Historical droughts in Irish catchments 1767–2016

Abstract

Recent prolonged dry periods in summer 2018 and spring 2020 have reawakened interest in drought in Ireland, prompting questions regarding historical drought occurrence and potential long‐term risks. Employing 250 years of monthly precipitation and flow reconstructions, we investigate historical drought in Irish catchments evaluating the characteristics (number of events, duration, and deficits) of moderate, severe, and extreme droughts as well as the propagation of meteorological to hydrological drought. Using standardized indices, we identify three distinct catchment types. Cluster 1 catchments, located in the wetter northwest are characterized by small areas, low groundwater storage, and the highest frequency of hydrological drought relative to other catchments. Cluster 3 catchments, located in the drier east and southeast have larger areas, greater groundwater storage, the highest frequency of meteorological drought but the least hydrological droughts. However, once established, droughts in Cluster 3 tend to be more persistent with large accumulated deficits. Cluster 2 catchments, located in the southwest and west, are intermediate to Clusters 1 and 3, with hydrological droughts typically of shorter durations, reduced accumulated deficits but greater mean deficits. The most extreme droughts based on accumulated deficits across all catchments occurred in 1803–1806, 1854–1859, 1933–1935, 1944–1945, 1953–1954, and 1975–1977. Although not as severe, droughts in 1887–1888, 1891–1894, and 1971–1974 also appear as significant extremes. Changes in drought characteristics reveal a complex picture with the direction, magnitude, and significance of trends dependent on the accumulation period used to define drought, the period of record analysed, and the reference period used to standardize indices. Of particular note is a tendency towards shorter, more intense meteorological and hydrological droughts. Our findings offer important insight for drought and water management in Ireland given the paucity of extreme droughts in short observed river flow records.

Relating drought indices to impacts reported in newspaper articles

“…These findings are consistent with O'Connor et al . (2022) who identify Cluster 1 catchments as being the most susceptible to hydrological drought in summer due to the lack of groundwater storage. Cluster 3 catchments show the highest likelihoods of impact reporting from September through April where even in December at more extreme deficits there exists a very high chance of drought related impact reports occurring.…”

“…(2021) also produced uncertainty estimates for flow reconstructions by applying different model structures and parameter sets; here we use the available ensemble median flow reconstruction for each catchment. Previous hierarchical cluster analysis of the SSI‐1, 3, 6 and 12 during 1767–2016 identified three dominant catchment clusters for Ireland from the same 51 test catchments (O'Connor et al ., 2022). To allow for a comparison of results between the studies, we conduct our analysis using the same cluster groupings (see Figure 1a).…”

“…The 70‐year reference period (1930–1999) and the Tweedie distribution, both found by O'Connor et al . (2022) to perform best at fitting SPI and SSI in Irish catchments, were employed to generate indices. Extracted monthly SPI and SSI series for 1767–2016 were subsequently truncated to 1900–2016, concurrent with the derived drought impact data discussed next.…”

“…Median monthly flow and precipitation were extracted for catchments comprising each identified cluster. As per O'Connor et al (2022), standardized drought indices for 1767-2016 were applied to median monthly precipitation and flow data in each cluster. These were the Standardized Precipitation Index (SPI; McKee et al, 1993) and Standardized Streamflow Index (SSI; Vicente-Serrano et al, 2012) over accumulations of 1, 2, 3, 4, 5, 6, 9, 12 and 18 months.…”

“…Drought indicators have been extracted from reconstructed flows to assess historical droughts in a number of studies (e.g., Caillouet et al ., 2017; Rudd et al ., 2017; Hanel et al ., 2018; Moravec et al ., 2019; Erfurt et al ., 2020; O'Connor et al ., 2022). However, knowledge of drought characteristics alone does not necessarily translate into socio‐economic impacts.…”