Global Ensemble Forecast System Research Articles

A novel hybrid artificial neural network - Parametric scheme for postprocessing medium-range precipitation forecasts

Many present-day statistical schemes for postprocessing weather forecasts, in particular precipitation forecasts, rely on calibration using prescribed statistical models to relate forecast statistics to distributional parameters. The efficacy of such schemes is often constrained not only by prescribed predictor-predictand relation, but also by arbitrary choices of temporal window and lead time range for training. To address this limitation, we propose an end-to-end, computationally efficient hybrid postprocessing scheme capable of producing full predictive distributions of precipitation accumulation without explicit stratification of forecast-observation pairs by forecast lead time and season. The proposed framework uses the censored, shifted gamma distribution (CSGD) as the predictive distribution but uses an artificial neural network (ANN) to estimate the distributional parameters of CSGD through a unified approach. This approach, referred to as ANN-CSGD, allows for simultaneous estimation of distributional parameters over multiple lead times and seasons in a single model by incorporating the latter variables as predictors to the ANN. We test our proposed ANN-CSGD model for postprocessing of ensemble mean forecasts of 24-h precipitation totals over selected river basins in California, at one- to seven-day lead times, from the Global Ensemble Forecast System (GEFS). The probabilistic quantitative precipitation forecasts (PQPFs) from the ANN-CSGD, are more skillful overall than those from the benchmark CSGD and the Mixed-type meta-Gaussian distribution (MMGD) models. The ANN-CSGD PQPFs highly improve the performance of those from CSGD in predicting the probability of precipitation (PoP) and are also much sharper and reliable at higher precipitation thresholds. We demonstrate how the hybrid approach, by using the entire available training data and its modified formulation, efficiently represents interactions between GEFS forecasts and season/lead times, thus leading to enhanced predictive performance.

Impact of Land Surface Processes on a Record‐Breaking Rainfall Event on May 06–07, 2017, in Guangzhou, China

AbstractA record‐breaking extreme rainfall event with a maximum rainfall amount over 24 h of 524.1 mm occurred in Guangzhou, China, on May 06–07, 2017. To study the impact of land surface processes on this extreme rainfall, two 21‐member convective‐permitting ensemble forecasts over South China were performed based on two land surface models (LSMs), Noah and Community Land Model (CLM), and 21 forecasting members of the Global Ensemble Forecast System (GEFS). The results showed that, in general, members using the Noah LSM could better simulate urban heat islands (UHIs) and urban convection than members using the CLM LSM in this case. By investigating the ensemble member that most resembled the observations, it was found that the high temperature center in the urban area caused a thermal low in the early stage. As the southerly winds strengthened, the low‐level convergence line continued moving northward and eventually triggered convection in the mountainous region. A sensitivity experiment showed that the impact of land surface heterogeneity on precipitation could be reflected on a finer scale, and heavy rainfall was very sensitive to the changes in small‐scale land surface forces, including terrain and land use. Slight variations in small‐scale land surface conditions caused great responses in the total precipitation, indicating that for the occurrence of such quasi‐stationary extreme rainfall, a subtle balance between different atmospheric and land surface factors may be required.

Comparison of different land-surface perturbation methods in short-range ensemble forecasts

In order to compare the sensitivity of short-range ensemble forecasts to different land-surface parameters in the South China region, three perturbation experiments related to the land surface model (LSM), initial soil moisture (ISM), and land–atmosphere coupling coefficient (LCC) were designed, and another control experiment driven by the Global Ensemble Forecast System (GEFS) was also performed. All ensemble members were initiated at 0000 UTC each day, and integrated for 24 h for a total of 40 days from the period 1 April to 10 May 2019 based on the Weather Research and Forecasting model. The results showed that the perturbation experiment of the LSM (LSMPE) had the largest ensemble spread, as well as the lowest ensemble-mean root-mean-square error among the three sets of land-surface perturbed experiments, which indicated that it could represent more uncertainty and less error. The ensemble spread of the perturbation experiment of the ISM (ISMPE) was generally less than that of LSMPE but greater than that of LCCPE (the perturbation experiment of the LCC). In particular, although the perturbation of the LCC could not produce greater spread, it had an effective influence on the intensity of precipitation. However, the ensemble spread of all the land-surface perturbed experiments was smaller than that of GEFSPE (the control experiment). Therefore, in future, land-surface perturbations and atmospheric perturbations should be combined in the design of ensemble forecasting systems to make the model represent more uncertainties.摘要为了比较不同陆面扰动方法对短时集合预报的影响, 本研究设计了陆面模式扰动实验 (LSMPE), 初始土壤湿度扰动实验 (ISMPE), 陆面-大气耦合系数扰动实验 (LCCPE) 以及大气扰动对照实验 (GEFSPE). 结果表明, 在三组陆面扰动实验中, LSMPE能代表最大的不确定性且误差最小;ISMPE的离散度要比LCCPE稍大, 但是耦合系数的减弱能有效降低区域平均降水量.然而, 所有陆面扰动实验产生的离散度均小于GEFSPE, 这表明未来有必要将陆面扰动与大气扰动相结合, 以使得模式能代表更多的不确定性.

A study of predictability of coupled ocean\u2013atmosphere system using attractor radius and global attractor radius

While ocean–atmosphere coupled models play an increasingly important role in weather-climate simulation and prediction, the predictability theory based on an atmosphere-only model has significant limitations in interpreting prediction results and guiding predictability studies. Here we use a conceptual ocean–atmosphere coupled model that describes the typical interactions of a synoptic-scale atmosphere with a seasonally-interannually varying upper ocean as well as a deep ocean that varies on decadal timescales to systematically study the predictability of a coupled system. Moving from an atmosphere-only system to an ocean–atmosphere coupled system, the initial-value predictability problem becomes a joint initial-value and boundary-value problem. Although the coupling process increases the uncertainties of the boundary, ocean signals with longer timescales are added to the atmosphere system, thus increasing its predictability. We then investigate the predictability characteristics of the National Centers for Environmental Prediction coupled Climate Forecast System (CFS) and the uncoupled Global Ensemble Forecast System (GEFS). In the coupled CFS system, the practical predictability limit of the lower troposphere is significantly longer than in the uncoupled GEFS due to the contribution of low-frequency boundary signals from air-sea interactions. While further deep and thorough examination is necessary for understanding ocean predictability in the climate system, a preliminary discussion for the predictability of the upper and deep oceans within a coupled ocean–atmosphere framework is also presented in this study.

Assimilation of GPM Microwave Imager Radiance data with the WRF hybrid 3DEnVar system for the prediction of Typhoon Chan-hom (2015)

The impact of assimilating radiances from the global satellite precipitation program (GPM) Microwave Imager (GMI) on the track prediction of typhoon Chan-hom was studied with a hybrid three-dimensional ensemble-variational (3DEnVar) data assimilation (DA) system. It is found that the data assimilation of clear-sky GMI radiance is able to depict the structure of typhoon well by warming the typhoon inner core area. The bias correction coefficients obtained by the off-line model along with the variational bias correction method can reduce the bias. GMI radiance data assimilation experiment can effectively adjust the core area of Typhoon “Chan-hom” and systematically correct the typhoon position in the background of the model. The sensitivity of the hybrid 3DEnVar DA behaviors on the choice of the ensemble members is also investigated. Results show that the 3DEnVar Global Ensemble Forecast System (3DEnVar_GEFS)-based ensemble spread is essentially dominated in the vicinity of Typhoon Chan-hom, which is less disturbed by random errors. In general, the track forecasts from 3DEnVar_RCV match better with the best track by adding Gaussian noise randomly based on the background error than those from the 3DVAR, and 3DEnVar_GEFS.

An Investigation of Large Cross-Track Errors in North Atlantic Tropical Cyclones in the GEFS and ECMWF Ensembles

AbstractThe largest medium-range (72–120 h) cross-track errors (CTE) of tropical cyclone (TC) forecasts from the Global Ensemble Forecast System (GEFS) over the northern Atlantic Ocean are examined for the 2008–16 seasons. The 38 unique forecasts within the upper quartile of most negative CTEs (i.e., left-of-track bias larger than 250 km by 72 h) do not have a clear common source of steering error, although 12 of the forecasts involve the underprediction of a weak upper-level trough to the west of the TC by 36 h. Meanwhile, at least 18 of the 36 most positive CTEs (right-of-track bias) are associated with TCs embedded in the southwest extent of a subtropical ridge, the strength of which is increasingly underpredicted during the first 24 h of the forecast. Excessive height falls north of the TC are driven by overpredicted divergence aloft, which corresponds to overpredicted TC outer-core convection. The convection is triggered by a 5%–20% overprediction of near-TC moisture and instability in the initial conditions. Weather Research and Forecasting (WRF) Model simulations are run at 36-, 12-, and 4-km grid spacing for select right-of-track cases, using the GEFS for initial and lateral boundary conditions. The 36-km WRF reproduces the same growth of errors as the GEFS because of, in part, sharing the same stability and moisture errors in the initial conditions. Changes in the convective parameterization affect how quickly these errors grow by affecting how much convection spins up. The addition of a 4-km nest with no convective parameterization causes the errors to grow ~20% faster, resulting in an even larger right-of-track error.

Dynamical–Statistical Prediction of Week-2 Severe Weather for the United States

AbstractA dynamical–statistical model is developed for forecasting week-2 severe weather (hail, tornadoes, and damaging winds) over the United States. The supercell composite parameter (SCP) is used as a predictor, which is derived from the 16-day dynamical forecasts of the National Centers for Environmental Prediction (NCEP) Global Ensemble Forecast System (GEFS) model and represents the large-scale convective environments influencing severe weather. The hybrid model forecast is based on the empirical relationship between GEFS hindcast SCP and observed weekly severe weather frequency during 1996–2012, the GEFS hindcast period. Cross validations suggest that the hybrid model has a low skill for week-2 severe weather when applying simple linear regression method at 0.5° × 0.5° (latitude × longitude) grid data. However, the forecast can be improved by using the 5° × 5° area-averaged data. The forecast skill can be further improved by using the empirical relationship depicted by the singular value decomposition method, which takes into account the spatial covariations of weekly severe weather. The hybrid model was tested operationally in spring 2019 and demonstrated skillful forecasts of week-2 severe weather frequency over the United States.

Toward Unifying Short-Term and Next-Day Convection-Allowing Ensemble Forecast Systems with a Continuously Cycling 3-km Ensemble Kalman Filter over the Entire Conterminous United States

AbstractUsing the Weather Research and Forecasting Model, 80-member ensemble Kalman filter (EnKF) analyses with 3-km horizontal grid spacing were produced over the entire conterminous United States (CONUS) for 4 weeks using 1-h continuous cycling. For comparison, similarly configured EnKF analyses with 15-km horizontal grid spacing were also produced. At 0000 UTC, 15- and 3-km EnKF analyses initialized 36-h, 3-km, 10-member ensemble forecasts that were verified with a focus on precipitation. Additionally, forecasts were initialized from operational Global Ensemble Forecast System (GEFS) initial conditions (ICs) and experimental “blended” ICs produced by combining large scales from GEFS ICs with small scales from EnKF analyses using a low-pass filter. The EnKFs had stable climates with generally small biases, and precipitation forecasts initialized from 3-km EnKF analyses were more skillful and reliable than those initialized from downscaled GEFS and 15-km EnKF ICs through 12–18 and 6–12 h, respectively. Conversely, after 18 h, GEFS-initialized precipitation forecasts were better than EnKF-initialized precipitation forecasts. Blended 3-km ICs reflected the respective strengths of both GEFS and high-resolution EnKF ICs and yielded the best performance considering all times: blended 3-km ICs led to short-term forecasts with similar or better skill and reliability than those initialized from unblended 3-km EnKF analyses and ~18–36-h forecasts possessing comparable quality as GEFS-initialized forecasts. This work likely represents the first time a convection-allowing EnKF has been continuously cycled over a region as large as the entire CONUS, and results suggest blending high-resolution EnKF analyses with low-resolution global fields can potentially unify short-term and next-day convection-allowing ensemble forecast systems under a common framework.

Air quality and aerosol predictions at National Oceanic and Atmospheric Administration (NOAA) and their applications

Operational air quality predictions produced by NOAA/National Weather Service include predictions of ozone, fine particulate matter (PM2.5) and wildfire smoke for the United States and predictions of windblown dust for the contiguous 48 states. Prediction maps are distributed at https://airquality.weather.gov/ and as a web service at https://idpgis.ncep.noaa.gov/arcgis/rest/services/NWS_Forecasts_Guidance_Warnings. Community Multiscale Air Quality modeling system (CMAQ) used for ozone and PM2.5 predictions was most recently updated in December 2018, when a unified bias-correction procedure was introduced for ozone and PM2.5. Wildfire smoke dispersion is operationally predicted using the Hybrid Single Particle Lagrangian Integrated Trajectory Model (HYSPLIT). Wildfire smoke dispersion is included in version 4 of the Eulerian High Resolution Rapid Refresh (HRRR) model that is being evaluated for operational implementation. One of the largest uncertainties for prediction of air quality over the U.S. involves specification of wildfire smoke emissions. Different smoke emission approaches used in these systems will be discussed.NOAA is testing ozone and PM2.5 predictions extended from 48 to 72 hours. Global aerosol predictions, which include biomass burning aerosols, are being tested for inclusion in one member of the Global Ensemble Forecast System (GEFS) to replace the current standalone global aerosol prediction system. An overview of this testing will be presented.NOAA’s operational air quality predictions contribute to protection of lives and health in the United States. These predictions are used by state and local air quality forecasters to issue official air quality forecasts for their respective areas. General public accesses NOAA’s hourly predictions directly at https://airquality.weather.gov/. These predictions are also used in the applications about wildfire and health as well as in smoke vulnerability assessments lead by our partners at the U. S. Environmental Protection Agency (EPA) and Centers for Disease Control and Prevention (CDC).

Downscaling of seasonal ensemble forecasts to the convection‐permitting scale over the Horn of Africa using the WRF model

AbstractThe new SEAS5 global ensemble forecast system was dynamically downscaled over the Horn of Africa for summer (June‐July‐August) 2018. For this purpose, a multi‐physics ensemble was designed with a grid increment of 3 km and without any intermediate nest based on the Weather Research and Forecasting model (WRF). The WRF and the SEAS5 model output were compared with each other and reference datasets to assess the biases in 4 different regions of Ethiopia. Also, the WRF ensemble variability was investigated in relation to model parameterization and lateral boundary conditions. Over the summer, the SEAS5 has a positive temperature bias of 0.17°C compared to ECMWF analysis average for the study domain, while the WRF bias is +1.14°C. Concerning precipitation, the WRF model had average accumulated values of 264 mm, compared to 248 mm for SEAS5 and 236 mm for the observations. Over south Ethiopia, however, the downscaling produced over 50% more precipitation than the other datasets. The maximum northward extension of the tropical rain belt was reduced by about 2° in both models when compared to observations. Downscaling increased reliability for precipitation, correcting the SEAS5 underdispersion: ensemble spread for precipitation was increased by about 70% in the WRF ensemble in three of the four Ethiopian sub‐regions, whereas the very dry Somali region remained unaffected. The WRF ensemble analysis revealed that the ensemble spread is mainly caused by the perturbed boundary conditions, as their effect is often 50% larger than the physics‐induced variability in the mountainous part of Ethiopia for precipitation and temperature.

A Nonhomogeneous Regression-Based Statistical Postprocessing Scheme for Generating Probabilistic Quantitative Precipitation Forecast

AbstractThis paper introduces a new, two-part scheme for postprocessing single-valued precipitation forecast to create probabilistic quantitative precipitation forecast (PQPF). This scheme, herein referred to as the mixed-type nonhomogeneous regression (MNHR), combines the use of logistic regression for estimating rainfall intermittency and nonhomogeneous regression for estimation of additional parameters of the conditional distribution. The performance of MNHR is evaluated relative to operational mixed-type meta-Gaussian distribution (MMGD) and the censored, shifted gamma distribution (CSGD) in postprocessing Global Ensemble Forecast System (GEFS) reforecasts averaged over 25 watersheds in the American River basin in California. The results point to superior performance of MNHR relative to MMGD and CSGD in terms of the skill of postprocessed PQPFs at 24- and 96-h accumulation windows. In addition, it is observed that the performance of CSGD tends to trail behind MNHR and MMGD at least for the 24-h window, though the performance differences tend to narrow at higher forecast amounts and longer lead times. Our analyses suggest that CSGD’s underperformance arises partly from its tendency to inflate the shift parameter estimates, which is pronounced over the study site possibly because of infrequent rainfall occurrence. By contrast, MNHR’s use of logistic regression helps avoid such bias, and its formulation of conditional distribution addresses the lack of skewness of MMGD for higher forecast amounts. Moreover, MHNR-based PQPF exhibits both superior calibration and relatively high sharpness at short lead times and on an unconditional sense, whereas it features lower sharpness relative to the other two suites when conditioned on higher forecast amount. This trade-off between calibration andconditionalsharpness warrants further research.

Post-Processing and Evaluation of Precipitation Ensemble Forecast under Multiple Schemes in Beijiang River Basin

Precipitation is one of the most important factors affecting the accuracy and uncertainty of hydrological forecasting. Considerable progress has been made in numerical weather prediction after decades of development, but the forecast products still cannot be used directly for hydrological forecasting. This study used ensemble pro-processor (EPP) to post-process the Global Ensemble Forecast System (GEFS) and Climate Forecast System version 2 (CFSv2) with four designed schemes, and then integrated them to investigate the forecast accuracy in longer time scales based on the best scheme. Many indices such as correlation coefficient, Nash efficiency coefficient, rank histogram, and continuous ranked probability skill score were used to evaluate the results in different aspects. The results show that EPP can improve the accuracy of raw forecast significantly, and the scheme considering cumulative forecast precipitation is better than that only considers single-day forecast. Moreover, the scheme that considers some observed precipitation would help to improve the accuracy and reduce the uncertainty. In terms of medium- and long-term forecasts, the integrated forecast based on GEFS and CFSv2 after post-processed would be better than CFSv2 significantly. The results of this study would be a very important demonstration to remove the deviation of ensemble forecast and improve the accuracy of hydrological forecasting in different time scales.

Beyond univariate calibration: verifying spatial structure in ensembles of forecast fields

Abstract. Most available verification metrics for ensemble forecasts focus on univariate quantities. That is, they assess whether the ensemble provides an adequate representation of the forecast uncertainty about the quantity of interest at a particular location and time. For spatially indexed ensemble forecasts, however, it is also important that forecast fields reproduce the spatial structure of the observed field and represent the uncertainty about spatial properties such as the size of the area for which heavy precipitation, high winds, critical fire weather conditions, etc., are expected. In this article we study the properties of the fraction of threshold exceedance (FTE) histogram, a new diagnostic tool designed for spatially indexed ensemble forecast fields. Defined as the fraction of grid points where a prescribed threshold is exceeded, the FTE is calculated for the verification field and separately for each ensemble member. It yields a projection of a – possibly high-dimensional – multivariate quantity onto a univariate quantity that can be studied with standard tools like verification rank histograms. This projection is appealing since it reflects a spatial property that is intuitive and directly relevant in applications, though it is not obvious whether the FTE is sufficiently sensitive to misrepresentation of spatial structure in the ensemble. In a comprehensive simulation study we find that departures from uniformity of the FTE histograms can indeed be related to forecast ensembles with biased spatial variability and that these histograms detect shortcomings in the spatial structure of ensemble forecast fields that are not obvious by eye. For demonstration, FTE histograms are applied in the context of spatially downscaled ensemble precipitation forecast fields from NOAA's Global Ensemble Forecast System.

Projecting Northern Hemisphere Flow Regime Transition Using Integrated Enstrophy

Integrated enstrophy (IE) is the square of vorticity integrated over an entire hemisphere at a particular level in the atmosphere. Previous work has shown this quantity is correlated to the positive Lyapunov Exponent for hemispheric flow, and as such is a measure of flow stability or predictability. In this study, IE is calculated at 500 hPa over an area that encompasses 0° to 70° in the Northern Hemisphere. The data sets used were the 500 hPa initial and forecast fields for the Global Ensemble Forecasting System (GEFS) (on a 1 × 1 latitude-longitude grid) provided by the National Oceanic and Atmospheric Administration (NOAA) Weather Prediction Center (WPC) and the National Centers for Environmental Prediction/NOAA reanalyses (on a 2.5 × 2.5 latitude-longitude grid) archived in Boulder, CO. The GEFS forecast fields were provided every 24 h out to 240 h. By examining these forecasts over a year, it was found here that significant changes in the calculated IE values, as quantitatively determined, are a good predictor of flow regime transition, and 34 cases were found. We also found that the model IE forecasts identified these regime transitions reliably out to about seven days, however, the probability of detection and the skill decreased after this time. Additionally, a threshold for changes in IE was found for the cases studied here.

Prediction Skill of Extended Range 2-m Maximum Air Temperature Probabilistic Forecasts Using Machine Learning Post-Processing Methods

The extended range temperature prediction is of great importance for public health, energy and agriculture. The two machine learning methods, namely, the neural networks and natural gradient boosting (NGBoost), are applied to improve the prediction skills of the 2-m maximum air temperature with lead times of 1–35 days over East Asia based on the Environmental Modeling Center, Global Ensemble Forecast System (EMC-GEFS), under the Subseasonal Experiment (SubX) of the National Centers for Environmental Prediction (NCEP). The ensemble model output statistics (EMOS) method is conducted as the benchmark for comparison. The results show that all the post-processing methods can efficiently reduce the prediction biases and uncertainties, especially in the lead week 1–2. The two machine learning methods outperform EMOS by approximately 0.2 in terms of the continuous ranked probability score (CRPS) overall. The neural networks and NGBoost behave as the best models in more than 90% of the study area over the validation period. In our study, CRPS, which is not a common loss function in machine learning, is introduced to make probabilistic forecasting possible for traditional neural networks. Moreover, we extend the NGBoost model to atmospheric sciences of probabilistic temperature forecasting which obtains satisfying performances.

Ensemble-Guided Tropical Cyclone Track Forecasting for Optimal Satellite Remote Sensing

Within the realm of satellite remote sensing, optimal data acquisition to study natural phenomena under time, resource, and cost constraints is a well-known problem. Furthermore, since the sensors themselves are at remote locations with sparse ground connectivity, the optimal method must use a computationally light forecasting algorithm, which assimilates information from the observations at possibly irregular intervals, in near real time. In this article, we propose and demonstrate the ensemble -guided cyclone track forecasting (EGCTF) method for application in remote tropical cyclone tracking. The algorithm uses ensemble data produced by numerical weather prediction models to guide the forecasting process while assimilating measured cyclone center positions. The algorithm was tested and analyzed with the Global Ensemble Forecasting System (GEFS) data and the National Hurricane Center data for the 2018 year hurricanes within the Atlantic basin. Compared with a baseline method that uses the GEFS-issued mean ensemble track (AEMN) for forecasting and no data assimilation, the proposed algorithm exhibited positive forecast skill for more than 290 test cases over forecast periods spanning 6-48 h. The skill is seen to improve with lengthening forecast periods, with five test cases showing greater than 75% skill for a forecast period of 6 h to 247 test cases for the forecast period of 48 h.

Prediction of Flash Droughts over the United States

AbstractWe examine reforecasts of flash droughts over the United States for the late spring (April–May), midsummer (June–July), and late summer/early autumn (August–September) with lead times up to 3 pentads based on the NOAA second-generation Global Ensemble Forecast System reforecasts version 2 (GEFSv2). We consider forecasts of both heat wave and precipitation deficit (P deficit) flash droughts, where heat wave flash droughts are characterized by high temperature and depletion of soil moisture and P deficit flash droughts are caused by lack of precipitation that leads to (rather than being the cause of) high temperature. We find that the GEFSv2 reforecasts generally capture the frequency of occurrence (FOC) patterns. The equitable threat score (ETS) of heat wave flash drought forecasts for late spring in the regions where heat wave flash droughts are most likely to occur over the north-central and Pacific Northwest regions is statistically significant up to 2 pentads. The GEFSv2 reforecasts capture the basic pattern of the FOC of P-deficit flash droughts and also are skillful up to lead about 2 pentads. However, the reforecasts overestimate the P-deficit flash drought FOC over parts of the Southwest in late spring, leading to large false alarm rates. For autumn, the reforecasts underestimate P-deficit flash drought occurrence over California and Nevada. The GEFSv2 reforecasts are able to capture the approximately linear relationship between evaporation and soil moisture, but the lack of skill in precipitation forecasts limits the skill of P-deficit flash drought forecasts.

Intercomparison of multiple statistical methods in post‐processing ensemble precipitation and temperature forecasts

AbstractEnsemble weather forecasting generally suffers from bias and under‐dispersion, which limit its predictive power. Several post‐processing methods have been developed to overcome these limitations, and an intercomparison is needed to understand their performance. Four state‐of‐the‐art methods are compared in post‐processing precipitation and air temperature of the Global Ensemble Forecasting System (GEFS) reforecasts using a simple bias correction (BC) method as a reference. These methods include extended logistic regression (ExLR), generator‐based post‐processing (GPP), Bayesian model averaging (BMA) and affine kernel dressing (AKD). All these methods are tested over 659 national standard meteorological stations in China. The research concerns are the influence of region and forecast date and the role of BC on ensemble weather forecasting. It was found that: (1) the deterministic methods (GPP and ExLR) are more skilful than the probabilistic methods (BMA and AKD) in obtaining the well‐calibrated and skilful ensemble forecasts; (2) the forecast skill of the post‐processed ensemble weather forecasts is comparably high in the northern arid areas for precipitation, while the forecast skill for air temperature is only low in the Qinghai‐Tibetan Plateau area; (3) the skill difference of the post‐processed forecasts on different forecast date is only evident for air temperature, while not apparent for precipitation; and (4) only correcting bias for the ensemble weather forecasts can achieve about 0–70% (for precipitation) and 30–100% (for air temperature) forecast skill improvement for deterministic methods.

Recalibrating wind‐speed forecasts using regime‐dependent ensemble model output statistics

AbstractRaw output from deterministic numerical weather prediction models is typically subject to systematic biases. Although ensemble forecasts provide invaluable information regarding the uncertainty in a prediction, they themselves often misrepresent the weather that occurs. Given their widespread use, the need for high‐quality wind‐speed forecasts is well‐documented. Several statistical approaches have therefore been proposed to recalibrate ensembles of wind‐speed forecasts, including a heteroscedastic truncated regression approach. An extension to this method that utilises the prevailing atmospheric flow is implemented here in a quasigeostrophic simulation study and on Global Ensemble Forecasting System (GEFS) reforecast data, in the hope of alleviating errors owing to changes in the synoptic‐scale atmospheric state. When the wind speed depends strongly on the underlying weather regime, the resulting forecasts have the potential to provide substantial improvements in skill relative to conventional post‐processing techniques. This is particularly pertinent at longer lead times, where there is more improvement to be gained over current methods, and in weather regimes associated with wind speeds that differ greatly from climatology. In order to realise this potential, an accurate prediction of the future atmospheric regime is required.

Forecasting Severe Weather with Random Forests

AbstractUsing nine years of historical forecasts spanning April 2003–April 2012 from NOAA’s Second Generation Global Ensemble Forecast System Reforecast (GEFS/R) ensemble, random forest (RF) models are trained to make probabilistic predictions of severe weather across the contiguous United States (CONUS) at Days 1–3, with separate models for tornado, hail, and severe wind prediction at Day 1 in an analogous fashion to the Storm Prediction Center’s (SPC’s) convective outlooks. Separate models are also trained for the western, central, and eastern CONUS. Input predictors include fields associated with severe weather prediction, including CAPE, CIN, wind shear, and numerous other variables. Predictor inputs incorporate the simulated spatiotemporal evolution of these atmospheric fields throughout the forecast period in the vicinity of the forecast point. These trained RF models are applied to unseen inputs from April 2012 to December 2016, and their forecasts are evaluated alongside the equivalent SPC outlooks. The RFs objectively make statistical deductions about the relationships between various simulated atmospheric fields and observations of different severe weather phenomena that accord with the community’s physical understandings about severe weather forecasting. Using these quantified flow-dependent relationships, the RF outlooks are found to produce calibrated probabilistic forecasts that slightly underperform SPC outlooks at Day 1, but significantly outperform their outlooks at Days 2 and 3. In all cases, a blend of the SPC and RF outlooks significantly outperforms the SPC outlooks alone, suggesting that use of RFs can improve operational severe weather forecasting throughout the Day 1–3 period.