2020 Hurricane Outlook

Tropical Storm Arthur

On May 12, the National Hurricane Center (NHC) began issuing special Tropical Weather Outlooks for the potential development of a subtropical cyclone from a low expected to develop over the southwest Atlantic within the next five days. Strong global model consensus for development of a tropical or subtropical cyclone yielded high confidence in development odds from the NHC during the next day or so as increasing showers and cloudiness began to concentrate across the Florida Straits in association with a decaying cold front. This entity gradually became more concentrated over time with the formation of a trough of low pressure on May 14 that produced disorganized shower activity and gusty winds across the Florida Keys, portions of southeast Florida, and the northwestern Bahamas. By May 16, the low had attained sufficient organization off the coast of southeastern Florida to warrant an Air Force reconnaissance mission. This mission determined that the system had attained a well-defined circulation and advisories on Tropical Depression One were issued by the NHC shortly thereafter. Over the next six hours, the minimum central pressure gradually fell, and Air Force reconnaissance determined that the depression’s winds had increased sufficiently to be upgraded to Tropical Storm Arthur at 0300 UTC May 17. Arthur continued North-northeastward off the coast of the southeastern United States through the following 24 hours with little intensification. Around 1600 UTC on May 18, Arthur made its closest approach to the mainland United States with its center passing approximately 20 miles east of Cape Hatteras, North Carolina. While much of the heaviest wind and rainfall from Arthur remained offshore, tropical storm-force winds were recorded across parts of the Outer Banks of North Carolina during the day. By this time, Arthur was beginning to experience increasing southwesterly wind shear from an approaching upper-level trough over the eastern United States as well as cooling sea surface temperatures. However, this wind shear was initially flow-parallel enough to enhance baroclinic deepening of the cyclone and Arthur reached its peak intensity of 50 knots and a minimum central pressure of 991 mb as it began moving eastward away from North America later that day. By May 19, Arthur began to lose its tropical characteristics and the final NHC advisory was issued at 1500 UTC declaring the system as post-tropical.

Tropical Storm Bertha

The second preseason named storm of 2020 in the North Atlantic originated from an elongated surface trough that interacted with an upper-level disturbance. On May 26, the NHC began issuing special Tropical Weather Outlooks on a weak surface trough over the southern Florida peninsula that was expected to lift northward and over the adjacent waters of the southeastern United States later that day. Due to the apparent marginal environmental conditions and anticipated limited time over water, the entity was given a low chance of development. However, overnight on May 27, more persistent deep convection developed along the eastern side of the surface trough axis where a compact area of low pressure developed and became better defined. The increase in organization of the low and tropical-storm force winds observed by the KCAE Charleston, SC dual-pol doppler radar were sufficient for advisories to be initiated on Tropical Storm Bertha at 1230 UTC. Shortly thereafter, Bertha reached its estimated peak intensity of 45 knots as it made landfall in Mount Pleasant, SC at approximately 1330 UTC. The storm then moved inland and weakened over South Carolina where it produced locally heavy rainfall and gusty winds. Less than 24 hours after its formation, Bertha became post-tropical at 0900 UTC on May 28 as it continued northward through central Appalachia.

This marked the sixth consecutive year of a preseason named storm being observed in the North Atlantic basin since record-keeping in the basin began in 1851. Additionally, 2020 is the sixth known year on record with two named storms observed prior to the official start of hurricane season joining 1887, 1908, 1951, 2012, and 2016. While these occurrences appear to be becoming more common in recent, it is noted that the assistance of improved satellite and surface observation capabilities has greatly improved the detection of tropical and subtropical cyclones in the North Atlantic basin over the last several decades. Furthermore, early season activity does not hold a strong correlation to total seasonal hurricane activity in the North Atlantic.

Tropical cyclone (TC) activity has an adverse and perennial impact on life and property in countries adjacent to these ocean basins worldwide (Blake et al. 2011). Therefore, research into tropical meteorology over the last several decades has been vital in increasing understanding of the mechanisms that drive TC formation, steering, and intensification in addition to increasing our capabilities to track these storms. This includes accurate record-keeping of seasonal activity across all TC basins with activity quantified in the North Atlantic and eastern North Pacific (east of 180⁰) through number of named storms (storms that attain winds ≥ 34 knots), number of hurricanes (≥ 64 knots), number of major hurricanes (≥ 96 knots), and Accumulated Cyclone Energy (ACE). ACE is the approximation of wind energy used by a tropical system over its lifetime. This is calculated by summing the squares of the estimated maximum sustained velocity in knots at each observed 6-hour interval for storms that are at least tropical storm intensity (≥ 34 knots). Thus, the seasonal ACE value is the summation of the ACE of all storms that were observed during a given year.

For this discussion, the primary basin of investigation will be the North Atlantic Ocean which observes high interannual variability in TC activity when compared to other tropical basins (Bell & Muthuvel 2006). Consequently, there is greater value in assessing what factors can influence TC variability to determine the potential magnitude of TC activity that could occur during hurricane season. The North Atlantic hurricane season spans from June 1st to November 30th and averages twelve named storms, six hurricanes, three major hurricanes, and an ACE value of 106 (*10^4 knots^2) based on the 1981-2010 mean. These TCs typically form in regions of copious moisture, strong atmospheric instability, light vertical wind shear, and sea surface temperatures of 26.5 ⁰C or greater (Gray 1979). Considering these environmental variables are necessary, but not exclusive, for TC development, there is an expanding aspect of research focused on the understanding of how these environments are modulated at temporal scales on the order of weeks, months, and years. Specifically, the investigation of how climate teleconnections influence TC activity worldwide when certain modes of variability are present. By leveraging the established literature on the relationships of these climate teleconnections to TC activity in the North Atlantic, scientists through the years have been able to forecast the magnitude of activity expected in the basin in the months preceding the season with reasonable accuracy (Klotzbach et al. 2019). With this, these same methods will be used to analyze a multitude of highly correlated variables to North Atlantic TC activity to provide a qualitative discussion and reasoning for the quantitative seasonal hurricane forecast outlined in the conclusion.

I. North Atlantic SST

Sea surface temperature (SST) anomalies in the North Atlantic leading up to hurricane season have been correlated to hurricane activity given the high specific heat of ocean water allowing temperature profiles and atmospheric feedbacks to be maintained into hurricane season. The highest positively correlated region for SSTs exists along the Canary Current region and eastern Tropical Atlantic. This region coincides with the location of the Atlantic Meridional Mode (AMM) and thus the AMM is commonly invoked in seasonal forecasts of hurricane activity in the basin (Appendix: Figure 1). The AMM describes the meridional difference in SST anomalies between the tropical and equatorial Atlantic (Chiang & Vimont 2004). When the AMM is positive, the Canary Current region is warmer than average, and the equatorial Atlantic is cooler. The steepening of the temperature gradient between these regions during hurricane season has been attributed to the enhancement of the African Easterly Jetstream which is a key component in the production of hurricanes in the North Atlantic (see Section IV).

When evaluating the entire North Atlantic basin SST profile, one of the most well-known modes of variability in these SST anomalies that have been linked to TC activity is the Atlantic Multidecadal Oscillation (AMO). The AMO refers to the variance of SST anomalies across the North Atlantic Ocean and follows a 65 to 80-year period based on the observed data available to researchers (Kerr 2000). A positive AMO phase is associated with above average tropical cyclone activity due to the above average SSTs observed in the North Atlantic’s Main Development Region (MDR; 10⁰N-20⁰N, 85⁰W-20⁰W) where a majority of tropical cyclones form. The negative phase is associated with below average tropical cyclone activity in the North Atlantic due to the MDR being cooler than average. While the AMO has been valuable in understanding TC variability at extended temporal scales, recent modulation of the North Atlantic SST configuration has raised questions regarding its reliability in seasonal TC forecasting. Specifically, with record Arctic sea ice melt in recent years and notable freshening (decrease in ocean salinity) of the far North Atlantic, there is growing evidence that this is slowing the Atlantic Meridional Overturning Circulation (AMOC). The AMOC is regarded as a main component of heat transport via surface and deep ocean currents across the South and North Atlantic basins (Frajka-Williams et al. 2019). Considering these recent findings, and their implication on altering the climate of the North Atlantic, the persistent below average SSTs in the Far North Atlantic and warmer subtropical North Atlantic and Canadian Maritime in recent years appear to validate these claims. The modulation of the SST regime in this regard has resulted in a more persistent positive North Atlantic Oscillation (NAO) pattern during the winter months. A positive NAO typically favors stronger high pressure over the northeast Atlantic which enhances easterly trade winds and cooling across the tropical Atlantic. Consequently, the frequency of negative AMO configurations observed in the spring SST profile appear to have increased recently (2014, 2015, and 2018). Further, even years where the spring base state of the tropical Atlantic has been warmer, such as 2016 and 2017, the SST configuration was not traditional to that of a positive AMO because the subtropics and Canadian Maritime remained warm. Despite these clear differences in SSTs during the months preceding hurricane season, warming of the tropical Atlantic through the peak of hurricane season was observed in each of the last six hurricane seasons (2014-2019). Most notably, the start of the 2018 hurricane season observed near-record cold Atlantic MDR SSTs and still warmed significantly to near average by peak hurricane season. This was likely in large part due to the anomalous zonal westerlies enhanced by a stronger West African Monsoon, or WAM (see Section IV for more). Therefore, the seasonality of these recent trends in North Atlantic SSTs can be explained by tropical Atlantic cooling in the winter months due to enhanced trade winds via positive NAO and warming of this region during summer months when the WAM is active.

The spring 2020 North Atlantic SST profile maintains some of these irregularities with near average SSTs in the MDR, above average SSTs across a region between 20⁰N to 30⁰N and well-below average SSTs observed in the subtropical Atlantic and near the Canadian Maritime (Figure 2). While not initially apparent, the evolution of the SST profile in the North Atlantic has progressively become more comparable to a traditional positive AMO. The March-April-May (MAM) mean 850mb zonal wind pattern can be used to explain the appearance of the May SST profile (Figure 3). The presence of anomalous easterly winds in the tropical Atlantic has inhibited warming in the region. Additionally, anomalous westerlies in the lower-latitudes and mid-latitudes of the east Atlantic have enhanced warming in the areas just north and south of the MDR. This pattern is indicative of a particularly strong WAM and was observed in 2018 and 2019, but with much stronger anomalous zonal easterlies which caused more significant cooling of the MDR during those years. A zonal wind pattern such as this had not been particularly common since the return of the positive AMO regime in 1995. However, with the recent uptrend in WAM strength for the first time since the 1960’s, the prior active era in North Atlantic hurricane activity (1945-1969) was investigated to determine if the current zonal wind regime was analogous to what was witnessed then. The most noteworthy finding of this analysis was that several of the most active years of the 1950’s and 1960’s featured an 850mb zonal wind pattern like 2020 and observed a migration of the pattern northward as the WAM amplified into peak hurricane season (Figure 4). This behavior yielded further warming of the tropical Atlantic during hurricane season and enhanced positive (cyclonic) relative vorticity in the region. Considering the analogous pattern and enhancement in TC activity observed in recent years during peak hurricane season because of such a regime leads to the conclusion that Atlantic hurricane activity will be enhanced in 2020.

II. ENSO

The El Niño Southern Oscillation (ENSO) is defined as the variance of the SSTs in the eastern Equatorial Pacific Ocean. The period for ENSO is two to seven years and can vary upwards of 2 to 3⁰C during this period. The fluctuation of the SSTs in this region is considered as one of the most influential in global climate variability and is associated with significant variations of precipitation in the continents adjacent to the Pacific Ocean. ENSO is classified into three different phases (La Niña, neutral, and El Niño) depending on the mean SST anomaly of the Niño 3.4 ENSO region (5⁰N-5⁰S, 170⁰W-120⁰W). La Niña is the term used for the cool phase of ENSO (≤ -0.5⁰C) and is linked to weaker vertical wind shear across the Atlantic MDR due to the displacement of the Subtropical Jetstream (STJ) further North during the summer months and overall less convective activity observed in the eastern Pacific. El Niño is known as the warm phase of ENSO (≥ +0.5⁰C) and typically favors upward motion across the eastern Pacific and American continents. This upward motion not only enhances convection across the eastern Pacific which promotes a more active summer STJ, but also generally promotes downward motion and inhibition of convection over the North Atlantic and Maritime Continent (Enfield & Mestas-Nuñez 1999).

In a neutral ENSO state (-0.5⁰C to +0.5⁰C), climatologically average atmospheric conditions favor easterly trade winds across the Equatorial Pacific. This is in association with a zonal overturning circulation known as the Walker circulation which is caused by the east-west imbalance in oceanic heat in the deep tropics of the Pacific. This imbalance in heat is accentuated by the eastern Equatorial Pacific cool tongue which is caused by the diffluent Ekman transport of ocean water away from the equator causing cooler subsurface waters to replace it. The upwelling of these subsurface waters explains the asymmetric appearance of the Equatorial Pacific’s subsurface thermocline profile in neutral conditions. Therefore, when the Walker circulation is disrupted, easterly trade winds weaken or reverse (known as a Westerly Wind Burst) and upwelling in the eastern Equatorial Pacific ceases. The lack of upwelling, or existence of downwelling, in the eastern Equatorial Pacific causes anomalous warming of the region which can lead to genesis of an El Niño event. Contrarily, when the Walker circulation is strengthened, easterly trade winds strengthen (Easterly Wind Burst) causing more significant upwelling in the eastern Pacific and cooling of the region which can be associated with the genesis of La Niña.

Understanding the evolution of ENSO to improve climate prediction is still regarded as a challenging task with many researched methods proposed as factors in ENSO’s variability (Wang 2018). One method that can be used in real-time application for ENSO prediction is the delayed-oscillator theory. This theory is based on the oceanic response to surface wind stress across the Equatorial Pacific. When a Westerly Wind Burst (WWB) occurs, the ocean responses by generating a downwelling Kelvin wave (eastward propagating) and an upwelling Rossby wave (westward propagating) near the International Date Line (180⁰). Depending on the persistence of this WWB over the span of a couple of months the downwelling Kelvin wave (DKW) that is generated will propagate into the eastern Equatorial Pacific over a few months causing warming of ENSO and potentially an El Niño event. While this is occurring, the upwelling Rossby wave (URW) propagates westward toward the west Pacific at a slower rate and cools this region. Once these waves reach the periphery of the Pacific basin, they are then refracted by the east-west boundaries of the basin (continents) or terminated by the atmospheric forcing present at that time. If no new DKW is generated, then the refracted Rossby wave (now known as an upwelling Kelvin wave) will then propagate eastward causing cooling of the eastern Equatorial Pacific. Conversely, if an Easterly Wind Burst (EWB) occurs, then an upwelling Kelvin wave (UKW) and a downwelling Rossby wave (DRW) are generated and can lead to the development of La Niña (Roundy & Kiladis 2007).

During the 2019 hurricane season, the weak El Niño event that had developed through boreal winter 2018-19 weakened gradually which yielded neutral conditions for a brief period before a new DKW arrived to warm the ENSO regions once more. While this DKW did renew warm neutral conditions, the warmest part of the Equatorial Pacific remained weighted in the central Pacific where SST anomalies averaged around +0.9⁰C through much of boreal winter 2019-20. This led to an atmospheric response comparable to that of Modoki El Niño, or central Pacific-based El Niño. In this global atmospheric pattern, a new WWB was observed which generated an appreciable DKW which likely would have been capable of keeping ENSO in a warm state if it propagated eastward unabated. However, notable changes in atmospheric behavior began in February and March with the a series of EWBs that began in the eastern Pacific and propagated westward with time which greatly attenuated the existing warm subsurface and also generated an UKW near the date line. The emergence of this pattern and cooling oceanic subsurface enhanced by these EWBs began to diminish the likelihood of a persistent warm ENSO state into hurricane season. By April, with a strengthening UKW now dominating the subsurface and the remnant DKW meeting its demise in the east Pacific, it became increasingly apparent that such a subsurface profile is analogous to many traditional developing La Niña events. This observation was further strengthened by the apparent lack of Madden-Julian Oscillation (MJO) activity in the Pacific basin which is a mechanism typically associated with enhancing WWBs and El Niño growth. Now, as hurricane season begins, the dominant Indian Ocean and Africa standing wave, classic to a La Niña atmosphere, only grows more apparent in addition to the continued EWBs across the Pacific. This has caused significant ENSO cooling during the month of May in which ENSO region Niño 3.4 went from +0.2⁰C to -0.5⁰C in just four weeks. With the UKW now arriving to the eastern Pacific, further cooling is expected and there is a high likelihood of La Niña conditions being present during the peak of hurricane season. The anticipated strength of this La Niña during hurricane season will be dependent on the emergence of a reinforcing UKW, however, it is plausible that a moderate La Niña (-1.0⁰C to -1.4⁰C) may be present by the fall of 2020 given current trends.

This ENSO forecast is supported by the North American Multi-Model Ensemble (NMME) which is a suite of dynamical climate prediction models. This forecast is also backed by the European Centre for Medium Range Forecasts (ECMWF) and Meteo-France climate models. The consensus of these models predicts weak La Niña conditions will be present during peak hurricane season; however, the level of uncertainty encompasses solutions for both cool neutral from models like the ECMWF and moderate La Niña in solutions such as the CFSv2 (Figure 5). With this, it is likely that ENSO will be favorable for North Atlantic hurricane activity given the expected cooling of ENSO will inhibit convective development over the Pacific and provide a below-average wind shear environment across the MDR.

III. PDO & PMM

Another aspect of the Pacific’s impacts on Atlantic tropical cyclone activity is the Pacific Decadal Oscillation (PDO). The PDO refers to the variation in SSTs in the mid-latitude Eastern Pacific. In the positive phase of the PDO, waters are warmer than average across the eastern part of the mid-latitude Pacific near the west coast of North America with cooler waters in the central North Pacific. In a negative PDO, the SST profile is the opposite with cooler than average waters off the west coast of North Pacific. The Pacific Meridional Mode (PMM) refers to the variance in SSTs between the subtropical Eastern Pacific and equatorial Pacific. More specifically, the PMM refers to how the region from Baja California to Hawaii compares with the equatorial Pacific and is typically in phase with the PDO. Additionally, this coincides with the Western Development Region (WDR; 10°–20°N, 116°W–180°) of the eastern North Pacific and has a strong correlation to hurricane activity with the most active hurricane seasons occurring when the PMM was positive (Collins et al. 2016). Given the known inverse relationship between eastern North Pacific and North Atlantic hurricane activity, accounting for the PMM is important in anticipating how the eastern North Pacific could impact potential activity in the North Atlantic.

Since 2014, the PDO and PMM have been consistently positive which has fueled a notable uptick in eastern North Pacific hurricane activity including record-breaking years such as 2015 and 2018. With this increase in activity, the Caribbean Sea (western MDR) has been greatly inhibited due to the presence of stronger vertical wind shear. Despite this, the tropical and subtropical Atlantic has remained active in recent years with the enhancement of the WAM and warmer subtropical Atlantic. In contrast to the last several years, persistent troughing off the west coast of North America provided notable cooling of the PMM and PDO during early 2020 with both observing their most negative values since 2013. With a cooler WDR in combination with the increasing likelihood of La Niña developing by peak hurricane season, it appears that the eastern North Pacific is slated for one of its most inactive years since 2011. In conclusion, the cooling of the PMM and PDO, to their coolest since 2013, would serve to increase the favorability of the MDR for hurricane activity especially as ENSO cools further into peak hurricane season.

IV. AEJ & Sahel Precipitation

The African Easterly Jet (AEJ) refers to the 700mb Jetstream that exists over the African Sahel region (15⁰N-20⁰N, 15⁰W-35⁰E) where 700mb winds are enhanced due to the contrast in temperature and moisture between the Sahara Desert and Gulf of Guinea during boreal summer (Sheen et al. 2017). This Jetstream is known for enhancing and transporting African Easterly Waves (AEWs) across the African continent and into the Atlantic Ocean. AEWs that emerge into the Atlantic account for most of the tropical cyclone activity that occurs in the basin on average (Landsea 1993). Given this, the differences in placement of the AEJ can have significant effects on the favorability of development of tropical waves into tropical cyclones. When the placement of the AEJ is further north, AEWs are enhanced due to the increased latitudinal distance from the equator that enhances vorticity via the Coriolis force. However, this can generate stronger Saharan Air Layer outbreaks especially during the early part of hurricane season (June to August) due to the enhanced easterly trade winds which can inhibit immediate TC formation as AEWs emerge. In contrast, when the AEJ is displaced south, tropical cyclone development from AEWs is suppressed once they have emerged into the Atlantic.

Useful late spring indicators of the favored placement and strength of the AEJ during hurricane season are African Sahel rainfall and the AMO (Martin & Thorncroft 2014). The wetness of the African Sahel can be quantified in the NCEP/NCAR dataset (1948-present) through the columnar precipitable water, or PWAT, values (kg/m3) which estimates the amount of condensed water vapor is present in a vertical column of air. The AMO, as mentioned in Section I, refers to the variance of SST anomalies across the North Atlantic Ocean. During a positive AMO (and AMM), the stronger meridional temperature gradient between the equatorial and tropical Atlantic enhances the AEJ and the West African Monsoon (WAM) which favors stronger zonal westerly anomalies across the tropical Atlantic during hurricane season. This can assist in a positive feedback of a wetter African Sahel which then enhances the WAM further and reinforces positive AMO growth during monsoon season.

Given the recent wetness in the African Sahel region, it was of interest to investigate if 2020 would continue this trend. Based on NCEP/NCAR PWAT data, 2020 observed the second wettest African Sahel since the last wet Sahel period in the 1960’s (seventh wettest on record) for March-April behind only 2019 (Figure 6). Further, the AMO configuration has rebounded from its negative winter state to a more traditionally positive appearance with a warming Canary Current and tropical Atlantic. With these elements in place, it is likely the North Atlantic will reap the same benefits as 2018 and 2019 with an enhanced AEJ and amplified WAM. This will further enhance the North Atlantic’s hurricane activity especially in the tropical Atlantic where these AEWs could develop quicker. The enhancement in activity in this part of the basin because of these factors has been exemplified in recent by Hurricane Helene in 2018 and Hurricane Lorenzo in 2019. Therefore, the state of the African Sahel suggests that the existing favorable state of the AEJ and WAM will be yet another factor among the overwhelming signals for an active North Atlantic hurricane season.

V. Climate Background State

Another variable of use in monitoring climate teleconnections in the tropics is the 200mb velocity potential pattern. 200mb velocity potential measures the convergent and divergent component of the wind field aloft which is useful in identifying regions of upward and downward atmospheric motions given the known principles of mass continuity in the atmosphere. Considering that tropical convection typically favors regions of upward motion (divergence aloft), it is valuable to better understand the evolution of these features in real-time. Since the conclusion of the Strong El Niño of 2015-16, the mean low-frequency (temporal scale of months to years) velocity potential pattern has favored a wavenumber three configuration with the strongest rising motion present over the African continent and weaker rising cells over the maritime continent and east Pacific (Figure 7). This configuration can be associated to the amplification of the West African Monsoon and wetter Sahel in recent years. This rising over Africa will only be further strengthened by the transition of the low frequency state toward a wavenumber one pattern in the developing La Niña atmosphere which has already become apparent in recent months (Figure 8). Rising over the African continent and Indian Ocean favors strong 850mb zonal easterly anomalies across the tropical and equatorial Pacific and enhanced 850mb zonal westerly anomalies across the tropical and equatorial Atlantic. The induced slowing of easterly trade winds in the Atlantic serve to enhance surface convergence. Further, the existence of the easterly trade wind belt directly north of the deep tropics has been observed as an enhancing factor of relative vorticity and tropical cyclogenesis accordingly in the MDR during hurricane season. Thus, it is no surprise that a velocity potential pattern such as this is consistent with the most active (seasonal ACE ≥150% of average) North Atlantic hurricane seasons from 1950 to present (Figure 9). With this pattern in place, and a less convectively active east Pacific due to La Niña, this could make the North Atlantic less dependent on intra-seasonal atmospheric forcing (such as the MJO) to induce TC outbreaks compared to recent years.

VI. Analog Years

With advancement in understanding of TC variability in the North Atlantic, the variables above that possess strong relationships to this variability can also be used to find commonalities in history to better understand the potential outcomes of the upcoming season. Through this approach, the identification of years which featured similar modes and climate backgrounds can be used to deduce which outcomes seem most likely.

The base characteristics used to identify analog years were La Niña or cooling ENSO into peak hurricane season, a quasi-negative PMM and PDO, and an Atlantic SST configuration where the warmest SST anomalies were between 20⁰N to 30⁰N and a cooler Canadian Maritime. Further, given the evident seasonality of the AMM recently, there is an element of anticipated warming expected in the MDR during the summer months which would yield a more positive AMO configuration. Another component that was accounted for in identifying analog years was the presence of a wet African Sahel. The years that best fit these characteristics are 1950, 1955, 1961, 1964, 1995, and 2007. The analog average activity yielded from these years was: 14.9 named storms, 8.4 hurricanes, 4.3 major hurricanes, and an ACE of 165.2 (*10^4 knots^2) (Table 1). This average would be considered a well-above average hurricane season with only 2007 not featuring above average activity out of the six years. The abundance of years selected that are prior to the current North Atlantic active era (1995-Present) were driven primarily by the similarity in the North Atlantic SST configuration and African Sahel wetness. The key difference in the regime observed in recent compared to the 1995-2013 period is the existence of a cooler tropical Atlantic and a stronger WAM. These characteristics are comparable to that of the 1950-69 active period according to Bell & Muthuvel (2006). However, this period differs in the placement of mean rising motion further east over the date line instead of the Maritime Continent and Indian Ocean during peak hurricane season which is not expected this year. The inclusion of 1995 was motivated by the similar evolution of ENSO and strengthening of a positive AMO SST configuration in the North Atlantic. 2007 is considered as a fail mode in the selected analog years primarily due to its particularly strong easterly trade winds in the tropical Atlantic enhanced by the onset of a strong La Niña, but differs from 2020 with regards to the wetness of the African Sahel. Other noteworthy years that differed in certain facets but were similar in evolution of ENSO and climate background state included 1954, 1959, 1988, 1998, and 2010.

The 2019 seasonal hurricane outlook introduced a new technique in forecasting seasonal ACE which invoked the highly correlated linear relationship between numbers of named storms, hurricanes, and major hurricanes versus ACE from 1950 to 2018. This was accomplished through a statistical linear regression model which could then be used in a predictive format for seasonal hurricane forecasts. When evaluating the relationship of each category to ACE, hurricane and major hurricane held the strongest correlations (r = 0.86 and r = 0.87). Further, there were some acknowledged biases in the model which were apparent when the proportion of named storms to hurricanes differed greatly from the mean. Thus, to correct some of this inherent bias, the hurricane and major hurricane categories were summed to create a new variable which held a linear correlation of 0.92 with respect to ACE from 1950 to 2019 (Figure 10). This new variable was then averaged with the output of the original model that included number of named storms to create the latest version, ACE Model v2, which will be used in this outlook.

Linear model relationship between ACE and number of major hurricanes (r = 0.87)

Seasonal ACE = 31.1 + 28.1 * MH

Linear model relationship between ACE and number of hurricanes (r = 0.86)

Seasonal ACE = -14.9 + 18.8 * HUR

Linear model relationship between ACE and number of named storms (r = 0.73)

Seasonal ACE = -12.8 + 9.9 * NS

The seasonal ACE forecast for 2019 yielded a value of 151 (*10^4 knots^2) from the 17 named storms, 9 hurricanes, and 4 major hurricanes predicted. With the official seasonal ACE finalized at 132.2, the percent error for the 2019 forecast was approximately -12%. By comparison, the prior two years of seasonal ACE forecasts in 2017 and 2018 prior to the use of this technique had forecast percent errors near 90% despite errors being relatively comparable in the named storm, hurricane, and major hurricane categories. Additionally, when applying the ACE model to the forecasts for 2017 and 2018, percent errors dropped to near 60% and 39%, respectively. Given the initial success of this technique in improving seasonal ACE forecasts, its use will be continued in seasonal hurricane outlooks henceforth (Table 2 & 3).