• Alves, J., , Y. Y. Chao, , and H. L. Tolman, 2005: The operational North Atlantic hurricane wind-wave forecasting system at NOAA/NCEP. Tech. Note 244, NOAA/NWS/NCEP/MMAB, 59 pp. + appendices.

  • Alves, J., and Coauthors, 2013: The NCEP–FNMOC combined wave ensemble product: Expanding benefits of interagency probabilistic forecasts to the oceanic environment. Bull. Amer. Meteor. Soc., 94, 1893–1905, doi:10.1175/BAMS-D-12-00032.1.

    • Search Google Scholar
    • Export Citation
  • Alves, J., , A. Chawla, , H. Tolman, , D. Schwab, , G. Lang, , and G. Mann, 2014: The operational implementation of a Great Lakes wave forecasting system at NOAA/NCEP. Wea. Forecasting, 29, 1473–1497, doi:10.1175/WAF-D-12-00049.1.

    • Search Google Scholar
    • Export Citation
  • Amante, C., , and B. W. Eakins, 2009: ETOPO1 1 arc-minute global relief model: Procedures, data sources and analysis. Tech. Memo. 24, NOAA/NESDIS/NGDC, 19 pp.

  • Ardhuin, F., and Coauthors, 2010: Semiempirical dissipation source functions for ocean waves. Part I: Definition, calibration, and validation. J. Phys. Oceanogr., 40, 19171941, doi:10.1175/2010JPO4324.1.

    • Search Google Scholar
    • Export Citation
  • Blake, E. S., , T. B. Kimberlain, , R. J. Berg, , J. Cangialosi, , and J. L. Beven II, 2013: Tropical cyclone report: Hurricane Sandy (22–29 October 2012). National Hurricane Center Rep. AL182012, 157 pp. [Available online at http://www.nhc.noaa.gov/data/tcr/AL182012_Sandy.pdf.]

  • Cao, D., , H. S. Chen, , and H. L. Tolman, 2007: Verification of ocean wave ensemble forecasts at NCEP. Proc. 10th Int. Workshop on Wave Hindcasting and Forecasting and Coastal Hazards Symp., Oahu, HI, NOAA/NCEP, Paper G1, 10 pp. [Available online at http://polar.ncep.noaa.gov/mmab/papers/tn261/mmab261.pdf.]

  • Cardone, V., , R. Jensen, , D. Resio, , V. Swail, , and A. Cox, 1996: Evaluation of contemporary ocean wave models in rare extreme events: The “Halloween storm” of October 1991 and the Storm of the Century of March 1993. J. Atmos. Oceanic Technol., 13, 198230, doi:10.1175/1520-0426(1996)013<0198:EOCOWM>2.0.CO;2.

    • Search Google Scholar
    • Export Citation
  • Chao, Y. Y., , and H. L. Tolman, 2010: Performance of NCEP regional wave models in predicting peak sea states during the 2005 North Atlantic hurricane season. Wea. Forecasting, 25, 15431567, doi:10.1175/2010WAF2222309.1.

    • Search Google Scholar
    • Export Citation
  • Chao, Y. Y., , J.-H. G. Alves, , and H. L. Tolman, 2005: An operational system for predicting hurricane-generated wind waves in the North Atlantic Ocean. Wea. Forecasting, 20, 652671, doi:10.1175/WAF851.1.

    • Search Google Scholar
    • Export Citation
  • Chawla, A., and Coauthors, 2013: A multigrid wave forecasting model: A new paradigm in operational wave forecasting. Wea. Forecasting, 28, 1057–1078, doi:10.1175/WAF-D-12-00007.1.

    • Search Google Scholar
    • Export Citation
  • Chen, H. S., 2006: Ensemble prediction of ocean waves at NCEP. Proc. 28th Ocean Engineering Conf., Taiwan, National Sun Yat-Sen University, 18.

  • Division of Natural Sciences and Mathematics, 2013: Beach-dune performance assessment after Hurricane Sandy related to FEMA disaster. Annual Rep. DR-NJ 4086, Division of Natural Sciences and Mathematics, Richard Stockton College, New Jersey, 26 pp.

  • Fanelli, C., , P. Fanelli, , and D. Wolcott, 2013: Hurricane Sandy. Water level and meteorological data report, National Ocean Services, NOS/NOAA, 60 pp.

  • Funakoshi, Y., , J. Feyen, , F. Aikman, , H. Tolman, , A. van der Westhuysen, , A. Chawla, , I. Rivin, , and A. Taylor, 2012: Development of Extratropical Surge and Tide Operational Forecast System (ESTOFS). Proc. 12th ASCE Int. Conf. on Estuarine and Coastal Modeling, St. Augustine, FL, ASCE, 201212, doi: 10.1061/9780784412411.00012.

  • Gladwin, H., , B. Morrow, , and J. Lazo, 2013: Communication and understanding of Hurricane Sandy storm surge forecast and warning information. Florida International University Natural Hazards Workshop, NSF CMMI 1322088, Miami, FL, Florida International University. [Available online at http://www.academia.edu/4316097/Communication_and_Understanding_of_Hurricane_Sandy_Storm_Surge_Forecast_and_Warning_Information.]

  • Hsu, S., 2013: Storm surges in New York during Hurricane Sandy in 2012: A verification of the wind-stress tide relation. Bound.-Layer Meteor., 148, 593598, doi:10.1007/s10546-013-9830-1.

    • Search Google Scholar
    • Export Citation
  • Impact Forecasting, 2013: Hurricane Sandy event recap report. Rep., Aon Benfield, 50 pp. [Available online at http://thoughtleadership.aonbenfield.com/Documents/20130514_if_hurricane_sandy_event_recap.pdf.]

  • Kurihara, Y., , M. A. Bender, , R. E. Tuleya, , and R. J. Ross, 1995: Improvements in the GFDL hurricane prediction system. Mon. Wea. Rev., 123, 27912801, doi:10.1175/1520-0493(1995)123<2791:IITGHP>2.0.CO;2.

    • Search Google Scholar
    • Export Citation
  • Landsea, C. W., , and J. L. Franklin, 2013: Atlantic hurricane database uncertainty and presentation of a new database format. Mon. Wea. Rev., 141, 35763592, doi:10.1175/MWR-D-12-00254.1.

    • Search Google Scholar
    • Export Citation
  • Magnusson, L., , J.-R. Bidlot, , S. T. Lang, , A. Thorpe, , N. Wedi, , and M. Yamaguchi, 2014: Evaluation of medium-range forecasts for Hurricane Sandy. Mon. Wea. Rev., 142, 19621981, doi:10.1175/MWR-D-13-00228.1.

    • Search Google Scholar
    • Export Citation
  • McCallum, B. E., and Coauthors, 2013: Monitoring storm tide and flooding from Hurricane Sandy along the Atlantic coast of the United States, October 2012. U.S. Geological Survey Open-File Rep. 2013–1043, 42 pp. [Available online at http://pubs.usgs.gov/of/2013/1043/.]

  • Sela, J. G., 1980: Spectral modeling at the National Meteorological Center. Mon. Wea. Rev., 108, 12791292, doi:10.1175/1520-0493(1980)108<1279:SMATNM>2.0.CO;2.

    • Search Google Scholar
    • Export Citation
  • Skamarock, W. C., and Coauthors, 2008: A description of the Advanced Research WRF version 3. NCAR Tech. Note NCAR/TN-475+STR, 113 pp. [Available online at http://www.mmm.ucar.edu/wrf/users/docs/arw_v3_bw.pdf.]

  • Sullivan, D., , and L. Uccellini, 2013: Hurricane/post-Tropical Cyclone Sandy, October 22–29, 2012. NOAA Service Assessment, NOAA, 46 pp. + appendices.

  • Tallapragada, V., , N. Surgi, , Q. Liu, , Y. Kwon, , R. Tuleya, , and W. O’Connor, 2008: Performance of the advanced operational HWRF modeling system during pre-implementation testing and in real-time 2007 hurricane season. 28th Conf. on Hurricanes and Tropical Meteorology, Orlando, FL, Amer. Meteor. Soc., 4A.5. [Available online at https://ams.confex.com/ams/28Hurricanes/techprogram/paper_138066.htm.]

  • Tolman, H. L., 2009: User manual and system documentation of WAVEWATCH III version 3.14. Tech. Note 276, NOAA/NWS/NCEP/MMAB, 194 pp. + appendices.

  • Tolman, H. L., , and D. Chalikov, 1996: Source terms in a third-generation wind wave model. J. Phys. Oceanogr., 26, 24972518, doi:10.1175/1520-0485(1996)026<2497:STIATG>2.0.CO;2.

    • Search Google Scholar
    • Export Citation
  • Van der Westhuysen, A. J., and Coauthors, 2014: Development and validation of the nearshore wave prediction system. 11th Symp. on the Coastal Environment, Austin, TX, Amer. Meteor. Soc., 4.5. [Available online at https://ams.confex.com/ams/93Annual/webprogram/Paper222877.html.]

  • Wei, M., , Z. Toth, , R. Wobus, , and Y. Zhu, 2008: Initial perturbations based on the ensemble transform (ET) technique in the NCEP global operational forecast system. Tellus, 60A, 6279, doi:10.1111/j.1600-0870.2007.00273.x.

    • Search Google Scholar
    • Export Citation
  • View in gallery

    Hurricane Sandy’s best track. Also indicated are the NDBC buoys selected for wave model validation, as follows: 41002 (A), 41009 (B), 41010 (C), 41012 (D), 41013 (E), 41046 (F), 41047 (G), 41048 (H), 44005 (I), 44008 (J), 44098 (K), and 44025 (L). Alphabetic labels in parentheses are used to identify these buoys in Fig. 3.

  • View in gallery

    Time series of modeled and measured at NDBC buoys 41001, 41048, 44008, and 44025. Model data are from full forecast cycles approximately 72 and 24 h before the storm peak at each location, as indicated under respective horizontal axis labels. Buoy data are hourly, whereas model data are 3-hourly.

  • View in gallery

    Target plots showing the normalized bias of maximum relative to NDBC buoys (vertical axis), and the lag between its modeled and observed time of occurrence (horizontal axis). Solid and dashed lines indicate mean values and standard deviations, respectively. Results for (top) GWGFS, (middle) GWHUR, and (bottom) GWES. Buoy labels are as in Fig. 1.

  • View in gallery

    ESA poststorm wind speed analysis (m s−1) at 2305 UTC 27 Oct, from the SMOS mission. Sandy’s trajectory is indicated by the thin black line.

  • View in gallery

    NHC sea-state analysis for 0600 UTC 28 Oct overlayed on 0655 UTC GOES-East IR satellite image. GWGFS contours every 3 ft are in white, observations from regional buoys are in yellow, and measurements from 0655 UTC Jason-1/2 altimeter passes, with the color legend at top. The red hurricane symbol represents the location of Sandy.

  • View in gallery

    NHC poststorm analysis of at 0000 UTC 28 Oct. The contours of 12 ft (yellow line), 20 ft (magenta line), and 30 ft (red stippled area).

  • View in gallery

    GWHUR H00 forecast at 1200 UTC 2 Sep 2010 for Hurricane Earl (in ft).

  • View in gallery

    Swath of maximum at four forecast ranges prior to 2330 UTC 20 Oct 2012, when Sandy made landfall on the New Jersey coast: (top left) 96, (top right) 72, (bottom left) 48, and (bottom right) 24 h. The shown swath is simply the largest found at any grid point over the entire simulation period at a given cycle (180 h for GWGFS and GWHUR; 240 h for GWES).

  • View in gallery

    Time series of (left) and (right) from the NCEP operational system in 2012 (OPER) and 2014 (HIRES). Ensemble mean from the GWES at forecast ranges approximately (top) 72 and (bottom) 48 h before the peak waves reached NDBC buoys (top two rows) 41048 (western Bermuda) and (bottom two rows) 44008 (southern Nantucket). These two buoys recorded the highest waves during the passage of the hurricane near the eastern coast of the United States.

All Time Past Year Past 30 Days
Abstract Views 0 0 0
Full Text Views 44 44 5
PDF Downloads 29 29 4

Operational Wave Guidance at the U.S. National Weather Service during Tropical/Post–Tropical Storm Sandy, October 2012

View More View Less
  • 1 Systems Research Group, Inc., and Environmental Modeling Center, NOAA/NCEP, College Park, Maryland
  • 2 National Hurricane Center, NOAA/NCEP, Miami, Florida
  • 3 Environmental Modeling Center, NOAA/NCEP, College Park, Maryland
© Get Permissions
Full access

Abstract

Waves generated during Hurricane Sandy (October 2012) contributed significantly to life and property losses along the eastern U.S. seaboard. Extreme waves generated by Sandy propagated inland riding high water levels, causing direct destruction of property and infrastructure. High waves also contributed to the observed record-breaking storm surges. Operational wave-model guidance provided by the U.S. National Weather Service, via numerical model predictions made at NOAA’s National Centers for Environmental Prediction (NCEP), gave decision makers accurate information that helped mitigate the severity of this historical event. The present study provides a comprehensive performance assessment of operational models used by NCEP during Hurricane Sandy, and makes a brief review of reports issued by government agencies, private industry, and universities, indicating the importance of the interplay of waves and surges during the hurricane. Performance of wave models is assessed through validation made relative to western Atlantic NOAA/NDBC buoys that recorded significant wave heights exceeding 6 m (19.7 ft). Bulk validation statistics indicate a high skill of operational wave forecasts up to and beyond the 3-day range. Event-based validation reveals a remarkably high skill of NCEP’s wave ensemble system, with significant added value in its data for longer forecasts beyond the 72-h range. The study concludes with considerations about the extent of severe sea-state footprints during Sandy, the dissemination of real-time wave forecasts, and its impacts to emergency management response, as well as recent upgrades and future developments at NCEP that will improve the skill of its current wave forecasting systems, resulting in more reliable wave forecasts during life-threatening severe storm events in the future.

National Centers for Environmental Prediction Marine Modeling and Analysis Branch Contribution Number 323.

Corresponding author address: Jose-Henrique G. M. Alves, Environmental Modeling Center, NOAA/NCEP, Center for Weather and Climate Prediction, 5830 University Research Court, College Park, MD 20740. E-mail: henrique.alves@noaa.gov

Abstract

Waves generated during Hurricane Sandy (October 2012) contributed significantly to life and property losses along the eastern U.S. seaboard. Extreme waves generated by Sandy propagated inland riding high water levels, causing direct destruction of property and infrastructure. High waves also contributed to the observed record-breaking storm surges. Operational wave-model guidance provided by the U.S. National Weather Service, via numerical model predictions made at NOAA’s National Centers for Environmental Prediction (NCEP), gave decision makers accurate information that helped mitigate the severity of this historical event. The present study provides a comprehensive performance assessment of operational models used by NCEP during Hurricane Sandy, and makes a brief review of reports issued by government agencies, private industry, and universities, indicating the importance of the interplay of waves and surges during the hurricane. Performance of wave models is assessed through validation made relative to western Atlantic NOAA/NDBC buoys that recorded significant wave heights exceeding 6 m (19.7 ft). Bulk validation statistics indicate a high skill of operational wave forecasts up to and beyond the 3-day range. Event-based validation reveals a remarkably high skill of NCEP’s wave ensemble system, with significant added value in its data for longer forecasts beyond the 72-h range. The study concludes with considerations about the extent of severe sea-state footprints during Sandy, the dissemination of real-time wave forecasts, and its impacts to emergency management response, as well as recent upgrades and future developments at NCEP that will improve the skill of its current wave forecasting systems, resulting in more reliable wave forecasts during life-threatening severe storm events in the future.

National Centers for Environmental Prediction Marine Modeling and Analysis Branch Contribution Number 323.

Corresponding author address: Jose-Henrique G. M. Alves, Environmental Modeling Center, NOAA/NCEP, Center for Weather and Climate Prediction, 5830 University Research Court, College Park, MD 20740. E-mail: henrique.alves@noaa.gov

1. Introduction

Hurricane Sandy evolved from a classic late-season tropical cyclone that formed in the southwestern Caribbean Sea on 22 October 2014, and quickly intensified to a hurricane on 24 October, before crossing Jamaica (Blake et al. 2013). A series of interactions with upper-tropospheric troughs occurred as Sandy moved northward through the Bahamas, and then offshore of the East Coast of the United States, which led to both an unusual thermodynamic structure and significant expansion of the wind field. This contributed to Sandy becoming one of the largest Atlantic hurricanes on record.1 Sandy completed the transition to a post–tropical cyclone shortly before making landfall along the coast of New Jersey on 2330 UTC 29 October 2012. As it propagated northward across the western North Atlantic, Sandy left behind an unprecedented area larger than 500 000 km2 with significant wave heights exceeding 7–8 m, which peaked at more than 10 m near its radius of maximum winds. In association with record-breaking storm surges, extreme waves generated by the massive wind field of Sandy contributed significantly to the destruction of property along the coasts of New Jersey and New York.

Inspection of data disseminated by the Federal Emergency Management Agency (FEMA) and the U.S. Geophysical Survey (USGS) suggest that a massive increase in water levels caused by Hurricane Sandy’s passage, ranging from 1.5 to 3 m in a vast extension of the Delaware, New Jersey, New York, and Connecticut coasts, allowed penetration of waves as far as 10 km inland in several coastal sites (McCallum et al. 2013). As a consequence, extreme waves generated by Sandy propagated inland riding the unusually high water levels, ultimately causing direct destruction of property and infrastructure. High waves also contributed directly to the observed record-breaking storm surges, as exemplified in Hsu (2013). Presenting a simple study of the contributions of Sandy’s extreme waves to storm surges observed in the New York Harbor, that author concludes that 93% of the total observed surge was caused by the convergence of high waves at the harbor entrance.

Operational wave model guidance provided by the National Weather Service (NWS), using numerical model predictions made at NOAA’s National Centers for Environmental Prediction (NCEP), gave decision makers accurate information that helped mitigate the severity of this historical event. Early reports issued by NOAA evaluating forecasts and observations made during Hurricane/post–Tropical Storm Sandy (Blake et al. 2013; Fanelli et al. 2013), while focusing on storm surge predictions and damage, have not provided much insight on the importance of wave forecasts to prevention, emergency management operations, and mitigation of damage caused by Sandy to coastal communities. Therefore, in an effort to expand the record about forecasts and observations of relevant environmental conditions during Sandy, the present study provides a comprehensive assessment of the performance of operational models used for wave guidance issued by NCEP during Sandy, and discusses issues that became relevant in supporting emergency management services (EMS).

In section 2 we provide a brief review of reports issued by government agencies, private industry, and universities indicating the importance of the interplay of waves and surges during Sandy. Section 3 provides an overview of the wave model systems that were operational at NCEP during Sandy’s passage, and that supported wave guidance issued by the National Weather Service. Section 4 describes observations made during Sandy of severe sea-state conditions, focusing on wave heights exceeding 12 ft (3.66 m), which corresponds to a reference level used to define high-seas radii in NHC tropical cyclone marine advisories. Section 5 is dedicated to an assessment of model skill relative to measured data. A description of the spatial extent of severe sea states during Sandy, and a discussion about the challenges faced by NWS marine forecasters in their effort to disseminate high-quality wave guidance are presented in section 6. A brief overview of recent upgrades to NCEP’s operational models is given in section 7, followed by our conclusions.

2. Reports of wave damage

Reports issued by research institutions and private industry, mostly in the insurance sector, recognize that the main cause of unexpected destruction caused by Hurricane Sandy was primarily due to the combination of extreme waves riding on a massive storm surge. The more fundamental dynamical coupling of waves and surges, which led to compound destruction, is simple to understand: record-breaking storm surges caused by the superposition of high tides, combined with wind and wave setup, elevated the water line over land in the coastal zone, in a way that a massive surf zone with very large breaking waves developed, submerging urban areas along the eastern U.S. coast. Breaking waves relentlessly battered piers, bridges, subways, power lines, and several types of oceanfront property, including thousands of homes. But the interplay between waves and surge goes beyond simple superposition, and may have included additional processes, such as water level hikes due to wave setup and steepening of waves due to localized surge-induced currents, making matters worse during this massive storm.

Several private sector reports describe Sandy’s impact to the insurance sector in the United States. A typical example is the damage survey in the states of New York and New Jersey published by Impact Forecasting (2013). The report indicates that “record storm surge and excessive wave heights were the predominant cause of damage, as opposed to wind.” In the Caribbean region, the report describes that in Cuba, for example, “tremendous waves up to 10.0 meters in height and a two-meter storm surge caused substantial coastal flooding.” The Division of Natural Sciences and Mathematics, from The Richard Stockton College of New Jersey (Division of Natural Sciences and Mathematics 2013), provides a more detailed assessment of the destructive effects caused predominantly by waves riding on top of the surge, to the New Jersey coast. Particularly impressive is their description of the impact of waves on Monmouth County, a location just south of the New York Bay: “Damage seen in Deal and Elberon demanded that waves exceeded 30 feet in NAVD 88 elevation levels2 on breaking on the bluff. These huge breakers essentially bulldozed the berm, beach and irregular dune system all along the Monmouth county Atlantic shoreline. Damages to oceanfront property (public and private) increased dramatically northward.”

In contrast with earlier reports issued by NOAA which focused exclusively on storm surge/water levels (Blake et al. 2013; Fanelli et al. 2013), the more recent NOAA service assessment report by Sullivan and Uccellini (2013), recognizes the major role shared by waves and surges to make Sandy a particularly damaging event. The report describes that “damaging wave action atop the storm surge increased flooding and battered the south and east facing shores of Long Island and northern New Jersey, respectively.” In this context, the report further indicates that “many EMS expressed surprise at the large and damaging waves Sandy caused.” Citing a survey by Gladwin et al. (2013), indeed, Sullivan and Uccellini (2013) emphasize that of coastal residents surveyed after Sandy, 77% described the impact of waves as more than they expected, matching the 79% that thought the impact of surges was higher than expected. Such expectations reflect the fact that even small to moderate storm surges can cause life-threatening and damaging conditions whenever extreme waves approach the coast, riding atop the surge.

The report by Sullivan and Uccellini (2013) provides 23 recommendations to improve NWS operations and services in the context of extreme waves and surges caused by hurricanes. Results presented in the next sections provide useful data in a way that such recommendations may be expanded to better deal with the dissemination of information about damaging waves. Given the importance of waves in extreme storms such as Sandy, as reported above, such expanded recommendations, covering a more complete range of processes that are of importance to EMS and the general public, may help to increase the capabilities toward protecting life and property in future storms.

3. Operational wave models at NCEP

NCEP operational models were at the heart and forefront of the ocean and wave guidance provided during Hurricane Sandy to American forecasters, and were readily available to the general public. NCEP’s wave guidance for the Atlantic Ocean is made using three major wave forecasting systems:

  • A global wave model system forced with GFS winds (GWGFS).
  • A global wave model system forced with a blend of GFS and hurricane winds from the GFDL model (GWHUR).
  • A global wave ensemble system (GWES), with 21 members.

Both GWGFS and GWHUR run a mosaic spatial grid, with spatial resolutions ranging from 4 min near the U.S. coast to 30 min in deeper-water, offshore areas. During Sandy’s passage, the GWES was running a regular 1° global grid—spatial resolution has since then changed to ½°, as described in section 7. All models run four daily cycles, with GWHUR running out to 5 days, GWGFS to 7 days, and GWES to 10 days. At the center of NCEP’s operational wave forecasting systems are implementations of the wave model WAVEWATCH III (hereafter WW3; Tolman 2009), which all currently use regular grids on spherical coordinates (curvilinear and triangular grids are also available in WW3, but not used by NCEP’s operational models).

NCEP’s multigrid global wave model (GWGFS) was implemented operationally in November 2007. This single-system model consists of a mosaic of 9 two-way nested grids, with resolution varying from ½°–°. The GWGFS was designed to run side by side with the atmospheric GFS model, so that atmospheric and wave models step through time jointly, and the latter intakes 10-m winds from the former as they become available. The wave model also uses sea surface temperature and ice concentrations based on passive microwave analysis made at NCEP. A recent upgrade in May 2012 included changing the physics package of Tolman and Chalikov (1996) with the more recent parameterizations of Ardhuin et al. (2010). Detailed information on the system and validation results are provided in Chawla et al. (2013).

NCEP’s multigrid hurricane wave model (GWHUR) uses a grid layout similar to the global operational model GWGFS, with some minor differences. The Arctic Ocean grid is not used in this model; instead, there are two ¼° regional grids in the Atlantic and Pacific Oceans. The GWHUR system is driven by a combination of GFS and hurricane winds, which are provided from NCEP’s operational hurricane model, whenever active tropical systems are present. Details on interpolation of hurricane winds on GFS winds can be found in Chao et al. (2005). During Sandy’s passage, the GWHUR system was using the GFDL hurricane model (Kurihara et al. 1995) for prescribing its hurricane winds, alongside the Tolman and Chalikov (1996) wave-generation physics package (recent upgrades including changes in the prescribed hurricane winds and in the physics package are described in section 7). In the absence of hurricanes, the system is driven solely by GFS winds.

NCEP’s global wave ensemble system (GWES) was implemented operationally in 2008, consisting of a 20-member ensemble forced with NCEP-GEFS bias-corrected wind data and one control run with NCEP’s deterministic GFS fields. More detailed description of NCEP’s GEFS and GFS are provided in Wei et al. (2008) and Sela (1980), respectively. A description of the initial GWES at NCEP is provided in Chen (2006), whereas an early date validation study is presented in Cao et al. (2007). A more recent investigation on the bulk performance of NCEP’s GWES, in association with a multicenter probabilistic forecast product issued jointly by NCEP and the Fleet Numerical Meteorology and Oceanography Center (FNMOC), is described in Alves et al. (2013). Post-Sandy upgrades to the GWES are described in section 7.

4. Measured waves and validation methods

A total of 23 buoys maintained by NOAA’s National Data Buoy Center (NDBC) in the western North Atlantic, measured wave systems with significant wave heights larger than 6 m (19.7 ft) during Hurricane Sandy. Buoys satisfying this criterion, along with their geographical locations, depth, and maximum measured during Sandy are listed in Table 1. Subsets of buoys in Table 1 were selected for the purpose of providing data for two alternative validation approaches: a bulk assessment based on data at fixed forecast ranges (e.g., 0–120 h); and an event-based assessment focusing on time series of the storm at a particular wave model run times (the so-called forecast cycles, typically running at NCEP four times daily, at 0000, 0600, 1200, and 1800 UTC). Wind and wave data used for validation was obtained from NDBC’s quality-controlled historic archives—no further QC was performed.

Table 1.

NDBC buoys that recorded wave fields with larger than 6 m (19.7 ft) in the western North Atlantic Ocean. Buoys are ranked from largest to smallest measured maximum wave heights. Maximum values of and are provided. For buoys used in bulk validation (BV), the number of data points above 12 ft (3.66 m) N is provided, along with associated mean values for and (in parentheses). Dates of maximum values are provided as the day in October, with UTC times indicated in parenthesis.

Table 1.

The first subset, used for bulk validation, included buoys at locations represented with sufficient resolution in all wave model spatial grids, and had available both measurements of wind speeds and wave heights. Because of the relatively coarse resolution of the GWES, seven buoys were discarded from Table 1, as they were located too close to the coast. Of the remaining 15, 3 buoys did not measure wind speeds during Sandy (41001, 44024, and 44097), so that 12 were finally retained. To allow a more accurate representation of the storm time series, the second subset was generated from site-specific, point outputs made at 1-h intervals for a smaller set of NDBC buoys, which included 11 buoys listed in Table 1. Buoys from each subset are indicated in the table by abbreviations for bulk validation (BV) and time series (TS). Both subsets were representative of peak wave heights measured during the hurricane in terms of offshore and nearshore positions, geographical location and wave-height range above 6 m (19.7 ft).

The bulk assessment was made using measurements from buoys in subset 1 (BV). Model data were extracted at the 15 selected buoy sites by interpolating gridded outputs from neighboring points. Gridded outputs from operational wave models at NCEP are made at 3-h intervals, which made possible extracting data points set to represent several forecast ranges and the nowcast. Bulk validation below is provided for the 24- and 72-h forecast ranges. For ensemble forecast data (GWES), the ensemble mean is used hereafter in all comparisons with measurements and other forecast systems.

5. Validation of NCEP wave guidance

a. Bulk assessment

Validation of NCEP’s global operational wave forecasting systems during Hurricane Sandy is presented here relative to 12 buoys selected from those listed in Table 1. Only buoys that were properly represented in all NCEP’s wave model spatial grids, and that had simultaneous measurements of and were retained. The location of NDBC buoys is shown in Fig. 1. Bulk validation of model results was made relative to observations at these buoys representing waves higher than 12 ft (3.66 m), measured within 5 days before or after the storm peak at each location. As mentioned previously, wave model skill at or above the 12-ft (3.66 m) level is an important reference for high-seas wave guidance issued by NWS marine forecasters during tropical storms.

Fig. 1.
Fig. 1.

Hurricane Sandy’s best track. Also indicated are the NDBC buoys selected for wave model validation, as follows: 41002 (A), 41009 (B), 41010 (C), 41012 (D), 41013 (E), 41046 (F), 41047 (G), 41048 (H), 44005 (I), 44008 (J), 44098 (K), and 44025 (L). Alphabetic labels in parentheses are used to identify these buoys in Fig. 3.

Citation: Monthly Weather Review 143, 5; 10.1175/MWR-D-14-00143.1

Results for and are summarized in Tables 2 and 3, respectively, for the 24- and 72-h forecast ranges. Relative values for bias (model − observation) and RMS errors are normalized by the mean observed and , which are included for each selected buoy in Table 1. Biases relative to the mean from all models at most validation sites are smaller than 10%, for both winds and waves, at both 24- and 72-h forecast horizons. Despite the low relative bias levels, their generally negative values for (positive for ) indicate a consistent underestimation of waves (overestimation of winds) by NCEP models during this extreme event, a discrepancy that warrants further investigation in the future. RMS errors and scatter indices3 from all models at most validation sites are smaller than 20%, for both winds and waves, at both 24- and 72-h forecast horizons. Correlation values from all models are mostly higher than 80% for both winds and waves, at both 24- and 72-h forecast horizons. Some deterioration is seen in shallower sites, where coastal effects not fully represented in the models become dominant.

Table 2.

Validation statistics comparing wave model data relative to NDBC buoys: bias, root-mean-square error , scatter index (SI), and correlation coefficients (r). Values are given for the 24- and 72-h forecast ranges, the latter in parentheses. Bias and are normalized with mean values indicated in Table 1.

Table 2.
Table 3.

Validation statistics comparing wave model data relative to selected NDBC buoys: bias, root-mean-square error , scatter index (SI), and correlation coefficient (r). Values are given for the 24- and 72-h forecast ranges, the latter in parentheses. Bias and are normalized with mean values indicated in Table 1.

Table 3.

A consistent feature in Table 3 is the poorer performance of GWHUR’s predictions relative to the other two global wave systems at NCEP. This is a consequence of GWHUR at the time of Sandy’s passage, still using an older version of WW3 with the Tolman and Chalikov (1996) physics package, which is now known to perform poorly in severe sea states and extreme storms (see Alves et al. 2014; Ardhuin et al. 2010). This shortcoming has been minimized with a recent upgrade to the system made in late 2015, when its physics package became identical to that used in GWGFS (see section 7 for details).

Validation statistics shown in Tables 2 and 3 reflect operational forecasts as they were issued before Sandy’s passage through the buoy sites. They indicate a generally high skill of operational wave forecasts up to the 3-day horizon, and are comparable to results obtained in wave models used in poststorm assessments, such as Cardone et al. (1996), involving manual calibration of winds, massive data intake, and sometimes wave model tuning. The remarkably good skill of NCEP’s operational numerical weather prediction models in representing accurately winds and waves during this major event, had a significant impact to ensure high standards of wave guidance issued by the NWS, which assisted EMS in ensuring the safety of life and property in one of the most populated and commercially active areas in the United States.

b. Time series assessment

Time series validation is based on point output from wave models at NDBC buoy sites, consisting of the full simulated two-dimensional wave spectra interpolated from neighboring grid points to the buoy location. Buoys used in the time series validation are indicated in Table 1 by the subset tag “TS.” The availability of full two-dimensional wave model spectra at buoy locations allows computing from them integral parameters matching the algorithms used in buoy data processing, leading to more representative comparisons with measured data.

Figure 2 illustrates the skill of NCEP’s global operational wave models (GWGFS, GWHUR, and GWES) in representing the evolution of over time at the three NDBC buoys with the highest measured , that are also well resolved in all wave model grids. Model time series were extracted from full forecast cycles at approximately 72 and 24 h prior to the peak of the storm, at each location. The good agreement also shows the high skill level observed in the bulk validation statistics summarized in Table 3. Model skill at other significant forecast ranges (0 and 120 h, in addition to 24 and 72 h) is illustrated in Fig. 3, showing target plots—a specialized diagram first introduced in Alves et al. (2005), and later modified by Chao and Tolman (2010). Target plots aggregate for all buoys, the normalized bias of maximum relative to the buoy and the lag between the modeled and observed peak occurrence time.

Fig. 2.
Fig. 2.

Time series of modeled and measured at NDBC buoys 41001, 41048, 44008, and 44025. Model data are from full forecast cycles approximately 72 and 24 h before the storm peak at each location, as indicated under respective horizontal axis labels. Buoy data are hourly, whereas model data are 3-hourly.

Citation: Monthly Weather Review 143, 5; 10.1175/MWR-D-14-00143.1

Fig. 3.
Fig. 3.

Target plots showing the normalized bias of maximum relative to NDBC buoys (vertical axis), and the lag between its modeled and observed time of occurrence (horizontal axis). Solid and dashed lines indicate mean values and standard deviations, respectively. Results for (top) GWGFS, (middle) GWHUR, and (bottom) GWES. Buoy labels are as in Fig. 1.

Citation: Monthly Weather Review 143, 5; 10.1175/MWR-D-14-00143.1

Figure 3 shows that at the short forecast range (24 h), the two deterministic model runs GWGFS and GWHUR provide the most accurate estimate of maximum , with biases of the order of ±15%, and peak occurrence lags ranging from 0 to ±6 h. The GWES system also provides accurate representation of maximum , but provides a larger occurrence time error ranging from 0 to 10 h. Model performance is similar for all global systems at the 72-h forecast range, with typical biases in the ±20% range, and storm peak occurrence lags in the ±12-h range. Results from both deterministic models deteriorate significantly at the longer 120-h forecast range, whereas the GWES provides predictions of maximum and its arrival times that are of greater quality. For the latter case, biases range from −35% to 20%, whereas occurrence time lags range from −12 to 14 h.

A remarkable feature in all figures for our time-domain analysis is the high skill of NCEP’s wave ensemble system GWES, even running a grid with resolutions 2–7 times coarser than the other deterministic systems (GWGFS and GWHUR). Such outcome is supported by a recent investigation of waves during Sandy, reported by Magnusson et al. (2014). At short forecast ranges, the GWES provides similar quality results relative to those from GWGFS and GWHUR. For the longer forecast range (120 h), there is a significant added value relative to the deterministic model runs. Further improvements to GWES have been realized with a recent upgrade to that system, as discussed in section 7.

6. Anomalous wave fields and guidance

Sandy’s spatial wave-height distribution in the days leading up to landfall had several features that presented challenges to the NWS. Officially disseminated information on the expected impacts of storm surge and large waves in the coastal zone were made difficult, mostly as a consequence of the unusual shape of Sandy’s maximum wave-height footprint, as it evolved within the western North Atlantic basin, and the resulting maximum wave height swath integrated over its lifetime. These were difficult concepts to explain to NWS users, and some of the discussions that allowed a successful outcome of the guidance issued by the NWS during Sandy, are provided next.

a. Real-time wave forecasts

NCEP wave guidance for Sandy proved to be extremely reliable 3–5 days in advance of landfall, and allowed the NWS to provide EMS and decision makers with advanced warning as to the extreme and highly unusual wave conditions expected. The information dissemination process that led to such highly reliable forecasts was challenging, because Sandy’s wave-height spatial distribution was highly unusual when compared to other northward-moving Atlantic tropical cyclones on record. Such an anomalously large wave field, coupled with seasonally high astronomical tides, led to extreme storm surge forecasts that initially caused confusion among some decision makers and the general public, in the days leading up to landfall. Internal discussions at NHC as to how best convey this message, however, ultimately resulted in higher-quality guidance.

Sandy’s intensity peaked at 100 kt (51.44 m s−1), making it a category 3 hurricane, on 25 October just before reaching the southeast coast of Cuba. Sandy quickly weakened to a 60-kt (30.87 m s−1) tropical storm during the next 48 h as it exited Cuba and moved through the central and northwest Bahamas. In this scenario, large wave generation for a tropical cyclone moving northward out of the Caribbean and into the Atlantic is typically disrupted, because of wind and wave interaction with the islands. After moving north of the Bahamas, Sandy veered northeast and paralleled the U.S. Southeast coast on 26 and 27 October, and wave growth was rapidly reestablished across the open Atlantic. During this time, Sandy interacted with an upper-level trough moving slowly eastward across the Gulf of Mexico. This complex interaction resulted in a gradual change in Sandy’s thermodynamic structure, a doubling in size of the wind field since emerging off of Cuba (Blake et al. 2013), and resulted in Sandy regaining hurricane strength. The hybrid structure and large size of Sandy’s wind field during this period generated an extremely large and unusual wave field rarely observed with Atlantic hurricanes.

The European Space Agency (ESA) poststorm wind speed analysis at 2305 UTC 27 October, from the Soil Moisture and Ocean Salinity (SMOS) mission, shown in Fig. 4, reveals the hybrid structure of Sandy following this upper trough interaction, with maximum winds well removed from the center, and wrapped cyclonically around the west through south semicircles. Comparison of this wind field analysis to the corresponding field (Fig. 5) shows how well the GWGFS forecast correlated with the observed wind distribution, with greater than 25 ft (7.62 m) found within the long fetch of wind speeds of 52.5 kt (27 m s−1) and greater.

Fig. 4.
Fig. 4.

ESA poststorm wind speed analysis (m s−1) at 2305 UTC 27 Oct, from the SMOS mission. Sandy’s trajectory is indicated by the thin black line.

Citation: Monthly Weather Review 143, 5; 10.1175/MWR-D-14-00143.1

Fig. 5.
Fig. 5.

NHC sea-state analysis for 0600 UTC 28 Oct overlayed on 0655 UTC GOES-East IR satellite image. GWGFS contours every 3 ft are in white, observations from regional buoys are in yellow, and measurements from 0655 UTC Jason-1/2 altimeter passes, with the color legend at top. The red hurricane symbol represents the location of Sandy.

Citation: Monthly Weather Review 143, 5; 10.1175/MWR-D-14-00143.1

NHC poststorm analysis of at 0000 UTC 28 October (Fig. 6), provides more detail on the areal extent and spatial distribution of wave heights resultant from the hybrid structure and large wind field of Sandy. The maximum of 38 ft (11.58 m) was analyzed across the southeastern quadrant, and well removed from the center of Sandy, which is in stark contrast to typical recurving Atlantic hurricanes, such as Hurricane Earl (2010). The latter is used here for comparison, as a category 4 hurricane with large wind field and relatively similar location and motion as Sandy, and one would expect reasonably similar wave conditions in both cases. However, with Earl, the GWGFS 0-h forecast at 1200 UTC 2 September 2010 shows the typical wave distribution associated with recurving Atlantic hurricanes (Fig. 7), where the highest values are confined exclusively across the northeast semicircle, maximum is located near the center, and across the southwest semicircle are generally less than half these values. Comparison of the wave fields of Sandy and Earl at similar latitudes and with similar motions illustrates the importance of the size of Sandy’s wind field in wave generation (i.e., the large size of a much weaker tropical cyclone was sufficient to yield very long fetch lengths) and the extended time for extreme wave growth.

Fig. 6.
Fig. 6.

NHC poststorm analysis of at 0000 UTC 28 Oct. The contours of 12 ft (yellow line), 20 ft (magenta line), and 30 ft (red stippled area).

Citation: Monthly Weather Review 143, 5; 10.1175/MWR-D-14-00143.1

Fig. 7.
Fig. 7.

GWHUR H00 forecast at 1200 UTC 2 Sep 2010 for Hurricane Earl (in ft).

Citation: Monthly Weather Review 143, 5; 10.1175/MWR-D-14-00143.1

b. Extent of severe sea-state footprint

As shown in previous sections, NCEPs operational model data were in very good agreement with buoy measurements during the passage of Sandy over a widespread region. In particular, measured data confirms the occurrence of exceptionally high waves across a large area, also supporting wave model evidence of widespread severe sea states. Figure 8 shows a bird’s-eye view of wave conditions that prevailed during the passage of Hurricane Sandy in the northwest Atlantic Ocean, associated with the unusual severe sea-state footprint described above.

Fig. 8.
Fig. 8.

Swath of maximum at four forecast ranges prior to 2330 UTC 20 Oct 2012, when Sandy made landfall on the New Jersey coast: (top left) 96, (top right) 72, (bottom left) 48, and (bottom right) 24 h. The shown swath is simply the largest found at any grid point over the entire simulation period at a given cycle (180 h for GWGFS and GWHUR; 240 h for GWES).

Citation: Monthly Weather Review 143, 5; 10.1175/MWR-D-14-00143.1

Maximum values of significant wave heights estimated by NCEPs global wave forecasting system are shown at all wave model grid points, for the period 26 October, when the hurricane changed course and followed a path parallel to the coast of the United States, until it made landfall in the New Jersey coast, on 29 October 2012. The figure indicates that an area larger than 500 000 km2 was exposed to larger than 10 m (with a potential for maximum wave heights exceeding 19 m), and that the entire deep-water basin in the western Atlantic had been exposed, at some moment, by in excess of 6 m (19.7 ft), with a potential of maximum waves in excess of 11 m. Considering this, it is fair to say that the entirety of the northwest Atlantic experienced for some time severe sea states during Sandy’s lifetime.

The anomalous wave distribution of Sandy, with extreme wave heights found in all quadrants, resulted in strong and powerful wave fields with energy propagating outward in all directions, affecting the entire North Atlantic Ocean. Large and powerful surf was observed along the eastern seaboard of the United States for periods of 24–72 h, producing significant to catastrophic beach erosion and coastal inundation along the entire Atlantic coast of North America, from the upper Florida Keys to the Canadian Maritime Provinces, as well as the Bahamas and the islands of the northeast Caribbean.

7. Changes to NWS numerical models

Alongside the regular schedule of upgrades to numerical weather prediction models at NCEP, the challenges brought up by Hurricane Sandy’s passage in the western North Atlantic inspired several changes to the NWS’s wave forecasting systems, some of which have been implemented since late 2012. In this section we discuss changes that are relevant for wave guidance in extreme storm events such as Sandy, focusing on upgrades to the global deterministic and probabilistic wave forecasting systems and the extension of wave forecasts to the nearshore zone.

a. Global wave ensemble upgrades

Several upgrades to NCEP’s Global Wave Ensemble System (GWES) relevant to hurricane wave forecasting became operational in July 2014, including a higher spatial resolution grid at ½°, and a more efficient wave-generation physics package proposed by Ardhuin et al. (2010). The upgrades were a result of a close interaction between three NCEP centers: EMC, NHC, and the Ocean Prediction Center (OPC). The higher-resolution spatial grids have provided more accurate output across the Atlantic basin, and near and in the lee of both the Caribbean and Bahamas Islands. Across the Pacific, the upgraded spatial grid has brought benefits to NHC and OPC warnings for high-sea areas, which often occur with gap wind events near the coasts. The upgraded physics package has shown critical improvement in the prediction of severe sea states since its implementation in two other NWS wave forecasting systems, the GWGFS and the Great Lakes wave model system (GLW), as reported in Alves et al. (2014) and Chawla et al. (2013).

To illustrate the impact of the recent GWES upgrades, wave forecasts for Hurricane Sandy were rerun, and the event-based, time series validation was repeated relative to the two NDBC buoys that recorded the largest wave heights: 41048 and 44008. Validation was extended to include peak periods , to investigate in more detail the effects of the new physics package in severe sea states. Figure 9 shows time series of mean and from the ensemble data computed with the previous operational model (OPER), using the Tolman and Chalikov (1996) physics package and 1° spatial resolution, and the upgraded system (HRES), with the Ardhuin et al. (2010) physics package and ½° spatial resolution, at two forecast ranges: 72–60 and 24–36 h before the storm peak. Improvements are striking. Both predictions of and are drastically improved. The upgraded system mean values are significantly closer to the measured data.

Fig. 9.
Fig. 9.

Time series of (left) and (right) from the NCEP operational system in 2012 (OPER) and 2014 (HIRES). Ensemble mean from the GWES at forecast ranges approximately (top) 72 and (bottom) 48 h before the peak waves reached NDBC buoys (top two rows) 41048 (western Bermuda) and (bottom two rows) 44008 (southern Nantucket). These two buoys recorded the highest waves during the passage of the hurricane near the eastern coast of the United States.

Citation: Monthly Weather Review 143, 5; 10.1175/MWR-D-14-00143.1

b. Upgrades to global deterministic wave models

Several upgrades to NCEP’s global deterministic wave forecasting systems were made in late 2014. The GWGFS system, for instance, had its bathymetric grids improved via using the ETOP01 dataset, a 1-arc-min global relief model of Earth’s surface distributed by Amante and Eakins (2009). A larger number of major changes happened to the GWHUR system, starting with its matching the current GWGFS physics package of Ardhuin et al. (2010), which has shown large improvements not only in severe sea-state conditions, but also in more typical scenarios (see Chawla et al. 2013). More critical to forecasts of hurricane wave conditions was the change from using GFDL atmospheric forecast winds to the new operational atmospheric model for hurricane forecasts at NCEP, the so-called HWRF system (Tallapragada et al. 2008), a specialized implementation of the WRF Model (Skamarock et al. 2008).

c. Extension to nearshore wave forecasts

NWS wave guidance has been further improved via the expansion of numerical wave models into the critical hurricane landfall zone. This has been achieved via their extension to include nearshore wave prediction systems, reconciling an important omission in the guidance that NOAA provided during Superstorm Sandy (Sullivan and Uccellini 2013). Since the 2013 hurricane season, NCEP’s Nearshore Wave Prediction System (NWPS) has supplemented the GWHUR at NHC with an Atlantic basin run forced with tropical cyclone–generated winds using guidance provided by the NHC (Van der Westhuysen et al. 2014). This provides additional data supporting wave guidance that is consistent with the official NHC hurricane forecast. This basin-scale run also provides wave boundary conditions to NWS marine forecasters at coastal Weather Forecast Offices (WFOs), which subsequently run NWPS on their local domains using their own specialized wind forcing. During tropical cyclone events, the NWPS simulations include probabilistic surge fields from P-Surge at a forecaster-selected level of exceedance. This can result in increased skill in predicting nearshore wave heights.

8. Conclusions

The present study provides a comprehensive assessment of the performance of operational models used for wave guidance issued by NCEP during Tropical Cyclone Sandy (October 2012), the largest North Atlantic hurricane on record. A brief review of reports issued by government agencies, private industry, and universities is provided, supporting the claim that the main cause of destruction caused by Hurricane Sandy was a consequence of the combination of high waves and storm surges. The study then focuses on an overview of the operational wave model systems at NCEP that provide data used in weather guidance disseminated by the National Weather Service. Validation of NCEP’s wave model systems that were operational during Sandy’s lifetime is made relative to NDBC buoys where significant wave heights in excess of 6 m (19.7 ft) were recorded. Finally, a brief overview is made of changes to NWS’s operational wave systems relevant to forecasting extreme events such as Sandy.

Validation of NWS global wave model forecasts indicate a high skill of operational wave forecasts up to the 3–5-day horizon. In particular, performance assessments indicate a remarkably good skill of NCEP’s wave ensemble system in forecasting severe sea states associated with Sandy. Model data indicate that Sandy generated waves with spatial distributions that were highly unusual, relative to other recurving North Atlantic hurricanes on record. Associated with its large size, this led to a unique situation where the entire northwest Atlantic Ocean basin was exposed to waves larger than 12 ft (3.66 m), a level typically associated with severe sea states, during Sandy’s lifetime. Despite this unusual and challenging scenario, the combination of high-quality wave model forecasts and internal discussion among NWS marine forecasters resulted in wave guidance that was extremely reliable 3–5 days in advance of landfall, allowing the NWS to provide EMS and decision makers with advanced warning to the extreme and highly unusual wave conditions, observed along the entire eastern seaboard of the United States.

Several improvements have been implemented operationally at NCEP since Sandy’s passage, which have shown positive impacts to the skill of its wave modeling suite in simulating severe sea states associated with hurricanes. Global deterministic wave models had implemented higher-quality bathymetric and coastal grids, wave-generation physics that are more adequate to simulating extreme events, and forcing wind fields provided by state-of-the-art atmospheric models specially developed for hurricane forecasting. NCEP’s global wave ensemble system, which had already shown high skill with its relatively low spatial resolution and older physics package during Sandy, was upgraded to include higher-resolution spatial grids, as well as the improved physics package now shared by all wave forecasting systems at NCEP. Finally, nearshore wave prediction capabilities have become operational, extending the high-quality deep-water wave forecasts and guidance to the more critically affected coastal communities.

Futures changes in wave forecasting capabilities, lined up to occur in the next few years at NCEP, include the implementation of transitional unstructured grid domains connecting global deep-water wave models to the nearshore wave prediction system; the addition of a curvilinear grid covering high-latitudes and the North Pole, which is expected to improve wave guidance in the Arctic region, as this becomes a focus of interest due to changing ice coverage and expanding navigation routes; improvements in nearshore wave prediction with the inclusion of water levels in coastal grids, and the implementation of on-demand forecast systems developed in conjunction with local marine forecasters at coastal WFOs; and upgrades to NCEP’s global wave probabilistic forecast system, which, partnering with the U.S. Navy and Environment Canada, will integrate a multicenter ensemble system, with the potential to further improve the predictability of extreme wave events. In association with improvements to NWS’s storm surge forecasting systems, as reported recently by Funakoshi et al. (2012), the continual improvement of NCEP’s operational wave forecasting systems will assist in increasing the quality of NWS’s wave guidance, as required in the guidelines established by Sullivan and Uccellini (2013).

Acknowledgments

The authors are thankful to the early reviews of this paper by Hugh Cobb and Chris Landsea, from the National Hurricane Center, as well as the insightful requirements and suggestions made by the three anonymous reviewers, chosen by Monthly Weather Review’s editorial board. All have contributed to a significant improvement in coverage, quality, and readability of the current paper.

REFERENCES

  • Alves, J., , Y. Y. Chao, , and H. L. Tolman, 2005: The operational North Atlantic hurricane wind-wave forecasting system at NOAA/NCEP. Tech. Note 244, NOAA/NWS/NCEP/MMAB, 59 pp. + appendices.

  • Alves, J., and Coauthors, 2013: The NCEP–FNMOC combined wave ensemble product: Expanding benefits of interagency probabilistic forecasts to the oceanic environment. Bull. Amer. Meteor. Soc., 94, 1893–1905, doi:10.1175/BAMS-D-12-00032.1.

    • Search Google Scholar
    • Export Citation
  • Alves, J., , A. Chawla, , H. Tolman, , D. Schwab, , G. Lang, , and G. Mann, 2014: The operational implementation of a Great Lakes wave forecasting system at NOAA/NCEP. Wea. Forecasting, 29, 1473–1497, doi:10.1175/WAF-D-12-00049.1.

    • Search Google Scholar
    • Export Citation
  • Amante, C., , and B. W. Eakins, 2009: ETOPO1 1 arc-minute global relief model: Procedures, data sources and analysis. Tech. Memo. 24, NOAA/NESDIS/NGDC, 19 pp.

  • Ardhuin, F., and Coauthors, 2010: Semiempirical dissipation source functions for ocean waves. Part I: Definition, calibration, and validation. J. Phys. Oceanogr., 40, 19171941, doi:10.1175/2010JPO4324.1.

    • Search Google Scholar
    • Export Citation
  • Blake, E. S., , T. B. Kimberlain, , R. J. Berg, , J. Cangialosi, , and J. L. Beven II, 2013: Tropical cyclone report: Hurricane Sandy (22–29 October 2012). National Hurricane Center Rep. AL182012, 157 pp. [Available online at http://www.nhc.noaa.gov/data/tcr/AL182012_Sandy.pdf.]

  • Cao, D., , H. S. Chen, , and H. L. Tolman, 2007: Verification of ocean wave ensemble forecasts at NCEP. Proc. 10th Int. Workshop on Wave Hindcasting and Forecasting and Coastal Hazards Symp., Oahu, HI, NOAA/NCEP, Paper G1, 10 pp. [Available online at http://polar.ncep.noaa.gov/mmab/papers/tn261/mmab261.pdf.]

  • Cardone, V., , R. Jensen, , D. Resio, , V. Swail, , and A. Cox, 1996: Evaluation of contemporary ocean wave models in rare extreme events: The “Halloween storm” of October 1991 and the Storm of the Century of March 1993. J. Atmos. Oceanic Technol., 13, 198230, doi:10.1175/1520-0426(1996)013<0198:EOCOWM>2.0.CO;2.

    • Search Google Scholar
    • Export Citation
  • Chao, Y. Y., , and H. L. Tolman, 2010: Performance of NCEP regional wave models in predicting peak sea states during the 2005 North Atlantic hurricane season. Wea. Forecasting, 25, 15431567, doi:10.1175/2010WAF2222309.1.

    • Search Google Scholar
    • Export Citation
  • Chao, Y. Y., , J.-H. G. Alves, , and H. L. Tolman, 2005: An operational system for predicting hurricane-generated wind waves in the North Atlantic Ocean. Wea. Forecasting, 20, 652671, doi:10.1175/WAF851.1.

    • Search Google Scholar
    • Export Citation
  • Chawla, A., and Coauthors, 2013: A multigrid wave forecasting model: A new paradigm in operational wave forecasting. Wea. Forecasting, 28, 1057–1078, doi:10.1175/WAF-D-12-00007.1.

    • Search Google Scholar
    • Export Citation
  • Chen, H. S., 2006: Ensemble prediction of ocean waves at NCEP. Proc. 28th Ocean Engineering Conf., Taiwan, National Sun Yat-Sen University, 18.

  • Division of Natural Sciences and Mathematics, 2013: Beach-dune performance assessment after Hurricane Sandy related to FEMA disaster. Annual Rep. DR-NJ 4086, Division of Natural Sciences and Mathematics, Richard Stockton College, New Jersey, 26 pp.

  • Fanelli, C., , P. Fanelli, , and D. Wolcott, 2013: Hurricane Sandy. Water level and meteorological data report, National Ocean Services, NOS/NOAA, 60 pp.

  • Funakoshi, Y., , J. Feyen, , F. Aikman, , H. Tolman, , A. van der Westhuysen, , A. Chawla, , I. Rivin, , and A. Taylor, 2012: Development of Extratropical Surge and Tide Operational Forecast System (ESTOFS). Proc. 12th ASCE Int. Conf. on Estuarine and Coastal Modeling, St. Augustine, FL, ASCE, 201212, doi: 10.1061/9780784412411.00012.

  • Gladwin, H., , B. Morrow, , and J. Lazo, 2013: Communication and understanding of Hurricane Sandy storm surge forecast and warning information. Florida International University Natural Hazards Workshop, NSF CMMI 1322088, Miami, FL, Florida International University. [Available online at http://www.academia.edu/4316097/Communication_and_Understanding_of_Hurricane_Sandy_Storm_Surge_Forecast_and_Warning_Information.]

  • Hsu, S., 2013: Storm surges in New York during Hurricane Sandy in 2012: A verification of the wind-stress tide relation. Bound.-Layer Meteor., 148, 593598, doi:10.1007/s10546-013-9830-1.

    • Search Google Scholar
    • Export Citation
  • Impact Forecasting, 2013: Hurricane Sandy event recap report. Rep., Aon Benfield, 50 pp. [Available online at http://thoughtleadership.aonbenfield.com/Documents/20130514_if_hurricane_sandy_event_recap.pdf.]

  • Kurihara, Y., , M. A. Bender, , R. E. Tuleya, , and R. J. Ross, 1995: Improvements in the GFDL hurricane prediction system. Mon. Wea. Rev., 123, 27912801, doi:10.1175/1520-0493(1995)123<2791:IITGHP>2.0.CO;2.

    • Search Google Scholar
    • Export Citation
  • Landsea, C. W., , and J. L. Franklin, 2013: Atlantic hurricane database uncertainty and presentation of a new database format. Mon. Wea. Rev., 141, 35763592, doi:10.1175/MWR-D-12-00254.1.

    • Search Google Scholar
    • Export Citation
  • Magnusson, L., , J.-R. Bidlot, , S. T. Lang, , A. Thorpe, , N. Wedi, , and M. Yamaguchi, 2014: Evaluation of medium-range forecasts for Hurricane Sandy. Mon. Wea. Rev., 142, 19621981, doi:10.1175/MWR-D-13-00228.1.

    • Search Google Scholar
    • Export Citation
  • McCallum, B. E., and Coauthors, 2013: Monitoring storm tide and flooding from Hurricane Sandy along the Atlantic coast of the United States, October 2012. U.S. Geological Survey Open-File Rep. 2013–1043, 42 pp. [Available online at http://pubs.usgs.gov/of/2013/1043/.]

  • Sela, J. G., 1980: Spectral modeling at the National Meteorological Center. Mon. Wea. Rev., 108, 12791292, doi:10.1175/1520-0493(1980)108<1279:SMATNM>2.0.CO;2.

    • Search Google Scholar
    • Export Citation
  • Skamarock, W. C., and Coauthors, 2008: A description of the Advanced Research WRF version 3. NCAR Tech. Note NCAR/TN-475+STR, 113 pp. [Available online at http://www.mmm.ucar.edu/wrf/users/docs/arw_v3_bw.pdf.]

  • Sullivan, D., , and L. Uccellini, 2013: Hurricane/post-Tropical Cyclone Sandy, October 22–29, 2012. NOAA Service Assessment, NOAA, 46 pp. + appendices.

  • Tallapragada, V., , N. Surgi, , Q. Liu, , Y. Kwon, , R. Tuleya, , and W. O’Connor, 2008: Performance of the advanced operational HWRF modeling system during pre-implementation testing and in real-time 2007 hurricane season. 28th Conf. on Hurricanes and Tropical Meteorology, Orlando, FL, Amer. Meteor. Soc., 4A.5. [Available online at https://ams.confex.com/ams/28Hurricanes/techprogram/paper_138066.htm.]

  • Tolman, H. L., 2009: User manual and system documentation of WAVEWATCH III version 3.14. Tech. Note 276, NOAA/NWS/NCEP/MMAB, 194 pp. + appendices.

  • Tolman, H. L., , and D. Chalikov, 1996: Source terms in a third-generation wind wave model. J. Phys. Oceanogr., 26, 24972518, doi:10.1175/1520-0485(1996)026<2497:STIATG>2.0.CO;2.

    • Search Google Scholar
    • Export Citation
  • Van der Westhuysen, A. J., and Coauthors, 2014: Development and validation of the nearshore wave prediction system. 11th Symp. on the Coastal Environment, Austin, TX, Amer. Meteor. Soc., 4.5. [Available online at https://ams.confex.com/ams/93Annual/webprogram/Paper222877.html.]

  • Wei, M., , Z. Toth, , R. Wobus, , and Y. Zhu, 2008: Initial perturbations based on the ensemble transform (ET) technique in the NCEP global operational forecast system. Tellus, 60A, 6279, doi:10.1111/j.1600-0870.2007.00273.x.

    • Search Google Scholar
    • Export Citation
1

Following the hurricane database described in Landsea and Franklin (2013).

2

Relative to the North American vertical datum of 1988, which is approximately 20 ft (6.10 m) above mean sea level.

3

Scatter index here defined as the standard deviation of differences normalized by the observed mean.

Save