What is Hurricane Climatology?

Hurricane Climatology on Prezi

Hurricane clusters in the vicinity of Florida

Models that predict annual U.S. hurricane activity assume a Poisson distribution for the counts. This assumption leads to a forecast that under predicts both the number of years without hurricanes and the number of years with three or more Florida hurricanes. The under dispersion in forecast counts arises from a tendency for hurricanes to arrive in groups along this part of the U.S. coastline. We recently developed an extension to our earlier Poisson model that assumes the rate of hurricane clusters follows a Poisson distribution with cluster size capped at two hurricanes.

Hindcasts from the cluster model better fit the distribution of Florida hurricanes conditional on the climate covariates including the NAO and SOI. Results are similar to models that parameterize the extra-Poisson variation in the observed counts including the negative binomial and the Poisson inverse Gaussian, but we argue that the cluster model is physically consistent with the way Florida hurricanes tend to arrive in groups.

This research is done in collaboration with Thomas H. Jagger (Climatek) and is supported by the Risk Prediction Initiative. It is currently in review with the AMS Journal of Applied Meteorology and Climatology.

Statistical models for tropical cyclone activity

In collaboration with Gabriele Villarini and as part of the U.S. CLIVAR working group on hurricanes, I've written a short summary paper on statistical models for tropical cyclone activity. The paper is available here. The data for the R code are available here. The document was created using Sweave, LaTeX and R. The work is part of a larger project to publish a book on this topic with Oxford University Press. Comments and suggestions are certainly welcome.

Spatial grids for hurricane climate research

We demonstrate a new framework for studying hurricane climatology. The framework consists of a spatial tessellation of the hurricane basin using equal-area hexagons. The hexagons are efficient in covering hurricane tracks and provide a scaffolding to combine attribute data from tropical cyclones with spatial climate data. The framework's utility is demonstrated using examples from recent hurricane seasons. Seasons that have similar tracks are quantitatively assessed and grouped. Regional cyclone frequency and intensity variations are mapped. A geographically-weighted regression of cyclone intensity on frequency and SST (results shown here) emphasizes the importance of a warm ocean in the intensification of cyclones over regions where the heat content is greatest. The largest differences between model predictions and observations occur near the coast. The framework would be ideally suited for comparing tropical cyclones generated from different numerical simulations (see U.S. CLIVAR hurricane working group). The hexagons have equal area and are plotted on a map using the Lambert conformal conic projection with standard parallels of 23 and 38 degrees.

Geographers poised to lead a new revolution in hurricane climate research

In my talk this week to the West Florida chapter of the American Meteorological Society I make a case that geographers are well positioned to lead a new revolution in hurricane climate research. Watch here.

Shorebirds benefit from hurricanes

Tiny threatened shorebirds on Florida’s west coast not only survive hurricanes, they seem to benefit from the storms’ aftereffects, according to new research findings that contradict conventional wisdom.

The findings could have implications for beach nourishment efforts throughout the world and how they affect wildlife.---Mickie Anderson, UF media relations. Read more. Photo credit: Western Snowy Plover, Mike Baird, Morro Bay, CA USA.

How Can Solar Variability Affect Hurricanes?

An inverse relationship between hurricane activity over the Caribbean and the number of sunspots has recently been identified. Here we investigate this relationship using daily observations and find support for the hypothesis that changes in ultraviolet (UV) radiation are the cause. The relationship is statistically significant after accounting for annual variation in ocean heat and the El Nino cycle. A warming response in the upper troposphere to increased solar UV forcing, as measured by the Mg II core-to-wing ratio, decreases the atmosphere's convective available potential energy (CAPE) leading to a weaker cyclone. The response amplitude at a hurricane intensity of 44 m/s is 6.7 m/s +/- 2.56 m/s per 0.01 Mg II units (s.d.), which compares with 4.6 m/s estimated from the heat-engine theory using a temperature trend derived from observations. An increasing response sensitivity with increasing hurricane strength is found in the observations and in an application of the theory. Read more. Citation: Elsner, J. B., T. H. Jagger, and R. E. Hodges (2010), Daily tropical cyclone intensity response to solar ultraviolet radiation, Geophys. Res. Lett., 37, L09701, doi:10.1029/2010GL043091.

Oil Spills and Hurricanes

The oil slick over the northern Gulf of Mexico will reflect more incoming sunlight so will warm slower than the surrounding ocean. If this delayed warming continues into the hurricane season, then a tropical cyclone that visits the region might have slightly weaker winds. Reduced water evaporation from any remaining oil film at the time of the hurricane will contribute to the decrease in wind speeds assuming the high winds do not immediately disperse the oil.

Frequency and Intensity Changes

Today 20% of the strongest cyclones exceed 49 m/s on average globally. With a 1C rise in SST, 20% of the strongest cyclones could exceed 51 m/s according to Elsner et al. (2008). Thus the 80th percentile increases from 49 to 51 m/s. Today, on average, 17 cyclones/yr exceed 49 m/s and 13 exceed 51 m/s. If 51 m/s is the new 80th percentile (after a 1C warming) then, without a change in the overall number of cyclones, 13 becomes 17.

Environmental Signals in Property Damage Losses

The strongest Atlantic hurricanes are getting stronger as ocean temperatures warm (Elsner et al. 2008) and the strengthening is expected to continue (Knutson et al. 2010). However, along the U.S. coast the intensity of hurricanes has not gone up and there is considerable debate about potential future damage losses from these catastrophic events. Here we model the historical damage losses and show the magnitude of losses at a return period of 50 years is largest under a scenario featuring a warm Atlantic Ocean, a weak North Atlantic surface pressure gradient, El Nino, and few sunspots. Results are consistent with our current understanding of hurricane climate variability and they suggest a future of greater long-term loss potential if seas continue to warm.

Old Hurricanes

The record of past tropical cyclones provides an important means to evaluate the hurricane hazard. Historical chronologies are a source of information about tropical cyclones prior to the modern era. Chenoweth (2006) describes an archive of 383 tropical cyclones occurring during the eighteenth and nineteenth centuries, largely before the official hurricane record.

We demonstrate a novel way this archive can be used to articulate historical tropical cyclone activity across space. First, an event in the archive is assigned a series of latitude/longitude coordinates approximating the descriptive locations of the cyclone’s affect. Second, tropical cyclones from the modern record that approach these locations (modern analogs) are mapped. Third, a probable pathway and a realistic track of the archived event is created by averaging the modern analog tracks. As an example, the procedure is used to generate a map showing the tracks of the Atlantic tropical cyclones of 1766. Sensitivity of the methodology to changes in event location and event timing are considered.

Results show historical hurricane chronologies when combined with a history of cyclone tracks can provide new information about the older events not directly related to where the original information was gathered. When this new information is available for all cyclones it should help climatologists better understand long-term variations in tropical cyclone activity.

For more information see here.

Alternative Risk Transfer

Imagine an insurer of coastal properties concerned about the next hurricane season. A severe storm happens once every few decades; if the severe storm hits this year, she’ll have to pay out most of her money in claims. To help remain solvent, she could buy reinsurance. Alternatively, she could issue a catastrophe (cat) bond, which would pass the risk on to an investor. An investor could buy the bond valued at, say, $100,000; over time, the insurer would repay the bond with, say, 15% interest. If no hurricane hits during the year, the investor makes 15% on his investment. The insurer also turns a profit because she continues to collect premiums. But if this low-probability severe hurricane does hit, then the investor loses his $100,000, which is used by the insurer to pay claims.

A cat bond triggers payments based on the occurrence of a specified catastrophic event. Most cat bonds to date have been linked to hurricanes and earthquakes, but some have been issued to respond to mortality events. Capital raised by issuing a cat bond is invested in a safe security like a treasury bill, which is held by a special-purpose vehicle (SPV). A SPV is often a company created to execute specific financial transactions. The bond issuer holds a call option on the bond principal (option to buy all or part of the principal) in the SPV with triggers that are specified in the bond contract.

The triggers can be defined in terms of the insurance company's total losses from the catastrophe or some hazard event characteristic. If the defined catastrophic event occurs, the bond issuer can withdraw bond funds from the SPV to pay claims, and part or all of the interest and principal payments are forgiven. If the catastrophe does not occur the investor receives the principal plus interest equal to the risk-free rate (e.g., London Inter-Bank Offered Rate--LIBOR), plus a spread above LIBOR. Cat bond maturity is typically on the order of 1 to 5 years.

Catastrophe Finance: An Emerging Academic Discipline

The recent and on-going events in the world's financial markets demonstrate that finance theory remains far from perfected. Meanwhile, the threat of natural disasters continues to increase due to population growth, economic development, climate changes, geologic activity, and political unrest. To better understand and predict natural disasters and their consequences research and training are needed at the interface of geoscience and economics. New academic programs for graduate students in the area of catastrophe finance would help fill this need and could provide better tools and models for risk management and assessment. In turn, greater awareness of the geosciences by market professionals could help assist the spread of scientific knowledge. Importantly, such programs would train the next generation of professionals in finance and environmental organizations to use markets to the advantage of environmental programs and to anticipate the adverse consequences of financial innovation necessary for creating a sustainable future.

Eos, v90, 281-282. [membership required].

Listen to a BBC Radio 4 Podcast interview with Quentin Cooper.

A New Way to Define Anomalous Years

Recently we used networks to examine year-to-year relationships in hurricane activity. This requires mapping the time series of hurricane counts onto a network. In this way the network is physically related to the variation of hurricanes from one year to the next. This idea is relatively new and was introduced by Lacasa et al. [2008]. By doing this we address the following two questions: How can the occurrence of hurricane landfalls over time be examined from the perspective of network analysis? And, what advantages are gained from this perspective? The intellectual merit of the work is an advance in our understanding of historical coastal hurricane activity and the broader impact is a new method for identifying anomalies from time series data. The paper will appear in a forthcoming issue of Geophysical Research Letters. It is coauthored with Thomas Jagger and Emily Fogarty.

The picture shows the visibility network based on the time series of U.S. hurricane counts over the period 1851--2008. The colors indicate the node degree (number of links); 2 or less (red), 3--5 (orange), 6--10 (yellow), 11--20 (green), 21--30 (blue), and more than 30 (dark blue). The network suggests a novel way to think about anomalies in a time series. Years are anomalous not in a statistical sense of violating a Poisson assumption, but in the sense that the temporal ordering of the counts identifies a year that is unique in that it has a large count but is surrounded in time by years with low counts. Thus we contend that node degree is a useful indicator of an anomalous year. That is, a year that stands above most of the other years, but particularly above its "neighboring" years represents more of an anomaly in physical terms than does a year that is simply well-above the average. Node degree captures information about the frequency of hurricanes for a given year and information about the relationship of that frequency to the frequencies over the given year's recent history and near future. With this definition 1985 stands out as the most anomalous of the hurricane years with 1933, 1886, and 1964 also unusual.

Lacasa, L., B. Luque, F. Ballesteros, J. Luque, and J.C. Nuno (2008), From time series to complex networks: The visibility graph. Proc. Nat. Acad. Sci., USA, 105, 4972--4875.

Summit Summary

The 2nd International Summit on Hurricanes and Climate Change was held May 31-Jun 5th in Corfu, Greece. Judging by the level of science and by the participants enthusiasm for doing it again, the Summit was an overwhelming success. There were participants from 17 countries. The relative frequency of participants by country is plotted as a word cloud (using wordle). There were 25 participants from the USA, 8 from Germany and China, 7 from France, 6 from Australia and Bermuda, and 5 from the UK. A word cloud illustrating the relative frequency of keywords from the participants abstract titles is also shown.





Summary points arising from the talks and discussions (with help from Rick Murnane):

1. Although there were a few skeptics, most in attendance would agree that there appears to be an upward trend in the intensity of the strongest tropical cyclones worldwide. Although there does not appear to be a trend associated with the global frequency of tropical cyclones.
2. There is a growing appreciation among the participants for a significant feedback relationship between TC activity and weather and climate events on the inter-annual time scale and on spatial scales that extend across latitudes.
3. The availability of quality archival and proxy records of hurricane activity is increasing. These data hold the potential to provide new insights into the linkages between climate variation and tropical cyclone activity.
4. The idea that Atlantic tropical cyclone activity is controlled by changes in relative sea-surface temperature rather than local sea-surface temperature was discounted on physical and statistical grounds.

The Summit was sponsored by Aegean Conferences, the Risk Prediction Initiative, and Climatek.

Travel awards to help defray some of the costs for students were given to N. Jourdain, R. Hodges, H. Kim, S. Lavender, M. Lenard, J. Malmstadt, K. Sheitlin, A. Suzuki-Parker, E. Vincent, A. Werner. The talks are being made available here.

Plans are to have Springer publish an edited book of the proceedings.

Plans are underway to hold the 3rd International Summit on Hurricanes and Climate Change in 2011 on the spectacularly beautiful island of Santorini.

Follow the Summit

The 2nd International Summit on Hurricanes and Climate Change takes place in Corfu starting next Monday. We will archive podcasts of the talks here. The speaker list includes:

June 1
Session 1: M Chenoweth, K Walsh, JL McBride, F Chauvin, R Romero
Session 2: JB Elsner, JD Woodruff, AV Fedorov, K Oouchi

June 2
Session 3: RE Hart, B Soden, ME Mann, A Khain, J Done
Session 4: GJ Holland, YM Tourre, JP Kossin, S Gentile

June 3
Session 5: K Emanuel, SL Lavender, D Liu, B Owens, C Liu
Session 6: J Nott, C Welker, A Osso, AS Daloz

June 4
Session 7: AA Tsonis, K Scheitlin, DJ Vimont, C Wang, J Rumpf
Session 8: PS Chu, R Kumar, S Rahmstorf, MR Lowry

Prelude to the 2nd Summit on Hurricanes & Climate Change, Corfu, Greece

Florida Hurricanes and Damage Costs

Comments & Replies

Our paper entitled: "The increasing intensity of the strongest tropical cyclones" published last September in Nature (henceforth EKJ08) garnered considerable attention in the scientific community. Here are replies to some of the comments we received about this work.

Comment: "If the total number of storms has not changed, and the number of strong storms has increased from 13 to 17, then surely the number of weak storms must have decreased proportionately."
Reply: We used statistics rather than arithmetic. The rate of strong tropical cyclones is much smaller than the rate of weak tropical cyclones. As the ocean warms the stronger tropical cyclones can "borrow" a few cyclones from below the threshold intensity that will significantly increase the rate of the stronger tropical cyclones while not significantly reducing the rate of weaker cyclones.

Comment: "Intuitively the number of tropical cyclones exceeding the mean rate plus 2 times the standard deviation as was shown in previous studies (e.g., Kossin et al. 2007) should be equivalent to number of tropical cyclones exceeding some upper quantile level as shown in EKJ08."
Reply: The number of cyclones exceeding plus 2 times the standard deviation is positively correlated to the rate of cyclones. A basin with a lower rate of tropical cyclones will have fewer cyclones exceeding plus 2 times the standard deviation compared with a basin with a higher rate, while the number of cyclones exceeding the 90th percentile is independent of the rate. Thus it is not appropriate to compare the differences between Kossin et al. (2007) and EKJ08 using this approach.

Comment: "I've regressed the most intense TC per season on year and my results do not match those presented in EKJ08."
Reply: A regression of the most intense TC per season is not the same as quantile regression on year as was done in EKJ08 for the following reasons. a) Quantile regression minimizes a linear absolute deviation statistic rather than a quadratic statistic, and b) quantile regression treats each intensity value equally; no wind speed contributes more to the model fit.

Comment: "Your results are only marginally significant and there are many factors contributing to hurricane intensification."
Reply: That is correct, but all else being equal, a warming of the tropical oceans where tropical cyclones form should increase their intensity. Since the strongest tropical cyclones are, on average, closest to there theoretical maximum potential intensity it stands to reason that if there is a warming signal it should be most apparent in the tendency of the strongest cyclones. Moreover, statistical inference is concerned with drawing conclusions based on data together with prior assumptions. Arguments that include the basic physics of the role ocean heat plays in tropical cyclone intensity have more weight before the data are examined.

Comment: "The authors claim that the increasing trend is consistent with theory, yet numerical modeling studies suggest a different sensitivity of tropical cyclone intensity to warming."
Reply: Numerical models are not theory. They are based on theory, but require many ad hoc empirical arguments that put them into the realm of "scenario generators." The theory we have in mind is the 2nd law of thermodynamics.

Comment: "I'm surprised that the relationship between intensity and sea-surface temperature is not stronger."
Reply: The physics of cyclone intensification works against the correlative relationship. An active year of tropical cyclones will effectively remove warmth from the ocean so that a seasonal average temperature will not correlate as strongly with tropical cyclone activity as one might expect even though the physical causality is strong.

Comment: "Yet when you look at scatter plots of these SST series versus number of intense TC’s there is no relationship in the warmer SST, more intense TC’s direction."
Reply: We did not look at the number of TCs; we looked at the intensity. There is no theory for TC formation. However, given a TC, there is a nice theory for the efficiency of intensification. So, we focused on intensity rather than on frequency. Given a TC in a nearly optimal dynamic environment, we should expect to see it reach a higher intensity with warmer SST. If on average 10% of the storms get within 5% of their MPI and the MPI increases then we would see the strongest storms getting stronger, assuming all else stays the same.

Comment: "Here's a hypothetical, what if the predictor had been another quantity that also shows a significant trend over the period 1981-2006, I don't know...my weight, perhaps...would one be discussing what the physical meaning of a non-significant correlation between the two was?"
Reply: This example has little to do with the relationship of TCs to warming seas since in the latter there is a theory linking the two, whereas with your weight and TCs there is none. In science this makes a big difference.

Proceedings of the 1st International Summit on Hurricanes & Climate Change

Springer has announced the availability of a proceedings volume of the 1st International Summit on Hurricanes and Climate Change titled: Hurricanes and Climate Change edited by myself and Thomas Jagger. The volume is available from Springer.

A 4-day international summit on hurricanes and climate change took place in 2007 to discuss and debate various scientific issues related to this important topic. There were 77 attendees from 18 different countries who participated in the summit.

This volume is a sample collection of papers from talks that were presented. The chapters are organized around the broad hurricane-climate themes of empirical evidence, statistical methods, and numerical models. A major focus of the conference was the importance of statistical models for understanding how hurricane activity is changing and may change in the future. Emphasis was also placed on evidence of low-frequency swings in hurricane activity using historical and geological records. Results from various high-resolution numerical models, including a 20-km mesh model, were consistent in showing stronger hurricanes in a warmer future. Most numerical models indicate an overall decrease in the number of storms attributable to greater atmospheric stability and to a decrease in vertical mass flux.

The volume draws from an international community of scholars in the field of hurricane climate science. The science addresses a variety of perspectives ranging from the micro-physics of lightning to geological evidence of prehistoric storminess.