scholarly journals Bivariate Gaussian models for wind vectors in a distributional regression framework

2019 ◽  
Vol 5 (2) ◽  
pp. 115-132 ◽  
Author(s):  
Moritz N. Lang ◽  
Georg J. Mayr ◽  
Reto Stauffer ◽  
Achim Zeileis
Keyword(s):  
Wind Direction ◽  
Wind Component ◽  
Smooth Functions ◽  
Ensemble Forecasts ◽  
Alpine Valley ◽  
Explanatory Variables ◽  
Predictive Skill ◽  
Regression Framework ◽  
Distribution Parameters ◽  
Distributional Regression

Abstract. A new probabilistic post-processing method for wind vectors is presented in a distributional regression framework employing the bivariate Gaussian distribution. In contrast to previous studies, all parameters of the distribution are simultaneously modeled, namely the location and scale parameters for both wind components and also the correlation coefficient between them employing flexible regression splines. To capture a possible mismatch between the predicted and observed wind direction, ensemble forecasts of both wind components are included using flexible two-dimensional smooth functions. This encompasses a smooth rotation of the wind direction conditional on the season and the forecasted ensemble wind direction. The performance of the new method is tested for stations located in plains, in mountain foreland, and within an alpine valley, employing ECMWF ensemble forecasts as explanatory variables for all distribution parameters. The rotation-allowing model shows distinct improvements in terms of predictive skill for all sites compared to a baseline model that post-processes each wind component separately. Moreover, different correlation specifications are tested, and small improvements compared to the model setup with no estimated correlation could be found for stations located in alpine valleys.

scholarly journals Bayesian Model Averaging for Wind Speed Ensemble Forecasts Using Wind Speed and Direction

2017 ◽  
Vol 32 (6) ◽  
pp. 2217-2227 ◽  
Author(s):  
Siri Sofie Eide ◽  
John Bjørnar Bremnes ◽  
Ingelin Steinsland
Keyword(s):  
Wind Speed ◽  
Wind Direction ◽  
Bayesian Model Averaging ◽  
Weather Prediction ◽  
Model Averaging ◽  
Strong Wind ◽  
Ensemble Forecasts ◽  
Wind Speeds ◽  
Low Wind Speeds ◽  
The One

Abstract In this paper, probabilistic wind speed forecasts are constructed based on ensemble numerical weather prediction (NWP) forecasts for both wind speed and wind direction. Including other NWP variables in addition to the one subject to forecasting is common for statistical calibration of deterministic forecasts. However, this practice is rarely seen for ensemble forecasts, probably because of a lack of methods. A Bayesian modeling approach (BMA) is adopted, and a flexible model class based on splines is introduced for the mean model. The spline model allows both wind speed and wind direction to be included nonlinearly. The proposed methodology is tested for forecasting hourly maximum 10-min wind speeds based on ensemble forecasts from the European Centre for Medium-Range Weather Forecasts at 204 locations in Norway for lead times from +12 to +108 h. An improvement in the continuous ranked probability score is seen for approximately 85% of the locations using the proposed method compared to standard BMA based on only wind speed forecasts. For moderate-to-strong wind the improvement is substantial, while for low wind speeds there is generally less or no improvement. On average, the improvement is 5%. The proposed methodology can be extended to include more NWP variables in the calibration and can also be applied to other variables.

scholarly journals Multiple causes of nonstationarity in the Weihe annual low-flow series

2018 ◽  
Vol 22 (2) ◽  
pp. 1525-1542 ◽  
Author(s):  
Bin Xiong ◽  
Lihua Xiong ◽  
Jie Chen ◽  
Chong-Yu Xu ◽  
Lingqi Li
Keyword(s):  
Frequency Analysis ◽  
Driving Forces ◽  
Potential Evapotranspiration ◽  
Aridity Index ◽  
Base Flow ◽  
Low Flow ◽  
Distribution Model ◽  
Flow Index ◽  
Explanatory Variables ◽  
Distribution Parameters

Abstract. Under the background of global climate change and local anthropogenic activities, multiple driving forces have introduced various nonstationary components into low-flow series. This has led to a high demand on low-flow frequency analysis that considers nonstationary conditions for modeling. In this study, through a nonstationary frequency analysis framework with the generalized linear model (GLM) to consider time-varying distribution parameters, the multiple explanatory variables were incorporated to explain the variation in low-flow distribution parameters. These variables are comprised of the three indices of human activities (HAs; i.e., population, POP; irrigation area, IAR; and gross domestic product, GDP) and the eight measuring indices of the climate and catchment conditions (i.e., total precipitation P, mean frequency of precipitation events λ, temperature T, potential evapotranspiration (EP), climate aridity index AIEP, base-flow index (BFI), recession constant K and the recession-related aridity index AIK). This framework was applied to model the annual minimum flow series of both Huaxian and Xianyang gauging stations in the Weihe River, China (also known as the Wei He River). The results from stepwise regression for the optimal explanatory variables show that the variables related to irrigation, recession, temperature and precipitation play an important role in modeling. Specifically, analysis of annual minimum 30-day flow in Huaxian shows that the nonstationary distribution model with any one of all explanatory variables is better than the one without explanatory variables, the nonstationary gamma distribution model with four optimal variables is the best model and AIK is of the highest relative importance among these four variables, followed by IAR, BFI and AIEP. We conclude that the incorporation of multiple indices related to low-flow generation permits tracing various driving forces. The established link in nonstationary analysis will be beneficial to analyze future occurrences of low-flow extremes in similar areas.

Multiyear Prediction of Monthly Mean Atlantic Meridional Overturning Circulation at 26.5°N

Science ◽  
2012 ◽  
Vol 335 (6064) ◽  
pp. 76-79 ◽  
Author(s):  
Daniela Matei ◽  
Johanna Baehr ◽  
Johann H. Jungclaus ◽  
Helmuth Haak ◽  
Wolfgang A. Müller ◽  
...  
Keyword(s):  
Atlantic Meridional Overturning Circulation ◽  
Meridional Overturning Circulation ◽  
Density Difference ◽  
Upper Ocean ◽  
Overturning Circulation ◽  
Ensemble Forecasts ◽  
Climate Predictability ◽  
Predictive Skill ◽  
Meridional Overturning

Attempts to predict changes in Atlantic Meridional Overturning Circulation (AMOC) have yielded little success to date. Here, we demonstrate predictability for monthly mean AMOC strength at 26.5°N for up to 4 years in advance. This AMOC predictive skill arises predominantly from the basin-wide upper-mid-ocean geostrophic transport, which in turn can be predicted because we have skill in predicting the upper-ocean zonal density difference. Ensemble forecasts initialized between January 2008 and January 2011 indicate a stable AMOC at 26.5°N until at least 2014, despite a brief wind-induced weakening in 2010. Because AMOC influences many aspects of climate, our results establish AMOC as an important potential carrier of climate predictability.

Verification of Two Years of CNR-ISAC Subseasonal Forecasts

2019 ◽  
Vol 34 (2) ◽  
pp. 331-344 ◽  
Author(s):  
Daniele Mastrangelo ◽  
Piero Malguzzi
Keyword(s):  
Research Council ◽  
National Research Council ◽  
Mean Square ◽  
Weekly Basis ◽  
Ensemble Forecasts ◽  
Atmospheric Sciences ◽  
Discrimination Capability ◽  
Predictive Skill ◽  
Anomaly Correlation Coefficient ◽  
Forecasting System

Abstract The monthly forecasting system of the Institute of Atmospheric Sciences and Climate of the National Research Council (CNR-ISAC) of Italy is operationally run on a weekly basis in the framework of the Subseasonal-to-Seasonal (S2S) project to produce 41-member ensemble forecasts. The first two years of forecasts, covering 106 weeks from April 2015, are verified against ERA-Interim as weekly averages starting from the first forecast day. Nonprobabilistic scores of 500-hPa geopotential height and 850-hPa temperature anomalies are computed for the extratropical hemispheres. The anomaly correlation coefficient shows enhanced predictive skill during the cold months, when favorable values are occasionally obtained beyond week 2. The root-mean-square error saturates toward the climatological value between weeks 2 and 3. Reliability diagrams are used to evaluate the probabilistic forecast skill of 2-m temperature over Northern Hemisphere extratropical land points, in terms of above- and below-normal events. The forecasting system loses reliability and resolution beyond week 2, but well reproduces the observed 2-yr mean frequency up to week 4, proving to be unbiased. The reliability of the forecasting system systematically outperforms that obtained by persisting the previous week forecast. Beyond week 2, the forecast distribution of below-normal events shows low confidence. However, a reliability diagram based on equally populated bins of forecast probabilities highlights residual resolution up to week 4 at low probabilities. ROC diagrams confirm that the modeling system has greater discrimination capability for below-normal events. The reliability analysis of accumulated precipitation shows minor differences between below- and above-normal events, with lower skill than 2-m temperature.

scholarly journals On the ROC Area of Ensemble Forecasts for Rare Events

2021 ◽  
Author(s):  
Zied Ben Bouallegue ◽  
David S. Richardson
Keyword(s):  
Roc Curve ◽  
Rare Events ◽  
Ensemble Forecast ◽  
Imprecise Probabilities ◽  
Forecast Verification ◽  
Ensemble Forecasts ◽  
Relative Operating Characteristic ◽  
Discrimination Ability ◽  
Predictive Skill ◽  
Roc Area

The relative operating characteristic (ROC) curve is a popular diagnostic tool in forecast verification, with the area under the ROC curve (AUC) used as a verification metric measuring the discrimination ability of a forecast. Along with calibration, discrimination is deemed as a fundamental probabilistic forecast attribute. In particular, in ensemble forecast verification, AUC provides a basis for the comparison of potential predictive skill of competing forecasts. While this approach is straightforward when dealing with forecasts of common events (e.g. probability of precipitation), the AUC interpretation can turn out to be oversimplistic or misleading when focusing on rare events (e.g. precipitation exceeding some warning criterion). How should we interpret AUC of ensemble forecasts when focusing on rare events? How can changes in the way probability forecasts are derived from the ensemble forecast affect AUC results? How can we detect a genuine improvement in terms of predictive skill? Based on verification experiments, a critical eye is cast on the AUC interpretation to answer these questions. As well as the traditional trapezoidal approximation and the well-known bi-normal fitting model, we discuss a new approach which embraces the concept of imprecise probabilities and relies on the subdivision of the lowest ensemble probability category.

Exploring winter predictability in Europe using the ECMWF hindcasts

2021 ◽  
Author(s):  
Daniele Mastrangelo ◽  
Ignazio Giuntoli ◽  
Piero Malguzzi
Keyword(s):  
Lead Time ◽  
High Impact ◽  
Temperature Fields ◽  
Lead Times ◽  
Economic Sectors ◽  
Ensemble Forecasts ◽  
Modes Of Variability ◽  
Predictive Skill ◽  
Extended Range ◽  
Impact Events

<p>The accuracy and reliability in predicting winter anomalies, particularly high-impact events, is crucial for economic sectors like energy production and trade. Sub-seasonal predictions can provide a useful tool for early detection of these events. In this context, this study aims to target atmospheric patterns leading to skillful winter predictions at S2S lead times.</p><p>With a focus on Europe, we explore extended range predictability (up to 35 days) in the ECMWF hindcast dataset (1999-2018). This dataset provides a sizable sample for assessing the winter months predictive skill of the model and can be considered as a preparatory step to the use of the more comprehensive real-time ensemble forecasts.</p><p>The verification is performed on geopotential and temperature fields against the ERA5 reanalysis. We first identify the most skillful predictions both in terms of lead-time and period of initialization. Later we assess whether these skillful predictions correspond to high-impact events, especially cold spells. Finally, in the attempt to identify the potential drivers of improved predictability, we track back to the dominant Euro-Atlantic modes of variability present in the initial atmospheric states of the well predicted events.</p>

scholarly journals GAMLSS: A distributional regression approach

2018 ◽  
Vol 18 (3-4) ◽  
pp. 248-273 ◽  
Author(s):  
Mikis D Stasinopoulos ◽  
Robert A Rigby ◽  
Fernanda De Bastiani
Keyword(s):  
Regression Analysis ◽  
General Framework ◽  
Generalized Additive Models ◽  
Additive Models ◽  
Explanatory Variables ◽  
Regression Approach ◽  
The Mean ◽  
Distributional Regression

Abstract: A tutorial of the generalized additive models for location, scale and shape (GAMLSS) is given here using two examples. GAMLSS is a general framework for performing regression analysis where not only the location (e.g., the mean) of the distribution but also the scale and shape of the distribution can be modelled by explanatory variables.

scholarly journals Can We See the Wind? Statistical Downscaling of Historical Sea Surface Winds in the Subarctic Northeast Pacific

2012 ◽  
Vol 25 (5) ◽  
pp. 1511-1528 ◽  
Author(s):  
Adam H. Monahan
Keyword(s):  
Wind Speed ◽  
Sea Surface ◽  
Wind Component ◽  
Surface Winds ◽  
Northeast Pacific ◽  
Predictive Skill ◽  
Sea Surface Winds ◽  
Mean Wind Speed ◽  
Mean Vector ◽  
Vector Wind

The statistical predictability of wintertime (December–February) monthly-mean sea surface winds (both vector wind components and wind speed) in the subarctic northeast Pacific off the west coast of Canada is considered, in the context of surface wind downscaling. Predictor fields (zonal wind, meridional wind, wind speed, and temperature) are shown to carry predictive information on the large scales (both vertical and horizontal) that are well simulated by numerical weather prediction and global climate models. It is found that, in general, the monthly mean vector wind components are more predictable by indices of the large-scale flow than by the monthly mean wind speed, with no systematic vertical variation in predictive skill for either across the depth of the troposphere. The difference in predictive skill between monthly-mean vector wind components and wind speed is interpreted in terms of an idealized model of the vector wind speed probability distribution, which demonstrates that for the conditions in the subarctic northeast Pacific, the sensitivity of mean wind speed to the standard deviations of vector wind component fluctuations (which are not well predicted) is greater than that to the mean vector wind components. It is demonstrated that this sensitivity is state dependent, and it is suggested that monthly mean wind speeds may be inherently more predictable in regions where the sensitivity to the vector wind component means is greater than that to the standard deviations. It is also demonstrated that daily wind fluctuations (both vector wind and wind speed) are generally more predictable than monthly-mean variability, and that monthly averages of the predicted daily winds generally represent the monthly-mean surface winds better than the predictions directly from monthly mean predictors.

Sign in / Sign up

Export Citation Format

Share Document