Search Results

You are looking at 1 - 10 of 42 items for

  • Author or Editor: Fanyou Kong x
  • Refine by Access: All Content x
Clear All Modify Search
Fanyou Kong, Kelvin K. Droegemeier, and Nicki L. Hickmon

Abstract

In Part I, the authors used a full physics, nonhydrostatic numerical model with horizontal grid spacing of 24 km and nested grids of 6- and 3-km spacing to generate the ensemble forecasts of an observed tornadic thunderstorm complex. The principal goal was to quantify the value added by fine grid spacing, as well as the assimilation of Doppler radar data, in both probabilistic and deterministic frameworks. The present paper focuses exclusively on 3-km horizontal grid spacing ensembles and the associated impacts on the forecast quality of temporal forecast sequencing, the construction of initial perturbations, and data assimilation. As in Part I, the authors employ a modified form of the scaled lagged average forecasting technique and use Stage IV accumulated precipitation estimates for verification. The ensemble mean and spread of accumulated precipitation are found to be similar in structure, mimicking their behavior in global models. Both the assimilation of Doppler radar data and the use of shorter (1–2 versus 3–5 h) forecast lead times improve ensemble precipitation forecasts. However, even at longer lead times and in certain situations without assimilated radar data, the ensembles are able to capture storm-scale features when the associated control forecast in a deterministic framework fails to do so. This indicates the potential value added by ensembles although this single case is not sufficient for drawing general conclusions. The creation of initial perturbations using forecasts of the same grid spacing shows no significant improvement over simply extracting perturbations from forecasts made at coarser spacing and interpolating them to finer grids. However, forecast quality is somewhat dependent upon perturbation amplitude, with smaller scaling values leading to significant underdispersion. Traditional forecast skill scores show somewhat contradictory results for accumulated precipitation, with the equitable threat score most consistent with qualitative performance.

Full access
Jeffrey D. Duda, Xuguang Wang, Fanyou Kong, and Ming Xue

Abstract

Two approaches for accounting for errors in quantitative precipitation forecasts (QPFs) due to uncertainty in the microphysics (MP) parameterization in a convection-allowing ensemble are examined. They include mixed MP (MMP) composed mostly of double-moment schemes and perturbing parameters within the Weather Research and Forecasting single-moment 6-class microphysics scheme (WSM6) MP scheme (PPMP). Thirty-five cases of real-time storm-scale ensemble forecasts produced by the Center for Analysis and Prediction of Storms during the NOAA Hazardous Weather Testbed 2011 Spring Experiment were examined.

The MMP ensemble had better fractions Brier scores (FBSs) for most lead times and thresholds, but the PPMP ensemble had better relative operating characteristic (ROC) scores for higher precipitation thresholds. The pooled ensemble formed by randomly drawing five members from the MMP and PPMP ensembles was no more skillful than the more accurate of the MMP and PPMP ensembles. Significant positive impact was found when the two were combined to form a larger ensemble.

The QPF and the systematic behaviors of derived microphysical variables were also examined. The skill of the QPF among different members depended on the thresholds, verification metrics, and forecast lead times. The profiles of microphysics variables from the double-moment schemes contained more variation in the vertical than those from the single-moment members. Among the double-moment schemes, WDM6 produced the smallest raindrops and very large number concentrations. Among the PPMP members, the behaviors were found to be consistent with the prescribed intercept parameters. The perturbed intercept parameters used in the PPMP ensemble fell within the range of values retrieved from the double-moment schemes.

Full access
Eswar R. Iyer, Adam J. Clark, Ming Xue, and Fanyou Kong

Abstract

Previous studies examining convection-allowing models (CAMs), as well as NOAA/Hazardous Weather Testbed Spring Forecasting Experiments (SFEs) have typically emphasized “day 1” (12–36 h) forecast guidance. These studies find a distinct advantage in CAMs relative to models that parameterize convection, especially for fields strongly tied to convection like precipitation. During the 2014 SFE, “day 2” (36–60 h) forecast products from a CAM ensemble provided by the Center for Analysis and Prediction of Storms (CAPS) at the University of Oklahoma were examined. Quantitative precipitation forecasts (QPFs) from the CAPS ensemble, known as the Storm Scale Ensemble Forecast (SSEF) system, are compared to NCEP’s operational Short Range Ensemble Forecast (SREF) system, which provides lateral boundary conditions for the SSEF, to see if the CAM ensemble outperforms the SREF through forecast hours 36–60. Equitable threat scores (ETSs) were computed for precipitation thresholds ranging from 0.10 to 0.75 in. for each SSEF and SREF member, as well as ensemble means, for 3-h accumulation periods. The ETS difference between the SSEF and SREF peaked during hours 36–42. Probabilistic forecasts were evaluated using the area under the receiver operating characteristic curve (ROC area). The SSEF had higher values of ROC area, especially at thresholds ≥ 0.50 in. Additionally, time–longitude diagrams of diurnally averaged rainfall were constructed for each SSEF/SREF ensemble member. Spatial correlation coefficients between forecasts and observations in time–longitude space indicated that the SSEF depicted the diurnal cycle much better than the SREF, which underforecasted precipitation with a peak that had a 3-h phase lag. A minority of SREF members performed well.

Full access
Caren Marzban, Ranran Wang, Fanyou Kong, and Stephen Leyton

Abstract

The rank histogram (RH) is a visual tool for assessing the reliability of ensemble forecasts (i.e., the degree to which the forecasts and the observations have the same distribution). But it is already known that in certain situations it conveys misleading information. Here, it is shown that a temporal correlation can lead to a misleading RH, but such a correlation contributes only to the sampling variability of the RH, and so it is accounted for by producing a RH that explicitly displays sampling variability. A simulation is employed to show that the variance within each ensemble member (i.e., climatological variance), the correlation between ensemble members, and the correlation between the observations and the forecasts, all have a confounding effect on the RH, making it difficult to use the RH for assessing the climatological component of forecast reliability. It is proposed that a “residual” quantile–quantile plot (denoted R-Q-Q plot) is better suited than the RH for assessing the climatological component of forecast reliability. Then, the RH and R-Q-Q plots for temperature and wind speed forecasts at 90 stations across the continental United States are computed. A wide range of forecast reliability is noted. For some stations, the nonreliability of the forecasts can be attributed to bias and/or under-or overclimatological dispersion. For others, the difference between the distributions can be traced to lighter or heavier tails in the distributions, while for other stations the distributions of the forecasts and the observations appear to be completely different. A spatial signature is also noted and discussed briefly.

Full access
Aaron Johnson, Xuguang Wang, Ming Xue, and Fanyou Kong

Abstract

Twenty-member real-time convection-allowing storm-scale ensemble forecasts with perturbations to model physics, dynamics, initial conditions (IC), and lateral boundary conditions (LBC) during the NOAA Hazardous Weather Testbed Spring Experiment provide a unique opportunity to study the relative impact of different sources of perturbation on convection-allowing ensemble diversity. In Part II of this two-part study, systematic similarity/dissimilarity of hourly precipitation forecasts among ensemble members from the spring season of 2009 are identified using hierarchical cluster analysis (HCA) with a fuzzy object-based threat score (OTS), developed in . In addition to precipitation, HCA is also performed on ensemble forecasts using the traditional Euclidean distance for wind speed at 10 m and 850 hPa, and temperature at 500 hPa.

At early lead times (3 h, valid at 0300 UTC) precipitation forecasts cluster primarily by data assimilation and model dynamic core, indicating a dominating impact of models, with secondary clustering by microphysics. There is an increasing impact of the planetary boundary layer (PBL) scheme on clustering relative to the microphysics scheme at later lead times. Forecasts of 10-m wind speed cluster primarily by the PBL scheme at early lead times, with an increasing impact of LBC at later lead times. Forecasts of midtropospheric variables cluster primarily by IC at early lead times and LBC at later lead times. The radar and Mesonet data assimilation (DA) show its impact, with members without DA in a distinct cluster, through the 12-h lead time (valid at 1200 UTC) for both precipitation and nonprecipitation variables. The implication for optimal ensemble design for storm-scale forecasts is also discussed.

Full access
Rebecca Cintineo, Jason A. Otkin, Ming Xue, and Fanyou Kong

Abstract

In this study, the ability of several cloud microphysical and planetary boundary layer parameterization schemes to accurately simulate cloud characteristics within 4-km grid-spacing ensemble forecasts over the contiguous United States was evaluated through comparison of synthetic Geostationary Operational Environmental Satellite (GOES) infrared brightness temperatures with observations. Four double-moment microphysics schemes and five planetary boundary layer (PBL) schemes were evaluated. Large differences were found in the simulated cloud cover, especially in the upper troposphere, when using different microphysics schemes. Overall, the results revealed that the Milbrandt–Yau and Morrison microphysics schemes tended to produce too much upper-level cloud cover, whereas the Thompson and the Weather Research and Forecasting Model (WRF) double-moment 6-class (WDM6) microphysics schemes did not contain enough high clouds. Smaller differences occurred in the cloud fields when using different PBL schemes, with the greatest spread in the ensemble statistics occurring during and after daily peak heating hours. Results varied somewhat depending upon the verification method employed, which indicates the importance of using a suite of verification tools when evaluating high-resolution model performance. Finally, large differences between the various microphysics and PBL schemes indicate that large uncertainties remain in how these schemes represent subgrid-scale processes.

Full access
Aaron Johnson, Xuguang Wang, Fanyou Kong, and Ming Xue

Abstract

Convection-allowing ensemble forecasts with perturbations to model physics, dynamics, and initial (IC) and lateral boundary conditions (LBC) generated by the Center for the Analysis and Prediction of Storms for the NOAA Hazardous Weather Testbed (HWT) Spring Experiments provide a unique opportunity to understand the relative impact of different sources of perturbation on convection-allowing ensemble diversity. Such impacts are explored in this two-part study through an object-oriented hierarchical cluster analysis (HCA) technique.

In this paper, an object-oriented HCA algorithm, where the dissimilarity of precipitation forecasts is quantified with a nontraditional object-based threat score (OTS), is developed. The advantages of OTS-based HCA relative to HCA using traditional Euclidean distance and neighborhood probability-based Euclidean distance (NED) as dissimilarity measures are illustrated by hourly accumulated precipitation ensemble forecasts during a representative severe weather event.

Clusters based on OTS and NED are more consistent with subjective evaluation than clusters based on traditional Euclidean distance because of the sensitivity of Euclidean distance to small spatial displacements. OTS improves the clustering further compared to NED. Only OTS accounts for important features of precipitation areas, such as shape, size, and orientation, and OTS is less sensitive than NED to precise spatial location and precipitation amount. OTS is further improved by using a fuzzy matching method. Application of OTS-based HCA for regional subdomains is also introduced. Part II uses the HCA method developed in this paper to explore systematic clustering of the convection-allowing ensemble during the full 2009 HWT Spring Experiment period.

Full access
Marc Berenguer, Madalina Surcel, Isztar Zawadzki, Ming Xue, and Fanyou Kong

Abstract

This second part of a two-paper series compares deterministic precipitation forecasts from the Storm-Scale Ensemble Forecast System (4-km grid) run during the 2008 NOAA Hazardous Weather Testbed (HWT) Spring Experiment, and from the Canadian Global Environmental Multiscale (GEM) model (15 km), in terms of their ability to reproduce the average diurnal cycle of precipitation during spring 2008. Moreover, radar-based nowcasts generated with the McGill Algorithm for Precipitation Nowcasting Using Semi-Lagrangian Extrapolation (MAPLE) are analyzed to quantify the portion of the diurnal cycle explained by the motion of precipitation systems, and to evaluate the potential of the NWP models for very short-term forecasting.

The observed diurnal cycle of precipitation during spring 2008 is characterized by the dominance of the 24-h harmonic, which shifts with longitude, consistent with precipitation traveling across the continent. Time–longitude diagrams show that the analyzed NWP models partially reproduce this signal, but show more variability in the timing of initiation in the zonal motion of the precipitation systems than observed from radar.

Traditional skill scores show that the radar data assimilation is the main reason for differences in model performance, while the analyzed models that do not assimilate radar observations have very similar skill.

The analysis of MAPLE forecasts confirms that the motion of precipitation systems is responsible for the dominance of the 24-h harmonic in the longitudinal range 103°–85°W, where 8-h MAPLE forecasts initialized at 0100, 0900, and 1700 UTC successfully reproduce the eastward motion of rainfall systems. Also, on average, MAPLE outperforms radar data assimilating models for the 3–4 h after initialization, and nonradar data assimilating models for up to 5 h after initialization.

Full access
Fanyou Kong, Kelvin K. Droegemeier, and Nicki L. Hickmon

Abstract

Using a nonhydrostatic numerical model with horizontal grid spacing of 24 km and nested grids of 6- and 3-km spacing, the authors employ the scaled lagged average forecasting (SLAF) technique, developed originally for global and synoptic-scale prediction, to generate ensemble forecasts of a tornadic thunderstorm complex that occurred in north-central Texas on 28–29 March 2000. This is the first attempt, to their knowledge, in applying ensemble techniques to a cloud-resolving model using radar and other observations assimilated within nonhorizontally uniform initial conditions and full model physics. The principal goal of this study is to investigate the viability of ensemble forecasting in the context of explicitly resolved deep convective storms, with particular emphasis on the potential value added by fine grid spacing and probabilistic versus deterministic forecasts. Further, the authors focus on the structure and growth of errors as well as the application of suitable quantitative metrics to assess forecast skill for highly intermittent phenomena at fine scales.

Because numerous strategies exist for linking multiple nested grids in an ensemble framework with none obviously superior, several are examined, particularly in light of how they impact the structure and growth of perturbations. Not surprisingly, forecast results are sensitive to the strategy chosen, and owing to the rapid growth of errors on the convective scale, the traditional SLAF methodology of age-based scaling is replaced by scaling predicated solely upon error magnitude. This modification improves forecast spread and skill, though the authors believe errors grow more slowly than is desirable.

For all three horizontal grid spacings utilized, ensembles show both qualitative and quantitative improvement relative to their respective deterministic control forecasts. Nonetheless, the evolution of convection at 24- and 6-km spacings is vastly different from, and arguably inferior to, that at 3 km because at 24-km spacing, the model cannot explicitly resolve deep convection while at 6 km, the deep convection closure problem is ill posed and clouds are neither implicitly nor explicitly represented (even at 3-km spacing, updrafts and downdrafts only are marginally resolved). Despite their greater spatial fidelity, the 3-km grid spacing experiments are limited in that the ensemble mean reflectivity tends to be much weaker in intensity, and much broader in aerial extent, than that of any single 3-km spacing forecast owing to amplitude reduction and spatial smearing that occur when averaging is applied to spatially intermittent phenomena. The ensemble means of accumulated precipitation, on the other hand, preserve peak intensity quite well.

Although a single case study obviously does not provide sufficient information with which to draw general conclusions, the results presented here, as well as those in Part II (which focuses solely on 3-km grid spacing experiments), suggest that even a small ensemble of cloud-resolving forecasts may provide greater skill, and greater practical value, than a single deterministic forecast using either the same or coarser grid spacing.

Full access
Jiangshan Zhu, Fanyou Kong, Lingkun Ran, and Hengchi Lei

Abstract

To study the impact of training sample heterogeneity on the performance of Bayesian model averaging (BMA), two BMA experiments were performed on probabilistic quantitative precipitation forecasts (PQPFs) in the northern China region in July and August of 2010 generated from an 11-member short-range ensemble forecasting system. One experiment, as in many conventional BMA studies, used an overall training sample that consisted of all available cases in the training period, while the second experiment used stratified sampling BMA by first dividing all available training cases into subsamples according to their ensemble spread, and then performing BMA on each subsample. The results showed that ensemble spread is a good criterion to divide ensemble precipitation cases into subsamples, and that the subsamples have different statistical properties. Pooling the subsamples together forms a heterogeneous overall sample. Conventional BMA is incapable of interpreting heterogeneous samples, and produces unreliable PQPF. It underestimates the forecast probability at high-threshold PQPF and local rainfall maxima in BMA percentile forecasts. BMA with stratified sampling according to ensemble spread overcomes the problem reasonably well, producing sharper predictive probability density functions and BMA percentile forecasts, and more reliable PQPF than the conventional BMA approach. The continuous ranked probability scores, Brier skill scores, and reliability diagrams of the two BMA experiments were examined for all available forecast days, along with a logistic regression experiment. Stratified sampling BMA outperformed the raw ensemble and conventional BMA in all verifications, and also showed better skill than logistic regression in low-threshold forecasts.

Full access