Real-Time Forecasting of Snowfall Using a Neural Network

Roebber, Paul J.; Butt, Melissa R.; Reinke, Sarah J.; Grafenauer, Thomas J.

doi:10.1175/waf1000.1

Cited by 34 publications

(28 citation statements)

References 7 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…This has implications for multiple forecasting applications where NN ensembles have been used. Some examples include diverse forecasting applications such as: economic modelling and policy making (McAdam and McNelis, 2005;Inoue and Kilian, 2008), financial and commodities trading (Zhang and Berardi, 2001;Chen and Leung, 2004;Versace et al, 2004;Bodyanskiy and Popov, 2006;Yu et al, 2008), fast-moving consumer goods (Trapero et al, 2012), tourism (Pattie and Snyder, 1996), electricity load (Hippert et al, 2001;Taylor and Buizza, 2002), temperature and weather (Roebber et al, 2007;Langella et al, 2010), river flood (Campolo et al, 1999) and hydrological modelling (Dawson and Wilby, 2001), climate (Fildes and Kourentzes, 2011), and ecology (Araújo and New, 2007) to name a few. Zhang et al (1998) lists multiple other forecasting applications where they have been employed successfully.…”

Section: Introductionmentioning

confidence: 99%

Neural network ensemble operators for time series forecasting

Kourentzes

Barrow

Crone

2014

Expert Systems with Applications

250

126

View full text Add to dashboard Cite

The combination of forecasts resulting from an ensemble of neural networks has been shown to outperform the use of a single "best" network model. This is supported by an extensive body of literature, which shows that combining generally leads to improvements in forecasting accuracy and robustness, and that using the mean operator often outperforms more complex methods of combining forecasts. This paper proposes a mode ensemble operator based on kernel density estimation, which unlike the mean operator is insensitive to outliers and deviations from normality, and unlike the median operator does not require symmetric distributions. The three operators are compared empirically and the proposed mode ensemble operator is found to produce the most accurate forecasts, followed by the median, while the mean has relatively poor performance. The findings suggest that the mode operator should be considered as an alternative to the mean and median operators in forecasting applications. Experiments indicate that mode ensembles are useful in automating neural network models across a large number of time series, overcoming issues of uncertainty associated with data sampling, the stochasticity of neural network training and the distribution of the forecasts.

show abstract

Section: Introductionmentioning

confidence: 99%

Neural network ensemble operators for time series forecasting

Kourentzes

Barrow

Crone

2014

Expert Systems with Applications

250

126

View full text Add to dashboard Cite

show abstract

“…, 인공신경망 모형을 이용하여 강수량, 온도 등을 입력자료로 훈련시켜 최심신적설량을 예측하거나 (Roebber et al, 2003;Roebber et al, 2007;Park et al, 2014), 회귀분석과 같은 통 계모형을 적용하여 최심신적설량을 예측하였다 (Lim and Hong, 2007;Choi et al, 2008;Kim et al, 2014;Joh et al, 2011 (1) = 회귀계수 x 1 = 강수량(mm)…”

Section: 서 론unclassified

Frequency Analysis of Future Maximum Fresh Snow Depth using Multiple Regression Model with Interaction

Park

Jeong

Chung³

2016

Journal of Korean Society of Hazard Mitigation

View full text Add to dashboard Cite

As the frequency of snowfall inducing life loss and property damages increases, many researches to calculate future maximum fresh snow depth and trend have been implemented. In this study, 400-year pre-industrial climate control simulated RCP 2.6, 4.5, 6.0, 8.5 climate change scenario for Korean Peninsula was applied to develop multiple regression models for the estimation of future probable maximum fresh snow depth. Precipitation, minimum, maximum, and average temperature was used as the independent variables, and the multiplication of precipitation and average temperature was inserted in the multiple regression model to consider the interaction between the two variables. Total 74 gauging stations data was applied to varify the developed model. The calculated probable maximum fresh snow depths for 10, 30, 50, and 100 years were compared. The model efficiency of the multiple regression model with interaction was superior to multiple linear regression model. The probable maximum fresh snow depth tend to decrease over the entire Korean Peninsula without the significant changes. The spatial distribution of the locations with highest maximum fresh snow depths remains the same. The results shows the overall variation trend rather than the exact snow depth. Using the developed model, the maximum fresh snow depth could be calculated using the precipitation and temperatures if the input variables are predicted accurately.

show abstract

“…It is not infrequent that mathematical theories that are well-known in some research areas strongly contribute to the creation of new methods and perspectives even in fields remote to those that motivated their introduction. This is the case of the so-called artificial neural networks (ANNs), which in recent years have been applied as a statistical data modeling tool to atmospheric sciences [23,45,1,30], energy systems [28,26,38,16], experimental and clinical medicine [6,43,3,48,31,44,20,19,4,37], medical diagnosis [29,5,7,41], and a huge variety of other settings. An ANN is a mathematical/computational model inspired by the structure and functional aspects of biological neural networks.…”

Section: Introductionmentioning

confidence: 99%

An artificial neural network approach for modeling the ward atmosphere in a medical unit

Schiavo

Prinari

Gronski

et al. 2015

Mathematics and Computers in Simulation

View full text Add to dashboard Cite

Artificial neural networks (ANNs) have been developed, implemented and tested on the basis of a four-year-long experimental data set, with the aim of analyzing the performance and clinical outcome of an existing medical ward, and predicting the effects that possible readjustments and/or interventions on the structure may produce on it. Advantages of the ANN technique over more traditional mathematical models are twofold: on one hand, this approach deals quite naturally with a large number of parameters/variables, and also allows to identify those variables which do not play a crucial role in the system dynamics; on the other hand, the implemented ANN can be more easily used by a staff of non-mathematicians in the unit, as an on-site predictive tool. As such, the ANN model is particularly suitable for the case study. The predictions from the ANN technique are then compared and contrasted with those obtained from a generalized kinetic approach previously proposed and tested by the authors. The comparison on the two case periods shows the ANN predictions to be somewhat closer to the experimental values. However, the mean deviations and the analysis of the statistical coefficients over a span of multiple years suggest the kinetic model to be more reliable in the long run, i.e., its predictions can be considered as acceptable even on periods that are quite far away from the two case periods over which the many parameters of the model had been optimized. The approach under study, referring to paradigms and methods of physical and mathematical models integrated with psychosocial sciences, has good chances of gaining the attention of the scientific community in both areas, and hence of eventually obtaining wider diffusion and generalization.

show abstract

Real-Time Forecasting of Snowfall Using a Neural Network

Cited by 34 publications

References 7 publications

Neural network ensemble operators for time series forecasting

Neural network ensemble operators for time series forecasting

Frequency Analysis of Future Maximum Fresh Snow Depth using Multiple Regression Model with Interaction

An artificial neural network approach for modeling the ward atmosphere in a medical unit

Contact Info

Product

Resources

About