This meeting, the first of its kind in recent years, has a number of talks focusing on a variety of research issues related to severe weather. Much discussion centered on data quality, and how errors, biases and uncertainty can be accounted for, in the effort to make some robust conclusions about the climate variability of severe weather and its ingredients. Some of the significant challenges with the datasets of tornado and severe weather reports include:
- population bias and variability in the sources of reports
- sparsity of meteorological measurements
- institutional data verification and archiving
- inconsistent attribution of Enhanced Fujita (EF) classification to events
- the scales themselves (e.g. EF scale) are based on damage rather than being parametric, compounding the population bias.
Sample size of reliable reports is often an issue when attempting to quantify risk, but the alternative to only utilizing the ‘reliable’ record is to inflate any biases inherent in the longer term record. This is amply demonstrated when one examines the record of tornado reports for any jurisdiction. Sudden jumps in the number of reports in the record are at least largely due to the increase in awareness of tornadoes in recent years (some of which is influenced by the media, better communications and popular culture), and technology enabling the reporting of precise locations and analyses of events in near-real time. This often comes hand-in-hand with increases in untrained personnel putting themselves at risk by ‘chasing’ tornadoes, sometimes mitigating the value of the larger number of reports.
These challenges cause biases and errors in the data records, which then may be inflated if these records are then used for calibration of stochastic modelling for the purposes of risk assessment, without first correcting for the biases and quantifying some uncertainty around both the original data, and the corrected.
Means of circumventing this reporting problem include improvements in consistency of reporting both in time and spatially, more accurate representation of physical processes in modelling exercises, the use of a framework to assess the reliability of reports, and the development of rigorous statistical techniques to account for these biases before conducting further analysis. Despite the challenges in data quality and reporting frequency, there is useful information that can be gleaned about climate-scale processes and impacts on tornadoes, when one uses these techniques to analyze and interpret the available data. These insights, coupled with improvements in numerical modelling in recent years indicate that there may be growing scope to make seasonal predictions of the precursors to the initiation of severe convective weather.
It was particularly gratifying to note that there is interest in severe weather impacting many jurisdictions worldwide, with research into severe weather on 4 continents being represented at this meeting, including work in developing nations, small island states, and countries with a longer experience of recording tornadoes. In addition, there were presentations of work being conducted to assess socio-economic factors which affect the vulnerability to tornado impacts, and also novel methodologies to objectively assess damage which has been inflicted by tornadoes.
Download the full report from the right of this page where, we present summaries of the talks of most interest, grouped by subject. For more information, please contact us at firstname.lastname@example.org.