Communicating Probability Information in Hurricane Forecasts: Assessing Statements that Forecasters Use on Social Media and Implications for Public Assessments of Reliability

Forecasters are responsible for predicting the weather and communicating risk with stakeholders and members of the public. This study investigates the statements that forecasters use to communicate probability information in hurricane forecasts and the impact these statements may have on how members of the public evaluate forecast reliability. We use messages on Twitter to descriptively analyze probability statements in forecasts leading up to Hurricanes Harvey, Irma, Maria, and Florence from forecasters in three different groups: the National Hurricane Center, local Weather Forecast Offices, and in the television broadcast community. We then use data from a representative survey of United States adults to assess how members of the public wish to receive probability information and the impact of information format on assessments of forecast reliability. Results from the descriptive analysis indicate forecasters overwhelmingly use words and phrases in place of numbers to communicate probability information. In addition, the words and phrases forecasters use are generally vague in nature -- they seldom include rank adjectives (e.g., “low” or “high”) to qualify blanket expressions of uncertainty (e.g., “there is a chance of flooding”). Results from the survey show members of the public generally prefer both words/phrases and numbers when receiving forecast information. They also show information format affects public judgments of forecast reliability; on average, people believe forecasts are more reliable when they include numeric probability information.

Comparison of ARCHER MPERC to NHC Analysis

Eyewall replacement cycles (ERCs) are events that occur in intense tropical cyclones (TCs) and are difficult to predict. An ERC event involves a secondary outer eyewall that surrounds the inner eyewall. The outer eyewall slowly moves towards the eye and weakens the inner eyewall, eventually replacing the inner eyewall. During this process, wind speeds lower and the structure of a TC becomes disorganized, further weakening the storm. TCs often restrengthen after an ERC. Little is known about the process and as such, poses an obstacle to forecasters. The Automated Rotational Center Hurricane Eye Retrieval (ARCHER) Microwave-based Probability of Eyewall Replacement Cycle (MPERC) is an algorithm that uses 89-95 GHz passive microwave imagery and intensity estimates from the National Hurricane Center (NHC), Central Pacific Hurricane Center (CPHC), or the Joint Typhoon Warning Center (JTWC) to predict the possibility of an ERC. The effectiveness and ability of ARCHER MPERC was analyzed and compared to the NHC’s official reports on all Atlantic Basin tropical cyclones from 2017 to 2019. MPERC ultimately predicted seventeen ERCs in nine tropical cyclones. Of those, seven were valid ERCs. The algorithm works well, predicting approximately 41% of the total number of predictions correctly. However, MPERC did not predict five ERCs that were cited by the NHC. It was further found that it was true that MPERC produces incorrect results in sheared and dry environments.

End-user Preference for and Understanding of Hurricane Forecast Graphs

Hurricane forecast graphics have the challenging task of communicating information about spatio-temporal uncertainty. This study assesses the impact of graph literacy and graph format on user preference and understanding. In a laboratory setting, we compared user responses to official National Hurricane Center advisory maps and alternative visualizations. Results indicate that prior experience with a visualization drives preference and that graph literacy, visualization format, and tropical cyclone characteristics, in combination, influence interpretations of hurricane forecast track. The findings from this study are expected to inform redesign efforts of hurricane risk communication products.

ACCOMPLISHMENTS OF NOAA’S AIRBORNE HURRICANE FIELD PROGRAM AND A BROADER FUTURE APPROACH TO FORECAST IMPROVEMENT

Since 2005, NOAA has conducted the annual Intensity Forecasting Experiment (IFEX), led by scientists from the Hurricane Research Division at NOAA’s Atlantic Oceanographic andMeteorological Laboratory. They partner with NOAA’s Aircraft Operations Center, who maintain and operate the WP-3D and G-IV Hurricane Hunter aircraft, and NCEP’s National Hurricane Center and Environmental Modeling Center, who task airborne missions to gather data used by forecasters for analysis and forecasting and for ingest into operational numerical weather prediction models. The goal of IFEX is to improve tropical cyclone (TC) forecasts using an integrated approach of analyzing observations from aircraft, initializing and evaluating forecast models with those observations, and developing new airborne instrumentation and observing strategies targeted at filling observing gaps and maximizing the data’s impact in model forecasts. This summary article not only highlights recent IFEX contributions towards improved TC understanding and prediction, but also reflects more broadly on the accomplishments of the program during the 16 years of its existence. It describes how IFEX addresses high-priority forecast challenges, summarizes recent collaborations, describes advancements in observing systems monitoring structure and intensity, as well as in assimilation of aircraft data into operational models, and emphasizes key advances in understanding of TC processes, particularly those that lead to rapid intensification. The article concludes by laying the foundation for the “next generation” of IFEX as it broadens its scope to all TC hazards, particularly rainfall, storm-surge inundation, and tornadoes, that have gained notoriety during the last few years after several devastating landfalling TCs.

Performance Evaluation of Numerical Tools for Hurricane Forecast (NTHF) System during 2020 North Atlantic Tropical Cyclones Season

This study evaluates the performance of the Numerical Tools for Hurricane Forecast (NTHF) system during the 2020 North Atlantic (NATL) tropical cyclones (TCs) season. The system is configured to provide 5-day forecasts with basic input from the National Hurricane Center (NHC) and the Global Forecast System. For the NTHF validation, the NHC operational best track was used. The average track errors for 2020 NATL TCs ranged from 62 km at 12 h to 368 km at 120 h. The NTHF track forecast errors displayed an improvement over 60% above the guidance Climatology and Persistence (CLIPER) model from 36 h to 96 h, although the NTHF was better than the CLIPER in all forecast periods. The forecast errors for the maximum wind speed (minimum central pressure) ranged between 20 km/h and 25 km/h (4 hPa to 8 hPa), but the NTHF model intensity forecasts showed only marginal improvement of less than 20% after 78 h over the baseline Decay Statistical Hurricane Intensity Prediction Scheme (D-SHIPS) model. Nevertheless, the NTHF’s ability to provide accurate intensity forecasts for the 2020 NATL TCs was higher than the NTHF’s average ability during the 2016–2019 period.

Operational Forecasting of Tropical Cyclone Rapid Intensification at the National Hurricane Center

Although some recent progress has been made in operational tropical cyclone (TC) intensity forecasting, the prediction of rapid intensification (RI) remains a challenging problem. To document RI forecast progress, deterministic and probabilistic operational intensity models used by the National Hurricane Center (NHC) are briefly reviewed. Results show that none of the deterministic models had RI utility from 1991 to about 2015 due to very low probability of detection, very high false alarm ratio, or both. Some ability to forecast RI has emerged since 2015, with dynamical models being the best guidance for the Atlantic and statistical models the best RI guidance for the eastern North Pacific. The first probabilistic RI guidance became available in 2001, with several upgrades since then leading to modest skill in recent years. A tool introduced in 2018 (DTOPS) is currently the most skillful among NHC’s probabilistic RI guidance. To measure programmatic progress in forecasting RI, the Hurricane Forecast Improvement Program has introduced a new RI metric that uses the traditional mean absolute error but restricts the sample to only those cases where RI occurred in the verifying best track or was forecast. By this metric, RI forecasts have improved by ~20–25% since the 2015–2017 baseline period.

Influência dos Distúrbios Tropicais na Formação de Fenômenos Adversos de Atmosfera Estável no Nordeste Brasileiro

Resumo Este estudo tem como principal objetivo avaliar a influência dos Distúrbios Tropicais (DT) do Hemisfério Norte (HN), associados com Ciclones Tropicais (CT), nos fenômenos de atmosfera estável no Nordeste do Brasil (NEB), de 2013 a 2015, a fim de melhorar o método de previsão do tempo nessa área. Informações sobre CT encontram-se no National Hurricane Center. Os fenômenos adversos no NEB foram investigados usando dados METAR e mapas SYNOP. A análise dos sistemas sinóticos baseou-se nas imagens dos satélites no canal infravemelho e nos mapas de diferentes variáveis. Os dados de reanálise foram do European Center for Medium-Range Weather Forecasts. Foram observados 9 DT durante a passagem dos CT. No mesmo período, nevoeiros e nevoas úmidas com precipitação foram registrados no NEB. Estes fenômenos foram associados aos cavados nos alísios resultado da passagem dos DT. Sobre NEB em altos níveis foi observada convergência entre as correntes do DT no HN e do cavado do ciclone extratropical no HS. Esse processo criou movimentos descendentes sobre NEB, os quais contribuíram no acúmulo de umidade em baixos níveis e formação dos fenômenos adversos. Os resultados confirmam o efeito dos DT no tempo do NEB e sua importância na previsão do tempo nesta região.

On the Prospects for Improved Tropical Cyclone Track Forecasts

The success story of numerical weather prediction is often illustrated with the dramatic decrease of errors in tropical cyclone track forecasts over the past decades. In a recent essay, Landsea and Cangialosi, however, note a diminishing trend in the reduction of perceived positional error (PPE; difference between forecast and observed positions) in National Hurricane Center tropical cyclone (TC) forecasts as they contemplate whether “the approaching limit of predictability for tropical cyclone track prediction is near or has already been reached.” In this study we consider a different interpretation of the PPE data. First, we note that PPE is different from true positional error (TPE; difference between forecast and true positions) as it is influenced by the error in the observed position of TCs. PPE is still customarily used as a proxy for TPE since the latter is not directly measurable. As an alternative, TPE is estimated here with an inverse method, using PPE measurements and a theoretically based assumption about the exponential growth of TPE as a function of lead time. Eighty-nine percent variance in the behavior of 36–120-h lead-time 2001–17 seasonally averaged PPE measurements is explained with an error model using just four parameters. Assuming that the level of investments, and the pace of improvements to the observing, modeling, and data assimilation systems continue unabated, the four-parameter error model indicates that the time limit of predictability at the 181 nautical mile error level (n mi; 1 n mi = 1.85 km), reached at day 5 in 2017, may be extended beyond 6 and 8 days in 10 and 30 years’ time, respectively.

Comparison of Tropical Cyclone Center Positions Determined from Satellite Observations at Infrared and Microwave Frequencies

Determining tropical cyclone (TC) center positions is of interest to many researchers who conduct TC analysis and forecasts. In this study, we develop and apply a TC centering technique to Cross-Track Infrared Sounder (CrIS) and Advanced Technology Microwave Sounder (ATMS) observations of brightness temperature and report on an improvement of accuracy by adding a TC spectral analysis to the state of the art [Automated Rotational Center Hurricane Eye Retrieval (ARCHER)], especially for ATMS. We show that the ARCHER TC center-fixing algorithm locates TC centers more successfully based on the infrared channel with center frequency at 703.75 cm−1 (channel 89) of the CrIS than the ATMS channel 22 (183.31 ± 1.0 GHz) due to small-scale features in ATMS channel’s brightness temperature field associated with strong convective clouds. We propose to first apply the ARCHER TC center-fixing algorithm to ATMS channel 4 (51.76 GHz) that is less affected by small-scale convective clouds, and then to perform a set of the azimuthal spectral analysis of the ATMS channel-22 observations with tryout centers within a squared box centered at the ATMS channel-4-determined center. The center that gives the largest symmetric component is the final ATMS-determined center. Compared to the National Hurricane Center best track, the root-mean-square center-fixing errors determined from the two ATMS channels (one single CrIS channel) are 29.9 km (35.8 km) and 28.0 km (30.9 km) for 104 tropical storm and 81 hurricane cases, respectively, in the 2019 hurricane season.

Community Response to Hurricane Threat: Estimates of Warning Issuance Time Distributions

Hurricane evacuation warnings from local officials are one of the most significant determinants of households’ evacuation departure times. Consequently, it is important to know how long after the National Hurricane Center (NHC) issues a hurricane watch or warning that local officials wait to issue evacuation warnings. The distribution of local evacuation warning issuance delays determined from poststorm assessment data shows a wide range of warning issuance delay times over an 85-h time span, although the vast majority of times fall within a 40-h window. Nearly 30% of the jurisdictions issued evacuation warnings before an NHC hurricane warning. Only 5% delayed the decision for more than 25 h after the NHC hurricane warning. The curves for warning issuance delays, using both the NHC watch and NHC warning issuance times as reference points, are very different from the warning issuance curves observed for the rapid-onset events. The hurricane data exhibit much more of an “S shape” than the exponential shape that is seen for rapid-onset data. Instead, curves for three different types of storm tracks, defined by a perpendicular/parallel dimension and a straight/meandering dimension, follow three noticeably different logistic distributions. The data also indicate that warnings were issued significantly earlier for coastal counties than for inland counties. These results have direct practical value to analysts that are calculating evacuation time estimates for coastal jurisdictions. Moreover, they suggest directions for future research on the reasons for the timing of local officials’ hurricane evacuation decisions.