ALBERT

Description: This paper presents a method to bias correct and downscale wind speed over water bodies that are unresolved by numerical weather prediction (NWP) models and analyses. The dependency of wind speeds over water bodies to fetch length is investigated as a predictor of model wind speed error. Because model bias is found to be related to the forecast wind direction, a statistical method that uses the forecast fetch to remove wind speed bias is developed and tested. The method estimates wind speed bias using recent forecast errors from similar stations (i.e., those with comparable fetch lengths). As a result, the bias correction is not tied to local observations but instead to locations with similar land–water characteristics. Thus, it can also be used to downscale wind fields over inland and coastal water bodies. The fetch method is compared to four reference bias correction methods using one year’s worth of wind speed output from three NWP analyses in Florida. The fetch method yields a bias error near zero and results in a reduction of the mean absolute error that is comparable to the reference methods. The fetch method is then used to bias correct and downscale a coarse analysis to 500-m grid spacing over a coastal estuary in central Florida.

Description: A computationally efficient method is developed that performs gridded postprocessing of ensemble 10-m wind vector forecasts. An expansive set of idealized WRF Model simulations are generated to provide physically consistent, high-resolution winds over a coastal domain characterized by an intricate land/water mask. The ensemble model output statistics (EMOS) technique is used to calibrate the ensemble wind vector forecasts at observation locations. The local EMOS predictive parameters (mean and variance) are then spread throughout the grid utilizing flow-dependent statistical relationships extracted from the downscaled WRF winds. In a yearlong study, the method is applied to 24-h wind forecasts from the Global Ensemble Forecast System (GEFS) at 28 east-central Florida stations. Compared to the raw GEFS, the approach improves both the deterministic and probabilistic forecast skill. Analysis of multivariate rank histograms indicates that the postprocessed forecasts are calibrated. A downscaling case study illustrates the method as applied to a quiescent easterly flow event. Strengths and weaknesses of the approach are presented and discussed.

Description: A sea surface temperature (SST) analysis system designed to initialize short-term atmospheric model forecasts is evaluated for a month-long, relatively clear period in May 2004. System inputs include retrieved SSTs from the Geostationary Operational Environmental Satellite (GOES)-East and the Moderate Resolution Imaging Spectroradiometer (MODIS). The GOES SSTs are processed via a sequence of quality control and bias correction steps and are then composited. The MODIS SSTs are bias corrected and checked against the background field (GOES composites) prior to assimilation. Buoy data, withheld from the analyses, are used to bias correct the MODIS and GOES SSTs and to evaluate both the composites and analyses. The bias correction improves the identification of residual cloud-contaminated MODIS SSTs. The largest analysis system improvements are obtained from the adjustments associated with the creation of the GOES composites (i.e., a reduction in buoy/GOES composite rmse on the order of 0.3°–0.5°C). A total of 120 analyses (80 night and 40 day) are repeated for different experimental configurations designed to test the impact of the GOES composites, MODIS cloud mask, spatially varying background error covariance and decorrelation length scales, data reduction, and anisotropy. For the May 2004 period, the nighttime MODIS cloud mask is too conservative, at times removing good SST data and degrading the analyses. Nocturnal error variance estimates are approximately half that of the daytime and are relatively spatially homogeneous, indicating that the nighttime composites are, in general, superior. A 30-day climatological SST gradient is used to create anisotropic weights and a spatially varying length scale. The former improve the analyses in regions with significant SST gradients and sufficient data while the latter reduces the analysis rmse in regions where the innovations tend to be well correlated with distinct and persistent SST gradients (e.g., Loop Current). Data thinning reduces the rmse by expediting analysis convergence while simultaneously enhancing the computational efficiency of the analysis system. Based on these findings, an operational analysis configuration is proposed.

Description: High- and low-resolution sea surface temperature (SST) analysis products are used to initialize the Weather Research and Forecasting (WRF) Model for May 2004 for short-term forecasts over Florida and surrounding waters. Initial and boundary conditions for the simulations were provided by a combination of observations, large-scale model output, and analysis products. The impact of using a 1-km Moderate Resolution Imaging Spectroradiometer (MODIS) SST composite on subsequent evolution of the marine atmospheric boundary layer (MABL) is assessed through simulation comparisons and limited validation. Model results are presented for individual simulations, as well as for aggregates of easterly- and westerly-dominated low-level flows. The simulation comparisons show that the use of MODIS SST composites results in enhanced convergence zones, earlier and more intense horizontal convective rolls, and an increase in precipitation as well as a change in precipitation location. Validation of 10-m winds with buoys shows a slight improvement in wind speed. The most significant results of this study are that 1) vertical wind stress divergence and pressure gradient accelerations across the Florida Current region vary in importance as a function of flow direction and stability and 2) the warmer Florida Current in the MODIS product transports heat vertically and downwind of this heat source, modifying the thermal structure and the MABL wind field primarily through pressure gradient adjustments.

Description: Probabilistic wind speed forecasts for tropical cyclones from Monte Carlo–type simulations are assessed within a theoretical framework for a simple unbiased Gaussian system that is based on feature size and location error that mimic tropical cyclone wind fields. Aspects of the wind speed probability data distribution, including maximum expected probability and forecast skill, are assessed. Wind speed probability distributions are shown to be well approximated by a bounded power-law distribution when the feature size is smaller than the location error and tends toward a U-shaped distribution as the location error becomes small. Forecast skill (i.e., true and Heidke skill scores) is shown to be highly dependent on the probability forecast data distribution. Forecasts from the National Hurricane Center (NHC) Wind Speed Probability Forecast Product are used to assess the applicability of the simple system in the interpretation and evaluation of a more advanced system.

Description: A wind-wave forecast system, designed with the intention of generating unbiased ensemble wave forecasts for extreme wind events, is assessed. Wave hindcasts for 12 tropical cyclones (TCs) are forced using a wind analysis produced from a combination of the North American Regional Reanalysis (NARR) and a parametric wind model. The default drag parameterization is replaced by one that is more in line with recent studies where a cap at weak-to-moderate wind speeds is applied. Quadrant-based significant wave height (Hs) statistics are composited in a storm-relative reference frame and stratified by the radius of maximum wind, storm speed, and storm intensity. Improvements in Hs are gleaned from both downscaling the NARR winds and tuning the wave model. However, the paradigm whereby the drag coefficient depends solely on the wind speed is limiting. Results indicate that Hs is biased low in the right quadrants (for all statistical subcategories). Conversely, Hs is high biased in the left-rear quadrant even though the analysis wind field is underforecast there. At radii less than 100 nautical miles, the model peak wave direction is offset from the observed, with the model (buoy) peak more in line with (to the left of) the direction of the tropical cyclone motion. As a result, the predominant storm-relative wind direction, which is northwesterly in the left-rear quadrant, opposes that of the buoy peak wave direction, while the model peak is more crosswise with respect to the wind. This will likely reduce the magnitude of the wind stress in the model.

Description: A computationally efficient method of producing tropical cyclone (TC) wind analyses is developed and tested, using a hindcast methodology, for 12 Gulf of Mexico storms. The analyses are created by blending synthetic data, generated from a simple parametric model constructed using extended best-track data and climatology, with a first-guess field obtained from the NCEP–NCAR North American Regional Reanalysis (NARR). Tests are performed whereby parameters in the wind analysis and vortex model are varied in an attempt to best represent the TC wind fields. A comparison between nonlinear and climatological estimates of the TC size parameter indicates that the former yields a much improved correlation with the best-track radius of maximum wind rm. The analysis, augmented by a pseudoerror term that controls the degree of blending between the NARR and parametric winds, is tuned using buoy observations to calculate wind speed root-mean-square deviation (RMSD), scatter index (SI), and bias. The bias is minimized when the parametric winds are confined to the inner-core region. Analysis wind statistics are stratified within a storm-relative reference frame and by radial distance from storm center, storm intensity, radius of maximum wind, and storm translation speed. The analysis decreases the bias and RMSD in all quadrants for both moderate and strong storms and is most improved for storms with an rm of less than 20 n mi. The largest SI reductions occur for strong storms and storms with an rm of less than 20 n mi. The NARR impacts the analysis bias: when the bias in the former is relatively large, it remains so in the latter.

Description: A tropical cyclone (TC) wind speed probability forecast product developed at the Cooperative Institute for Research in the Atmosphere (CIRA) and adopted by the National Hurricane Center (NHC) is evaluated for U.S. land-threatening and landfalling events over four hurricane seasons from 2004 to 2007. A key element of this work is the discernment of risk associated with the interval forecast probabilities for the three wind speed categories (i.e., 34, 50, and 64 kt, where 1 kt = 0.52 m s−1). A quantitative assessment of the interval probabilities (0–12, 12–24, 24–36, 36–48, 48–72, 72–96, and 96–120 h) is conducted by converting them into binary (yes–no) forecasts using decision thresholds that are selected using the true skill statistic (TSS) and the Heidke skill score (HSS). The NHC product performs well as both the HSS and TSS demonstrate skill out to the 48–72- and 72–120-h intervals, respectively. Overall, reliability diagrams and bias scores indicate that the NHC product has a tendency to overforecast event likelihood for cases where the forecast probabilities exceed 60%. Specifically, the NHC product tends to overforecast for the 34-kt category but underforecasts for the 64-kt category, especially at later forecast intervals. Results for the 50-kt category are mixed but also exhibit a tendency to underforecast during the latter intervals. Decision thresholds range from 1% to 55% depending on the selection method, wind speed category, and time interval. Given that the average forecast probabilities decrease with forecast hour, small forecast probabilities may be meaningful. The HSS is recommended over the TSS for decision threshold selection because the use of the TSS introduces significant bias and the HSS is less sensitive to filtering of correct negatives.

Description: Data reduction tools are developed and evaluated using a data analysis framework. Simple (nonadaptive) and intelligent (adaptive) thinning algorithms are applied to both synthetic and real data and the thinned datasets are ingested into an analysis system. The approach is motivated by the desire to better represent high-impact weather features (e.g., fronts, jets, cyclones, etc.) that are often poorly resolved in coarse-resolution forecast models and to efficiently generate a set of initial conditions that best describes the current state of the atmosphere. As a precursor to real-data applications, the algorithms are applied to one- and two-dimensional synthetic datasets. Information gleaned from the synthetic experiments is used to create a thinning algorithm that combines the best aspects of the intelligent methods (i.e., their ability to detect regions of interest) while reducing the impacts of spatial irregularities in the data. Both simple and intelligent thinning algorithms are then applied to Atmospheric Infrared Sounder (AIRS) temperature and moisture profiles. For a given retention rate, background, and observation error, the optimal 1D analyses (i.e., lowest MSE) tend to have observations that are near regions of large curvature and gradients. Observation error leads to the selection of spurious data in homogeneous regions of the intelligent algorithms. In the 2D experiments, simple thinning tends to perform better within the homogeneous data regions. Analyses produced using AIRS data demonstrate that observations selected via a combination of the simple and intelligent approaches reduce clustering, provide a more even distribution along the satellite swath edges, and, in general, have lower error and comparable computational requirements compared to standard operational thinning methodologies.