Uncertainty decomposition in climate-change impact assessments: a Bayesian perspective

Ohn, Ilsang,Seo, Seung Beom,Kim, Seonghyeon,Kim, Young-Oh,Kim, Yongdai The Korean Statistical Society 2020 Communications for statistical applications and me Vol.27 No.1

A climate-impact projection usually consists of several stages, and the uncertainty of the projection is known to be quite large. It is necessary to assess how much each stage contributed to the uncertainty. We call an uncertainty quantification method in which relative contribution of each stage can be evaluated as uncertainty decomposition. We propose a new Bayesian model for uncertainty decomposition in climate change impact assessments. The proposed Bayesian model can incorporate uncertainty of natural variability and utilize data in control period. We provide a simple and efficient Gibbs sampling algorithm using the auxiliary variable technique. We compare the proposed method with other existing uncertainty decomposition methods by analyzing streamflow data for Yongdam Dam basin located at Geum River in South Korea.

Uncertainty decomposition in water resources projection considering interaction effects

Ohn Ilsang,Kim Yongdai,Kim Young-Oh 한국수자원학회 2018 한국수자원학회논문집 Vol.51 No.11

기후변화로 인한 수자원 전망은 배출 시나리오, 전지구적 순환모형, 상세화 기법, 수문 모형 등 여러 전망 단계를 거쳐 이루어지며, 각 단계는 수자 원 전망의 총 불확실성의 원천이 된다. 몇몇 연구를 통해 개별 전망 단계의 총 불확실성에 대한 상대적인 기여를 계량화하는 방법이 제안되었으며며, 이러한 분석을 불확실성 분해라고 한다. 불확실성 분해 분석은 큰 불확실성을 발생시키는 단계를 진단하고, 이를 반영한 불확실성 저감 계획을 수립할 수 있게 한다. 전망 단계 간의 교호작용은 불확실성 분해 시 고려되어야 하는 중요한 문제 중 하나이다. 본 연구는 교호작용 효과로 인한 불 확실성을 계량화하고 이를 불확실성 분해에 반영하는 새로운 방법을 제안한다. 제안한 방법은 전망 단계별 불확실성을 주효과와 교호작용 효과를 모두 고려하여 계량화함과 동시에 총 불확실성에서 개별 전망 단계가 차지하는 상대적인 비중을 제시할 수 있다는 장점이 있다. 제안한 방법을 충 주댐 유량 전망의 불확실성 분석에 적용하였다. 충주댐 유역의 불확실성 분석 결과 여름과 겨울 두 계절 모두에서 교호작용 효과의 불확실성은 주 효과의 불확실성에 비해 그 크기가 작은 것으로 나타났다. 교호작용 효과를 고려하여 불확실성을 분해한 결과 배출 시나리오, 전지구적 순환모형, 상세화 기법, 수문 모형의 네 단계 중 여름철은 전지구적 순환모형의 불확실성이, 겨울철은 상세화 기법의 불확실성이 가장 큰 것으로 분석되었다. Water resources projection typically consists of several stages including emission scenarios, global circulation models (GCMs), downscaling techniques, and hydrological models, and each stage is a source of total uncertainty in water resources projection. Several studies proposed methods to quantify the relative contribution of each stage to total uncertainty, and we call such analysis uncertainty decomposition. Uncertainty decomposition enables us to investigate the stages yielding large uncertainties and to establish the uncertainty reduction plan that reflects them. Interactions between stages is one of the important issues to be considered in uncertainty decomposition. This study suggests a new uncertainty decomposition method considering interaction effect. The proposed method has an advantage of decomposing the total uncertainty to the uncertainty from each stage considering both the main and interactions effects. We apply the proposed method to streamflow projection for Chungju Dam basin. The results show that the uncertainties from the main effects are larger than the uncertainties from interaction effects in both summer and winter. Using the proposed uncertainty decomposition method, we show that the GCM stage is the largest source of the total uncertainty in summer and the downscaling technique stage is the one in winter among the following four stages: emission scenarios, GCMs, downscaling techniques, and hydrological models.

수자원 전망의 시나리오별 불확실성 분해

이지수(Jisu Lee),온일상(Ilsang Ohn),김용대(Yongdai Kim) 한국데이터정보과학회 2018 한국데이터정보과학회지 Vol.29 No.6

기후변화로 인한 수자원 전망은 배출 시나리오, 전지구적 순환모형, 상세화 기법, 수문 모형 등 여러 전망 단계를 거쳐 이루어지며, 각 단계는 수자원 전망의 총 불확실성의 원천이 된다. 근래에 들어 수자원 전망의 총 불확실성을 개별 전망 단계로, 이를 다시 전망 단계에 속하는 여러 시나리오의 불확실성으로 분해하는 방법의 필요성이 제기되고 있다. 그러나 그 필요성에 비하여 관련 연구는 전무한 실정이다. 본 연구에서는 수자원 전망의 총 불확실성을 시나리오의 불확실성으로 분해나는 방법을 제안한다. 제안하는 방법은 개별 시나리오의 불확실성이 총 불확실성에 기여하는 상대적 비중을 제시함으로써 시나리오 간의 불확실성 비교를 가능하게 한다. 제안한 방법을 이용하여 실제 수자원 전망 자료를 분석한 결과를 제시한다. Water resource projections considering climate change are based on combinations of scenarios in multiple stages including the emission scenarios, global circulation models, downscaling techniques, and hydrological models. Each scenario is a source of total uncertainty of the water resource projections. Decomposing total uncertainty into components from individual scenarios has recently emerged as an important issue due to their usefulness. But there is no comprehensive research. In this study, we propose a method to decompose the total uncertainty into the individual uncertainty of scenarios. The proposed method makes it possible to compare the uncertainties of the scenarios by providing the relative contribution of each scenario to the total uncertainty. The results of analyzing the real data by the proposed method are presented.

Leave-one-out Bayesian model averaging for probabilistic ensemble forecasting

Kim, Yongdai,Kim, Woosung,Ohn, Ilsang,Kim, Young-Oh The Korean Statistical Society 2017 Communications for statistical applications and me Vol.24 No.1

Over the last few decades, ensemble forecasts based on global climate models have become an important part of climate forecast due to the ability to reduce uncertainty in prediction. Moreover in ensemble forecast, assessing the prediction uncertainty is as important as estimating the optimal weights, and this is achieved through a probabilistic forecast which is based on the predictive distribution of future climate. The Bayesian model averaging has received much attention as a tool of probabilistic forecasting due to its simplicity and superior prediction. In this paper, we propose a new Bayesian model averaging method for probabilistic ensemble forecasting. The proposed method combines a deterministic ensemble forecast based on a multivariate regression approach with Bayesian model averaging. We demonstrate that the proposed method is better in prediction than the standard Bayesian model averaging approach by analyzing monthly average precipitations and temperatures for ten cities in Korea.

기후변화에 따른 소양호의 수온 장기 모의 및 불확실성 정량화

윤여정 ( Yeojeong Yun ),박형석 ( Hyungseok Park ),정세웅 ( Sewoong Chung ),김용대 ( Yongda Kim ),온일상 ( Ilsang Ohn ),이서로 ( Seoro Lee ) 한국물환경학회(구 한국수질보전학회) 2020 한국물환경학회지 Vol.36 No.1

Future climate change may affect the hydro-thermal and biogeochemical characteristics of dam reservoirs, the most important water resources in Korea. Thus, scientific projection of the impact of climate change on the reservoir environment, factoring uncertainties, is crucial for sustainable water use. The purpose of this study was to predict the future water temperature and stratification structure of the Soyanggang Reservoir in response to a total of 42 scenarios, combining two climate scenarios, seven GCM models, one surface runoff model, and three wind scenarios of hydrodynamic model, and to quantify the uncertainty of each modeling step and scenario. Although there are differences depending on the scenarios, the annual reservoir water temperature tended to rise steadily. In the RCP 4.5 and 8.5 scenarios, the upper water temperature is expected to rise by 0.029 ℃ (±0.012)/year and 0.048 ℃ (±0.014)/year, respectively. These rise rates are correspond to 88.1 % and 85.7 % of the air temperature rise rate. Meanwhile, the lower water temperature is expected to rise by 0.016 ℃ (±0.009)/year and 0.027 ℃ (±0.010)/year, respectively, which is approximately 48.6 % and 46.3 % of the air temperature rise rate. Additionally, as the water temperatures rises, the stratification strength of the reservoir is expected to be stronger, and the number of days when the temperature difference between the upper and lower layers exceeds 5 ℃ increases in the future. As a result of uncertainty quantification, the uncertainty of the GCM models showed the highest contribution with 55.8 %, followed by 30.8 % RCP scenario, and 12.8 % W2 model.