Predicting spring tornado activity in the Central Great Plains by March 1st

Here we illustrate a statistical model for predicting tornado activity in the central Plains by March 1st.  The model predicts the number of tornado reports during April--June using February sea-surface temperature (SST) data from the Gulf of Alaska (GAK) and the western Caribbean region (WCA).  The model uses a Bayesian formulation where the likelihood on the counts is a negative binomial distribution and where the non-stationarity in tornado reporting is included as a trend term plus first-order autocorrelation.  Posterior densities for the model parameters are generated using the method of integrated nested Laplacian approximation (INLA).  The model yields a 51% increase in the number of tornado reports per degree C increase in SST over the WCA and a 15% decrease in the number of reports per degree C increase in SST over the GAK.  These significant relationships are broadly consistent with a physical understanding of large scale atmospheric patterns conducive to severe convective storms across the Great Plains.  The SST covariates explain 11% of the out-of-sample variability in observed F1--F5 tornado reports.  The paper demonstrates the utility of INLA for fitting Bayesian models to tornado climate data.  The research was conducted in the Department of Geography at Florida State University in collaboration with Holly Widen.  It will be published later this year in the American Meteorological Society's Monthly Weather Review.  The code is available from http://rpubs.com/jelsner/4745.

Predicting the distribution of violent tornadoes

Here we illustrate a statistical point process model that uses the spatial occurrence of non-violent tornadoes to predict the distribution of the rare, violent tornadoes during springtime across the U.S. central Great Plains. The average rate of non-violent tornadoes is 55 per 10000 square km per 62 years which compares with an average rate of only 1.5 violent tornadoes per 10000 square km over the same period (less than 3%). Violent tornado report density peaks at 2.6 per 10000 square km (62 yr) in the city to 0.7 per 10000 km in the countryside. 

The risk of a violent tornado is higher by a factor of 1.5, on average, in the vicinity of less violent tornadoes after accounting for the population bias. The model for the occurrence rate of violent tornadoes indicates that rates are lower by 10.3 (3.6, 16.5)% (95% CI) for every 1 km increase in distance from nearest non-violent tornado controlling for distance from nearest city. Model significance and distance-from-nearest non-violent tornado parameter are not sensitive to population threshold or definition of violent tornado. We show that the model is useful for generating a catalogue of touchdown points that can be used as a component to a tornado catastrophe model.

The research was done in collaboration with Richard Murnane, Thomas Jagger, and Holly Widen at Florida State University and will be published later this year in the journal Mathematical Geosciences.

Hurricane Climatology: A Modern Statistical Guide Using R

Our newest book is now available.  Our website provides the code used to produce all the figures.  Learn to code.  Enjoy!

Decreasing Population Bias in Tornado Reports

Tornado-hazard assessment is hampered by a population bias in the available data.  We demonstrate a way to statistically quantify this bias using the ratio of city to country report densities. The expected report densities come from a model of the number of reports as a function of distance from nearest city center. On average since 1950 reports near cities with populations of at least 1000 in a 5.5 deg latitude by 5.5 deg longitude region centered on Russell, KS exceed those in the country by 70% (54%, 84%) [95% CI].

The model is applied to 10-year moving windows to show that the percentage is decreasing with time (see Figure).  Over the most recent period (2002-2011) the tornado report density in the city is slightly less than 3 reports per 100 square km per 100 years and this value is statistically indistinguishable from the report density in the country. On average the population bias is less pronounced for F0 tornadoes, but the bias disappears more quickly over time for the F1 and stronger tornadoes.  We show evidence that this decline could be related in part to an increase in the number of storm chasers. The population-bias model can enhance the usefulness of the Storm Prediction Center's tornado database and help create more meaningful spatial climatologies.

The research was done in collaboration with Laura E. Michaels, Kelsey N. Scheitlin, and Ian J. Elsner.  It will be published in the American Meteorological Society's Weather, Climate and Society journal later this year.

Consensus on Climate Trends in Western North Pacific Tropical Cyclones

Research on trends in western North Pacific tropical cyclone (TC) activity is limited by problems associated with different wind speed conversions used by the various meteorological agencies.  Here we use a quantile method to effectively overcome this conversion problem. Following the assumption that the intensity ranks of TCs are the same among agencies, quantiles at the same probability level in different data sources are regarded as having the same wind speed level. Tropical cyclone data from the Joint Typhoon Warning Center (JTWC) and Japan Meteorological Agency (JMA) are chosen for research and comparison. Trends are diagnosed for the upper 45% of the strongest TCs annually. The 27-yr period beginning with 1984, when the JMA began using the Dvorak (1982) technique, is determined to be the most reliable for achieving consensus among the two agencies regarding these trends. The start year is a compromise between including as many years in the data as possible, but not so many that the period includes observations that result in inconsistent trend estimates. The consensus of TC trends between the two agencies over the period is interpreted as fewer but stronger events since 1984, even with the lower power dissipation index (PDI) in the western North Pacific in recent years.  Read more.

The Spatial Pattern of the Sun-Hurricane Connection

We define the spatial response of hurricanes to extremes in the solar cycle. Using an equal-area hexagon tessellation, regional hurricane counts are examined during the period 1851–2010. The response features fewer hurricanes across the Caribbean, Gulf of Mexico, and along the eastern seaboard of the United States when sunspots are numerous. In contrast fewer hurricanes are observed in the central North Atlantic when sunspots are few. The sun-hurricane connection is as important as the El Niño Southern Oscillation in statistically explaining regional hurricane occurrences.  Read more.

Hurricane Sandy and climate change

While the SSTs did not cause Sandy to curve into New Jersey, they quite likely caused Sandy to be stronger. Our new research shows that the limiting intensity of hurricanes (how strong hurricanes can get as a statistical limit) relates to SST at about 8 m/s/C.  With SSTs in the path of Sandy that were 2-3 C warmer than is typical, we would predict a strong hurricane to be twice as strong on average.  Read more.