• 한국수자원학회:학술대회논문집
• /
• 한국수자원학회 2017년도 학술발표회
• /
• pp.188-188
• /
• 2017

Rainfall is an important input to hydrological models. The accuracy of hydrological studies for water resources and floods management depend primarily on the estimation of rainfall. Thailand is among the countries that have regularly affected by floods. Flood forecasting and warning are necessary to prevent or mitigate loss and damage. Merging near real time satellite-based precipitation estimation with relatively high spatial and temporal resolutions to ground gauged precipitation data could contribute to reducing uncertainty and increasing efficiency for flood forecasting application. This study tested the applicability of satellite-based rainfall for water resources management and flood forecasting. The objectives of the study are to assess uncertainty associated with satellite-based rainfall estimation, to perform bias correction for satellite-based rainfall products, and to evaluate the performance of the bias-corrected rainfall data for the prediction of flood events. This study was conducted using a case study of Thai catchments including the Chao Phraya, northeastern (Chi and Mun catchments), and the eastern catchments for the period of 2006-2015. Data used in the study included daily rainfall from ground gauges, telegauges, and near real time satellite-based rainfall products from TRMM, GSMaP and PERSIANN CCS. Uncertainty in satellite-based precipitation estimation was assessed using a set of indicators describing the capability to detect rainfall event and efficiency to capture rainfall pattern and amount. The results suggested that TRMM, GSMaP and PERSIANN CCS are potentially able to improve flood forecast especially after the process of bias correction. Recommendations for further study include extending the scope of the study from regional to national level, testing the model at finer spatial and temporal resolutions and assessing other bias correction methods.

• 대기
• /
• 제29권3호
• /
• pp.241-255
• /
• 2019

Bias correction (BC) and quality control (QC) are essential steps for the proper use of satellite observations in data assimilation (DA) system. BC should be calculated over quality controlled observation. And also QC should be performed for bias corrected observation. In the Korea Institute of Atmospheric Prediction Systems (KIAPS) Package for Observation Processing (KPOP), we adopted an adaptive BC method that calculates the BC coefficients with background at the analysis time rather than using static BC coefficients. In this study, we have developed an iterative QC-BC method for Advanced Microwave Sounding Unit-A (AMSU-A) to reduce the negative feedback from the interaction between BC and QC. The new iterative QC-BC is evaluated in the KIAPS 3-dimensional variational (3DVAR) DA cycle for January 2016. The iterative QC-BC method for AMSU-A shows globally significant benefits for error reduction of the temperature. The positive impacts for the temperature were predominant at latitudes of $30^{\circ}{\sim}90^{\circ}$ of both hemispheres. Moreover, the background warm bias across the troposphere is decreased. Even though AMSU-A is mainly designed for atmospheric temperature sounding, the improvement of AMSU-A pre-processing module has a positive impact on the wind component over latitudes of $30^{\circ}S$ near upper-troposphere, respectively. Consequently, the 3-day-forecast-accuracy is improved about 1% for temperature and zonal wind in the troposphere.

• 한국지리정보학회지
• /
• 제2권2호
• /
• pp.69-78
• /
• 1999

본 논문에서는 한반도의 지역별 강수량 예측을 위한 신경망 기법을 소개한다. 시계열 패턴 예측 문제에 적용될 수 있는 기존의 다양한 신경망 모델의 특성을 분석하고 이로부터 강수량 예측 문제에 적합한 모델 및 학습 알고리즘을 제시한다. 본 논문에서 제시하는 모델은 계층적구조의 신경망으로 각 노드의 출력값은 일정기간동안 버퍼에 저장되어 상위계층에 입력으로 작용한다. 본 연구에서는 제안된 모델에 대하여 이중연결형태의 시냅스 구조를 채택하고, 이에 대한 네트워크의 동작특성과 학습알고리즘 등을 정의한다. 이러한 이중연결구조는 기존의 다층퍼셉트론에서 바이어스 노드의 역할을 담당하며, 노드가 갖는 특징들간의 관계를 효과적으로 반영함으로써 기존의 전형적인 시계열 예측 신경망인 FIR(Finite Impulse Response) 네트워크와 비교할 때 학습의 효율을 개선시킨다. 제시된 이론은 월별 및 계절별 강수량 예측 실험에 적용하였다. 신경망 예측기의 학습자료로서 과거 수십년동안 관측된 강수량 데이터와 해수표면온도 데이터를 사용하며 예측 실험결과로부터 제시된 이론의 타당성을 고찰한다.

• 한국대기환경학회지
• /
• 제19권5호
• /
• pp.595-610
• /
• 2003

The statistical indexes such as RMSE (Root Mean Square Error), Mean Bias error, and IOA (Index of agreement) are used to evaluate 3 Dimensional wind and temperature fields predicted by operational meteorological model RAMS (Regional Atmospheric Meteorological System) implemented in CARIS (Chemical Accident Response Information System) for the dispersion forecast of hazardous chemicals in case of the chemical accidents in Korea. The operational atmospheric model, RAMS in CARIS are designed to use GDAPS, GTS, and AWS meteorological data obtained from KMA (Korean Meteorological Administration) for the generation of 3-dimensional initial meteorological fields. The predicted meteorological variables such as wind speed, wind direction, temperature, and precipitation amount, during 19 ∼ 23, August 2002, are extracted at the nearest grid point to the meteorological monitoring sites, and validated against the observations located over the Korean peninsula. The results show that Mean bias and Root Mean Square Error are 0.9 (m/s), 1.85 (m/s) for wind speed at 10 m above the ground, respectively, and 1.45 ($^{\circ}C$), 2.82 ($^{\circ}C$) for surface temperature. Of particular interest is the distribution of forecasting error predicted by RAMS with respect to the altitude; relatively smaller error is found in the near-surface atmosphere for wind and temperature fields, while it grows larger as the altitude increases. Overall, some of the overpredictions in comparisons with the observations are detected for wind and temperature fields, whereas relatively small errors are found in the near-surface atmosphere. This discrepancies are partly attributed to the oversimplified spacing of soil, soil contents and initial temperature fields, suggesting some improvement could probably be gained if the sub-grid scale nature of moisture and temperature fields was taken into account. However, IOA values for the wind field (0.62) as well as temperature field (0.78) is greater than the 'good' value criteria (> 0.5) implied by other studies. The good value of IOA along with relatively small wind field error in the near surface atmosphere implies that, on the basis of current meteorological data for initial fields, RAMS has good potentials to be used as a operational meteorological model in predicting the urban or local scale 3-dimensional wind fields for the dispersion forecast in association with hazardous chemical releases in Korea.

• 한국수자원학회논문집
• /
• 제51권11호
• /
• pp.1031-1041
• /
• 2018

본 연구에서는 ANFIS 기반 GloSea5 앙상블 기상전망 개선 기법을 개발하고 평가하였다. 대상유역은 국내 주요 다목적댐인 충주댐 유역을 선정하였으며, 개선 기법은 ANFIS 기반의 전 후처리기법으로 구성된다. 전처리 기법에서 GloSea5의 앙상블 멤버에 가중치를 부여하며(OWM), 후처리 과정에서는 전처리결과를 편의보정 한다(MOS). 평가결과 편의보정된 GloSea5에 비해 예측성능이 개선되었으며, CASE3, CASE1, CASE2 순으로 모의성능이 우수하였다. 전처리 기법은 강수의 변동성이 큰 계절에 개선효과가 우수하였으며, 후처리 기법은 전처리로 개선하지 못한 오차를 줄 일 수 있는 것으로 나타났다. 따라서 본 연구에서 개발한 ANFIS 기반 GloSea5 앙상블 기상전망 개선 기법은 전 후처리 기법을 함께 사용하는 것이 가장 좋으며, 특히 여름철과 같이 강수의 변동성이 큰 계절에 활용성이 높을 것으로 판단된다.

• 한국수자원학회논문집
• /
• 제52권7호
• /
• pp.475-482
• /
• 2019

2주에서 2개월까지 선행기간을 가지는 계절내-계절(Subseasonal-to-Seasonal, S2S) 예측결과는 산업전반에 걸쳐 다양한 분야에 활용이 가능할 것으로 기대되고 있으나, 일기예보나 중장기 예보대비 낮은 예측성으로 인하여 현재까지 활용성이 매우 낮은 실정이다. 본 연구에서는 기계학습 기법중 비선형회귀 분야에서 좋은 결과를 보여주는 다층퍼셉트론 기법을 이용하여 S2S 예측자료의 후처리를 통한 국내 영역에서의 강수예측성 향상에 관한 연구를 수행하였다. 후처리 모형의 학습을 위한 입력자료로는 ECMWF의 S2S 과거예측(Hindcast) 정보를 이용하였으며 양분예보기법에 기반하여 학습된 다층퍼셉트론 모델을 이용한 후처리 결과와의 비교 분석이 수행되었다. 비교분석 결과 편차도(Bias score)는 평균 59.7% 감소하였고, 정확도(Accuracy)는 124.3% 증가하였으며, 임계성공지수(Critical Success Index)는 88.5% 향상된 것으로 분석되었다. 탐지확률(Probability of detection)의 경우 원자료 대비 평균 9.5% 감소하였으나 이는 ECMWF의 예측모델이 강수의 발생일을 과도하게 예측하였기 때문인 것으로 분석되었다. 본 연구 수행 결과 비록 ECMWF의 S2S 예측자료의 예측성이 낮더라도 후처리를 통해 예측성을 향상 시킬 수 있음을 확인하였으며, 본 연구 결과는 향후 수자원과 농업 분야에서 S2S 자료의 활용성을 높이는데 도움이 될 수 있을 것으로 판단된다.

• 대기
• /
• 제26권4호
• /
• pp.599-615
• /
• 2016

This study investigates the accuracy of short-term ocean predictions during the development of ocean stratification for the Korea Meteorological Administration (KMA) Global Seasonal Forecast System version 5 (GloSea5) as well as the effect of atmosphere-ocean coupling on the predictions through a series of sensitive numerical experiments. Model performance is evaluated using the marine meteorological buoys at seas around the Korean peninsular (KP), Tropical Atmosphere Ocean project (TAO) buoys over the tropical Pacific ocean, and ARGO floats data over the western North Pacific for boreal winter (February) and spring (May). Sensitive experiments are conducted using an ocean-atmosphere coupled model (i.e., GloSea5) and an uncoupled ocean model (Nucleus for European Modelling of the Ocean, NEMO) and their results are compared. The verification results revealed an overall good performance for the SST predictions over the tropical Pacific ocean and near the Korean marginal seas, in which the Root Mean Square Errors (RMSE) were $0.31{\sim}0.45^{\circ}C$ and $0.74{\sim}1.11^{\circ}C$ respectively, except oceanic front regions with large spatial and temporal SST variations (the maximum error reached up to $3^{\circ}C$). The sensitive numerical experiments showed that GloSea5 outperformed NEMO over the tropical Pacific in terms of bias and RMSE analysis, while NEMO outperformed GloSea5 near the KP regions. These results suggest that the atmosphere-ocean coupling substantially influences the short-term ocean forecast over the tropical Pacific, while other factors such as atmospheric forcing and the accuracy of simulated local current are more important than the coupling effect for the KP regions being far from tropics during the development of ocean stratification.

• 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.

• 대기
• /
• 제20권2호
• /
• pp.73-89
• /
• 2010

An updateable model output statistics (UMOS) system for daily maximum and minimum temperature ($T_M$ and $T_m$) over South Korea based on the Canadian UMOS system were developed and validated. RDAPS (regional data assimilation and prediction system) and KWRF (Korea WRF) which have quite different physics and dynamics were used for the development of UMOS system. The 20 most frequently selected potential predictors for each season, station, and forecast projection time from the 68 potential predictors of the MOS system, were used as potential predictors of the UMOS system. The UMOS equations were developed through the weighted blending of the new and old model data, with weights chosen to emphasize the new model data while including enough old model data to ensure stable equations and a smooth transition of dependency from the old model to the new model. The UMOS equations are being updated by every 7 days. The validation results of $T_M$ and $T_m$ showed that seasonal mean bias, RMSE, and correlation coefficients for the total forecast projection times are -0.41-0.17 K, 1.80-2.46 K, and 0.80-0.97, respectively. The performance is slightly better in autumn and winter than in spring and summer. Also the performance of UMOS system are clearly dependent on location, better at the coastal region than inland area. As in the MOS system, the performance of UMOS system is degraded as the forecast day increases.

• 한국환경과학회지
• /
• 제30권12호
• /
• pp.1053-1065
• /
• 2021

In this study, the prediction technology of Hydrological Quantitative Precipitation Forecast (HQPF) was improved by optimizing the weather predictors used as input data for machine learning. Results comparison was conducted using bias and Root Mean Square Error (RMSE), which are predictive accuracy verification indicators, based on the heavy rain case on August 21, 2021. By comparing the rainfall simulated using the improved HQPF and the observed accumulated rainfall, it was revealed that all HQPFs (conventional HQPF and improved HQPF 1 and HQPF 2) showed a decrease in rainfall as the lead time increased for the entire grid region. Hence, the difference from the observed rainfall increased. In the accumulated rainfall evaluation due to the reduction of input factors, compared to the existing HQPF, improved HQPF 1 and 2 predicted a larger accumulated rainfall. Furthermore, HQPF 2 used the lowest number of input factors and simulated more accumulated rainfall than that projected by conventional HQPF and HQPF 1. By improving the performance of conventional machine learning despite using lesser variables, the preprocessing period and model execution time can be reduced, thereby contributing to model optimization. As an additional advanced method of HQPF 1 and 2 mentioned above, a simulated analysis of the Local ENsemble prediction System (LENS) ensemble member and low pressure, one of the observed meteorological factors, was analyzed. Based on the results of this study, if we select for the positively performing ensemble members based on the heavy rain characteristics of Korea or apply additional weights differently for each ensemble member, the prediction accuracy is expected to increase.