NCEP Global Ensemble Forecast System Research Articles

Evaluation of Multiweek Tropical Cyclone Forecasts in the Philippines

Abstract In the pursuit of providing tropical cyclone (TC) forecasts beyond the conventional time scales covered by weather forecasting in the Philippines, this study has examined the multiweek (i.e., from week 1 to week 4) TC forecast skill in the country. TC forecasts derived from three ensemble models, namely, the NCEP Climate Forecast System version 2 (CFSv2), the European Centre for Medium-Range Weather Forecasts Ensemble Prediction System (ECMWF), and the NCEP Global Ensemble Forecast System version 12 (GEFSv12) from 6 October 2020 to 31 October 2021 were verified. Results revealed that the ECMWF model is consistently the most skillful in multiweek TC prediction over the domain bounded by 110°–155°E and 0°–27°N in the western North Pacific. The ECMWF obtained hit rates ranging from 0.25 to 0.31, low false alarm rates of 0–0.33, and the highest equitable threat scores among the models. In contrast to this, the GEFSv12 and CFSv2 models had varying skills, with the former performing better in the first two weeks and the latter in longer lead times. It is further revealed that the three models generally underestimate the observed number of storms, storm days, and accumulated cyclone energy. Moreover, the study shows that the forecast TC tracks have a significant (p < 0.05) positional bias toward the right of observed tracks beyond week 1, and that they tend to propagate slower than observations especially in week 1 and week 2. These findings contribute to better understanding the strengths and limitations of these ensemble models useful for eventual provision of multiweek TC forecasts in the Philippines.

Development and Evaluation of Multi-week Tropical Cyclone Strike Probability Forecasts in the Philippines

Tropical cyclone (TC) forecast, provided a few weeks in advance, can be beneficial for the preparation and mitigation of disaster risks in TC-vulnerable countries such as the Philippines. In this study, TC strike probability forecasts with a lead time of up to 4 wk were derived by combining the TC tracks from each ensemble member of the three models – namely, NCEP Coupled Forecast System version 2 (CFSv2), European Centre for Medium-Range Weather Forecasts Ensemble Prediction System (ECMWF), and NCEP Global Ensemble Forecast System version 12 (GEFSv12) – to show how likely it is for a TC to form or strike an area within the 300-km radius of the TC center. To assess its performance, real-time-derived TC strike probability forecasts over the Tropical Cyclone Information Domain (TCID) of the Philippines (bounded by 0–27 °N and 110–155 °E) covering the period from 06 Oct 2020–31 Oct 2022 were evaluated. Verification metrics revealed that the skill and reliability of the TC forecasts vary with lead time and the TC being forecasted. Week 1 forecasts are reliable and can be helpful for decision-making, whereas Week 2 forecasts are considered most reliable only up to the 51–60% probability interval. On the other hand, forecasts with 3–4-wk lead times are reliable for probabilities less than 20%. The case study performed using five TCs with different intensity classifications has shown that generally, forecasts for TCs with stronger intensity have higher skill than forecasts for relatively weaker ones. It was also observed that the magnitude of the probability values varies with the intensity changes within the validity period. These findings suggest that multi-model ensemble forecasts can be utilized for the improvement and eventual operationalization of multi-week TC strike probability forecasts over the Philippines.

Evaluating GFS and ECMWF Ensemble Forecasts of Integrated Water Vapor Transport along the U.S. West Coast

Abstract Atmospheric rivers (ARs) are long and narrow regions in the atmosphere of enhanced integrated water vapor transport (IVT) and can produce extreme precipitation and high societal impacts. Reliable and skillful forecasts of landfalling ARs in the western United States are critical to hazard preparation and aid in decision support activities, such as Forecast-Informed Reservoir Operations (FIRO). The purpose of this study is to compare the cool-season water year skill of the NCEP Global Ensemble Forecast System (GEFS) and ECMWF Ensemble Prediction System (EPS) forecasts of IVT along the U.S. West Coast for 2017–20. The skill is analyzed using probability-over-threshold forecasts of IVT magnitudes ≥ 250 kg m−1 s−1 (P250) using contingency table skill metrics in coastal Northern California and along the west coast of North America. Analysis of P250 with lead time (dProg/dt) found that the EPS provided ∼1 day of additional lead time for situational awareness over the GEFS at lead times of 6–10 days. Forecast skill analysis highlights that the EPS leads over the GEFS with success ratios 0.10–0.15 higher at lead times > 6 days for P250 thresholds of ≥25% and ≥50%, while event-based skill analysis using the probability of detection (POD) found that both models were largely similar with minor latitudinal variations favoring higher POD for each model in different locations along the coast. The relative skill of the EPS over the GEFS is largely attributed to overforecasting by the GEFS at longer lead times and an increase in the false alarm ratio. Significance Statement The purpose of this study is to evaluate the efficacy of the NCEP Global Ensemble Forecast System (GEFS) and the ECMWF Ensemble Prediction System (EPS) in forecasting enhanced water vapor transport along the U.S. West Coast commonly associated with landfalling atmospheric rivers and heavy precipitation. The ensemble models allow us to calculate the probability that enhanced water vapor transport will occur, thereby providing situational awareness for decision-making, such as in hazard mitigation and water resource management. The results of this study indicate that the EPS model is on average more skillful than the GEFS model at lead times of ∼6–10 days with a higher success ratio and lower false alarm ratio.

The Development of the NCEP Global Ensemble Forecast System Version 12

Abstract The Global Ensemble Forecast System (GEFS) is upgraded to version 12, in which the legacy Global Spectral Model (GSM) is replaced by a model with a new dynamical core—the Finite Volume Cubed-Sphere Dynamical Core (FV3). Extensive tests were performed to determine the optimal model and ensemble configuration. The new GEFS has cubed-sphere grids with a horizontal resolution of about 25 km and an increased ensemble size from 20 to 30. It extends the forecast length from 16 to 35 days to support subseasonal forecasts. The stochastic total tendency perturbation (STTP) scheme is replaced by two model uncertainty schemes: the stochastically perturbed physics tendencies (SPPT) scheme and stochastic kinetic energy backscatter (SKEB) scheme. Forecast verification is performed on a period of more than two years of retrospective runs. The results show that the upgraded GEFS outperforms the operational-at-the-time version by all measures included in the GEFS verification package. The new system has a better ensemble error–spread relationship, significantly improved skills in large-scale environment forecasts, precipitation probability forecasts over CONUS, tropical cyclone track and intensity forecasts, and significantly reduced 2-m temperature biases over North America. GEFSv12 was implemented on 23 September 2020.

A Summary of GFS Ensemble Integrated Water Vapor Transport Forecasts and Skill along the U.S. West Coast during Water Years 2017–20

AbstractThe ability to provide accurate forecasts and improve situational awareness of atmospheric rivers (ARs) is key to impact-based decision support services and applications such as forecast-informed reservoir operations. The purpose of this study is to quantify the cool-season water year skill for 2017–20 of the NCEP Global Ensemble Forecast System forecasts of integrated water vapor transport along the U.S. West Coast commonly observed during landfalling ARs. This skill is summarized for ensemble probability-over-threshold forecasts of integrated water vapor transport magnitudes ≥ 250 kg m−1 s−1 (referred to as P250). The P250 forecasts near North-Coastal California at 38°N, 123°W were reliable and successful at lead times of ~8–9 days with an average success ratio > 0.5 for P250 forecasts ≥ 50% at lead times of 8 days and Brier skill scores > 0.1 at a lead time of 8–9 days. Skill and accuracy also varied as a function of latitude and event characteristics. The highest (lowest) success ratios and probability of detection values for P250 forecasts ≥ 50% occurred on average across Northern California and Oregon (Southern California), whereas the average probability of detection of more intense and longer duration landfalling ARs was 0.1–0.2 higher than weaker and shorter duration events at lead times of 3–9 days. The potential for these forecasts to enhance situational awareness may also be improved, depending on individual applications, by allowing for flexibility in the location and time of verification; the success ratios increased 10%–30% at lead times of 5–10 days allowing for flexibility of ±1.0° latitude and ±6 h in verification.

Evaluation of Tropical Cyclone Genesis Precursors with Relative Operating Characteristics (ROC) in Highresolution Ensemble Forecasts: Hurricane Ernesto

ABSTRACT Identifying the environmental conditions that control tropical cyclone (TC) genesis is a challenging problem. This study examines a new method to evaluate the precursors of TC genesis using high-resolution ensemble forecasts and relative operating characteristic (ROC) diagrams. With an advanced research version of the Weather Research and Forecasting (WRF) model, high-resolution ensemble forecasts (at 5 km horizontal resolution) are conducted in various configurations using a bred vector method to form a set of 140 ensemble members for predicting Hurricane Ernesto’s genesis. Basic evaluation shows that high-resolution ensemble forecasts are able to predict well-developed TCs, whereas the NCEP Global Ensemble Forecast System (GEFS) fails to do so. This set of 140 ensemble members is employed to study the precursors of Hurricane Ernesto’s genesis by contrasting the genesis and nongenesis cases. Specifically, ROC curves, composite figures for genesis and nongenesis cases, and Kolmogorov-Smirnov tests are applied to characterize the relationship between important environmental parameters near the beginning of the simulation and genesis likelihood 15-18 h later. It is found that moist conditions at 850 hPa, vertical wind shear, the strength of the 850 hPa pre existing wave, and upper-level warming play notable roles in Ernesto’s genesis.

Evaluating the MJO prediction skill from different configurations of NCEP GEFS extended forecast

NOAA is accelerating its efforts to improve the numerical guidance and prediction capability for extended range (weeks 3 and 4) prediction in its seamless forecast system. Madden Julian Oscillation (MJO) is the dominant mode of sub-seasonal variability in tropics and prediction skill of MJO is investigated in this paper. We used different configurations of the NCEP Global Ensemble Forecast System (GEFS) to perform the experiments. The configurations include (1) the operational version of the stochastic perturbation forced with operational Sea Surface Temperatures (SSTs); (2) an updated stochastic physics forced with operational SSTs; (3) an updated stochastic physics forced with bias-corrected SSTs that are from Climate Forecast System (Version 2); and (4) as in (3) but with the addition of a scale aware-convection scheme. We evaluated MJO prediction skill from the experiments using Wheeler–Hendon indices and also examined the performance of the forecast system on prediction of key MJO components. We found that using the updated stochastic scheme improved the MJO prediction lead-time by about 4 days. Further updating the underlying SSTs with the bias corrected CFSv2 forecast increased the MJO prediction lead time by another 1.7 days. The best configuration of the four experiments is the last configuration which extends forecast lead time by ~ 9 days. Further investigation shows that upper and lower level zonal wind over the tropics has larger improvement than the outgoing longwave radiation (OLR). The improvement of the MJO prediction skill appears to be related primarily to the improvement in the representation associated circulations and OLR over the tropical West Pacific.

Predictive Skill and Predictability of North Atlantic Tropical Cyclogenesis in Different Synoptic Flow Regimes

Practical predictability of tropical cyclogenesis over the North Atlantic is evaluated in different synoptic flow regimes using the NCEP Global Ensemble Forecast System (GEFS) reforecasts with forecast lead time up to two weeks. Synoptic flow regimes are represented by tropical cyclogenesis pathways defined in a previous study based on the low-level baroclinicity and upper-level forcing of the genesis environmental state, including nonbaroclinic, low-level baroclinic, trough-induced, weak tropical transition (TT), and strong TT pathways. It is found that the strong TT and weak TT pathways have lower predictability than the other pathways, linked to the lower predictability of vertical wind shear and midlevel humidity in the genesis vicinity of a developing TT storm. Further analysis suggests that stronger extratropical influences contribute to lower genesis predictability. It is also shown that the regional and seasonal variations of the genesis predictive skill in the GEFS can be largely explained by the relative frequency of occurrence of each pathway and the predictability differences among pathways. Predictability of tropical cyclogenesis is further discussed using the concept of the genesis potential index.

Performance of the New NCEP Global Ensemble Forecast System in a Parallel Experiment

Abstract A new version of the Global Ensemble Forecast System (GEFS, v11) is tested and compared with the operational version (v10) in a 2-yr parallel run. The breeding-based scheme with ensemble transformation and rescaling (ETR) used in the operational GEFS is replaced by the ensemble Kalman filter (EnKF) to generate initial ensemble perturbations. The global medium-range forecast model and the Global Forecast System (GFS) analysis used as the initial conditions are upgraded to the GFS 2015 implementation version. The horizontal resolution of GEFS increases from Eulerian T254 (~52 km) for the first 8 days of the forecast and T190 (~70 km) for the second 8 days to semi-Lagrangian T574 (~34 km) and T382 (~52 km), respectively. The sigma pressure hybrid vertical layers increase from 42 to 64 levels. The verification of geopotential height, temperature, and wind fields at selected levels shows that the new GEFS significantly outperforms the operational GEFS up to days 8–10 except for an increased warm bias over land in the extratropics. It is also found that the parallel system has better reliability in the short-range probability forecasts of precipitation during warm seasons, but no clear improvement in cold seasons. There is a significant degradation of TC track forecasts at days 6–7 during the 2012–14 TC seasons over the Atlantic and eastern Pacific. This degradation is most likely a sampling issue from a low number of TCs during these three TC seasons. The results for an extended verification period (2011–14) and the recent two hurricane seasons (2015 and 2016) are generally positive. The new GEFS became operational at NCEP on 2 December 2015.

A Comparison of Perturbations from an Ensemble Transform and an Ensemble Kalman Filter for the NCEP Global Ensemble Forecast System

Abstract Two perturbation generation schemes, the ensemble transformation with rescaling (ETR) and the ensemble Kalman filter (EnKF), are compared for the NCEP operational environment for the Global Ensemble Forecast System (GEFS). Experiments that utilize each of the two schemes are carried out and evaluated for two boreal summer seasons. It is found that these two schemes generally have comparable performance. Experiments utilizing both perturbation methods fail to generate sufficient spread at medium-range lead times beyond day 8. In general, the EnKF-based experiment outperforms the ETR in terms of the continuous ranked probability skill score (CRPSS) in the Northern Hemisphere (NH) for the first week. In the SH, the ensemble mean forecast is more skillful from the ETR perturbations. Additional experiments are performed with the stochastic total tendency perturbation (STTP) scheme, in which the total tendencies of all model variables are perturbed to represent the uncertainty in the forecast model. An improved spread–error relationship is found for the ETR-based experiments, but the STTP increases the ensemble spread for the EnKF-based experiment that is already overdispersive at early lead times, especially in the SH. With STTP employed, an increase in the EnKF-based CRPSS in the NH is reduced with a larger degradation in both the probability and ensemble-mean forecast skills in the SH. The results indicate that a rescaling of the EnKF initial perturbations and/or tuning of the STTP scheme is required when STTP is applied using the EnKF-based perturbations. This study provided guidance for the replacement of ETR with EnKF perturbations as part of the 2015 GEFS implementation.

Sensitivity of Calibrated Week-2 Probabilistic Forecast Skill to Reforecast Sampling of the NCEP Global Ensemble Forecast System

AbstractCPC requires the reforecast-calibrated Global Ensemble Forecast System (GEFS) to support the production of their official 6–10- and 8–14-day temperature and precipitation forecasts. While a large sample size of forecast–observation pairs is desirable to generate the necessary model climatology and variances, and covariances to observations, sampling by reforecasts could be done to use available computing resources most efficiently. A series of experiments was done to assess the impact on calibrated forecast skill of using a smaller sample size than the current available reforecast dataset. This study focuses on the skill of week-2 probabilistic forecasts of the 7-day-mean 2-m temperature and accumulated precipitation. The tercile forecasts are expressed as being below-, near-, and above-normal temperature/median precipitation over the continental United States (CONUS). Calibration statistics were calculated using an ensemble regression technique from 25 yr of daily, 11-member GEFS reforecasts for 1986–2010, which were then used to postprocess the GEFS model forecasts for 2011–13. In assessing the skill of calibrated model output using a reforecast dataset with fewer years and ensemble members, and an ensemble run less frequently than daily, it was determined that reductions in the number of ensemble members to six or fewer and reductions in the frequency of reforecast runs from daily to once a week were achievable with minimal loss of skill. However, reducing the number of years of reforecasts to less than 25 resulted in a greater skill degradation. The loss of skill was statistically significant using only 18 yr of reforecasts from 1993 to 2010 to generate model statistics.

Climatology and Environmental Characteristics of Extreme Precipitation Events in the Southeastern United States

Abstract This paper documents the characteristics of extreme precipitation events (EPEs) in the southeastern United States (SEUS) during 2002–11. The EPEs are identified by applying an object-based method to 24-h precipitation analyses from the NCEP stage-IV dataset. It is found that EPEs affected the SEUS in all months and occurred most frequently in the western portion of the SEUS during the cool season and in the eastern portion during the warm season. The EPEs associated with tropical cyclones, although less common, tended to be larger in size, more intense, and longer lived than “nontropical” EPEs. Nontropical EPEs in the warm season, relative to those in the cool season, tended to be smaller in size and typically involved more moist, conditionally unstable conditions but weaker dynamical influences. Synoptic-scale composites are constructed for nontropical EPEs stratified by the magnitude of vertically integrated water vapor transport (IVT) to examine distinct scenarios for the occurrence of EPEs. The composite results indicate that “strong IVT” EPEs occur within high-amplitude flow patterns involving strong transport of moist, conditionally unstable air within the warm sector of a cyclone, whereas “weak IVT” EPEs occur within low-amplitude flow patterns featuring weak transport but very moist and conditionally unstable conditions. Finally, verification of deterministic precipitation forecasts from a reforecast dataset based on the NCEP Global Ensemble Forecast System reveals that weak-IVT EPEs were characteristically associated with lower forecast skill than strong-IVT EPEs. Based on these results, it is suggested that further research should be conducted to investigate the forecast challenges associated with EPEs in the SEUS.

Multivariate Probabilistic Analysis and Predictability of Medium-Range Ensemble Weather Forecasts

Abstract Ensemble weather forecasting has been operational for two decades now. However, the related uncertainty analysis in terms of probabilistic postprocessing still focuses on single variables, grid points, or stations. Inevitable dependencies in space and time and between variables are often ignored. To address this problem, two probabilistic postprocessing methods are presented, which are multivariate versions of Gaussian fit and kernel dressing, respectively. The multivariate case requires the estimation of a full rank, invertible covariance matrix. For this purpose, a Graphical Least Absolute Shrinkage and Selection Operators (GLASSO) estimator has been employed that is based on sparse undirected graphical models regularized by an L1 penalty term in order to parameterize the full rank inverse covariance. In all cases, the result is a multidimensional probability density. The forecasts used to test the approach are station forecasts of 2-m temperature and surface pressure from four main global ensemble prediction systems (EPS) with medium-range weather forecasts: the NCEP Global Ensemble Forecast System (GEFS), the Met Office Global and Regional Ensemble Prediction System (MOGREPS), the Canadian Meteorological Centre (CMC) Global Ensemble Prediction System (GEPS), and the ECMWF EPS. To evaluate the multivariate probabilistic postprocessing, especially the uncertainty estimates, common verification methods such as the analysis rank histogram and the continuous ranked probability score (CRPS) are applied. Furthermore, a multivariate extension of the CRPS, the energy score, allows for the verification of a complete medium-range forecast as well as for determining its predictability. It is shown that the predictability is similar for all of the examined ensemble prediction systems, whereas the GLASSO proved to be a useful tool for calibrating the commonly observed underdispersion of ensemble forecasts during the first few lead days by using information from the full covariance matrix.

Ensemble Transform with 3D Rescaling Initialization Method

Abstract The ensemble transform with rescaling (ETR) method has been used to produce fast-growing components of analysis error in the NCEP Global Ensemble Forecast System (GEFS). The rescaling mask contained in the ETR method constrains the amplitude of perturbations to reflect regional variations of analysis error. However, because of a lack of suitable three-dimensional (3D) analysis error estimation, in the operational GEFS the mask is based on the estimated analysis error at 500 hPa and is not flow dependent but changes monthly. With the availability of an ensemble-based data assimilation system at NCEP, a 3D mask can be computed. This study generates initial perturbations by the ensemble transform with 3D rescaling (ET_3DR) and compares the performance with the ETR. Meanwhile, the ET_3DR is also applied within the ensemble Kalman filter (EnKF) method (hereafter EnKF_3DR). Results from a set of experiments indicate that the 3D mask suppresses perturbations less in unstable regions. Relative to the ETR, the large amplitudes of the ET_3DR initial perturbations at 500 hPa better reflect areas of baroclinic instability over the extratropics and deep convection over the tropics. Furthermore, the maxima of the vertical distribution for the ET_3DR initial perturbations correspond to the heights of the subtropical westerly and tropical easterly jet regions. Such perturbations produce faster spread growths. Results with EnKF_3DR also show benefits from an orthonormalization by the ensemble transform algorithm and amplitude constraint by the 3D mask rescaling. Thus, the EnKF_3DR forecasts outperform the EnKF.

Skill of the MJO and Northern Hemisphere Blocking in GEFS Medium-Range Reforecasts

Abstract Forecast characteristics of Northern Hemisphere atmospheric blocking and the Madden–Julian oscillation (MJO) were diagnosed using an extensive time series (December–February 1985–2012) of daily medium-range ensemble reforecasts based on a version of the NCEP Global Ensemble Forecast System (GEFS). For blocking, (i) interannual variability of analyzed blocking frequency was quite large, (ii) the GEFS slightly underforecasted blocking frequency at longer leads in the Euro-Atlantic sector, (iii) predictive skill of actual blocking was substantially smaller than its perfect-model skill, (iv) block onset and cessation were forecast less well than overall blocking frequency, (v) there was substantial variability of blocking skill between half-decadal periods, and (vi) the reliability of probabilistic blocking forecasts degraded with increasing lead time. For the MJO, (i) forecasts of strong Indian Ocean MJOs propagated too slowly, especially the component associated with outgoing longwave radiation (OLR), that is, convection; (ii) tropical precipitation was greatly overforecast at early lead times; (iii) the ensemble predictions were biased and/or underdispersive, manifested in U-shaped rank histograms of MJO indices (magnitude forecasts were especially U shaped); (iv) MJO correlation skill was larger for its wind than for its OLR component, and was larger for the higher-amplitude MJO events; (v) there was some half-decadal variability in skill; and (vi) probabilistic skill of the MJO forecast was modest, and skill was larger when measured relative to climatology than when measured relative to a lagged persistence forecast. For longer-lead forecasts, the GEFS demonstrated little ability to replicate the changes in blocking frequency due to a strong MJO that were noted in analyzed data.

Tropical Cyclone–like Vortices Detection in the NCEP 16-Day Ensemble System over the Western North Pacific in 2008: Application and Forecast Evaluation

Abstract An automated technique has been developed for the detection and tracking of tropical cyclone–like vortices (TCLVs) in numerical weather prediction models, and especially for ensemble-based models. A TCLV is detected in the model grid when selected dynamic and thermodynamic fields meet specified criteria. A backward-and-forward extension from the mature stage of the track is utilized to complete the track. In addition, a fuzzy logic approach is utilized to calculate the TCLV fuzzy combined-likelihood value (TFCV) for representing the TCLV characteristics in the ensemble forecast outputs. The primary objective of the TCLV tracking and TFCV maps is for use as an evaluation tool for the operational forecasters. It is demonstrated that this algorithm efficiently extracts western North Pacific TCLV information from the vast amount of ensemble data from the NCEP Global Ensemble Forecast System (GEFS). The predictability of typhoon formation and activity during June–December 2008 is also evaluated. The TCLV track numbers and TFCV averages around the formation locations during the 0–96-h period are more skillful than for the 102–384-h forecasts. Compared to weak tropical cyclones (TCs; maximum intensity ≤ 50 kt), the storms that eventually become stronger TCs do have larger TFCVs. Depending on the specified domain size and the ensemble track numbers to define a forecast event, some skill is indicated in predicting the named TC activity. Although this evaluation with the 2008 typhoon season indicates some potential, an evaluation with a larger sample is necessary to statistically verify the reliability of the GEFS forecasts.

Tracking and Verification of East Atlantic Tropical Cyclone Genesis in the NCEP Global Ensemble: Case Studies during the NASA African Monsoon Multidisciplinary Analyses

Abstract This study evaluates the performance of the NCEP global ensemble forecast system in predicting the genesis and evolution of five named tropical cyclones and two unnamed nondeveloping tropical systems during the NASA African Monsoon Multidisciplinary Analyses (NAMMA) between August and September 2006. The overall probabilities of the ensemble forecasts of tropical cyclone genesis are verified relative to a genesis time defined to be the first designation of the tropical depression from the National Hurricane Center (NHC). Additional comparisons are also made with high-resolution deterministic forecasts from the NCEP Global Forecast System (GFS). It is found that the ensemble forecasts have high probabilities of genesis for the three strong storms that formed from African easterly waves, but failed to accurately predict the pregenesis phase of two weaker storms that formed farther west in the Atlantic Ocean. The overall accuracy for the genesis forecasts is above 50% for the ensemble forecasts initialized in the pregenesis phase. The forecast uncertainty decreases with the reduction of the forecast lead time. The probability of tropical cyclone genesis reaches nearly 90% and 100% for the ensemble forecasts initialized near and in the postgenesis phase, respectively. Significant improvements in the track forecasts are found in the ensemble forecasts initialized in the postgenesis phase, possibly because of the implementation of the NCEP storm relocation scheme, which provides an accurate initial storm position for all ensemble members. Even with coarser resolution (T126L28 for the ensemble versus T384L64 for the GFS), the overall performance of the ensemble in predicting tropical cyclone genesis is compatible with the high-resolution deterministic GFS. In addition, false alarm rates for nondeveloping waves were low in both the GFS and ensemble forecasts.

A Nonparametric Postprocessor for Bias Correction of Hydrometeorological and Hydrologic Ensemble Forecasts

AbstractThis paper describes a technique for quantifying and removing biases from ensemble forecasts of hydrometeorological and hydrologic variables. The technique makes no a priori assumptions about the distributional form of the variables, which is often unknown or difficult to model parametrically. The aim is to estimate the conditional cumulative distribution function (ccdf) of the observed variable given a (possibly biased) real-time ensemble forecast. This ccdf represents the “true” probability distribution of the forecast variable, subject to sampling uncertainties. In the absence of a known distributional form, the ccdf should be estimated nonparametrically. It is noted that the probability of exceeding a threshold of the observed variable, such as flood stage, is equivalent to the expectation of an indicator variable defined for that threshold. The ccdf is then modeled through a linear combination of the indicator variables of the forecast ensemble members. The technique is based on Bayesian optimal linear estimation of indicator variables and is analogous to indicator cokriging (ICK) in geostatistics. By developing linear estimators for the conditional expectation of the observed variable at many thresholds, ICK provides a discrete approximation of the full ccdf. Since ICK minimizes the conditional error variance of the indicator variable at each threshold, it effectively minimizes the continuous ranked probability score (CRPS) when infinitely many thresholds are employed. The technique is used to bias-correct precipitation ensemble forecasts from the NCEP Global Ensemble Forecast System (GEFS) and streamflow ensemble forecasts from the National Weather Service (NWS) River Forecast Centers (RFCs). Split-sample validation results are presented for several attributes of ensemble forecast quality, including reliability and discrimination. In general, the forecast biases were substantially reduced following ICK. Overall, the technique shows significant potential for bias-correcting ensemble forecasts whose distributional form is unknown or nonparametric.