• 대기
• /
• 제22권2호
• /
• pp.245-257
• /
• 2012

The Asian dust (Hwangsa) forecasting model, Asian Dust Aerosol Model (ADAM) has been modified by using satelliate monitoring of surface vegetation, which enables to simulate dusts occuring not only in springtime but also for all-year-round period. Coupled with the Unified Model (UM), the operational weather forecasting model at KMA, UM-ADAM2 was implemented for operational dust forecasting since 2010, with an aid of development of Meteorology-Chemistry Interface Processor (MCIP) for usage UM. The performance analysis of the ADAM2 forecast was conducted with $PM_{10}$ concentrations observed at monitoring sites in the source regions in China and the downstream regions of Korea from March to December in 2010. It was found that the UM-ADAM2 model was able to simulate quite well Hwangsa events observed in spring and wintertime over Korea. In the downstream region of Korea, the starting and ending times of dust events were well-simulated, although the surface $PM_{10}$ concentration was slightly underestimated for some dust events. The general negative bias less than $35{\mu}g\;m^{3}$ in $PM_{10}$ is found and it is likely to be due to other fine aerosol species which is not considered in ADAM2. It is found that the correlation between observed and forecasted $PM_{10}$ concentration increases as forecasting time approaches, showing stably high correlation about 0.7 within 36 hr in forecasting time. This suggests the possibility that there is potential for the UM-ADAM2 model to be used as an operational Asian dust forecast model.

• 한국정보통신학회논문지
• /
• 제24권1호
• /
• pp.8-14
• /
• 2020

미세먼지의 인체 영향이 밝혀지며 예보정확도 개선에 대한 요구가 증가하고 있다. 이에 기계 학습 기법을 도입하여 예측 정확성을 높이려는 노력이 수행되고 있으나, 저농도 발생 비율이 매우 큰 미세먼지 데이터로 인해 전체 예측 성능이 떨어지는 문제가 있다. 본 논문에서는 PM10 미세먼지 예보 정확도 향상을 위해 농도별 분리 예측 모델을 제안한다. 이를 위해 천안 지역의 기상 및 대기오염 인자를 활용하여 저, 고농도별 예측 모델을 설계하고 전 영역 예측 모델과의 성능 비교를 수행하였다. RMSE, MAPE, 상관계수 및 AQI 정확도를 통한 성능 비교 결과, 전체 기준에서 예측 성능이 향상됨을 확인하였으며, AQI 고농도 예측 성능의 경우 20.62%의 성능 향상이 나타났음을 확인하였다.

• 한국대기환경학회지
• /
• 제33권6호
• /
• pp.554-569
• /
• 2017

Sensitivity analysis on $PM_{10}$ forecasting simulations was carried out by using two different initial and boundary conditions of meteorological fields: NCEP/FNL (National Centers for Environmental Prediction/Final Analysis) reanlaysis data and NCEP/GFS (National Centers for Environmental Prediction/Global Forecast System) forecasting data, and the comparisons were made between two different simulations. The two results both yielded lower $PM_{10}$ concentrations than observations, with relatively lower biased results by NCEP/FNL than NCEP/GFS. We explored the detailed individual meteorological variables to associate with $PM_{10}$ prediction performance. With the results of NCEP/FNL outperforming GFS, our conclusion is that no particular significant bias was found in temperature fields between NCEP/FNL and NCEP/GFS data, while the overestimated wind speed by NCEP/GFS data influenced on the lower $PM_{10}$ concentrations simulation than NCEP/FNL, by decreasing the duration time of high-$PM_{10}$ loaded air mass over both coastal and metropolitan areas. These comparative characteristics of FNL against GFS data such as maximum 3~4 m/s weaker wind speed, $PM_{10}$ concentration control with the highest possible factor of 1.3~1.6, and one or two hour difference of peak time for each case in this study, were also reflected into the results of statistical analysis. It is implying that improving the surface wind speed fluctuation is an important controlling factor for the better prediction of $PM_{10}$ over Korean Peninsula.

• 한국방재안전학회논문집
• /
• 제15권4호
• /
• pp.87-98
• /
• 2022

최근 짧은 시간 동안 많은 강우가 내리는 국지성 집중호우가 빈번히 발생하고 이로 인한 침수피해가 증가하고 있다. 국지성 집중호우로 인한 피해를 예방하기 위하여 기상청이 제공하는 지역 앙상블 예측시스템(Local ENsemble prediction System, LENS)과 관측자료와 동네예보 자료를 활용한 기계학습과 확률 매칭(Probability Matching, PM) 기법을 이용하여 수문학적 정량강우예측정보(Hydrological Quantative Precipitation Forecast, HQPF)을 개발하였다. 국지성 집중호우로 인한 침수피해 대비를 위한 호우 영향정보로 HQPF를 생산하고 있지만, 낮은 강우강도에 대하여 과대예측하는 경향이 나타났다. 본 연구에서는 HQPF의 예측정확도 향상과 과대예측 성향을 개선하기 위하여 머신러닝 학습자료 기간확대, 앙상블 기법 분석 및 확률매칭(PM) 기법 프로세스 변경을 통하여 HQPF 개선하였다. 개선된 HQPF의 예측성능을 평가하기 위해 2021년 8월 27일 ~ 2021년 9월 3일 장마전선으로 인한 호우 사례를 대상으로 예측성능 검증을 수행하였다. 10 mm 이하의 강우에 대하여 예측정확도가 크게 향상되었고, 관측과 유사한 발생가능성 및 강우영역을 예측하는 등 과대예측 성향이 개선되었음을 확인하였다.

• 대기
• /
• 제25권1호
• /
• pp.67-83
• /
• 2015

Predicting the location and intensity of precipitation still remains a main issue in numerical weather prediction (NWP). Resolution is a very important component of precipitation forecasts in NWP. Compared with a lower resolution model, a higher resolution model can predict small scale (i.e., storm scale) precipitation and depict convection structures more precisely. In addition, an ensemble technique can be used to improve the precipitation forecast because it can estimate uncertainties associated with forecasts. Therefore, NWP using both a higher resolution model and ensemble technique is expected to represent inherent uncertainties of convective scale motion better and lead to improved forecasts. In this study, the limited area ensemble prediction system for the convective-scale (i.e., high resolution) operational Unified Model (UM) in Korea Meteorological Administration (KMA) was developed and evaluated for the ensemble forecasts during August 2012. The model domain covers the limited area over the Korean Peninsula. The high resolution limited area ensemble prediction system developed showed good skill in predicting precipitation, wind, and temperature at the surface as well as meteorological variables at 500 and 850 hPa. To investigate which combination of horizontal resolution and ensemble member is most skillful, the system was run with three different horizontal resolutions (1.5, 2, and 3 km) and ensemble members (8, 12, and 16), and the forecasts from the experiments were evaluated. To assess the quantitative precipitation forecast (QPF) skill of the system, the precipitation forecasts for two heavy rainfall cases during the study period were analyzed using the Fractions Skill Score (FSS) and Probability Matching (PM) method. The PM method was effective in representing the intensity of precipitation and the FSS was effective in verifying the precipitation forecast for the high resolution limited area ensemble prediction system in KMA.

• 한국멀티미디어학회논문지
• /
• 제23권6호
• /
• pp.747-755
• /
• 2020

In this paper, we deal with outlier data problems that occur when constructing a PM_2.5 fine dust forecasting system using a neural network. In general, when learning a neural network, some of the data are not helpful for learning, but rather disturbing. Those are called outlier data. When they are included in the training data, various problems such as overfitting occur. In building a PM_2.5 fine dust concentration forecasting system using neural network, we have found several outlier data in the training data. We, therefore, remove them, and then make learning 3 ways. Over_outlier model removes outlier data that target concentration is low, but the model forecast is high. Under_outlier model removes outliers data that target concentration is high, but the model forecast is low. All_outlier model removes both Over_outlier and Under_outlier data. We compare 3 models with a conventional outlier removal model and non-removal model. Our outlier removal model shows better performance than the others.

• 한국멀티미디어학회논문지
• /
• 제22권10호
• /
• pp.1187-1198
• /
• 2019

In this study, we propose a new $PM_{10}$ forecasting model for Seoul region using DNN(Deep Neural Network) and secondary data. The previous numerical and Julian forecast model have been developed using primary data such as weather and air quality measurements. These models give excellent results for accuracy and false alarms, but POD is not good for the daily life usage. To solve this problem, we develop four secondary factors composed with primary data, which reflect the correlations between primary factors and high $PM_{10}$ concentrations. The proposed 4 models are A(Anomaly), BT(Back trajectory), CB(Contribution), CS(Cosine similarity), and ALL(model using all 4 secondary data). Among them, model ALL shows the best performance in all indicators, especially the PODs are improved.

• 한국정보통신학회:학술대회논문집
• /
• 한국정보통신학회 2021년도 춘계학술대회
• /
• pp.280-282
• /
• 2021

주관적 판단을 적용하여 예보되는 미세먼지 예보의 문제를 해결하기 위해 딥러닝 알고리즘을 이용하여 미세먼지 예측 모델을 설계하였다. 딥러닝 알고리즘 중 RNN과 LSTM을 이용하였으며, 하이퍼 파라미터 탐색을 통해 최적의 파라미터를 적용하여 설계하였다. RMSE와 예측 정확도를 통해 두 모델의 예측 성능을 평가하였다. 성능 평가 결과, RMSE와 전체 정확도에서 큰 차이는 없었으나 세부 예측 정확도의 차이가 있음을 확인하였다.

• 대기
• /
• 제22권1호
• /
• pp.13-28
• /
• 2012

The $PM_{10}$ concentration data is useful for indentifying intensity and a transport way of Asian dust. However, it is difficult to identify them properly due to the limited spatial resolution and coverage. Therefore, a methodology to estimate $PM_{10}$ concentration using visibility data obtained from synoptic observation was developed. To derive the converting function, correlation between visibility and $PM_{10}$ concentration is investigated using visibility and $PM_{10}$ concentration data observed at 20 stations in Korea from 2005 to 2009. To minimize bias due to atmospheric moisture, data with higher relative humidity over a critical value were eliminated while deriving $PM_{10}$-visibility relationship. As a result, an exponentially decreasing function of visibility is obtained under the condition that relative humidity is less than 82%. Verification of the visibility converting function to $PM_{10}$ concentration was carried out for the dust cases in 2010. It was found that spatial distributions of $PM_{10}$ calculated by visibility are in good agreement with the observed $PM_{10}$ distribution, especially for the strong dust cases in 2010. And correlation between the derived and observed $PM_{10}$ concentration was 0.63. We applied the function to obtain distributions of $PM_{10}$ concentration over North Korea, in which concentration data are not available, and compared them with satellite derived dust index, IODI distributions for dust cases in 2010. It is shown that the visibility function estimates quite similar patterns of dust concentration with IODI image, which suggests that it can contribute for prediction by indentifying transport route of Asian dust.

• Communications for Statistical Applications and Methods
• /
• 제29권3호
• /
• pp.319-331
• /
• 2022

In this paper, we develop a new statistical model to forecast the PM_2.5 level in Seoul, South Korea. The proposed model is based on the extreme quantile regression model with lasso penalty. Various meteorological variables and air pollution variables are considered as predictors in the regression model, and the lasso quantile regression performs variable selection and solves the multicollinearity problem. The final prediction model is obtained by combining various extreme lasso quantile regression estimators and we construct a binary classifier based on the model. Prediction performance is evaluated through the statistical measures of the performance of a binary classification test. We observe that the proposed method works better compared to the other classification methods, and predicts 'very bad' cases of the PM_2.5 level well.