Predicting ground-level ozone concentrations by adaptive Bayesian model averaging of statistical seasonal models

Mok, Kai Meng; Yuen, Ka‐Veng; Hoi, Ka In; Chao, K. M.; Lopes, Diogo

doi:10.1007/s00477-017-1473-1

Cited by 12 publications

(5 citation statements)

References 49 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…In more recent studies, there is a trend to use a method called Bayesian Model Averaging (BMA) for flood forecasting, which is a scheme for the model combination using likelihood measures as model weights (Yan & Moradkhani, ; Huo et al., ). BMA provides a coherent mechanism to account for model uncertainty and usually outperforms other multimodel combination methods for which it is gaining popularity (Mok, Yuen, Hoi, Chao, & Lopes, ; Samani, Moghaddam, & Ye, ). While the existing Bayesian models provide tools for examining flood vulnerability assessment and prediction, they do not consider the characteristics of flood control networks to inform flood infrastructure planning and prioritization.…”

Section: Literature Reviewmentioning

confidence: 99%

Bayesian modeling of flood control networks for failure cascade characterization and vulnerability assessment

Dong

Farahmand

et al. 2019

Computer aided Civil Eng

View full text Add to dashboard Cite

This paper presents a Bayesian network model to assess the vulnerability of the flood control infrastructure and to simulate failure cascade based on the topological structure of flood control networks along with hydrological information gathered from sensors. Two measures are proposed to characterize the flood control network vulnerability and failure cascade: (a) node failure probability (NFP), which determines the failure likelihood of each network component under each scenario of rainfall event, and (b) failure cascade susceptibility, which captures the susceptibility of a network component to failure due to failure of other links. The proposed model was tested in both single watershed and multiple watershed scenarios in Harris County, Texas using historical data from three different flooding events, including Hurricane Harvey in 2017. The proposed model was able to identify the most vulnerable flood control network segments prone to flooding in the face of extreme rainfall. The framework and results furnish a new tool and insights to help decision-makers to prioritize infrastructure enhancement investments and actions. The proposed Bayesian network modeling framework also enables simulation of failure cascades in flood control infrastructures, and thus could be used for scenario planning as well as near-real-time inundation forecasting to inform emergency response planning and operation, and hence improve the flood resilience of urban areas. 668 wileyonlinelibrary.com/journal/mice Comput Aided Civ Inf. 2020;35:668-684.

show abstract

Section: Literature Reviewmentioning

confidence: 99%

Bayesian modeling of flood control networks for failure cascade characterization and vulnerability assessment

Dong

Farahmand

et al. 2019

Computer aided Civil Eng

View full text Add to dashboard Cite

show abstract

“…Additionally, the spatial resolution of the model estimation results was generally coarser than 0.1 • × 0.1 • [6,16,21]. In terms of the statistical model, the machine learning model was used to estimate the MDA8 of China; however, most of these studies focused on the ozone prediction of a single city or an urban belt [13,26,[28][29][30][31][32][33][34][35], and the spatial distribution of these results is limited. For large-scale surface ozone estimation studies using the statistical model, the accuracy of the ozone estimation results is high [27,37,38].…”

Section: Discussionmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

“…As for China, surface ozone estimation has been implemented by many previous studies [17,18,[26][27][28][29][30][31][32][33][34][35][36]. Most of these studies focused on the ozone prediction of a single city or an urban belt such as Beijing, Guangzhou, Shanghai, Nanjing, Hong Kong, Lanzhou, Macau, Taipei, and Jinan in China [13,26,[28][29][30][31][32][33][34][35], and the spatial distribution of the surface ozone estimation results are limited. In terms of studies using deterministic models to estimate surface ozone at the regional and national scale, the time range of the surface ozone estimation results is short [17], the estimation accuracy needs to be improved [6,16,21], and the accuracy of the estimation results varies greatly in different regions [18].…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Spatiotemporal Patterns and Quantitative Analysis of Factors Influencing Surface Ozone over East China

Ma,

Liu,

Liu

et al. 2023

Sustainability

View full text Add to dashboard Cite

Surface ozone pollution in China has been persistently becoming worse in recent years; therefore, it is of great importance to accurately estimate ozone pollution and explore the spatiotemporal variations in surface ozone in East China. By using S5P-TROPOMI-observed NO2, HCHO data (7 km × 3.5 km), and other surface-ozone-influencing factors, including VOCs, meteorological data, NOX emission inventory, NDVI, DEM, population, land use and land cover, and hourly in situ surface ozone observations, an extreme gradient boosting model was used to estimate the daily 0.05° × 0.05° gridded maximum daily average 8 h ozone (MDA8) in East China during 2019–2021. Four surface ozone estimation models were established by combining NO2 and HCHO data from S5P-TROPOMI observations and CAMS reanalysis data. The sample-based validation R2 values of these four models were all larger than 0.92, while their site-based validation R2 values were larger than 0.82. The results revealed that the coverage ratio of the model using CAMS NO2 and CAMS HCHO was the highest (100%), while the coverage ratio of the model using S5P-TROPOMI NO2 and CAMS HCHO was the second highest (96.26%). Furthermore, the MDA8 estimation results of these two models were averaged to produce the final surface ozone estimation dataset. It indicated that O3 pollution in East China during 2019–2021 was susceptible to anthropogenic precursors such as VOCs (22.55%) and NOX (8.97%), as well as meteorological factors (27.35%) such as wind direction, temperature, and wind speed. Subsequently, the spatiotemporal patterns of ozone pollution were analyzed. Ozone pollution in East China is mainly concentrated in the North China Plain (NCP), the Pearl River Delta (PRD), and the Yangtze River Delta (YRD). Among these three regions, ozone pollution in the NCP mainly occurs in June (summer), ozone pollution in the YRD mainly occurs in May (spring), and ozone pollution in the PRD mainly occurs in April (spring) and September (autumn). In addition, surface O3 concentration in East China decreased by 3.74% in 2020 compared to 2019, which may have been influenced by the COVID-19 epidemic and the implementation of the policy of synergistic management of PM2.5 and O3 pollution. The regions mostly affected by the COVID-19 epidemic and the policy of the synergistic management of PM2.5 and O3 pollution were the NCP (−2~−8%), the Middle and Lower of Yangtze Plain (−6~−10%), and the PRD (−4~−10%). Overall, the estimated 0.05° × 0.05° gridded surface ozone in East China from 2019 to 2021 provides a promising data source and data analysis basis for the related researchers. Meanwhile, it reveals the spatial and temporal patterns of O3 pollution and the main influencing factors, which provides a good basis for the control and management of O3 pollution, and also provides technical support for the sustainable development of the environment in East China.

show abstract

“…natural and anthropogenic sources), and model structure and parameters have a significant impact on the reliability and accuracy of water quality forecasts (Moreno-Rodenas et al, 2019). Several techniques were commonly used to quantify the uncertainty of water quality forecasts, for instance, (1) pre-processing techniques: the Fuzzy Clustering (FC) method (Kim and Pachepsky, 2010), the Wavelet Transform (WT) (Barzegar et al, 2018) and the bias-correction method (Libera and Sankarasubramanian, 2018) and (2) post-processing techniques: the Multiple Linear Regression (MLR) (Wallace et al, 2016), the Kalman filtering (Rajakumar et al, 2019;Zhou et al, 2020), the Generalized Likelihood Uncertainty Estimation (GLUE) (Zhang et al, 2015), the Bayesian Model Averaging (BMA) (Mok et al, 2018) and the Bayesian Uncertainty Processor (BUP) (Borsuk et al, 2002;Arhonditsis et al, 2019). The creation of probabilistic forecast intervals could be taken as one of the effective approaches to quantify the impact of different uncertainties on water quality forecasting (Krapu et al, 2019).…”

Section: Introductionmentioning

confidence: 99%

Real-time probabilistic forecasting of river water quality under data missing situation: Deep learning plus post-processing techniques

Zhou

2020

Journal of Hydrology

137

View full text Add to dashboard Cite

Quantifying the uncertainty of probabilistic water quality forecasting induced by missing input data is fundamentally challenging. This study introduced a novel methodology for probabilistic water quality forecasting conditional on point forecasts.A Multivariate Bayesian Uncertainty Processor (MBUP) was adopted to probabilistically model the relationship between the point forecasts made by a deep learning artificial neural network (ANN) and their corresponding observed water quality. The methodology was tested using hourly water quality series at an island of Shanghai City in China. The novelties relied upon: firstly, the use of a transfer learning algorithm to overcome flatten-and under-prediction bottlenecks of river water quality raised in artificial neural networks, and secondly, the use of the MBUP to capture the dependence structure between observations and forecasts. Two deep learning ANNs were used to make the point forecasts. Then the MBUP approach driven by the point forecasts demonstrated its competency in improving the accuracy of probabilistic water quality forecasts significantly, where predictive distributions encountered in multi-step-ahead water quality forecasts were effectively reduced to small ranges. The results demonstrated that the deep learning plus the post-processing approach suitably extracted the complex dependence structure between the model's output and observed water quality so that model reliability (Containing Ratio > 85% and average Relative Band-width < 0.25) as well as forecast accuracy (Nash-Sutcliffe Efficiency coefficient > 0.8 and Root-Mean-Square-Error < 0.4 mg/l) for future horizons from 1 hour up to 10 hours were significantly improved, even if the input data missing rate reaches 50%.

show abstract

Predicting ground-level ozone concentrations by adaptive Bayesian model averaging of statistical seasonal models

Cited by 12 publications

References 49 publications

Bayesian modeling of flood control networks for failure cascade characterization and vulnerability assessment

Bayesian modeling of flood control networks for failure cascade characterization and vulnerability assessment

Spatiotemporal Patterns and Quantitative Analysis of Factors Influencing Surface Ozone over East China

Real-time probabilistic forecasting of river water quality under data missing situation: Deep learning plus post-processing techniques

Contact Info

Product

Resources

About