Regularized logistic models for probabilistic forecasting and diagnostics

Logistic models are studied as a tool to convert dynamical forecast information (deterministic and ensemble) into probability forecasts. A logistic model is obtained by setting the logarithmic odds ratio equal to a linear combination of the inputs. As with any statistical model, logistic models will suffer from overfitting if the number of inputs is comparable to the number of forecast instances. Computational approaches to avoid overfitting by regularization are discussed, and efficient techniques for model assessment and selection are presented. A logit version of the lasso (originally a linear regression technique), is discussed. In lasso models, less important inputs are identified and the corresponding coefficient is set to zero, providing an efficient and automatic model reduction procedure. For the same reason, lasso models are particularly appealing for diagnostic purposes.

Probabilistic forecasting: why model imperfection is a poison pill

This volume is a serious attempt to open up the subject of European philosophy of science to real thought, and provide the structural basis for the interdisciplinary development of its specialist fields, but also to provoke reflection on the idea of ‘European philosophy of science’. This efforts should foster a contemporaneous reflection on what might be meant by philosophy of science in Europe and European philosophy of science, and how in fact awareness of it could assist philosophers interpret and motivate their research through a stronger collective identity. The overarching aim is to set the background for a collaborative project organising, systematising, and ultimately forging an identity for, European philosophy of science by creating research structures and developing research networks across Europe to promote its development.

Uncertainty in wind energy forecasting

Wind energy is the energy source that contributes most to the renewable energy mix of European countries. While there are good wind resources throughout Europe, the intermittency of the wind represents a major problem for the deployment of wind energy into the electricity networks. To ensure grid security a Transmission System Operator needs today for each kilowatt of wind energy either an equal amount of spinning reserve or a forecasting system that can predict the amount of energy that will be produced from wind over a period of 1 to 48 hours. In the range from 5m/s to 15m/s a wind turbine’s production increases with a power of three. For this reason, a Transmission System Operator requires an accuracy for wind speed forecasts of 1m/s in this wind speed range. Forecasting wind energy with a numerical weather prediction model in this context builds the background of this work. The author’s goal was to present a pragmatic solution to this specific problem in the ”real world”. This work therefore has to be seen in a technical context and hence does not provide nor intends to provide a general overview of the benefits and drawbacks of wind energy as a renewable energy source. In the first part of this work the accuracy requirements of the energy sector for wind speed predictions from numerical weather prediction models are described and analysed. A unique set of numerical experiments has been carried out in collaboration with the Danish Meteorological Institute to investigate the forecast quality of an operational numerical weather prediction model for this purpose. The results of this investigation revealed that the accuracy requirements for wind speed and wind power forecasts from today’s numerical weather prediction models can only be met at certain times. This means that the uncertainty of the forecast quality becomes a parameter that is as important as the wind speed and wind power itself. To quantify the uncertainty of a forecast valid for tomorrow requires an ensemble of forecasts. In the second part of this work such an ensemble of forecasts was designed and verified for its ability to quantify the forecast error. This was accomplished by correlating the measured error and the forecasted uncertainty on area integrated wind speed and wind power in Denmark and Ireland. A correlation of 93% was achieved in these areas. This method cannot solve the accuracy requirements of the energy sector. By knowing the uncertainty of the forecasts, the focus can however be put on the accuracy requirements at times when it is possible to accurately predict the weather. Thus, this result presents a major step forward in making wind energy a compatible energy source in the future.

Current methods and advances in forecasting of wind power generation

Wind power generation differs from conventional thermal generation due to the stochastic nature of wind. Thus wind power forecasting plays a key role in dealing with the challenges of balancing supply and demand in any electricity system, given the uncertainty associated with the wind farm power output. Accurate wind power forecasting reduces the need for additional balancing energy and reserve power to integrate wind power. Wind power forecasting tools enable better dispatch, scheduling and unit commitment of thermal generators, hydro plant and energy storage plant and more competitive market trading as wind power ramps up and down on the grid. This paper presents an in-depth review of the current methods and advances in wind power forecasting and prediction. Firstly, numerical wind prediction methods from global to local scales, ensemble forecasting, upscaling and downscaling processes are discussed. Next the statistical and machine learning approach methods are detailed. Then the techniques used for benchmarking and uncertainty analysis of forecasts are overviewed, and the performance of various approaches over different forecast time horizons is examined. Finally, current research activities, challenges and potential future developments are appraised.

Probabilistic simulations of crop yield over western India using the DEMETER seasonal hindcast ensembles

Process-based integrated modelling of weather and crop yield over large areas is becoming an important research topic. The production of the DEMETER ensemble hindcasts of weather allows this work to be carried out in a probabilistic framework. In this study, ensembles of crop yield (groundnut, Arachis hypogaea L.) were produced for 10 2.5 degrees x 2.5 degrees grid cells in western India using the DEMETER ensembles and the general large-area model (GLAM) for annual crops. Four key issues are addressed by this study. First, crop model calibration methods for use with weather ensemble data are assessed. Calibration using yield ensembles was more successful than calibration using reanalysis data (the European Centre for Medium-Range Weather Forecasts 40-yr reanalysis, ERA40). Secondly, the potential for probabilistic forecasting of crop failure is examined. The hindcasts show skill in the prediction of crop failure, with more severe failures being more predictable. Thirdly, the use of yield ensemble means to predict interannual variability in crop yield is examined and their skill assessed relative to baseline simulations using ERA40. The accuracy of multi-model yield ensemble means is equal to or greater than the accuracy using ERA40. Fourthly, the impact of two key uncertainties, sowing window and spatial scale, is briefly examined. The impact of uncertainty in the sowing window is greater with ERA40 than with the multi-model yield ensemble mean. Subgrid heterogeneity affects model accuracy: where correlations are low on the grid scale, they may be significantly positive on the subgrid scale. The implications of the results of this study for yield forecasting on seasonal time-scales are as follows. (i) There is the potential for probabilistic forecasting of crop failure (defined by a threshold yield value); forecasting of yield terciles shows less potential. (ii) Any improvement in the skill of climate models has the potential to translate into improved deterministic yield prediction. (iii) Whilst model input uncertainties are important, uncertainty in the sowing window may not require specific modelling. The implications of the results of this study for yield forecasting on multidecadal (climate change) time-scales are as follows. (i) The skill in the ensemble mean suggests that the perturbation, within uncertainty bounds, of crop and climate parameters, could potentially average out some of the errors associated with mean yield prediction. (ii) For a given technology trend, decadal fluctuations in the yield-gap parameter used by GLAM may be relatively small, implying some predictability on those time-scales.

Probabilistic evaluation of time series models : a comparison of several approaches

Several methods are examined which allow to produce forecasts for time series in the form of probability assignments. The necessary concepts are presented, addressing questions such as how to assess the performance of a probabilistic forecast. A particular class of models, cluster weighted models (CWMs), is given particular attention. CWMs, originally proposed for deterministic forecasts, can be employed for probabilistic forecasting with little modification. Two examples are presented. The first involves estimating the state of (numerically simulated) dynamical systems from noise corrupted measurements, a problem also known as filtering. There is an optimal solution to this problem, called the optimal filter, to which the considered time series models are compared. (The optimal filter requires the dynamical equations to be known.) In the second example, we aim at forecasting the chaotic oscillations of an experimental bronze spring system. Both examples demonstrate that the considered time series models, and especially the CWMs, provide useful probabilistic information about the underlying dynamical relations. In particular, they provide more than just an approximation to the conditional mean.

Contrasting probabilistic scoring rules

There are several scoring rules that one can choose from in order to score probabilistic forecasting models or estimate model parameters. Whilst it is generally agreed that proper scoring rules are preferable, there is no clear criterion for preferring one proper scoring rule above another. This manuscript compares and contrasts some commonly used proper scoring rules and provides guidance on scoring rule selection. In particular, it is shown that the logarithmic scoring rule prefers erring with more uncertainty, the spherical scoring rule prefers erring with lower uncertainty, whereas the other scoring rules are indifferent to either option.

Visualizing probabilistic flood forecast information: expert preferences and perceptions of best practice in uncertainty communication

The aim of this article is to improve the communication of the probabilistic flood forecasts generated by hydrological ensemble prediction systems (HEPS) by understanding perceptions of different methods of visualizing probabilistic forecast information. This study focuses on interexpert communication and accounts for differences in visualization requirements based on the information content necessary for individual users. The perceptions of the expert group addressed in this study are important because they are the designers and primary users of existing HEPS. Nevertheless, they have sometimes resisted the release of uncertainty information to the general public because of doubts about whether it can be successfully communicated in ways that would be readily understood to nonexperts. In this article, we explore the strengths and weaknesses of existing HEPS visualization methods and thereby formulate some wider recommendations about the best practice for HEPS visualization and communication. We suggest that specific training on probabilistic forecasting would foster use of probabilistic forecasts with a wider range of applications. The result of a case study exercise showed that there is no overarching agreement between experts on how to display probabilistic forecasts and what they consider the essential information that should accompany plots and diagrams. In this article, we propose a list of minimum properties that, if consistently displayed with probabilistic forecasts, would make the products more easily understandable. Copyright © 2012 John Wiley & Sons, Ltd.

Continental and global scale flood forecasting systems

Floods are the most frequent of natural disasters, affecting millions of people across the globe every year. The anticipation and forecasting of floods at the global scale is crucial to preparing for severe events and providing early awareness where local flood models and warning services may not exist. As numerical weather prediction models continue to improve, operational centres are increasingly using the meteorological output from these to drive hydrological models, creating hydrometeorological systems capable of forecasting river flow and flood events at much longer lead times than has previously been possible. Furthermore, developments in, for example, modelling capabilities, data and resources in recent years have made it possible to produce global scale flood forecasting systems. In this paper, the current state of operational large scale flood forecasting is discussed, including probabilistic forecasting of floods using ensemble prediction systems. Six state-of-the-art operational large scale flood forecasting systems are reviewed, describing similarities and differences in their approaches to forecasting floods at the global and continental scale. Currently, operational systems have the capability to produce coarse-scale discharge forecasts in the medium-range and disseminate forecasts and, in some cases, early warning products, in real time across the globe, in support of national forecasting capabilities. With improvements in seasonal weather forecasting, future advances may include more seamless hydrological forecasting at the global scale, alongside a move towards multi-model forecasts and grand ensemble techniques, responding to the requirement of developing multi-hazard early warning systems for disaster risk reduction.

Analysis of eruptive and seismic sequences to improve the short-and long-term eruption forecasting

Forecasting the time, location, nature, and scale of volcanic eruptions is one of the most urgent aspects of modern applied volcanology. The reliability of probabilistic forecasting procedures is strongly related to the reliability of the input information provided, implying objective criteria for interpreting the historical and monitoring data. For this reason both, detailed analysis of past data and more basic research into the processes of volcanism, are fundamental tasks of a continuous information-gain process; in this way the precursor events of eruptions can be better interpreted in terms of their physical meanings with correlated uncertainties. This should lead to better predictions of the nature of eruptive events. In this work we have studied different problems associated with the long- and short-term eruption forecasting assessment. First, we discuss different approaches for the analysis of the eruptive history of a volcano, most of them generally applied for long-term eruption forecasting purposes; furthermore, we present a model based on the characteristics of a Brownian passage-time process to describe recurrent eruptive activity, and apply it for long-term, time-dependent, eruption forecasting (Chapter 1). Conversely, in an effort to define further monitoring parameters as input data for short-term eruption forecasting in probabilistic models (as for example, the Bayesian Event Tree for eruption forecasting -BET_EF-), we analyze some characteristics of typical seismic activity recorded in active volcanoes; in particular, we use some methodologies that may be applied to analyze long-period (LP) events (Chapter 2) and volcano-tectonic (VT) seismic swarms (Chapter 3); our analysis in general are oriented toward the tracking of phenomena that can provide information about magmatic processes. Finally, we discuss some possible ways to integrate the results presented in Chapters 1 (for long-term EF), 2 and 3 (for short-term EF) in the BET_EF model (Chapter 4).

Reliability, sufficiency, and the decomposition of proper scores

Scoring rules are an important tool for evaluating the performance of probabilistic forecasting schemes. A scoring rule is called strictly proper if its expectation is optimal if and only if the forecast probability represents the true distribution of the target. In the binary case, strictly proper scoring rules allow for a decomposition into terms related to the resolution and the reliability of a forecast. This fact is particularly well known for the Brier Score. In this article, this result is extended to forecasts for finite-valued targets. Both resolution and reliability are shown to have a positive effect on the score. It is demonstrated that resolution and reliability are directly related to forecast attributes that are desirable on grounds independent of the notion of scores. This finding can be considered an epistemological justification of measuring forecast quality by proper scoring rules. A link is provided to the original work of DeGroot and Fienberg, extending their concepts of sufficiency and refinement. The relation to the conjectured sharpness principle of Gneiting, et al., is elucidated.

Data assimilation for precipitation nowcasting using Bayesian inference

This work introduces a new variational Bayes data assimilation method for the stochastic estimation of precipitation dynamics using radar observations for short term probabilistic forecasting (nowcasting). A previously developed spatial rainfall model based on the decomposition of the observed precipitation field using a basis function expansion captures the precipitation intensity from radar images as a set of ‘rain cells’. The prior distributions for the basis function parameters are carefully chosen to have a conjugate structure for the precipitation field model to allow a novel variational Bayes method to be applied to estimate the posterior distributions in closed form, based on solving an optimisation problem, in a spirit similar to 3D VAR analysis, but seeking approximations to the posterior distribution rather than simply the most probable state. A hierarchical Kalman filter is used to estimate the advection field based on the assimilated precipitation fields at two times. The model is applied to tracking precipitation dynamics in a realistic setting, using UK Met Office radar data from both a summer convective event and a winter frontal event. The performance of the model is assessed both traditionally and using probabilistic measures of fit based on ROC curves. The model is shown to provide very good assimilation characteristics, and promising forecast skill. Improvements to the forecasting scheme are discussed

Probabilistic Models for Forecasting Earthquakes in the Northeast Region of India

Northeast India and its adjoining areas are characterized by very high seismic activity. According to the Indian seismic code, the region falls under seismic zone V, which represents the highest seismic-hazard level in the country. This region has experienced a number of great earthquakes, such as the Assam (1950) and Shillong (1897) earthquakes, that caused huge devastation in the entire northeast and adjacent areas by flooding, landslides, liquefaction, and damage to roads and buildings. In this study, an attempt has been made to find the probability of occurrence of a major earthquake (M-w > 6) in this region using an updated earthquake catalog collected from different sources. Thereafter, dividing the catalog into six different seismic regions based on different tectonic features and seismogenic factors, the probability of occurrences was estimated using three models: the lognormal, Weibull, and gamma distributions. We calculated the logarithmic probability of the likelihood function (ln L) for all six regions and the entire northeast for all three stochastic models. A higher value of ln L suggests a better model, and a lower value shows a worse model. The results show different model suits for different seismic zones, but the majority follows lognormal, which is better for forecasting magnitude size. According to the results, Weibull shows the highest conditional probabilities among the three models for small as well as large elapsed time T and time intervals t, whereas the lognormal model shows the lowest and the gamma model shows intermediate probabilities. Only for elapsed time T = 0, the lognormal model shows the highest conditional probabilities among the three models at a smaller time interval (t = 3-15 yrs). The opposite result is observed at larger time intervals (t = 15-25 yrs), which show the highest probabilities for the Weibull model. However, based on this study, the IndoBurma Range and Eastern Himalaya show a high probability of occurrence in the 5 yr period 2012-2017 with >90% probability.

Hybrid probabilistic wind power forecasting using temporally local Gaussian process

The demand for sustainable development has resulted in a rapid growth in wind power worldwide. Despite various approaches have been proposed to improve the accuracy and to overcome the uncertainties associated with traditional methods, the stochastic and variable nature of wind still remains the most challenging issue in accurately forecasting wind power. This paper presents a hybrid deterministic-probabilistic method where a temporally local ‘moving window’ technique is used in Gaussian Process to examine estimated forecasting errors. This temporally local Gaussian Process employs less measurement data while faster and better predicts wind power at two wind farms, one in the USA and the other in Ireland. Statistical analysis on the results shows that the method can substantially reduce the forecasting error while more likely generate Gaussian-distributed residuals, particularly for short-term forecast horizons due to its capability to handle the time-varying characteristics of wind power.

Global forecasting of thermal health hazards: the skill of probabilistic predictions of the Universal Thermal Climate Index (UTCI)

Although over a hundred thermal indices can be used for assessing thermal health hazards, many ignore the human heat budget, physiology and clothing. The Universal Thermal Climate Index (UTCI) addresses these shortcomings by using an advanced thermo-physiological model. This paper assesses the potential of using the UTCI for forecasting thermal health hazards. Traditionally, such hazard forecasting has had two further limitations: it has been narrowly focused on a particular region or nation and has relied on the use of single ‘deterministic’ forecasts. Here, the UTCI is computed on a global scale,which is essential for international health-hazard warnings and disaster preparedness, and it is provided as a probabilistic forecast. It is shown that probabilistic UTCI forecasts are superior in skill to deterministic forecasts and that despite global variations, the UTCI forecast is skilful for lead times up to 10 days. The paper also demonstrates the utility of probabilistic UTCI forecasts on the example of the 2010 heat wave in Russia.