Relating drought indices to impacts reported in newspaper articles

Relating drought indicators and real-world impacts is fundamental for understanding and addressing drought vulnerability. We link drought indices and impacts from newspapers compiled in the Irish Drought Impacts Database (IDID) for the period 1900–2016. For three catchment clusters across the island of Ireland we link the Standardized Precipitation Index (SPI) with land-based impacts and the Standardized Streamflow Index (SSI) with water-based impacts by matching total reported articles per month with concurrent drought indices. Using logistic regression we find SPI-3 links best with land based impact reports, whereas SSI-2 links best with water-based impact reports. Catchments in the east/southeast display the highest sensitivity to land- and water-based impacts; however, in summer months at low deficits northwestern catchments show a higher likelihood of impact reports. In winter months the likelihood of water-based impacts is considerably greater than the land-based equivalent, particularly in east/southeastern catchments. Moreover, the likelihood of news-worthy drought impacts has changed over the 117 year period. More severe deficits are required to induce a high likelihood (0.6) of land- and water-based impacts in east/southeastern and southwestern catchments during 1961–2016 compared with 1900–1960. Largest changes emerge in the southwest with SPI-3 values of −2.51 (<−3.00) required to reach the high impact likelihood threshold in the pre (post) 1961 period. Even greater reductions are found for water-based impacts in the southwest with SSI-2 values associated with high impact likelihoods changing from −2.04 to −2.58. Conversely, for catchments in the northwest more moderate drought deficits result in high impact likelihoods for both land-based (from <−3.00 to −2.32 SPI-3) and water-based impacts (from <−3.00 to −2.29 SSI-2) for the 1961– 2016 period. These findings show the value of newspaper archives for understanding regional sensitivities to drought plus their potential for underpinning a near real-time, drought monitoring and warning system in Ireland