• 한국수자원학회:학술대회논문집
• /
• 한국수자원학회 2012년도 학술발표회
• /
• pp.39-39
• /
• 2012

Data assimilation techniques have received growing attention due to their capability to improve prediction in various areas. Despite of their potentials, applicable software frameworks to probabilistic approaches and data assimilation are still limited because the most of hydrologic modelling software are based on a deterministic approach. In this study, we developed a hydrological modelling framework for sequential data assimilation, namely MPI-OHyMoS. MPI-OHyMoS allows user to develop his/her own element models and to easily build a total simulation system model for hydrological simulations. Unlike process-based modelling framework, this software framework benefits from its object-oriented feature to flexibly represent hydrological processes without any change of the main library. In this software framework, sequential data assimilation based on the particle filters is available for any hydrologic models considering various sources of uncertainty originated from input forcing, parameters and observations. The particle filters are a Bayesian learning process in which the propagation of all uncertainties is carried out by a suitable selection of randomly generated particles without any assumptions about the nature of the distributions. In MPI-OHyMoS, ensemble simulations are parallelized, which can take advantage of high performance computing (HPC) system. We applied this software framework for several catchments in Japan using a distributed hydrologic model. Uncertainty of model parameters and radar rainfall estimates is assessed simultaneously in sequential data assimilation.

• Journal of the korean society of oceanography
• /
• 제38권2호
• /
• pp.52-59
• /
• 2003

In this paper, a brief overview on data assimilation is provided in the context of oceanographic application. The ocean data assimilation needs to ingest various types of data such as satellites and floats, thus essentially requires dynamically-consistent assimilation methods. For such purpose, sequential and variational approaches are discussed and compared. The major advantage of the Kalman filter (KF) is that it can forecast error covariances at each time step. However, for models with very large dimension of state vector, the KF Is exceedingly expensive and computationally less efficient than four-dimensional variational assimilation (4D-Var). For operational application, simplified 4D-Var schemes as well as ensemble KF may be considered.

• 한국인터넷방송통신학회논문지
• /
• 제17권2호
• /
• pp.271-277
• /
• 2017

현대에 들어서면서 대기오염 물질이 심각하게 국민의 건강을 위협하는 단계에 이르렀기 때문에 이에 대한 예보의 중요성은 점점 높아지고 있다. 대기질을 예보하는데 있어서 예보 모델에 입력되는 초기장은 예보의 정확성에 영향을 미치는 요소이기 때문에 신뢰도 높은 초기장을 생성하는 것이 매우 중요하며, 이때 필요한 기법 중 하나가 자료동화이다. 자료동화는 대상 지역이 넓어지고, 관측소의 수가 증가될수록 더 많은 연산이 필요하기 때문에 그 수행시간이 길어진다. 때문에 예보 규모가 커질수록 기존의 순차처리 방식으로는 빠른 처리속도를 요구하는 현업에 적용하기 어렵다. 이에 본 논문에서는 자료동화 기법 중의 하나인 크레스만 방법을 CUDA를 이용하여 실시간으로 처리할 수 있는 방법을 제안하였다. 그 결과, 제안한 CUDA를 이용한 병렬처리 방법이 최소 35배 이상 속도가 향상되었다.

• 한국환경과학회:학술대회논문집
• /
• 한국환경과학회 2003년도 International Symposium on Clean Environment
• /
• pp.73-78
• /
• 2003

In this research, we implement Realtime Air Diffusion Prediction System which is a parallel Fortran model running on distributed-memory parallel computers. The system is designed for air diffusion simulations with four-dimensional data assimilation. For regional air quality forecasting a series of dynamic downscaling technique is adopted using the NCAR/Penn. State MM5 model which is an atmospheric model. The realtime initial data have been provided daily from the KMA (Korean Meteorological Administration) global spectral model output. It takes huge resources of computation to get 24 hour air quality forecast with this four step dynamic downscaling (27km, 9km, 3km, and lkm). Parallel implementation of the realtime system is imperative to achieve increased throughput since the realtime system have to be performed which correct timing behavior and the sequential code requires a large amount of CPU time for typical simulations. The parallel system uses MPI (Message Passing Interface), a standard library to support high-level routines for message passing. We validate the parallel model by comparing it with the sequential model. For realtime running, we implement a cluster computer which is a distributed-memory parallel computer that links high-performance PCs with high-speed interconnection networks. We use 32 2-CPU nodes and a Myrinet network for the cluster. Since cluster computers more cost effective than conventional distributed parallel computers, we can build a dedicated realtime computer. The system also includes web based Gill (Graphic User Interface) for convenient system management and performance monitoring so that end-users can restart the system easily when the system faults. Performance of the parallel model is analyzed by comparing its execution time with the sequential model, and by calculating communication overhead and load imbalance, which are common problems in parallel processing. Performance analysis is carried out on our cluster which has 32 2-CPU nodes.

• 한국수자원학회논문집
• /
• 제55권10호
• /
• pp.761-774
• /
• 2022

가뭄과 홍수의 예측, 기후변화가 유역 유출량, 더 나아가 수질 및 생태계에 미치는 영향의 정확한 분석을 위해서는 수문 모의 과정의 불확실성을 정량화하고 최소화하기 위한 노력이 필요하다. 수문자료동화는 수문모형의 상태량이나 매개변수를 갱신(update)하여 모의 초기 조건의 가장 가능성 있는 추정치를 생성하는 기법으로, 실시간 관측 정보를 이용하여 예측 정확도를 향상시킬 수 있는 방법이다. 본 연구에서는 airGRdatassim 모형을 이용하여 앙상블 기반 순차 자료동화 기법인 앙상블 칼만 필터와 파티클 필터로 용담댐 유역에 대해 일 유출을 모의하고, 자료동화 기법별 특성을 비교 및 분석하였다. 모의 결과, Kling-Gupta efficiency (KGE) 지표가 자료동화 적용 전 0.799에서 앙상블 칼만 필터와 파티클 필터 적용시 각각 0.826, 0.933으로 향상되었다. 또한 기상 강제력 노이즈의 범위, 갱신 대상 상태량 설정, 앙상블 수 등 수문자료동화의 설정과 관련된 하이퍼-매개변수(hyper-parameter)의 불확실성이 모의 예측 성능에 미치는 영향을 분석하였다. 강수 및 잠재 증발산 강제력의 오차 범위에 대한 민감도 분석 결과, 모든 모의 범위에서 파티클 필터가 앙상블 칼만 필터보다 예측 성능이 우수하였다. 파티클 필터는 기상 강제력 오차 크기가 작을수록 모의 성능이 향상되었으며, 앙상블 칼만 필터는 상대적으로 오차가 큰 경우 최적 성능이 확인되었다. 한편, 자료동화시 갱신되는 상태량의 종류를 줄일수록 자료동화에 의한 모의 성능은 감소하였다. 본 연구의 모의 실험 결과는 앙상블 자료동화를 이용하여 일 유출 모의 정확도 향상이 가능하지만, 최적 성능을 발휘하기 위해서는 수문자료동화 기법별 하이퍼-매개변수의 적정한 조정이 필요함을 함의한다.

• Smart Structures and Systems
• /
• 제22권4호
• /
• pp.399-411
• /
• 2018

The difficulty in modeling complex nonlinear structures lies in the presence of significant sources of uncertainties mainly attributed to sudden changes in the structure's behavior caused by regular aging factors or extreme events. Quantifying these uncertainties and accurately representing them within the complex mathematical framework of Structural Health Monitoring (SHM) are significantly essential for system identification and damage detection purposes. This study highlights the importance of uncertainty quantification in SHM frameworks, and presents a comparative analysis between intrusive and non-intrusive techniques in quantifying uncertainties for SHM purposes through two different variations of the Kalman Filter (KF) method, the Ensemble Kalman filter (EnKF) and the Polynomial Chaos Kalman Filter (PCKF). The comparative analysis is based on a numerical example that consists of a four degrees-of-freedom (DOF) system, comprising Bouc-Wen hysteretic behavior and subjected to El-Centro earthquake excitation. The comparison is based on the ability of each technique to quantify the different sources of uncertainty for SHM purposes and to accurately approximate the system state and parameters when compared to the true state with the least computational burden. While the results show that both filters are able to locate the damage in space and time and to accurately estimate the system responses and unknown parameters, the computational cost of PCKF is shown to be less than that of EnKF for a similar level of numerical accuracy.

• 한국지하수토양환경학회지:지하수토양환경
• /
• 제10권5호
• /
• pp.52-60
• /
• 2005

본 연구에서는 기존 연쇄 말콥체인(Coupled Markov Chain, CMC) 확률식의 연산 경직성을 개선하기 위하여 일반화 된 2차원 연쇄 말콥체인(Generalized Coupled Markov Chain, GCMC) 확률식이 개발되었다. 또한 개발된 확률식에 근거하여 평면상에서 무작위적으로 분포하는 참조정보를 효율적으로 활용하는 연산 알고리듬이 개발되었다. 개발된 모델은 대안적 지구통계 기법으로의 새로운 기능성을 제시한다. 본 연구를 통해 새롭게 개발된 GCMC 확률식은 기존 CMC 확률식에 비해 보다 유연한 참조 정보 활용 가능성을 가지며 특수한 경우로 기존 CMC 확률식이 유도되었다. 또한 순차적 연산의 인위적 오류 발생 기능성 및 실제 야외 데이터의 낮은 빈도를 고려하여 무작위로 추출된 위치에서 각 범위를 이용한 연산 알고리듬이 제안되었다. 개발된 모델은 가상의 2차원 토양도에 적용되었으며 기존 지구통계 기법인 SIS에 비하여 손색이 없는 새로운 지구통계 기법으로 토양 및 지질을 포함한 다양한 예측에 이용 될 수 있는 가능성을 보였다. 낮은 빈도로 샘플링 된 지시자에 대해서는 기존 지구통계 기법과 마찬가지로 저평가되는 현상을 보였으며 이를 보완하기 위하여 다양한 소스의 데이터 융합 등을 바탕으로 한 계속적인 연구가 요구된다.

• 한국농림기상학회지
• /
• 제8권2호
• /
• pp.68-76
• /
• 2006

A thermal time-based two-step phenological model was used to project flowering dates of Japanese cherry in South Korea from 1941 to 2100. The model consists of two sequential periods: the rest period described by chilling requirement and the forcing period described by heating requirement. Daily maximum and minimum temperature are used to calculate daily chill units until a pre-determined chilling requirement for rest release is met. After the projected rest release date, daily heat units (growing degree days) are accumulated until a pre-determined heating requirement for flowering is achieved. Model calculations using daily temperature data at 18 synoptic stations during 1955-2004 were compared with the observed blooming dates and resulted in 3.9 days mean absolute error, 5.1 days root mean squared error, and a correlation coefficient of 0.86. Considering that the phonology observation has never been fully standardized in Korea, this result seems reasonable. Gridded data sets of daily maximum and minimum temperature with a 270 m grid spacing were prepared for the climatological years 1941-1970 and 1971-2000 from observations at 56 synoptic stations by using a spatial interpolation scheme for correcting urban heat island effect as well as elevation effect. A 25km-resolution temperature data set covering the Korean Peninsula, prepared by the Meteorological Research Institute of Korea Meteorological Administration under the condition of Inter-governmental Panel on Climate Change-Special Report on Emission Scenarios A2, was converted to 270 m gridded data for the climatological years 2011-2040, 2041-2070 and 2071-2100. The model was run by the gridded daily maximum and minimum temperature data sets, each representing a climatological normal year for 1941-1970, 1971-2000, 2011-2040, 2041-2070, and 2071-2100. According to the model calculation, the spatially averaged flowering date for the 1971-2000 normal is shorter than that for 1941-1970 by 5.2 days. Compared with the current normal (1971-2000), flowering of Japanese cherry is expected to be earlier by 9, 21, and 29 days in the future normal years 2011-2040, 2041-2070, and 2071-2100, respectively. Southern coastal areas might experience springs with incomplete or even no Japanese cherry flowering caused by insufficient chilling for breaking bud dormancy.