본 연구에서는 2 016년부터 2 02 0년까지 내륙 관측소 중 안개 최다발 지역인 안동을 대상으로 XGBoost-DART 머신러닝 알고리즘을 이용하여 1 시간 후 안개 유무를 예측하였다. 기상자료, 농업관측자료, 추가 파생자료와 각 자료 를 오버 샘플링한 확장자료, 총 6개의 데이터 세트를 사용하였다. 목측으로 획득한 기상현상번호와 시정계 관측으로 측 정된 시정거리 자료를 각각 안개 유[1]무[0]로 이진 범주화하였다. 총 12개의 머신러닝 모델링 실험을 설계하였고, 안개 가 사회와 지역사회에 미치는 유해성을 고려하여 모델의 성능은 재현율과 AUC-ROC를 중심으로 평가하였다. 전체적으 로, 오버샘플링한 기상자료와 기상현상번호 기반의 예측 목표를 조합한 실험이 최고 성능을 보였다. 이 연구 결과는 머 신러닝 알고리즘을 활용한 안개 예측에 있어서, 목측으로 획득한 기상현상번호의 중요성을 암시한다.

The secular and seasonal variations of visibility of 24 cities in Korea are studied statistically from 1973 to 1986. The geographical distributions of annual change patterns of visibility are also studied by means of their characteristical variations. The secular variations of visibility show that annual totals of poor visibility($lt;10㎞) days increase significantly, while that of good visibility ($gt;30㎞) days decrease remarkably. Such tendencies are quite reversal to those appearing in several large cities in the U.S.A., England and Japan. Seasonal variations of visibility can be classified into three types : type A) The visibility is maximum in winter and minimum in summer, type B) The visibility is maximum in spring and fall, minimum in summer and winter, type C) The visibility is maximum in spring and fall, minimum is summer. Geographical distributions of annual mean visibility show that the poorest visibility appears in the Kyunggi inland region(mean 15㎞), while the best visibility occurs in mountainous north Kangwon-do and Cheju island(over 25㎞).

Fog can have severe impacts on human life (e.g., accidents in transportation systems) because it causes low visibility. Areas prone to fog near the sea or a river require accurate fog forecasts. In this study, we analyzed the characteristics of fog occurrence around Gumi Reservoir, a part of the inland Nakdong River. We also simulated and predicted visibility using the Weather Research and Forecasting (WRF) model. Visibility was estimated using several point measurements and several different methods: Stoelinga and Warner (SW99), Rapid Update Cycle (RUC), the Forecast System Laboratory (FSL), Decision Tree (DT), revised visibility (RVIS) and Gultepe_06 methods. The DT method showed a pattern similar to the observed fog occurrences. In this study, the Gulteppe_06 method produced the best performance under the lowest visibility situation, although there is some discrepancy in microphysics' simulated results, followed by DT and RUC.

The detailed characteristics of fog over South Korea were analyzed using the three-years quality controlled (QC) 237 visibility meter data operated by Korea Meteorological Administration. The fog (dense fog: DFog) frequency varies greatly with season and geographic location. The fog frequency at inland is highest in autumn, but at the West Coast in spring and summer. Fog occurs frequently from spring to autumn in the mountainous regions. Unlike the fog, the DFog is mostly prevalent in summer at land, mountain, and coastal regions. The large coefficients of variation of fog and DFog at the three regions and four seasons indicate that the locality of fog over South Korea is very high. The formation and dissipation (FaD) of fog show strong diurnal variations irrespective of geographic location and season, strongest at inland and weakest at sea. Fog usually occurs from night to sunrise and dissipates from early morning to late morning. The maximum FaD time of fog show seasonal variation with the seasonal change in solar elevation angle. The frequency of fog is inversely proportional to the duration time, mostly less than 3 hours regardless of season and geographic location. Also, the duration of DFog is mostly within 1-3 hours.

This study was conducted to investigate how PM10 concentration and Relative Humidity (RH) affected visibility in Jinju, Korea. A 9-yr dataset of 1 h averages for visibility, PM10, and RH data was analyzed to examine the correlation between these variables. On average, visibility decreased by 1.4 km for every 10 μg/㎥ increase in PM10 and by 2.1 km for every 10% increase in RH. In general, a negative correlation was observed between visibility and and PM10 concentration. However, under conditions of low PM10 concentration(< 15 μg/㎥) and visibility(< 2 km), there was a positive correlation between these two variables. In this case, RH levels were high (> 75%). A high correlation analysis between two variables need to be under control conditions with RH < 75%, PM10 15~100 μg/㎥, and visibility > 2 km.

강우, 강설, 안개발생 등 기상상태의 변화는 운전자의 주행환경에 영향을 미치는 기상요인으로 기상악화시 차량의 차두간격과 속도에 영향을 미치게 되어 도로용량을 감소시키고, 교통사고 발생으로 인한 차로감소 등의 상황을 유발하여 맑은 날보다 더 큰 혼잡을 발생시키는 것으로 분석되었다. 운전자의 시정을 감소시키는 기상악화는 통행속도가 높은 고속도로가 일반도로 보다 기상상태에 따른 통행속도 변화 민감도와 교통사고 심각도가 높게 나타나는 특성이 있다고 분석됨에 따라 고속도로의 교통류 특성변화에는 시정거리가 중요하게 작용하는 것으로 판단되었다. 따라서 본 연구에서는 통행속도가 높은 고속도로 기본구간의 교통류 특성에 영향을 미치는 주요 요인으로 교통량과 속도를 선정하였으며, 일정수준 이상의 교통량 확보가 가능한 수도권내 고속도로를 분석대상으로 선정하고 기상자료와 교통자료를 수집하여 시정거리 변화에 따른 고속도로 교통류 특성변화에 관한 연구를 수행하였다. 본 연구의 수행을 위하여 기존 문헌 고찰을 통해 자료수집 및 분석방법을 수립하고 고속도로의 시정거리 수준을 선정하며, 통계적 차이 검증을 수행하고 시정거리에 따른 고속도로의 교통류 특성 변화를 분석하여 용량 및 서비스 수준 분석 시 적용할 수 있는 방안을 강구하고자 한다.

The characteristics of meteorological conditions and air pollution were investigated in a valley city (Yangsan) on bad visibility days (from 05:00 to 09:00 LST) of the cold half year (November 2008 to April 2009). This analysis was performed using the hourly observed data of meteorological variables (temperature, wind speed and direction, relative humidity, and 2 m and 10 m temperature) and air pollutants (NO2, SO2, PM10, and O3). In addition, visibility data based on visual measurements and a visibility meter were used. The bad visibility days were classified into four types: fog, mist, haze, and the mixture (mist+haze). The results showed that the bad visibility days of the four types in the valley city were observed to be more frequently (about 50% of the total study period (99 days except for missing data)) than (27%) those near coastal metropolitan city (Busan). The misty days (39%) in the valley city were the most dominant followed by the hazy (37%), mixture (14%), and foggy days (10%). The visibility degradation on the misty days in Yangsan was closely related to the combined effect of high-level relative humidity due to the accumulation of water vapor from various sources (e.g. river, stream, and vegetation) and strong inversion due to the development of surface radiative cooling within the valley. On the hazy days, the visibility was mainly reduced by the increase in air pollutant (except for O3) concentrations from the dense emission sources under local conditions of weaker winds from the day before and stronger inversion than the misty days. The concentrations of NO2, PM10, and SO2 (up to +36 ppb, +25 ㎍/m3, and +7 ppb) on the hazy days were a factor of 1.4-2.3 higher than those (+25 ppb, +14 ㎍ /m3, and +3 ppb) on the misty days.

본 연구에서는 우리나라 시정의 분포 특성과 그 변화 경향을 분석하고자 하였다. 1973~2004년의 기간에 대해 국내 24개 관측지점의 시정자료를 분석한 결과, 평균 시정은 서울과 인천, 수도권이 위치한 중서부 지역에서 남부와 동부지역으로 갈수록 좋아진다. 계절별로는 가을철이 시정이 가장 좋으며, 겨울, 봄, 여름의 순이다. 양시정 일수는 우리나라 남서해안 지역으로 갈수록 증가하며 저시정 일수는 우리나라의 남동해안 지역으로 갈수록 감소한다. 시정의 변화경향을 분석한 결과 대부분의 지역에서 시정은 나빠지고 있다. 최근 10년 간 대부분의 지점에서 시정이 나빠졌지만, 수원, 포항, 대구, 울산, 부산 지점은 좋아지는 경향을 보였고 서울, 통영, 제주 지점은 변화가 없었다.

Visibility reduction is a barometer of air pollution, which people can notice easily. First of all, we need to measure quantified visibility continuously in order to examine visibility reduction. Prevailing visibility is not practical to measure visibility depending on observer's expertness. Scattering visibility using Forward Scattering Meter (Belfort Visibility Sensor 6230) has been measured at Kwangan-Dong in Pusan and analysed since July, 1998. According to the analysis, the correlation coefficient(R) between prevailing visibility and scattering visibility was 0.7235. The visibility appeared that each frequency of poor visibility (under 6㎞) and good visibility(over 25㎞) was 10.6%, 9.7% on summertime in Pusan and the visibility range from 10km to 20km ranked high frequency as a half of whole ranges. The order of correlation coefficients between visibility and air pollutants are ranking CO, PM10 and NO2, that values are 0.5878, 0.5369, 0.5284 respectively. In meteorological factor, the case of poor visibility presented more weakly wind speed and higher relative humidity than the case of good visibility. The correlation coefficient between calculated visibility of multiple linear regression model and observed visibility was 0.7215. But the trend of calculated and observed visibility variation was similar with the exception of several good visibility cases.

After analyzing the correlation between air pollution and visibility, TSP and NO2 is responsible for poor visibility in Pusan. After analyzing the correlation between meteorological factors and visibility, general pattern of humidity has clear negative correlation. The variation of wind speed has a positive correlation. In order to investigate the cause of poor visibility in Pusan area, the Andersen sampler and PM-2.5 are used to collect and analyze aerosol. This study was carried out to monitor the visibility using Forward scattering meter and to find out the characteristics and the cause of good visibility case and poor visibility case by measuring and analyzing a variety of parameters, such as particle size distributions, chemical compositions, and meteorological conditions in Pusan. According to the analysis of intensive sampling, NO3-, NH4+ ion concentration increased together with the mass concentration around 0.5∼2.5㎛ approximately during the case of poor visibility. NH4NO3, NH4Cl, and NaCl were thought to be the major components of fine particles.

To research the cause of the low and the poor visibility phenomena of Kimhae international airport statistically, we analyzed the field routine data for the last five years from 1985 to 1989. The poor visibilities below l㎞, 3㎞, and 5㎞ usually occurred at about 6 o`clock in the morning under the condition of calm or light wind from south to southwest direction, and lasted for from 1 to 3 hours. They were caused by the radiative cooling and the inflow of moisture from the South Sea. The frequency of the low visibility(below 9㎞) recorded 48.1% a year. And the low visibility below 8㎞ with relative humidity below 70% often occurred in the case of southeast, southwest, and northwest wind. And it reveals a peak at 11:00 a.m.. It is supposed to be caused by the pollutants flowing from the neighbouring industrial complex, Sasang, Jangnim and by the photochemical reaction. And, when the industrial complex is built in Sinho, Myongji, Noksan located southwest, and Changyu area located northwest direction from Kimhae international airport, the visibility and the air pollution may become worse and worse by the increased pollutants.

A berth assignment problem has a direct impact on assessment of charges made to ships and goods. In this paper, we concerned with of fuzzy mathematical programming models for a berth assignment problem to achieved an efficient berth operation in a fuzzy environment. In this paper, we focus on the berth assignment programming with fuzzy parameters which are based on personal opinions or subjective judgement. From the above point of view, assume that a goal and a constraint are given by fuzzy sets, respectively, which are characterized by membership functions. Let a fuzzy decision be defined as the fuzzy set resulting from the intersection of a goal and constraint. This paper deals with fuzziness in all parameters which are expressed by fuzzy numbers. A fuzzy parameter defined by a fuzzy number means a possibility distribution of the parameters. These fuzzy 0-1 integer programming problems are formulated by fuzzy functions whose concept is also called the extension principle. We deal with a berth assignment problem with triangular fuzzy coefficients and propose a branch and bound algorithm for solving the problem. We suggest three models of berth assignment to minimizing the objective functions such as total port time, total berthing time and maximum berthing time by using a revised Maximum Position Shift(MPS) concept. The berth assignment problem is formulated by min-max and fuzzy 0-1 integer programming. Finally, we gave the numerical solutions of the illustrative examples.

The Daechong dam is located nearby Chongju city, and the population alone in the Chongju- Chongwon area is over one half million. It is found that the number of days for fog occurrence after constructing the dam was much fewer than those before constructing it. It appears that the construction of a large dam did not cause to increase in the occurrence of fog in the Chongju area, and that was the increase due to topographical characteristics and wind direction. Low visibility in the Chongju-Chongwon area occurred frequently. In summer, Chongwon recorded more days of poor visibility than those observed at Chongju. It is thought that TSP. SO_2 and NO_2 were the main constituents of air pollutants causing the reduction in visibility. These air pollutants are produced by vehicles and industrial activities in the region.