2017 Atlantic Hurricane Season Outlook: Jul 28, 2017

Transcription

1 event Hurricane Season Outlook 2017 Atlantic Hurricane Season Outlook: Jul 28, 2017 This seasonal forecast briefing is the second in this season s series, released as new information becomes available. In this briefing, we provide an update on how the season appears to be shaping up, and feature an update on our experimental seasonal hurricane damage potential forecast. Current situation Since our early season outlook, we have seen tropical storms Bret and Cindy in June, and more recently tropical storm Don in mid-july. Cindy made landfall as a tropical storm at the Texas/Louisiana border, and despite its ragged appearance, brought 55mph winds ashore with some local flooding and a few tornadoes. Both Bret and Don formed over the tropical Atlantic and remained far away from the U.S. coast. However, the fact that Bret and Don developed over the tropical Atlantic so early in the season is evidence that conditions are primed for an active season this year. We also know that the most intense hurricanes have their origins in the tropical North Atlantic between the Caribbean and North Africa. Given these precursors to an active season, what levels of hurricane activity can we expect? Forecasts While earlier forecasts generally called for average activity, the latest round of forecasts have all shifted to above average activity for 2017 (Table 1), though there is some disagreement on just how far above average the activity will be. Notably, Colorado State University (CSU) have doubled the number of hurricanes from their early April forecast from 4 to 8, and almost doubled their forecast of seasonal Accumulated Cyclone Energy (a combined measure of storm frequency and intensity). Tropical Storm Risk (TSR) increased their hurricane numbers by similar amounts and now calculate an 86% likelihood of average or above average hurricane activity. The forecast numbers of named storms across the major forecasting centers (Table 1) range from 14 to 17 with an average of 14.8 (compared to a typical number of 12). For hurricanes the numbers range from 7 to 8.2 with an average of 7.4 (compared to a typical number of 6.5), and for major hurricanes (Saffir-Simpson category 3-5) they range from 3 to 3.3 (compared with a typical number of 2). This shift upwards reflects reduced odds of El Niño and the anomalously warm tropical and sub-tropical North Atlantic. A diverse array of methods is used to generate these forecasts. TSR, for example, use only the most recent decade to identify relationships, a nod to the potential change in the relationship from one decade to the next, whereas CSU uses 35 years of past data. TSR also develop predictions for sub-regions of the Atlantic ocean, each with a unique set of predictors, reflecting regional variation in the drivers of hurricane activity. CSU s statistical model uses two simple predictors that capture the Atlantic and Pacific controls on the hurricane season. This year, the warm Atlantic control suggests a very active season while the Pacific control suggests a moderate season. CSU also modify their forecasts using analog years and some subjective tuning to account for factors excluded from their algorithms. There is some disagreement on the credibility of U.S. landfall predictions. Forecasts of landfall are limited due to the small number of events and the complex and highly sensitive interactions between storm formation, track and coastal orientation that result in landfall. However, some centers do issue landfall forecasts. CSU are now predicting above average odds of at least one major hurricane landfall in the U.S. at 62% compared to an average likelihood of 52%. TSR also favor an enhanced likelihood of US landfall this season compared to normal.

2 Table 1: Atlantic seasonal hurricane forecasts as of Jul 28, 2017 Summary of 2017 Atlantic Seasonal Hurricane Forecasts 3 Data Source Median Value (Source CSU) Average of 6 analog Years (Source CSU) Date Issued # Named Storms # Hurricanes # Major Hurricanes ACE 1 Jul (118%) 8.2 (126%) 3.3 (165%) 133 (145%) Colorado State University Jul (125%) 8.0 (123%) 3.0 (150%) 135 (147%) NOAA/CPC 2 May (117%) 7.0 (108%) 3.0 (150%) 115 (124%) Tropical Storm Risk 2 Jul (142%) 7.0 (108%) 3.0 (150%) 116 (126%) The Weather Company May (117%) 7.0 (108%) 3.0 (150%) Average of the all of the above 14.8 (124%) 7.4 (114%) 3.1 (155%) 125 (136%) n/a Analog years An alternative view to forecast models is provided by hurricane activity in past years that had similar pre-season climate conditions and forecast conditions to this year. CSU uses this approach to qualitatively correct the output from their empirical forecast technique. CSU s selected analog years , 1969, 1979, 2004, 2006 and are characterized by neutral El Niño Southern Oscillation conditions, and a warm Atlantic. The average activity among these 6 analog years is shown in Table 1 and indicates above average number of storms, similar to CSU s official forecast. Climate signals: sea surface temperatures The waters of the North Atlantic continue to be warmer than normal and the earlier cool region of the Gulf of Mexico has now warmed up to be warmer than normal (Figure 1). The formation and development of hurricanes is highly dependent on the available energy in the upper layers of the ocean. This is why sea surface temperatures are often used as a proxy for available energy as a major factor in determining seasonal activity. CSU notes that a warmer than normal Atlantic generally supports weaker environmental winds (also predicted by TSR) and increased moisture in the atmosphere all favorable for hurricane formation. This is a supportive ocean state for the development of hurricanes from easterly waves in the deep tropics that can track towards the US. Figure 1: Departure of weekly average sea surface temperature from a longterm average ( C) on Jul (Source : : NCEP/NOAA). Given the tremendous inertia of the oceans, this general pattern is likely to persist throughout the hurricane season, though sub-seasonal variability will need to be monitored. Of course, ocean temperatures are not the only factor controlling seasonal activity. The orientation of the winds across the North Atlantic is also critical and to examine that we need to look at the El Niño Southern Oscillation (ENSO) 1 Accumulated Cyclone Energy (ACE) is a combined measure of hurricane intensity, duration and frequency. ACE is calculated as the sum of the square of the maximum wind speed in each 6-hour period during the life of a tropical cyclone from the time it reaches tropical storm strength (wind speeds 65 kmph (39 mph)) in units of 10 4 ; ACE = 10-4 v 2 max, where v is measured in knots. 2 NOAA/CPC, TSR and NCSU forecast likely ranges rather than single values. The values presented here are the middle of the forecasted ranges. 3 Forecasts generally include storms that occurred prior to the forecast issue date. 2

3 Climate signals: El Niño? The cooling and warming of the equatorial Pacific associated with ENSO exerts the strongest known control on Atlantic hurricane activity by driving an overturning of the tropical atmosphere; air rises over warm waters and sinks over cool waters. The warming this past Spring following the demise of La Niña (the cool phase of ENSO) has not continued and the central and eastern Pacific Oceans appear to have settled into a warmneutral state. Fluctuations about that state from month to month are not showing any warming trend indicative of a developing El Niño. For the first time this year, neutral conditions is the most likely ENSO state for the peak of this year s hurricane season, as indicated in the early July forecast from IRI/CPC (Figure 2). Neutral conditions now have a 56% likelihood while El Niño conditions have dropped to 38%. This Figure 2: The official CPC/IRI ENSO probability forecast, based on a consensus of forecasters using human judgment and model output. Source: International Research Institute for Climate and Society. Published Jul 13, expected neutral ENSO outlook is not likely to drive unfavorable winds aloft across the Atlantic, and will not hinder the warm Atlantic from generating above average hurricane activity this year. Climate signals: AMO Figure 3: Observed AMO index, defined as detrended 10-year low-pass filtered annual mean area-averaged SST anomalies over the North Atlantic basin (0N-65N, 80W-0E) for the period Source: NCAR. The term Atlantic Multi-decadal Oscillation (AMO) has been used to characterize historical swings in Atlantic Ocean temperatures every years. Historical hurricane activity has been higher in the AMO warm phase and lower in its cool phase. The most recent warm phase began in 1995 (Figure 3) and history suggests a swing to a cool phase is imminent, possibly heralding many years of low hurricane activity. However, there is low confidence in this occurring. The low confidence arises from our incomplete understanding of the AMO, evidenced by competing theories for it's driving mechanisms. Also, the likelihood of a flip in any year is small, and even if it does occur the cool phase will be warmer than it otherwise would be due to recent climate change. While ocean temperatures are warmer than normal for most of the Atlantic, the far North Atlantic remains cooler than normal (Figure 1). This is consistent with a negative AMO and bears watching over the coming years. In summary, the AMO is certainly something to keep monitoring this year and for the next few years, but there are other environmental drivers of hurricane activity we understand better, such as ENSO. We suggest looking first to ENSO before AMO as a guide for this year's hurricane season. 3

4 Confidence Despite being closer to the peak of the hurricane season, CSU still note there is considerable uncertainty in these forecasts. This arises from uncertainty in how warm the Atlantic will become during the peak of the hurricane season, the eventual outcome of ENSO, and the considerable range of hurricanes possible for a given set of environmental conditions. However, these forecasts are far more reliable than the early Spring outlooks due to the enhanced predictability of ENSO and inertia of the current Atlantic temperatures. Forecasts issued in early July are proven to be skillful at differentiating between active and inactive seasons. Seasonal Damage Potential The seasonal forecasts summarized above predict the number of tropical cyclones, but that does not necessarily correspond to damage. To get closer to a measure of damage, new science from the Willis Research Network has discovered a way to assess the damage potential of the upcoming hurricane season due to winds and coastal surge. In much the same way that current seasonal forecasts forecast hurricane numbers, this new view of the hurricane season uses predicted sea surface temperature patterns and environmental winds to infer the key drivers of hurricane damage: namely, hurricane intensity, hurricane size and hurricane translation speed. The resulting seasonal Cyclone Damage Potential (CDP) index uses a scale from 1 to 10 to assess likely seasonal damage potential relative to other seasons. As is the case with most seasonal forecasts of tropical cyclone numbers, we caution that this damage potential forecast is for the entire North Atlantic and does not necessarily correspond to landfalling damage potential. However, to assess actual seasonal storm damage to a given portfolio, the relationship between the index and damage is needed and can be obtained using historical loss and index values. Forecast ocean temperature and environmental wind forecasts were input into the CDP index from the National Centers for Environmental Prediction (NCEP) dynamical Climate Forecast System (CFS), and damage potential forecasts were produced using 100 CFS forecasts made during March 2017 and 120 CFS forecasts made during June Figure 4 shows the spread of these forecasts (red and yellow lines) compared to the long-term historical damage potential (black line). The most recent forecast issued on July 1 (yellow line) increases the forecast mean from 3.2 (issued on April 1, red line) to 5.4. This is also far above the historical mean of 3.3 meaning that the potential for damage is higher than normal this year. Likelihood flag CDP April 1 Forecast July 1 Forecast Historical Figure 4: Forecast distributions of cyclone damage potential, issued on April 1 (red line) and July 1 (yellow line). The historical distribution of cyclone damage potential (black line) over the period is shown for comparison. Summary The major forecasting centers have all increased their outlooks substantially to above average activity for There is also a suggestion that within this active season, environmental conditions could align to create extremely favorable conditions for tropical cyclone formation and development in early to mid August. Planning should account for this elevated risk. 4

5 Next briefing and additional advice This briefing will be updated in August to include the latest round of seasonal forecasts. These forecasts will offer moderate-to-good confidence as the environmental conditions for the peak of the hurricane season come into focus. The Willis Re Analytics Team will report on all tropical storms and hurricanes in the North Atlantic and the Gulf of Mexico. This includes briefings and updates to our clients during hurricane events. These will contain the latest information from the National Hurricane Center, commentary on likely tracks and intensities and, when available, updates and modeling guidance from the catastrophe modeling companies. Forecast information sources Klotzbach, P. J. and M. M. Bell: Forecasts of Atlantic Seasonal Hurricane Activity and Landfall Strike Probability for 2017, Jul 5, 2017, Department of Atmospheric Science, Colorado State University, Fort Collins CO, U.S. NOAA/CPC: NOAA 2017 Atlantic Hurricane Season Outlook, 25 May Available at Saunders, M. and A. Lea: July Forecast Update for Atlantic Hurricane Activity in 2017, July 4, 2017, Department of Space and Climate Physics, University College London, London, UK Wunderground blog: Category 6. Contact us Willis Research Network Dr. James Done Project Scientist and Willis Research Fellow Capacity Center for Climate and Weather Extremes, National Center for Atmospheric Research. P.O. Box 3000, Boulder, CO, U.S. Phone: Geoff Saville WRN Senior Research Manager Analytics Technology and Willis Research Network. 51 Lime Street, London, EC3M 7DQ. U.K. Phone: Roy Cloutier Executive Vice President, Willis Re Catastrophe Analytics 7760 France Avenue South Minneapolis, MN 55435, U.S. Phone: Willis Limited, Registered number: England and Wales. Registered address: 51 Lime Street, London, EC3M 7DQ. A Lloyd s Broker. Authorised and regulated by the Financial Conduct Authority for its general insurance mediation activities only. Copyright 2017 Willis Limited / Willis Re Inc. All rights reserved: No part of this publication may be reproduced, disseminated, distributed, stored in a retrieval system, transmitted or otherwise transferred in any form or by any means, whether electronic, mechanical, photocopying, recording, or otherwise, without the permission of Willis Limited / Willis Re Inc. Some information contained in this document may be compiled from third party sources and we do not guarantee and are not responsible for the accuracy of such. This document is for general information only and is not intended to be relied upon. Any action based on or in connection with anything contained herein should be taken only after obtaining specific advice from independent professional advisors of your choice. The views expressed in this document are not necessarily those of Willis Limited / Willis Re Inc., its parent companies, sister companies, subsidiaries or affiliates, Willis Towers Watson PLC and all member companies thereof (hereinafter Willis Towers Watson ). Willis Towers Watson is not responsible for the accuracy or completeness of the contents herein and expressly disclaims any responsibility or liability for the reader's application of any of the contents herein to any analysis or other matter, or for any results or conclusions based upon, arising from or in connection with the contents herein, nor do the contents herein guarantee, and should not be construed to guarantee, any particular result or outcome. Willis Towers Watson accepts no responsibility for the content or quality of any third party websites to which we refer. 5