Search

Ensemble forecasts from numerical weather prediction models show systematic errors that require correction via post-processing. While there has been substantial progress in flexible neural network-based post-processing methods over the past years, most station-based approaches still treat every input data point separately which limits the capabilities for leveraging spatial structures in the forecast errors. In order to improve information sharing across locations, we propose a graph neural network architecture for ensemble post-processing, which represents the station locations as nodes on a graph and utilizes an attention mechanism to identify relevant predictive information from neighboring locations. In a case study on 2-m temperature forecasts over Europe, the graph neural network model shows substantial improvements over a highly competitive neural network-based post-processing method., Comment: Accepted paper at ICML 2024 - Machine Learning for Earth System Modeling Workshop (https://leap-stc.github.io/ml4esm-workshop/)

Weather forecasts from numerical weather prediction models play a central role in solar energy forecasting, where a cascade of physics-based models is used in a model chain approach to convert forecasts of solar irradiance to solar power production, using additional weather variables as auxiliary information. Ensemble weather forecasts aim to quantify uncertainty in the future development of the weather, and can be used to propagate this uncertainty through the model chain to generate probabilistic solar energy predictions. However, ensemble prediction systems are known to exhibit systematic errors, and thus require post-processing to obtain accurate and reliable probabilistic forecasts. The overarching aim of our study is to systematically evaluate different strategies to apply post-processing methods in model chain approaches: Not applying any post-processing at all; post-processing only the irradiance predictions before the conversion; post-processing only the solar power predictions obtained from the model chain; or applying post-processing in both steps. In a case study based on a benchmark dataset for the Jacumba solar plant in the U.S., we develop statistical and machine learning methods for post-processing ensemble predictions of global horizontal irradiance and solar power generation. Further, we propose a neural network-based model for direct solar power forecasting that bypasses the model chain. Our results indicate that post-processing substantially improves the solar power generation forecasts, in particular when post-processing is applied to the power predictions. The machine learning methods for post-processing yield slightly better probabilistic forecasts, and the direct forecasting approach performs comparable to the post-processing strategies.

Reliable forecasts of quasi-stationary, recurrent, and persistent large-scale atmospheric circulation patterns (weather regimes) are crucial for various socio-economic sectors. Despite steady progress, probabilistic weather regime predictions still exhibit biases in the exact timing and amplitude of weather regimes. This study thus aims at advancing probabilistic weather regime predictions in the North Atlantic-European region through ensemble post-processing. Here, we focus on the representation of seven year-round weather regimes in the sub-seasonal to seasonal reforecasts of the European Centre for Medium-Range Weather Forecasts. The manifestation of each of the seven regimes can be expressed by a continuous weather regime index, representing the projection of the instantaneous 500-hPa geopotential height anomalies (Z500A) onto the respective mean regime pattern. We apply a two-step ensemble post-processing involving first univariate ensemble model output statistics and second ensemble copula coupling, which restores the multivariate dependency structure. Compared to current forecast calibration practices, which rely on correcting the Z500 field by the lead time dependent mean bias, our approach extends the forecast skill horizon for daily/instantaneous regime forecasts moderately by 1.2 days to 14.5 days. Additionally, to our knowledge our study is the first to systematically evaluate the multivariate aspects of forecast quality for weather regime forecasts. Our method outperforms current practices in the multivariate aspect, as measured by the energy and variogram score. Still our study shows, that even with advanced post-processing weather regime prediction becomes difficult beyond 14 days, which likely points towards intrinsic limits of predictability for daily/instantaneous regime forecasts. The proposed method can easily be applied to operational weather regime forecasts., Comment: submitted to Quarterly Journal of the Royal Meteorological Society (QJRMS)

Artificial intelligence (AI)-based data-driven weather forecasting models have experienced rapid progress over the last years. Recent studies, with models trained on reanalysis data, achieve impressive results and demonstrate substantial improvements over state-of-the-art physics-based numerical weather prediction models across a range of variables and evaluation metrics. Beyond improved predictions, the main advantages of data-driven weather models are their substantially lower computational costs and the faster generation of forecasts, once a model has been trained. However, most efforts in data-driven weather forecasting have been limited to deterministic, point-valued predictions, making it impossible to quantify forecast uncertainties, which is crucial in research and for optimal decision making in applications. Our overarching aim is to systematically study and compare uncertainty quantification methods to generate probabilistic weather forecasts from a state-of-the-art deterministic data-driven weather model, Pangu-Weather. Specifically, we compare approaches for quantifying forecast uncertainty based on generating ensemble forecasts via perturbations to the initial conditions, with the use of statistical and machine learning methods for post-hoc uncertainty quantification. In a case study on medium-range forecasts of selected weather variables over Europe, the probabilistic forecasts obtained by using the Pangu-Weather model in concert with uncertainty quantification methods show promising results and provide improvements over ensemble forecasts from the physics-based ensemble weather model of the European Centre for Medium-Range Weather Forecasts for lead times of up to 5 days.

Wind gust prediction plays an important role in warning strategies of national meteorological services due to the high impact of its extreme values. However, forecasting wind gusts is challenging because they are influenced by small-scale processes and local characteristics. To account for the different sources of uncertainty, meteorological centers run ensembles of forecasts and derive probabilities of wind gusts exceeding a threshold. These probabilities often exhibit systematic errors and require postprocessing. Model Output Statistics (MOS) is a common operational postprocessing technique, although more modern methods such as neural network-bases approaches have shown promising results in research studies. The transition from research to operations requires an exhaustive comparison of both techniques. Taking a first step into this direction, our study presents a comparison of a postprocessing technique based on linear and logistic regression approaches with different neural network methods proposed in the literature to improve wind gust predictions, specifically distributional regression networks and Bernstein quantile networks. We further contribute to investigating optimal design choices for neural network-based postprocessing methods regarding changes of the numerical model in the training period, the use of persistence predictors, and the temporal composition of training datasets. The performance of the different techniques is compared in terms of calibration, accuracy, reliability and resolution based on case studies of wind gust forecasts from the operational weather model of the German weather service and observations from 170 weather stations.

Statistical postprocessing is used to translate ensembles of raw numerical weather forecasts into reliable probabilistic forecast distributions. In this study, we examine the use of permutation-invariant neural networks for this task. In contrast to previous approaches, which often operate on ensemble summary statistics and dismiss details of the ensemble distribution, we propose networks that treat forecast ensembles as a set of unordered member forecasts and learn link functions that are by design invariant to permutations of the member ordering. We evaluate the quality of the obtained forecast distributions in terms of calibration and sharpness and compare the models against classical and neural network-based benchmark methods. In case studies addressing the postprocessing of surface temperature and wind gust forecasts, we demonstrate state-of-the-art prediction quality. To deepen the understanding of the learned inference process, we further propose a permutation-based importance analysis for ensemble-valued predictors, which highlights specific aspects of the ensemble forecast that are considered important by the trained postprocessing models. Our results suggest that most of the relevant information is contained in a few ensemble-internal degrees of freedom, which may impact the design of future ensemble forecasting and postprocessing systems., Comment: in press

Sub-seasonal weather forecasts are becoming increasingly important for a range of socio-economic activities. However, the predictive ability of physical weather models is very limited on these time scales. We propose several post-processing methods based on convolutional neural networks to improve sub-seasonal forecasts by correcting systematic errors of numerical weather prediction models. Our post-processing models operate directly on spatial input fields and are therefore able to retain spatial relationships and to generate spatially homogeneous predictions. They produce global probabilistic tercile forecasts for biweekly aggregates of temperature and precipitation for weeks 3-4 and 5-6. In a case study based on a public forecasting challenge organized by the World Meteorological Organization, our post-processing models outperform recalibrated forecasts from the European Centre for Medium-Range Weather Forecasts (ECMWF), and achieve improvements over climatological forecasts for all considered variables and lead times. We compare several model architectures and training modes and demonstrate that all approaches lead to skillful and well-calibrated probabilistic forecasts. The good calibration of the post-processed forecasts emphasizes that our post-processing models reliably quantify the forecast uncertainty based on deterministic input information in form of the ECMWF ensemble mean forecast fields only.

In many forecasting settings, there is a specific interest in predicting the sign of an outcome variable correctly in addition to its magnitude. For instance, when forecasting armed conflicts, positive and negative log-changes in monthly fatalities represent escalation and de-escalation, respectively, and have very different implications. In the ViEWS forecasting challenge, a prediction competition on state-based violence, a novel evaluation score called targeted absolute deviation with direction augmentation (TADDA) has therefore been suggested, which accounts for both for the sign and magnitude of log-changes. While it has a straightforward intuitive motivation, the empirical results of the challenge show that a no-change model always predicting a log-change of zero outperforms all submitted forecasting models under the TADDA score. We provide a statistical explanation for this phenomenon. Analyzing the properties of TADDA, we find that in order to achieve good scores, forecasters often have an incentive to predict no or only modest log-changes. In particular, there is often an incentive to report conservative point predictions considerably closer to zero than the forecaster's actual predictive median or mean. In an empirical application, we demonstrate that a no-change model can be improved upon by tailoring predictions to the particularities of the TADDA score. We conclude by outlining some alternative scoring concepts., Comment: 8 pages, 3 figures

We report on a course project in which students submit weekly probabilistic forecasts of two weather variables and one financial variable. This real-time format allows students to engage in practical forecasting, which requires a diverse set of skills in data science and applied statistics. We describe the context and aims of the course, and discuss design parameters like the selection of target variables, the forecast submission process, the evaluation of forecast performance, and the feedback provided to students. Furthermore, we describe empirical properties of students' probabilistic forecasts, as well as some lessons learned on our part.

Ensemble weather forecasts based on multiple runs of numerical weather prediction models typically show systematic errors and require post-processing to obtain reliable forecasts. Accurately modeling multivariate dependencies is crucial in many practical applications, and various approaches to multivariate post-processing have been proposed where ensemble predictions are first post-processed separately in each margin and multivariate dependencies are then restored via copulas. These two-step methods share common key limitations, in particular the difficulty to include additional predictors in modeling the dependencies. We propose a novel multivariate post-processing method based on generative machine learning to address these challenges. In this new class of nonparametric data-driven distributional regression models, samples from the multivariate forecast distribution are directly obtained as output of a generative neural network. The generative model is trained by optimizing a proper scoring rule which measures the discrepancy between the generated and observed data, conditional on exogenous input variables. Our method does not require parametric assumptions on univariate distributions or multivariate dependencies and allows for incorporating arbitrary predictors. In two case studies on multivariate temperature and wind speed forecasting at weather stations over Germany, our generative model shows significant improvements over state-of-the-art methods and particularly improves the representation of spatial dependencies.

An influential step in weather forecasting was the introduction of ensemble forecasts in operational use due to their capability to account for the uncertainties in the future state of the atmosphere. However, ensemble weather forecasts are often underdispersive and might also contain bias, which calls for some form of post-processing. A popular approach to calibration is the ensemble model output statistics (EMOS) approach resulting in a full predictive distribution for a given weather variable. However, this form of univariate post-processing may ignore the prevailing spatial and/or temporal correlation structures among different dimensions. Since many applications call for spatially and/or temporally coherent forecasts, multivariate post-processing aims to capture these possibly lost dependencies. We compare the forecast skill of different nonparametric multivariate approaches to modeling temporal dependence of ensemble weather forecasts with different forecast horizons. The focus is on two-step methods, where after univariate post-processing, the EMOS predictive distributions corresponding to different forecast horizons are combined to a multivariate calibrated prediction using an empirical copula. Based on global ensemble predictions of temperature, wind speed and precipitation accumulation of the European Centre for Medium-Range Weather Forecasts from January 2002 to March 2014, we investigate the forecast skill of different versions of Ensemble Copula Coupling (ECC) and Schaake Shuffle (SSh). In general, compared with the raw and independently calibrated forecasts, multivariate post-processing substantially improves the forecast skill. While even the simplest ECC approach with low computational cost provides a powerful benchmark method, recently proposed advanced extensions of the ECC and the SSh are found to not provide any significant improvements over their basic counterparts., Comment: 27 pages, 13 figures

Ensemble weather predictions typically show systematic errors that have to be corrected via post-processing. Even state-of-the-art post-processing methods based on neural networks often solely rely on location-specific predictors that require an interpolation of the physical weather model's spatial forecast fields to the target locations. However, potentially useful predictability information contained in large-scale spatial structures within the input fields is potentially lost in this interpolation step. Therefore, we propose the use of convolutional autoencoders to learn compact representations of spatial input fields which can then be used to augment location-specific information as additional inputs to post-processing models. The benefits of including this spatial information is demonstrated in a case study of 2-m temperature forecasts at surface stations in Germany., Comment: Accepted as conference paper at ICLR 2022 - AI for Earth and Space Science Workshop, https://ai4earthscience.github.io/iclr-2022-workshop/

The importance of accurately quantifying forecast uncertainty has motivated much recent research on probabilistic forecasting. In particular, a variety of deep learning approaches has been proposed, with forecast distributions obtained as output of neural networks. These neural network-based methods are often used in the form of an ensemble based on multiple model runs from different random initializations, resulting in a collection of forecast distributions that need to be aggregated into a final probabilistic prediction. With the aim of consolidating findings from the machine learning literature on ensemble methods and the statistical literature on forecast combination, we address the question of how to aggregate distribution forecasts based on such deep ensembles. Using theoretical arguments, simulation experiments and a case study on wind gust forecasting, we systematically compare probability- and quantile-based aggregation methods for three neural network-based approaches with different forecast distribution types as output. Our results show that combining forecast distributions can substantially improve the predictive performance. We propose a general quantile aggregation framework for deep ensembles that shows superior performance compared to a linear combination of the forecast densities. Finally, we investigate the effects of the ensemble size and derive recommendations of aggregating distribution forecasts from deep ensembles in practice.

Postprocessing ensemble weather predictions to correct systematic errors has become a standard practice in research and operations. However, only few recent studies have focused on ensemble postprocessing of wind gust forecasts, despite its importance for severe weather warnings. Here, we provide a comprehensive review and systematic comparison of eight statistical and machine learning methods for probabilistic wind gust forecasting via ensemble postprocessing, that can be divided in three groups: State of the art postprocessing techniques from statistics (ensemble model output statistics (EMOS), member-by-member postprocessing, isotonic distributional regression), established machine learning methods (gradient-boosting extended EMOS, quantile regression forests) and neural network-based approaches (distributional regression network, Bernstein quantile network, histogram estimation network). The methods are systematically compared using six years of data from a high-resolution, convection-permitting ensemble prediction system that was run operationally at the German weather service, and hourly observations at 175 surface weather stations in Germany. While all postprocessing methods yield calibrated forecasts and are able to correct the systematic errors of the raw ensemble predictions, incorporating information from additional meteorological predictor variables beyond wind gusts leads to significant improvements in forecast skill. In particular, we propose a flexible framework of locally adaptive neural networks with different probabilistic forecast types as output, which not only significantly outperform all benchmark postprocessing methods but also learn physically consistent relations associated with the diurnal cycle, especially the evening transition of the planetary boundary layer.

The amount, size, and complexity of astronomical data-sets and databases are growing rapidly in the last decades, due to new technologies and dedicated survey telescopes. Besides dealing with poly-structured and complex data, sparse data has become a field of growing scientific interest. A specific field of Astroinformatics research is the estimation of redshifts of extra-galactic sources by using sparse photometric observations. Many techniques have been developed to produce those estimates with increasing precision. In recent years, models have been favored which instead of providing a point estimate only, are able to generate probabilistic density functions (PDFs) in order to characterize and quantify the uncertainties of their estimates. Crucial to the development of those models is a proper, mathematically principled way to evaluate and characterize their performances, based on scoring functions as well as on tools for assessing calibration. Still, in literature inappropriate methods are being used to express the quality of the estimates that are often not sufficient and can potentially generate misleading interpretations. In this work we summarize how to correctly evaluate errors and forecast quality when dealing with PDFs. We describe the use of the log-likelihood, the continuous ranked probability score (CRPS) and the probability integral transform (PIT) to characterize the calibration as well as the sharpness of predicted PDFs. We present what we achieved when using proper scoring rules to train deep neural networks as well as to evaluate the model estimates and how this work led from well calibrated redshift estimates to improvements in probabilistic weather forecasting. The presented work is an example of interdisciplinarity in data-science and illustrates how methods can help to bridge gaps between different fields of application., Comment: Presented at ADASS XXX 2020

In order to enable the transition towards renewable energy sources, probabilistic energy forecasting is of critical importance for incorporating volatile power sources such as solar energy into the electrical grid. Solar energy forecasting methods often aim to provide probabilistic predictions of solar irradiance. In particular, many hybrid approaches combine physical information from numerical weather prediction models with statistical methods. Even though the physical models can provide useful information at intra-day and day-ahead forecast horizons, ensemble weather forecasts from multiple model runs are often not calibrated and show systematic biases. We propose a post-processing model for ensemble weather predictions of solar irradiance at temporal resolutions between 30 minutes and 6 hours. The proposed models provide probabilistic forecasts in the form of a censored logistic probability distribution for lead times up to 5 days and are evaluated in two case studies covering distinct physical models, geographical regions, temporal resolutions, and types of solar irradiance. We find that post-processing consistently and significantly improves the forecast performance of the ensemble predictions for lead times up to at least 48 hours and is well able to correct the systematic lack of calibration., Comment: 32 pages, 16 figures, 1 table

Capturing the uncertainty in probabilistic wind power forecasts is challenging, especially when uncertain input variables, such as the weather, play a role. Since ensemble weather predictions aim to capture the uncertainty in the weather system, they can be used to propagate this uncertainty through to subsequent wind power forecasting models. However, as weather ensemble systems are known to be biased and underdispersed, meteorologists post-process the ensembles. This post-processing can successfully correct the biases in the weather variables but has not been evaluated thoroughly in the context of subsequent forecasts, such as wind power generation forecasts. The present paper evaluates multiple strategies for applying ensemble post-processing to probabilistic wind power forecasts. We use Ensemble Model Output Statistics (EMOS) as the post-processing method and evaluate four possible strategies: only using the raw ensembles without post-processing, a one-step strategy where only the weather ensembles are post-processed, a one-step strategy where we only post-process the power ensembles, and a two-step strategy where we post-process both the weather and power ensembles. Results show that post-processing the final wind power ensemble improves forecast performance regarding both calibration and sharpness, whilst only post-processing the weather ensembles does not necessarily lead to increased forecast performance., Comment: Accepted by Wiley Wind Energy on April 19, 2022

Statistical postprocessing techniques are nowadays key components of the forecasting suites in many National Meteorological Services (NMS), with for most of them, the objective of correcting the impact of different types of errors on the forecasts. The final aim is to provide optimal, automated, seamless forecasts for end users. Many techniques are now flourishing in the statistical, meteorological, climatological, hydrological, and engineering communities. The methods range in complexity from simple bias corrections to very sophisticated distribution-adjusting techniques that incorporate correlations among the prognostic variables. The paper is an attempt to summarize the main activities going on this area from theoretical developments to operational applications, with a focus on the current challenges and potential avenues in the field. Among these challenges is the shift in NMS towards running ensemble Numerical Weather Prediction (NWP) systems at the kilometer scale that produce very large datasets and require high-density high-quality observations; the necessity to preserve space time correlation of high-dimensional corrected fields; the need to reduce the impact of model changes affecting the parameters of the corrections; the necessity for techniques to merge different types of forecasts and ensembles with different behaviors; and finally the ability to transfer research on statistical postprocessing to operations. Potential new avenues will also be discussed., Comment: This work has been submitted to the Bulletin of the American Meteorological Society. Copyright in this work may be transferred without further notice

Accurate and reliable forecasting of total cloud cover (TCC) is vital for many areas such as astronomy, energy demand and production, or agriculture. Most meteorological centres issue ensemble forecasts of TCC, however, these forecasts are often uncalibrated and exhibit worse forecast skill than ensemble forecasts of other weather variables. Hence, some form of post-processing is strongly required to improve predictive performance. As TCC observations are usually reported on a discrete scale taking just nine different values called oktas, statistical calibration of TCC ensemble forecasts can be considered a classification problem with outputs given by the probabilities of the oktas. This is a classical area where machine learning methods are applied. We investigate the performance of post-processing using multilayer perceptron (MLP) neural networks, gradient boosting machines (GBM) and random forest (RF) methods. Based on the European Centre for Medium-Range Weather Forecasts global TCC ensemble forecasts for 2002-2014 we compare these approaches with the proportional odds logistic regression (POLR) and multiclass logistic regression (MLR) models, as well as the raw TCC ensemble forecasts. We further assess whether improvements in forecast skill can be obtained by incorporating ensemble forecasts of precipitation as additional predictor. Compared to the raw ensemble, all calibration methods result in a significant improvement in forecast skill. RF models provide the smallest increase in predictive performance, while MLP, POLR and GBM approaches perform best. The use of precipitation forecast data leads to further improvements in forecast skill and except for very short lead times the extended MLP model shows the best overall performance., Comment: 24 pages, 7 figures

Ice nucleation from vapor presents a variety of challenges across a wide range of industries and applications including refrigeration, transportation, and energy generation. However, a rational comprehensive approach to fabricating intrinsically icephobic surfaces for frost formation-both from water condensation (followed by freezing) and in particular from desublimation (direct growth of ice crystals from vapor)-remains elusive. Here, guided by nucleation physics, we investigate the effect of material composition and surface texturing (atomically smooth to nanorough) on the nucleation and growth mechanism of frost for a range of conditions within the sublimation domain (0 C to -55C; partial water vapor pressures 6 to 0.02 mbar). Surprisingly, we observe that on silicon at very cold temperatures-below the homogeneous ice solidification nucleation limit (< -46 C)-desublimation does not become the favorable pathway to frosting. Furthermore we show that surface nanoroughness makes frost formation on silicon more probable. We experimentally demonstrate at temperatures between -48 C and -55 C that nanotexture with radii of curvature within one order of magnitude of the critical radius of nucleation favors frost growth, facilitated by capillary condensation, consistent with Kelvin's Equation. Our findings show that such nanoscale surface morphology imposed by design to impart desired functionalities-such as superhydrophobicity-or from defects, can be highly detrimental for frost icephobicity at low temperatures and water vapor partial pressures (< 0.05 mbar). Our work contributes to the fundamental understanding of heterogeneous phase transitions well within the equilibrium sublimation domain and has implications for applications such as travel, power generation, and refrigeration.

Ensemble weather predictions require statistical post-processing of systematic errors to obtain reliable and accurate probabilistic forecasts. Traditionally, this is accomplished with distributional regression models in which the parameters of a predictive distribution are estimated from a training period. We propose a flexible alternative based on neural networks that can incorporate nonlinear relationships between arbitrary predictor variables and forecast distribution parameters that are automatically learned in a data-driven way rather than requiring pre-specified link functions. In a case study of 2-meter temperature forecasts at surface stations in Germany, the neural network approach significantly outperforms benchmark post-processing methods while being computationally more affordable. Key components to this improvement are the use of auxiliary predictor variables and station-specific information with the help of embeddings. Furthermore, the trained neural network can be used to gain insight into the importance of meteorological variables thereby challenging the notion of neural networks as uninterpretable black boxes. Our approach can easily be extended to other statistical post-processing and forecasting problems. We anticipate that recent advances in deep learning combined with the ever-increasing amounts of model and observation data will transform the post-processing of numerical weather forecasts in the coming decade.

Probabilistic forecasts in the form of probability distributions over future events have become popular in several fields including meteorology, hydrology, economics, and demography. In typical applications, many alternative statistical models and data sources can be used to produce probabilistic forecasts. Hence, evaluating and selecting among competing methods is an important task. The scoringRules package for R provides functionality for comparative evaluation of probabilistic models based on proper scoring rules, covering a wide range of situations in applied work. This paper discusses implementation and usage details, presents case studies from meteorology and economics, and points to the relevant background literature.

In Bayesian inference, predictive distributions are typically in the form of samples generated via Markov chain Monte Carlo (MCMC) or related algorithms. In this paper, we conduct a systematic analysis of how to make and evaluate probabilistic forecasts from such simulation output. Based on proper scoring rules, we develop a notion of consistency that allows to assess the adequacy of methods for estimating the stationary distribution underlying the simulation output. We then provide asymptotic results that account for the salient features of Bayesian posterior simulators, and derive conditions under which choices from the literature satisfy our notion of consistency. Importantly, these conditions depend on the scoring rule being used, such that the choices of approximation method and scoring rule are intertwined. While the logarithmic rule requires fairly stringent conditions, the continuous ranked probability score (CRPS) yields consistent approximations under minimal assumptions. These results are illustrated in a simulation study and an economic data example. Overall, mixture-of-parameters approximations which exploit the parametric structure of Bayesian models perform particularly well. Under the CRPS, the empirical distribution function is a simple and appealing alternative option.

Statistical post-processing techniques are now widely used to correct systematic biases and errors in calibration of ensemble forecasts obtained from multiple runs of numerical weather prediction models. A standard approach is the ensemble model output statistics (EMOS) method, a distributional regression approach where the forecast distribution is given by a single parametric law with parameters depending on the ensemble members. Choosing an appropriate parametric family for the weather variable of interest is a critical, however, often non-trivial task, and has been the focus of much recent research. In this article, we assess the merits of combining predictive distributions from multiple EMOS models based on different parametric families. In four case studies with wind speed and precipitation forecasts from two ensemble prediction systems, we study whether state of the art forecast combination methods are able to improve forecast skill., Comment: 33 pages, 8 figures, 2 tables

In public discussions of the quality of forecasts, attention typically focuses on the predictive performance in cases of extreme events. However, the restriction of conventional forecast evaluation methods to subsets of extreme observations has unexpected and undesired effects, and is bound to discredit skillful forecasts when the signal-to-noise ratio in the data generating process is low. Conditioning on outcomes is incompatible with the theoretical assumptions of established forecast evaluation methods, thereby confronting forecasters with what we refer to as the forecaster's dilemma. For probabilistic forecasts, proper weighted scoring rules have been proposed as decision theoretically justifiable alternatives for forecast evaluation with an emphasis on extreme events. Using theoretical arguments, simulation experiments, and a real data study on probabilistic forecasts of U.S. inflation and gross domestic product growth, we illustrate and discuss the forecaster's dilemma along with potential remedies.

Weather forecasts are typically given in the form of forecast ensembles obtained from multiple runs of numerical weather prediction models with varying initial conditions and physics parameterizations. Such ensemble predictions tend to be biased and underdispersive and thus require statistical postprocessing. In the ensemble model output statistics (EMOS) approach, a probabilistic forecast is given by a single parametric distribution with parameters depending on the ensemble members. This article proposes two semi-local methods for estimating the EMOS coefficients where the training data for a specific observation station are augmented with corresponding forecast cases from stations with similar characteristics. Similarities between stations are determined using either distance functions or clustering based on various features of the climatology, forecast errors, ensemble predictions and locations of the observation stations. In a case study on wind speed over Europe with forecasts from the Grand Limited Area Model Ensemble Prediction System, the proposed similarity-based semi-local models show significant improvement in predictive performance compared to standard regional and local estimation methods. They further allow for estimating complex models without numerical stability issues and are computationally more efficient than local parameter estimation.

Ensemble model output statistics (EMOS) is a statistical tool for post-processing forecast ensembles of weather variables obtained from multiple runs of numerical weather prediction models in order to produce calibrated predictive probability density functions (PDFs). The EMOS predictive PDF is given by a parametric distribution with parameters depending on the ensemble forecasts. We propose an EMOS model for calibrating wind speed forecasts based on weighted mixtures of truncated normal (TN) and log-normal (LN) distributions where model parameters and component weights are estimated by optimizing the values of proper scoring rules over a rolling training period. The new model is tested on wind speed forecasts of the 50 member European Centre for Medium-Range Weather Forecasts ensemble, the 11 member Aire Limit\'ee Adaptation dynamique D\'eveloppement International-Hungary Ensemble Prediction System ensemble of the Hungarian Meteorological Service and the eight-member University of Washington mesoscale ensemble, and its predictive performance is compared to that of various benchmark EMOS models based on single parametric families and combinations thereof. The results indicate improved calibration of probabilistic and accuracy of point forecasts in comparison with the raw ensemble and climatological forecasts. The mixture EMOS model significantly outperforms the TN and LN EMOS methods, moreover, it provides better calibrated forecasts than the TN-LN combination model and offers an increased flexibility while avoiding covariate selection problems., Comment: 24 pages, 7 tables, 10 figures

Ensembles of forecasts are obtained from multiple runs of numerical weather forecasting models with different initial conditions and typically employed to account for forecast uncertainties. However, biases and dispersion errors often occur in forecast ensembles, they are usually under-dispersive and uncalibrated and require statistical post-processing. We present an Ensemble Model Output Statistics (EMOS) method for calibration of wind speed forecasts based on the log-normal (LN) distribution, and we also show a regime-switching extension of the model which combines the previously studied truncated normal (TN) distribution with the LN. Both presented models are applied to wind speed forecasts of the eight-member University of Washington mesoscale ensemble, of the fifty-member ECMWF ensemble and of the eleven-member ALADIN-HUNEPS ensemble of the Hungarian Meteorological Service, and their predictive performances are compared to those of the TN and general extreme value (GEV) distribution based EMOS methods and to the TN-GEV mixture model. The results indicate improved calibration of probabilistic and accuracy of point forecasts in comparison to the raw ensemble and to climatological forecasts. Further, the TN-LN mixture model outperforms the traditional TN method and its predictive performance is able to keep up with the models utilizing the GEV distribution without assigning mass to negative values., Comment: 24 pages, 10 figures

In weather forecasting, nonhomogeneous regression is used to statistically postprocess forecast ensembles in order to obtain calibrated predictive distributions. For wind speed forecasts, the regression model is given by a truncated normal distribution where location and spread are derived from the ensemble. This paper proposes two alternative approaches which utilize the generalized extreme value (GEV) distribution. A direct alternative to the truncated normal regression is to apply a predictive distribution from the GEV family, while a regime switching approach based on the median of the forecast ensemble incorporates both distributions. In a case study on daily maximum wind speed over Germany with the forecast ensemble from the European Centre for Medium-Range Weather Forecasts, all three approaches provide calibrated and sharp predictive distributions with the regime switching approach showing the highest skill in the upper tail.