Search Results
You are looking at 1 - 7 of 7 items for
- Author or Editor: Sebastian Lerch x
- Refine by Access: All Content x
Abstract
Ensemble weather predictions require statistical postprocessing 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 prespecified link functions. In a case study of 2-m temperature forecasts at surface stations in Germany, the neural network approach significantly outperforms benchmark postprocessing 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 postprocessing 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 postprocessing of numerical weather forecasts in the coming decade.
Abstract
Ensemble weather predictions require statistical postprocessing 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 prespecified link functions. In a case study of 2-m temperature forecasts at surface stations in Germany, the neural network approach significantly outperforms benchmark postprocessing 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 postprocessing 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 postprocessing of numerical weather forecasts in the coming decade.
Abstract
Postprocessing ensemble weather predictions to correct systematic errors has become a standard practice in research and operations. However, only a 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 6 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.
Abstract
Postprocessing ensemble weather predictions to correct systematic errors has become a standard practice in research and operations. However, only a 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 6 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.
Abstract
Precipitation is affected by soil moisture spatial variability. However, this variability is not well represented in atmospheric models that do not consider soil moisture transport as a three-dimensional process. This study investigates the sensitivity of precipitation to the uncertainty in the representation of terrestrial water flow. The tools used for this investigation are the Weather Research and Forecasting (WRF) Model and its hydrologically enhanced version, WRF-Hydro, applied over central Europe during April–October 2008. The model grid is convection permitting, with a horizontal spacing of 2.8 km. The WRF-Hydro subgrid employs a 280-m resolution to resolve lateral terrestrial water flow. A WRF/WRF-Hydro ensemble is constructed by modifying the parameter controlling the partitioning between surface runoff and infiltration and by varying the planetary boundary layer (PBL) scheme. This ensemble represents terrestrial water flow uncertainty originating from the consideration of resolved lateral flow, terrestrial water flow uncertainty in the vertical direction, and turbulence parameterization uncertainty. The uncertainty of terrestrial water flow noticeably increases the normalized ensemble spread of daily precipitation where topography is moderate, surface flux spatial variability is high, and the weather regime is dominated by local processes. The adjusted continuous ranked probability score shows that the PBL uncertainty improves the skill of an ensemble subset in reproducing daily precipitation from the E-OBS observational product by 16%–20%. In comparison to WRF, WRF-Hydro improves this skill by 0.4%–0.7%. The reproduction of observed daily discharge with Nash–Sutcliffe model efficiency coefficients generally above 0.3 demonstrates the potential of WRF-Hydro in hydrological science.
Abstract
Precipitation is affected by soil moisture spatial variability. However, this variability is not well represented in atmospheric models that do not consider soil moisture transport as a three-dimensional process. This study investigates the sensitivity of precipitation to the uncertainty in the representation of terrestrial water flow. The tools used for this investigation are the Weather Research and Forecasting (WRF) Model and its hydrologically enhanced version, WRF-Hydro, applied over central Europe during April–October 2008. The model grid is convection permitting, with a horizontal spacing of 2.8 km. The WRF-Hydro subgrid employs a 280-m resolution to resolve lateral terrestrial water flow. A WRF/WRF-Hydro ensemble is constructed by modifying the parameter controlling the partitioning between surface runoff and infiltration and by varying the planetary boundary layer (PBL) scheme. This ensemble represents terrestrial water flow uncertainty originating from the consideration of resolved lateral flow, terrestrial water flow uncertainty in the vertical direction, and turbulence parameterization uncertainty. The uncertainty of terrestrial water flow noticeably increases the normalized ensemble spread of daily precipitation where topography is moderate, surface flux spatial variability is high, and the weather regime is dominated by local processes. The adjusted continuous ranked probability score shows that the PBL uncertainty improves the skill of an ensemble subset in reproducing daily precipitation from the E-OBS observational product by 16%–20%. In comparison to WRF, WRF-Hydro improves this skill by 0.4%–0.7%. The reproduction of observed daily discharge with Nash–Sutcliffe model efficiency coefficients generally above 0.3 demonstrates the potential of WRF-Hydro in hydrological science.
Abstract
Deep-learning (DL) postprocessing methods are examined to obtain reliable and accurate probabilistic forecasts from single-member numerical weather predictions of integrated vapor transport (IVT). Using a 34-yr reforecast, based on the Center for Western Weather and Water Extremes West-WRF mesoscale model of North American West Coast IVT, the dynamically/statistically derived 0–120-h probabilistic forecasts for IVT under atmospheric river (AR) conditions are tested. These predictions are compared with the Global Ensemble Forecast System (GEFS) dynamic model and the GEFS calibrated with a neural network. In addition, the DL methods are tested against an established, but more rigid, statistical–dynamical ensemble method (the analog ensemble). The findings show, using continuous ranked probability skill score and Brier skill score as verification metrics, that the DL methods compete with or outperform the calibrated GEFS system at lead times from 0 to 48 h and again from 72 to 120 h for AR vapor transport events. In addition, the DL methods generate reliable and skillful probabilistic forecasts. The implications of varying the length of the training dataset are examined, and the results show that the DL methods learn relatively quickly and ∼10 years of hindcast data are required to compete with the GEFS ensemble.
Abstract
Deep-learning (DL) postprocessing methods are examined to obtain reliable and accurate probabilistic forecasts from single-member numerical weather predictions of integrated vapor transport (IVT). Using a 34-yr reforecast, based on the Center for Western Weather and Water Extremes West-WRF mesoscale model of North American West Coast IVT, the dynamically/statistically derived 0–120-h probabilistic forecasts for IVT under atmospheric river (AR) conditions are tested. These predictions are compared with the Global Ensemble Forecast System (GEFS) dynamic model and the GEFS calibrated with a neural network. In addition, the DL methods are tested against an established, but more rigid, statistical–dynamical ensemble method (the analog ensemble). The findings show, using continuous ranked probability skill score and Brier skill score as verification metrics, that the DL methods compete with or outperform the calibrated GEFS system at lead times from 0 to 48 h and again from 72 to 120 h for AR vapor transport events. In addition, the DL methods generate reliable and skillful probabilistic forecasts. The implications of varying the length of the training dataset are examined, and the results show that the DL methods learn relatively quickly and ∼10 years of hindcast data are required to compete with the GEFS ensemble.
Abstract
Prediction of weather is a main goal of atmospheric science. Its importance to society is growing continuously due to factors such as vulnerability to natural disasters, the move to renewable energy sources, and the risks of climate change. But prediction is also a major scientific challenge due to the inherently limited predictability of a chaotic atmosphere, and has led to a revolution in forecasting methods as we have moved to probabilistic prediction. These changes provide the motivation for Waves to Weather (W2W), a major national research program in Germany with three main university partners in Munich, Mainz, and Karlsruhe. We are currently in the second 4-yr phase of our planned duration of 12 years and employ 36 doctoral and postdoctoral scientists. In the context of this large program, we address what we have identified to be the most important and challenging scientific questions in predictability of weather, namely, upscale error growth, errors associated with cloud processes, and probabilistic prediction of high-impact weather. This paper presents some key results of the first phase of W2W and discusses how they have influenced our understanding of predictability. The key role of interdisciplinary research linking atmospheric scientists with experts in visualization, statistics, numerical analysis, and inverse methods will be highlighted. To ensure a lasting impact on research in our field in Germany and internationally, we have instituted innovative programs for training and support of early-career scientists, and to support education, equal opportunities, and outreach, which are also described here.
Abstract
Prediction of weather is a main goal of atmospheric science. Its importance to society is growing continuously due to factors such as vulnerability to natural disasters, the move to renewable energy sources, and the risks of climate change. But prediction is also a major scientific challenge due to the inherently limited predictability of a chaotic atmosphere, and has led to a revolution in forecasting methods as we have moved to probabilistic prediction. These changes provide the motivation for Waves to Weather (W2W), a major national research program in Germany with three main university partners in Munich, Mainz, and Karlsruhe. We are currently in the second 4-yr phase of our planned duration of 12 years and employ 36 doctoral and postdoctoral scientists. In the context of this large program, we address what we have identified to be the most important and challenging scientific questions in predictability of weather, namely, upscale error growth, errors associated with cloud processes, and probabilistic prediction of high-impact weather. This paper presents some key results of the first phase of W2W and discusses how they have influenced our understanding of predictability. The key role of interdisciplinary research linking atmospheric scientists with experts in visualization, statistics, numerical analysis, and inverse methods will be highlighted. To ensure a lasting impact on research in our field in Germany and internationally, we have instituted innovative programs for training and support of early-career scientists, and to support education, equal opportunities, and outreach, which are also described here.
Abstract
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 in 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 toward 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 are also discussed.
Abstract
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 in 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 toward 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 are also discussed.