Postulating that impactful events are not necessarily extreme and vice versa, we discuss the steps to construct a database of disruptive events for southern Québec by comparing indicators of impactful events, gleaned from newspapers, and extreme events, as determined by statistical analysis of meteorological station observations. How do the timelines from newspaper keyword searches correlate with meteorological time series?
For the newspaper analysis, we search for keywords in newspaper articles for two time periods, in both French and English, by looking at timelines of words counts (e.g., “snowstorm”, “fire”, “drought”, “high wind”) and time series of meteorological indicators for a historical period (1880-1899) and modern times (1995-2014). Some of the roadblocks to using newspaper articles are briefly examined and the influence of these roadblocks to producing accurate timelines is assessed. Identifying the most discussed events in the contemporary newspaper reports then allows us to examine both the historical station data and the large-scale meteorological patterns associated with these events. Analysis of the newspaper records suggests that flooding events are the most written about in southern Québec and have the highest disruptive impact in both historical and modern times. Snowstorms and winter storms, while of comparable severity, are more disruptive in modern times than in the past.
For the statistical analysis, instrumental meteorological data from historical and modern stations are assessed for reliability and accuracy, and extreme events meteorological events extracted from the observational data. We then examine the large-scale meteorological leading to these extreme events with reanalysis reconstructions. We also compare the station observations with the reanalysis estimates to assess reanalysis accuracy.
We argue that the combination of the three types of information — newspaper articles, station observations and reanalysis data — creates a comprehensive view of disruptive, high impact and extreme weather events from both the meteorological, data-driven view and the human, social angle. We wish to see if the newspaper timelines cohere to meteorological time series. We do this through case studies of flooding and winter storm events, which we examine in more detail as to the meteorological conditions and descriptions of their effects on the populace in the region of southern Québec. In these case studies, changes in impacts, vulnerabilities and resiliencie are seen through the vocabulary of the newspaper articles.

