerical analysis tools based on population sampling, which allows evaluation of performance, identifiability, regional sensitivity and etc. PDM is calibrated for five model parameters by using MCAT. The results show that the performance of model parameters(cmax and k(q)) indicates high identifiability and the others obtain equifinality. In addition, the multi-objective function is applied to PDM for seeking suitable model parameters. The solution of the multi-objective function consists of the Pareto solution accounting to various trade-offs between the different objective functions considering properties of hydrograph. The result indicated the performance of model and simulated hydrograph are acceptable in terms on Nash Sutcliffe Effciency*(=0.035), FSB(=0.161), and FDBH(=0.809) to calibration periods, validation periods as well.

• Journal of Digital Convergence
• /
• v.19 no.4
• /
• pp.25-31
• /
• 2021

Moving toward an aged society, traffic accidents involving elderly drivers have also attracted broader public attention. A rapid increase of senior involvement in crashes calls for developing appropriate crash-severity prediction models specific to senior drivers. In that regard, this study leverages machine learning (ML) algorithms so as to predict the severity of vehicle-pedestrian collisions induced by elderly drivers. Specifically, four ML algorithms (i.e., Logistic model, K-nearest Neighbor (KNN), Random Forest (RF), and Support Vector Machine (SVM)) have been developed and compared. Our results show that Logistic model and SVM have outperformed their rivals in terms of the overall prediction accuracy, while precision measure exhibits in favor of RF. We also clarify that driver education and technology development would be effective countermeasures against severity risks of senior driver-induced collisions. These allow us to support informed decision making for policymakers to enhance public safety.

• Imaging Science in Dentistry
• /
• v.52 no.3
• /
• pp.239-244
• /
• 2022

Purpose: Despite the proliferation of numerous morphometric and anthropometric methods for sex identification based on linear, angular, and regional measurements of various parts of the body, these methods are subject to error due to the observer's knowledge and expertise. This study aimed to explore the possibility of automated sex determination using convolutional neural networks(CNNs) based on lateral cephalometric radiographs. Materials and Methods: Lateral cephalometric radiographs of 1,476 Iranian subjects (794 women and 682 men) from 18 to 49 years of age were included. Lateral cephalometric radiographs were considered as a network input and output layer including 2 classes(male and female). Eighty percent of the data was used as a training set and the rest as a test set. Hyperparameter tuning of each network was done after preprocessing and data augmentation steps. The predictive performance of different architectures (DenseNet, ResNet, and VGG) was evaluated based on their accuracy in test sets. Results: The CNN based on the DenseNet121 architecture, with an overall accuracy of 90%, had the best predictive power in sex determination. The prediction accuracy of this model was almost equal for men and women. Furthermore, with all architectures, the use of transfer learning improved predictive performance. Conclusion: The results confirmed that a CNN could predict a person's sex with high accuracy. This prediction was independent of human bias because feature extraction was done automatically. However, for more accurate sex determination on a wider scale, further studies with larger sample sizes are desirable.

• Journal of The Geomorphological Association of Korea
• /
• v.26 no.3
• /
• pp.53-67
• /
• 2019

The purpose of this study is to identify the quality of landslide susceptibility in a landslide-prone area (Jinbu-myeon, Gangwon-do, South Korea) by spatial prediction modeling approach and compare the results obtained. For this goal, a landslide inventory map was prepared mainly based on past historical information and aerial photographs analysis (Daum Map, 2008), as well as some field observation. Altogether, 550 landslides were counted at the whole study area. Among them, 182 landslides are debris flow and each group of landslides was constructed in the inventory map separately. Then, the landslide inventory was randomly selected through Excel; 50% landslide was used for model analysis and the remaining 50% was used for validation purpose. Total 12 contributing factors, such as slope, aspect, curvature, topographic wetness index (TWI), elevation, forest type, forest timber diameter, forest crown density, geology, landuse, soil depth, and soil drainage were used in the analysis. Moreover, to find out the co-relation between landslide causative factors and incidents landslide, pixels were divided into several classes and frequency ratio for individual class was extracted. Eventually, six landslide susceptibility maps were constructed using the Bayesian Predictive Discriminant (BPD), Empirical Likelihood Ratio (ELR), and Linear Regression Method (LRM) models based on different category dada. Finally, in the cross validation process, landslide susceptibility map was plotted with a receiver operating characteristic (ROC) curve and calculated the area under the curve (AUC) and tried to extract success rate curve. The result showed that Bayesian, likelihood and linear models were of 85.52%, 85.23%, and 83.49% accuracy respectively for total data. Subsequently, in the category of debris flow landslide, results are little better compare with total data and its contained 86.33%, 85.53% and 84.17% accuracy. It means all three models were reasonable methods for landslide susceptibility analysis. The models have proved to produce reliable predictions for regional spatial planning or land-use planning.

• Journal of the Korea Society of Computer and Information
• /
• v.28 no.11
• /
• pp.53-63
• /
• 2023

This study developed a cognitive impairment predictive model as one of the screening tests for preventing dementia in the elderly by using Automated Machine Learning(AutoML). We used 'Wearable lifelog data for high-risk dementia patients' of National Information Society Agency, then conducted using PyCaret 3.0.0 in the Google Colaboratory environment. This study analysis steps are as follows; first, selecting five models demonstrating excellent classification performance for the model development and lifelog data analysis. Next, using ensemble learning to integrate these models and assess their performance. It was found that Voting Classifier, Gradient Boosting Classifier, Extreme Gradient Boosting, Light Gradient Boosting Machine, Extra Trees Classifier, and Random Forest Classifier model showed high predictive performance in that order. This study findings, furthermore, emphasized on the the crucial importance of 'Average respiration per minute during sleep' and 'Average heart rate per minute during sleep' as the most critical feature variables for accurate predictions. Finally, these study results suggest that consideration of the possibility of using machine learning and lifelog as a means to more effectively manage and prevent cognitive impairment in the elderly.

• Journal of the Korean earth science society
• /
• v.32 no.7
• /
• pp.770-783
• /
• 2011

In this study, we performed a downscaling of an ECHAM5 simulated dataset for the current and future climate produced under the Special Report on Emission Scenarios A1B (SRES A1B) by utilizing the National Centers for Environmental Prediction (NCEP) Regional Spectral Model (RSM). The current climate simulation was performed for the period 1980-2000 and the future climate run for the period 2040-2070 for the COordinated Regional climate Downscaling EXperiment (CORDEX)'s East Asia domain. The RSM is properly able to reproduce the climatological fields from the evaluation of the current climate simulation. Future climatological precipitation during the summer season is increased over the tropical Oceans, the maritime-continent, and Japan. In winter, on the other hand, precipitation is increased over the tropical Indian Ocean, the maritime-continents and the Western North Pacific, and decreased over the eastern tropical Indian Ocean. For the East Asia region few significant changes are detected in the precipitation climatological field. However, summer rainfall shows increasing trend after 2050 over the region. The future climate ground temperature shows a clear increasing trend in comparison with the current climate. In response to global warming, atmospheric warming is clearly detected, which strengthens the upper level trough.

• Journal of the Korean association of regional geographers
• /
• v.18 no.4
• /
• pp.473-493
• /
• 2012

Malaria infection is sensitively influenced by regional meteorological conditions along with global climate change. Remote sensing techniques have become an important tool for extraction of climatic and environmental factors, including rainfall, temperature, surface water, soil moisture, and land use, which are directly linked to the habitat qualities of malaria mosquitoes. Improvement of sensor fidelity with higher spatial and spectral resolution, new multinational sensor development, and decreased data cost have nurtured diverse remote sensing applications in malaria research. In 1984, eradication of endemic malaria was declared in Korea, but reemergence of malaria was reported in mid-1990s. Considering constant changes in malaria cases since 2000, the epidemiological management of the disease needs careful monitoring. Geographically, northmost counties neighboring North Korea have been ranked high in the number of malaria cases. High infection rates in these areas drew special attention and led to a hypothesis that malaria dispersion in these border counties might be caused by north-origin, malaria-bearing adult mosquitoes. Habitat conditions of malaria mosquitoes are important parameters for prediction of the vector abundance. However, it should be realized that malaria infection and transmission is a complex mechanism, where non-environmental factors, including human behavior, demographic structure, landscape structure, and spatial relationships between human residence and the vector habitats, are also significant considerations in the framework of medical geography.

• Journal of Korean Society of Coastal and Ocean Engineers
• /
• v.30 no.2
• /
• pp.69-83
• /
• 2018

The South China Sea (SCS) is a typical marginal sea characterized with the deep basin, shelf break, shallow shelf, many straits, and complex bathymetry. This study investigated the tidal characteristics and propagation, and reproduced typhoon-induced storm surge in this region using the regional real-time tide-surge model, which was based on the unstructured grid, resolving in detail the region of interest and forced by tide at the open boundary and by wind and air pressure at the surface. Typhoon Haiyan, which occurred in 2013 and caused great damage in the Philippines, was chosen as a case study to simulate typhoon's impact. Amplitudes and phases of four major constituents were reproduced reasonably in general, and the tidal distributions of four constituents were similar to the previous studies. The modelled tide seemed to be within the acceptable levels, considering it was difficult to reproduce the tide in this region based on the previous studies. The free oscillation experiment results described well the feature of tide that the diurnal tide is prevailing in the SCS. The tidal residual current and total energy dissipation were discussed to understand the tidal and sedimentary environments. The storm-surge caused by typhoon Haiyan was reasonably simulated using this modeling system. This study established the regional real-time barotropic tide/water level prediction system for the South China Sea including the seas around the Philippines through the validation of the model and the understanding of tidal characteristics.

• Korean Journal of Agricultural and Forest Meteorology
• /
• v.14 no.4
• /
• pp.143-154
• /
• 2012

As demand of water resources and attentions to changes in climate (e.g., due to ENSO) increase, long/short term prediction of precipitation is getting necessary in water planning. This research evaluated the ability of MM5 to predict precipitation in the Tampa Bay region over 23 year period from 1986 to 2008. Additionally MM5 results were statistically bias-corrected using observation data at 33 stations over the study area using CDF-mapping approach and evaluated comparing to raw results for each ENSO phase (i.e., El Ni$\tilde{n}$o and La Ni$\tilde{n}$a). The bias-corrected model results accurately reproduced the monthly mean point precipitation values. Areal average daily/monthly precipitation predictions estimated using block-kriging algorithm showed fairly high accuracy with mean error of daily precipitation, 0.8 mm and mean error of monthly precipitation, 7.1 mm. The results evaluated according to ENSO phase showed that the accuracy in model output varies with the seasons and ENSO phases. Reasons for low predictions skills and alternatives for simulation improvement are discussed. A comprehensive evaluation including sensitivity to physics schemes, boundary conditions reanalysis products and updating land use maps is suggested to enhance model performance. We believe that the outcome of this research guides to a better implementation of regional climate modeling tools in water management at regional/seasonal scale.