Current Forecast Models Research Articles

Statistical Forecast Model for Ice-Related Events in the Arctic

Abstract This paper presents a statistical ice event forecast model for the Arctic based on Fourier transforms and a mathematical filter. The results indicate that this model compares very well with both a multiple regression model and a human-made forecast. There seems to be a direct link between the period associated with the dominant spectral peak of the Fourier transform and the ease with which the date of events, such as fractures, bergy water, or open water, can be forecast. While useful for the normal timing of events, at this time, none of the current forecast models can predict events that occur before or beyond the usual or historical dates, which poses a forecast problem in the Arctic.

Integrated parallel forecasting model based on modified fuzzy time series and SVM

A dynamic parallel forecasting model is proposed, which is based on the problem of current forecasting models and their combined model. According to the process of the model, the fuzzy C-means clustering algorithm is improved in outliers operation and distance in the clusters and among the clusters. Firstly, the input data sets are optimized and their coherence is ensured, the region scale algorithm is modified and non-isometric multiscale region fuzzy time series model is built. At the same time, the particle swarm optimization algorithm about the particle speed, location and inertia weight value is improved, this method is used to optimize the parameters of support vector machine, construct the combined forecast model, build the dynamic parallel forecast model, and calculate the dynamic weight values and regard the product of the weight value and forecast value to be the final forecast values. At last, the example shows the improved forecast model is effective and accurate.

Predictability of Seasonal Streamflow in a Changing Climate in the Sierra Nevada

The goal of this work is to assess climate change and its impact on the predictability of seasonal (i.e., April–July) streamflow in major water supply watersheds in the Sierra Nevada. The specific objective is threefold: (1) to examine the hydroclimatic impact of climate change on precipitation and temperature at the watershed scale, as well as the variability and trends in the predictand (i.e., April–July streamflow runoff) and its operational predictors (including 1 April snow water equivalent, October–March precipitation and runoff, and April–June precipitation) in a changing climate; (2) to detect potential changes in the predictability of April–July streamflow runoff in response to climate change; and (3) to assess the relationship between April–July streamflow runoff and potential new predictors and the corresponding trend. Historical records (water year 1930–2015) of annual peak snow water equivalent, monthly full natural flow, monthly temperature and precipitation data from 12 major watersheds in the west side of the Sierra Nevada in California (which are of great water supply interest) are analyzed. The Mann-Kendall Trend-Free Pre-Whitening procedure is applied in trend analysis. The results indicate that no significant changes in both the predictand and predictors are detected. However, their variabilities tend to be increasing in general. Additionally, the predictability of the April–July runoff contributed from each predictor is generally increasing. The study further shows that standardized precipitation, runoff, and snow indices have higher predictability than their raw data counterparts. These findings are meaningful from both theoretical and practical perspectives, in terms of guiding the development of new forecasting models and enhancing the current operational forecasting model, respectively, for improved seasonal streamflow forecasting.

Two combined forecasting models based on singular spectrum analysis and intelligent optimized algorithm for short-term wind speed

Short-term wind speed forecasting has become an important technology to utilize sustainable energy, reduce the impact of wind power grid and improve the control of wind turbines. Current forecasting models based on individual algorithm and hybrid optimization algorithm could be applied to a variety of wind speed forecast. However, these algorithms ignored the limitation of optimizing parameter and the use of recent data, which may result in forecasting poor accuracy. In this paper, integrated approaches, combining singular spectrum analysis technique, cuckoo search algorithm (CSA) and harmony search algorithm (HSA), and back propagation neural network (BPNN), were introduced for conducting short-term wind speed forecasting. Firstly, the SSA technique is used for identifying and extracting instable components from raw wind speed signals. Then, the parameters of BPNN are employed to be optimized by the CSA/HSA, which improve precision of forecast results. Finally, the BPNN is utilized to deal with the wind speed series. The proposed hybrid model is conducted by the wind speed at three stations located in Penglai, China. Our experiment reveals that the proposed hybrid model can generate a more accurate, reliable and robust result and SSA technique is found to be superior technique to preprocess wind speed series. Furthermore, we compare to the forecast results with SSA and non SSA denoising procedure and found that hybrid model with SSA technique outperforms other prediction model.

Modulation of precipitation by conditional symmetric instability release

Although many theoretical and observational studies have investigated the mechanism of conditional symmetric instability (CSI) release and associated it with mesoscale atmospheric phenomena such as frontal precipitation bands, cloud heads in rapidly developing extratropical cyclones and sting jets, its climatology and contribution to precipitation have not been extensively documented. The aim of this paper is to quantify the contribution of CSI release, yielding slantwise convection, to climatological precipitation accumulations for the North Atlantic and western Europe. Case studies reveal that CSI release could be common along cold fronts of mature extratropical cyclones and the North Atlantic storm track is found to be a region with large CSI according to two independent CSI metrics. Correlations of CSI with accumulated precipitation are also large in this region and CSI release is inferred to be occurring about 20% of the total time over depths of over 1 km. We conclude that the inability of current global weather forecast and climate prediction models to represent CSI release (due to insufficient resolution yet lack of subgrid parametrization schemes) may lead to errors in precipitation distributions, particularly in the region of the North Atlantic storm track.

Bayesian regularisation neural network based on artificial intelligence optimisation

Stock prediction is generally considered to be challenging and known for its high noise and strong nonlinearities in financial time series analysis. However, current forecasting models ignore the importance of model parameter optimisation and the use of recent data. In this article, a novel forecasting approach with a Bayesian-regularised artificial neural networks (BR-ANN) was proposed. The weight of the proposed model (BR-ANN) is determined by the particle swarm optimisation (PSO) algorithm. Daily market prices and financial technical indicators are utilised as inputs to predict the one day future closing price of the Shanghai (in China) composite index. The Bayesian-regularised network uses a probabilistic nature for the network weights and can reduce the potential for over-fitting and over-training. Our empirical study and the results of our K-line theory analysis indicate that PSO is determined to be an effective algorithm to optimise the parameters of the Bayesian neural network compared with other well-known prediction algorithms. In particular, the PSO model is more reliable than the simple Bayesian regularisation neural network near the local maximum value.

Previsões multiescala de vazões para o sistema hidrelétrico brasileiro utilizando ponderação bayesiana de modelos (BMA)

RESUMO O uso de sistemas eficientes de previsão de afluências nas diversas escalas temporais permite otimizar a operação do conjunto de reservatórios hidrelétricos brasileiros, elevando o grau de segurança no fornecimento de energia elétrica e minimizando os custos operacionais. Entretanto, os modelos atuais de previsão utilizados pelo Operador Nacional do Sistema Elétrico (ONS) tendem a ser limitados no horizonte de previsão e na modelagem da dependência existente entre as diversas escalas de tempo, reduzindo a qualidade das previsões. Neste trabalho é proposta uma nova contribuição para os modelos de previsão de afluências em uso pelo ONS a partir do conceito de ponderação bayesiana de modelos (BMA), que permite integrar previsões mensais e semanais de vazões com objetivo de melhorar o desempenho das previsões semanais. As previsões mensais são obtidas por meio de um modelo periódico auto-regressivo exógeno (PARX), que busca captar a persistência das vazões na parte auto-regressiva e a contribuição do escoamento superficial na parcela exógena por meio do uso de informações climáticas de larga escala. Previsões semanais de afluência com até seis semanas de antecedência são obtidas a partir das informações disponibilizadas pelo ONS nos relatórios do Programa Mensal de Operação (PMO). A metodologia proposta é aplicada em séries de afluências semanais aos 28 principais reservatórios hidroelétricos brasileiros. Os resultados de previsão semanal de afluências obtidos com a ponderação das saídas dos modelos de previsão semanal e mensal indicam uma melhoria significativa em indicadores de desempenho de previsões (NS, MAPE e DM) quando comparados com os resultados de previsão oriundos do modelo semanal isolado. Os ganhos obtidos nos indicadores de desempenho são mais significativos a partir da segunda semana de antecedência. A abordagem proposta é flexível em termos de implementação, permitindo integrar outras escalas de previsão assim como diferentes modelos preditivos (por exemplo, modelos de base física).

The interaction between gravity waves and solar tides in a linear tidal model with a 4‐D ray‐tracing gravity‐wave parameterization

AbstractGravity waves (GWs) play an important role in atmospheric dynamics. Due to their short wavelengths, they must be parameterized in current weather and forecast models, which cannot resolve them explicitly. We are here the first to report the possibility and the implication of having an online GW parameterization in a linear but global model that incorporates their horizontal propagation, the effects of transients and of horizontal background gradients on GW dynamics. The GW parameterization is based on a ray‐tracer model with a spectral formulation that is safe against numerical instabilities due to caustics. The global model integrates the linearized primitive equations to obtain solar tides (STs), with a seasonally dependent reference climatology, forced by a climatological daily cycle of the tropospheric and stratospheric heating, and the (instantaneous) GW momentum and buoyancy flux convergences resulting from the ray tracer. Under a more conventional “single‐column” approximation, where GWs only propagate vertically and do not respond to horizontal gradients of the resolved flow, GW impacts are shown to be significantly changed in comparison with “full” experiments, leading to significant differences in ST amplitudes and phases, pointing at a sensitive issue of GW parameterizations in general. In the full experiment, significant semidiurnal STs arise even if the tidal model is only forced by diurnal heating rates. This indicates that an important part of the tidal signal is forced directly by GWs via their momentum and buoyancy deposition. In general, the effect of horizontal GW propagation and the GW response to horizontal large‐scale flow gradients is rather observed in nonmigrating than in migrating tidal components.

Improving the WRF model's (version 3.6.1) simulation over sea ice surface through coupling with a complex thermodynamic sea ice model (HIGHTSI)

Abstract. Sea ice plays an important role in the air–ice–ocean interaction, but it is often represented simply in many regional atmospheric models. The Noah sea ice scheme, which is the only option in the current Weather Research and Forecasting (WRF) model (version 3.6.1), has a problem of energy imbalance due to its simplification in snow processes and lack of ablation and accretion processes in ice. Validated against the Surface Heat Budget of the Arctic Ocean (SHEBA) in situ observations, Noah underestimates the sea ice temperature which can reach −10 °C in winter. Sensitivity tests show that this bias is mainly attributed to the simulation within the ice when a time-dependent ice thickness is specified. Compared with the Noah sea ice model, the high-resolution thermodynamic snow and ice model (HIGHTSI) uses more realistic thermodynamics for snow and ice. Most importantly, HIGHTSI includes the ablation and accretion processes of sea ice and uses an interpolation method which can ensure the heat conservation during its integration. These allow the HIGHTSI to better resolve the energy balance in the sea ice, and the bias in sea ice temperature is reduced considerably. When HIGHTSI is coupled with the WRF model, the simulation of sea ice temperature by the original Polar WRF is greatly improved. Considering the bias with reference to SHEBA observations, WRF-HIGHTSI improves the simulation of surface temperature, 2 m air temperature and surface upward long-wave radiation flux in winter by 6, 5 °C and 20 W m−2, respectively. A discussion on the impact of specifying sea ice thickness in the WRF model is presented. Consistent with previous research, prescribing the sea ice thickness with observational information results in the best simulation among the available methods. If no observational information is available, we present a new method in which the sea ice thickness is initialized from empirical estimation and its further change is predicted by a complex thermodynamic sea ice model. The ice thickness simulated by this method depends much on the quality of the initial guess of the ice thickness and the role of the ice dynamic processes.

A Novel Combined Model Based on an Artificial Intelligence Algorithm—A Case Study on Wind Speed Forecasting in Penglai, China

Wind speed forecasting plays a key role in wind-related engineering studies and is important in the management of wind farms. Current forecasting models based on different optimization algorithms can be adapted to various wind speed time series data. However, these methodologies cannot aggregate different hybrid forecasting methods and take advantage of the component models. To avoid these limitations, we propose a novel combined forecasting model called SSA-PSO-DWCM, i.e., particle swarm optimization (PSO) determined weight coefficients model. This model consisted of three main steps: one is the decomposition of the original wind speed signals to discard the noise, the second is the parameter optimization of the forecasting method, and the last is the combination of different models in a nonlinear way. The proposed combined model is examined by forecasting the wind speed (10-min intervals) of wind turbine 5 located in the Penglai region of China. The simulations reveal that the proposed combined model demonstrates a more reliable forecast than the component forecasting engines and the traditional combined method, which is based on a linear method.

Evolving granular analytics for interval time series forecasting

As a paradigm of data processing, granular computation concerns processing complex data entities called granules, which arise from data abstraction and derivation of knowledge from data. This paper addresses granular computation within the framework of interval time series forecasting and evolving intelligent systems. It develops a generalized interval evolving possibilistic fuzzy modeling algorithm as an analytics tool capable to process interval data stream and to produce interval forecasts. The algorithm uses interval arithmetic in its processing steps, employs the notion of data density to adapt the current forecasting model as data are input, and computes (dis)similarity between interval data using the Hausdorff distance. Computational experiments include forecasting of an interval time series data produced by a synthetic time-varying model with parameter drift, and forecasting of financial interval time series using actual daily minimum and maximum values of the US and Brazilian main equity indexes, S&P 500 and IBOVESPA, respectively. The results suggest that the generalized interval evolving possibilistic fuzzy algorithm is highly effective to model and forecast interval time series. It has comparable or better performance than alternative evolving fuzzy and benchmark interval-based approaches.

Development of an oil spill forecast system for offshore China

An oil spill forecast system for offshore China was developed based on Visual C++. The oil spill forecast system includes an ocean environmental forecast model and an oil spill model. The ocean environmental forecast model was designed to include timesaving methods, and comprised a parametrical wind wave forecast model and a sea surface current forecast model. The oil spill model was based on the “particle method” and fulfills the prediction of oil particle behavior by considering the drifting, evaporation and emulsification processes. A specific database was embedded into the oil spill forecast system, which contained fundamental information, such as the properties of oil, reserve of emergency equipment and distribution of marine petroleum platform. The oil spill forecast system was successfully applied as part of an oil spill emergency exercise, and provides an operational service in the Research and Development Center for Offshore Oil Safety and Environmental Technology.

Pacific Ocean SST and Z500 climate variability and western U.S. seasonal streamflow

ABSTRACTThe current study focuses on evaluating the relationship between the Pacific Ocean climate variability and western US streamflow for six hydrologic regions of the western United States, as defined by United States Geological Survey: Rio Grande, Upper Colorado, Lower Colorado, Great Basin, Pacific Northwest, and California. The singular‐valued decomposition (SVD) technique was applied on data for 50 years (1960–2010) of sea‐surface temperatures (SST), geopotential height index of 500 mbar (Z500), and 90 unimpaired western US streamflows; the results established a spatio‐temporal association for each major hydrologic region in the western United States with Pacific Oceanic variability. An approach using a 3‐ to 9‐month lead time was utilized, i.e. the previous year's July to August SST/Z500, the previous year's October to December SST/Z500 to predict streamflow for current year spring–summer (April to September), spring (April to June), and summer (July to September) seasons. Significant regions in the Pacific were identified that influence hydrology of the western United States. The traditional El Niño/Southern Oscillation (ENSO) and Pacific Decadal Oscillation regions were identified along with regions over eastern Russia, Canadian British Columbia, and the ‘Hondo’ region along the east coast of Japan. Z500 showed pronounced association with 3‐month lead time streamflow, whereas SST had better association with 6‐month lead time streamflow. The SVD results showed improvement in correlation values of smaller spatial regions over larger regions, and a lagged response of adjacent hydrologic regions to the same physical indicators. The results obtained in this study could be helpful in improving the current forecasting models for water management.

An Anomaly-Based Method for Identifying Signals of Spring and Autumn Low-Temperature Events in the Yangtze River Valley, China

AbstractAbnormally low temperature (LT) events in spring and autumn can cause severe damage to spring and autumn rice production in the mid- to lower Yangtze River valley in China. Advanced predictions of such events can help mitigate their damage. However, the current methods have limited success in describing and predicting those weather events. In this study, a new method is proposed to decompose any one of the meteorological variables into its climatic component and an anomaly, and the anomaly is used in identifying signals of the LT events. The method is used in 20 strong spring LT events and 44 autumn events during 1960–2008. The results show the advanced ability of this method to clearly describe the LT events as compared with the vague indications of such events that are produced by conventional methods currently in practice in China. In addition, the composite profile of vertical anomalies shows that a negative center of geopotential height anomalies at around 300 hPa, coexisting with a strong cold center of temperature anomalies at 850 hPa, is a signature for LT events. For the 44 autumn LT events and 20 spring LT events during 1960–2008, their early disturbances were identified up to 10.2 days and 6.9 days, respectively, before the occurrence of the LT events in the valley. This result suggests that identifying the early disturbances and extracting anomalous signals from the products of current medium-range weather forecast models may be a potential way to improve the prediction skill for LT events in the valley.

A Rolling Grey Model Optimized by Particle Swarm Optimization in Economic Prediction

Grey system theory has been widely used to forecast the economic data that are often nonlinear, irregular, and nonstationary. Current forecasting models based on grey system theory could adapt to various economic time series data. However, these models ignored the importance of the model parameter optimization and the use of recent data, which lead to poor forecasting accuracy. In this article, we propose a novel forecasting model, called particle swarm optimization rolling grey model (PSO‐RGM(1,1)), based on a rolling mechanism GM with optimized parameters by using the particle swarm optimization algorithm. The simple model is shown to be very effective in forecasting the tertiary industry data sequences, which are short and noisy but regular in secular trend. The experimental results show that PSO‐RGM(1,1) outperforms other commonly used forecasting models on three real economic data sets. Our empirical study shows that PSO is found to be the best overall algorithm to optimize the parameter of RGM compared with other well‐known metaheuristics. Furthermore, we evaluated other variant PSOs and found that single particle PSO outperforms others overall in terms of prediction accuracy, convergence speed, and degree of certainty.

Dynamic linear models for forecasting of radiation belt electrons and limitations on physical interpretation of predictive models

AbstractRelationships exist between radiation belt electron flux intensities and solar drivers such as solar wind speed, ion density, and magnetic fields. The particulars of these relationships, however, are not well understood. Many forecasting models have been developed in the last 25 years, attempting to make sense of these relationships and produce accurate forecasts for electron flux intensities. We discuss some of the inherent limitations that many forecasting models (e.g., static models) possess when trying to untangle the intricate and dynamic relationships between electron flux levels and solar wind drivers. Dynamics related to the solar cycle limit physical interpretations for static forecasting models to customized and narrow time windows. Furthermore, the interrelatedness of solar drivers severely limit the ability to uniquely partition and describe the relationship between any one solar driver with electron flux levels. We suggest an alternate approach using dynamic linear models (DLMs). DLMs avoid some of the inherent limitations of physical understanding static models possess. We compare the 1 day ahead forecast accuracy of a relatively simple DLM to the current NOAA relativistic electron forecast model (REFM). The REFM does produce a more favorable prediction efficiency averaged across years when compared to the relatively simple DLM (0.749 to 0.721). However, the competitiveness of the DLM suggests that further development may lead to more accurate and interpretable models in the future.

Climate change and ice hazards in the Beaufort Sea

Abstract Recent reductions in the summer extent of sea ice have focused the world’s attention on the effects of climate change. Increased CO2-derived global warming is rapidly shrinking the Arctic multi-year ice pack. This shift in ice regimes allows for increasing development opportunities for large oil and gas deposits known to occur throughout the Arctic. Here we show that hazardous ice features remain a threat to stationary and mobile infrastructure in the southern Beaufort Sea. With the opening up of the ice pack, forecasting of high-frequency oscillations or local eddy-driven ice motion will be a much more complex task than modeling average ice circulation. Given the observed reduction in sea ice extent and thickness this rather counterintuitive situation, associated with a warming climate, poses significant hazards to Arctic marine oil and gas development and marine transportation. Accurate forecasting of hazardous ice motion will require improved real-time surface wind and ocean current forecast models capable of ingesting local satellite-derived wind data and/or local, closely-spaced networks of anemometers and improved methods of determining high-frequency components of surface ocean current fields ‘up-stream’ from drilling and extraction operations.

Solar flare forecasting based on sequential sunspot data

It is widely believed that the evolution of solar active regions leads to solar flares. However, information about the evolution of solar active regions is not employed in most existing solar flare forecasting models. In the current work, a short-term solar flare forecasting model is proposed, in which sequential sunspot data, including three days of information about evolution from active regions, are taken as one of the basic predictors. The sunspot area, the McIntosh classification, the magnetic classification and the radio flux are extracted and converted to a numerical format that is suitable for the current forecasting model. Based on these parameters, the sliding-window method is used to form the sequential data by adding three days of information about evolution. Then, multi-layer perceptron and learning vector quantization are employed to predict the flare level within 48h. Experimental results indicate that the performance of the proposed flare forecasting model works better than previous models.

A Probabilistic Rip Current Forecast Model

Dusek, G. and Seim, H., 2013. A probabilistic rip current forecast modelA probabilistic rip current forecast model is created using a logistic regression formulation. Given input variables of nearshore significant wave height, vector mean wave direction, tidal elevation, and whether the forecast occurs in a 72-hour postwave event window, the probabilistic model predicts the likelihood of hazardous rip current occurrence. The model is trained using rip current observations made by lifeguards at Kill Devil Hills (KDH), North Carolina, in 2008–2009. Validation uses a hindcast at KDH over the summers of 2001–2007 and is compared with a hindcast of the National Weather Service (NWS) rip current model presently used at KDH. Using rip current rescues to indicate hazardous rip current occurrence, the probabilistic model has a Brier Score of 0.15 (0 is perfect prediction), compared with a minimum Brier Score of 0.45 for the NWS model, and represents a 67% or better improvement in prediction. Utilizing deepwater wave observations, instead of those collected nearshore, decreases model improvement by roughly half. The probabilistic model also outperforms the NWS model during high rescue instances, particularly on days with large tidal ranges. Functional improvements of the probabilistic model include the output of a true probabilistic forecast compared with a categorical forecast (low-medium-high) and the inclusion of only physically and statistically significant predictors.

On the environmental information for solar and wind energy facilities

Wind and solar energy are projected to be major sources of the world’s power in the coming decades. In this study, we first introduce satellite observations for surface solar irradiance and wind, and then discuss using the data for prediction and assessment of energy distribution. Using the current NCEP global forecast model (GFS), the global mean and standard deviation of solar power at the surface for the 5-d forecast are about 212 and 124 W/m2, respectively. For a 24-h forecast of a 3-h mean, the relative error is about 30%, and for a prediction of the daily mean, it is about 15%. The relative error of wind power forecasts for a 24-h forecast of a 3-h mean and a daily mean is 70% and 35%, respectively. The reanalysis results based on satellite observations and numerical weather prediction model are also used to study the distribution of solar and wind energy and the variation of the distribution related to climate change. No significant trend in downward solar radiation is found at the surface in the reanalysis data sets. Surface wind power however exhibits a significant positive trend as the global temperature is warming up. A comparison of a decadal mean wind energy between two decades (from 1949 to 1958 vs. from 1999 to 2008) shows that most of Asia had experienced a decrease in surface wind energy. Therefore, decisions about renewable energy developments need to consider such climate change scenarios.