https://orcid.org/0000-0001-6945-7306
Figures
Abstract
Intense weather events are projected to increase as a consequence of climate change. The summer 2018 drought in Europe impacted human health, ecosystems, and economic prosperity. Even locations with an abundance of fresh water, like Ireland, faced water restrictions due to depleted supplies. To characterize the effect of the 2018 drought on Irish rivers, we collected surface water samples from rivers across the island at the drought onset and termination. We analyzed samples for stable water isotopes δ18O and δ2H and calculated the fraction of evaporation from river groundwater and precipitation inflow (E/I) of rivers. We extended river δ18O and δ2H analysis to 2020 for rivers in two catchments, Corrib and Shannon, to investigate how Irish river systems respond to high precipitation events, and the role of loughs (lakes) in the system. River δ18O and δ2H values showed progressive depletion from west to east in response to precipitation depletion from airmasses arriving off the Atlantic Ocean. From onset to termination of the 2018 drought, river δ18O and δ2H values were enriched and the calculated E/I value increased for most rivers. D-excess were negatively correlated with E/I value, providing support for E/I calculations. Extended analysis of loughs along the Corrib and Shannon river systems showed that lough Corrib consistently induced isotopic enrichment, while loughs in the Shannon catchment inconsistently caused isotopic enrichment. Both systems exert control over river isotopic composition in hydrologic extremes. Findings promote additional research in hydrologic patterns in response to increasing frequency of floods and droughts.
Citation: Smith DF, Lyons WB, Henry T, Flynn R, Carey AE (2023) Irish surface water response to the 2018 drought. PLOS Water 2(11): e0000197. https://doi.org/10.1371/journal.pwat.0000197
Editor: Debora Walker, PLOS: Public Library of Science, UNITED STATES
Received: February 28, 2023; Accepted: October 11, 2023; Published: November 14, 2023
Copyright: © 2023 Smith et al. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Data Availability: All river δ18O and δ2H values are reported in this study. Upon publication content from S1 and S2 Tables will be available for open access through the Consortium of Universities for the Advancement of Hydrologic Science, Inc. HydroShare (https://www.hydroshare.org/). The data can be accessed upon manuscript publication at https://doi.org/10.4211/hs.aa3a731c51f44e4aba637e36e802117e. Data will also be uploaded to waterisotopes.org database. Data Further description of evaporation fraction calculation is provided in Diamond and Jack (2018).
Funding: Sample collection was conducted while WBL was supported Fulbright-GSI Fellowship at University of Galway. Funds through a Faculty Professional Leave from The Ohio State University College of Arts and Sciences also supported WBL.DFS was partially supported by internal fellowships at The Ohio State University. Funding for instrument purchase came from NSF GEO EAR/IF 0930016 and 1707989 (AEC, WBL). The authors received no specific funding for this work. The funders had no role in study design, data collection and analysis, decision to publish or preparation in the manuscript.
Competing interests: The authors have declared that no competing interests exist.
Introduction
As a result of future climate change, the incidence of extreme heat and drought around the world will increase [1]. Droughts have negative effects on terrestrial and aquatic communities, food and water resources, human health, socio-economic prosperity, infrastructure stability, and alter biogeochemical processes [2, 3]. Up to 50% of the world’s regions, both arid and humid, could experience more frequent drought conditions by 2050 according to modelled water balance changes [2]. However, the increase extent and severity is uncertain because it is difficult to quantify the temporal and spatial extent of drought [4, 5]. The onset of a drought can be slow, on the order of months to seasons, and the spatial distribution depends upon local climate and antecedent conditions [3]. Alternatively, flash droughts can occur, where the onset of drought is quick, on the order of weeks, caused by precipitation deficits and high evapotranspiration [6]. Because of the uncertainty in quantifying drought there is no universal definition for the term. Instead, droughts are classified by precipitation deficits from atmospheric evaporative demand indices (meteorological drought), soil moisture indices (agricultural drought), and hydrological indices (hydrological drought) [3, 7]. Less attention has been paid to drought in regions that are typically thought of as “water-rich”. The occurrence of water scarcity is not only determined by the total amount of water delivery, but it is also dependent on natural and human population water demand. Since drought is amplified by human demand on water resources, “water rich” countries can experience water scarcity in drought conditions even though the quantity of water delivered is greater than arid regions.
The island of Ireland, which includes the Republic of Ireland and Northern Ireland and will hereon be referred to as Ireland, can be classified as a “water rich” island, as it receives ~700 and >1200 mm precipitation annually. It is an energy-limited environment with no distinct wet or dry season, however precipitation amount is weighted toward winter months, and there is a precipitation a gradient that decreases from west to east [8, 9]. Despite annually consistent patterns and abundant precipitation, three centuries of droughts and multi-season precipitation deficits in Ireland have been reconstructed [10–13]. Research shows that Ireland will have a higher incidence of summer droughts and winter floods due to more uneven delivery of annual precipitation [14, 15], threatening water security in Ireland.
In 2018, the continent of Europe experienced extreme drought conditions, with the most severe precipitation deficits and elevated temperatures occurring in northern and central Europe [16]. In Ireland, precipitation deficits and heat waves were as extreme, and the conditions were categorized as a meteorological drought [17–19]. These conditions were consistent in the UK, which led to excess water demand that over extended supply [20]. Surface waters provide 80% of drinking water in the Republic of Ireland [21], and elevated water demand coincided with extreme low river discharge, which raises the question: how did surface waters respond to compounding environmental stress and human demand in the 2018 drought?
Stable water isotopes, δ18O and δ2H, are a powerful natural hydrologic tracers that can be used to trace water source and investigate hydrologic response to seasonal meteorological changes or events. We utilize stable water isotopes to investigate surface water response to the 2018 drought. Precipitation amount and temperature have strong control on precipitation event δ18O values, and North Atlantic Oscillation (NAO) has strong control on monthly precipitation δ18O values [8, 22]. Previous work conducted by Diefendorf & Patterson [23] presented a snapshot of Irish surface water isotopic composition in 2005, and showed that the rivers are a reflection of local precipitation. More recent work by Regan et al. [24] supported these results by showing that groundwater δ18O and δ2H values are also a reflection of local precipitation, but are influenced by orographic lifting, precipitation volume, biased winter recharge, and hydrogeologic setting. Global-scale modelling has shown that the isotopic composition of precipitation in Ireland is ~6‰ for δ18O and ~40‰ for δ2H [25]. These studies provide a foundation for the work presented herein.
The goal of this research is to evaluate the change in river water isotopic composition in relation to surface water deficit (supply < demand) and calculate potential evaporative water loss during the 2018 drought in Ireland. Sampled rivers were used to document shifts in Irish river isotopic composition at the onset and termination of the 2018 drought. We hypothesized that river water isotopic composition would not significantly change from the onset to termination of the 2018 drought from an evaporative effect. Objectives to test this hypothesis for this research were to (1) identify differences in Irish river δ18O and δ2H values and (2) calculate the evaporative fraction for Irish rivers in May and August 2018. We also present a δ18O and δ2H dataset from March 2018 –March 2020 for rivers in the Corrib and Shannon catchments to identify the control of loughs (lakes) on Irish surface water systems in drought and flood conditions. Rivers were sampled after extratropical storm Lorenzo passed over Ireland in October 2019 and after one of the wettest winter months on record in March 2020. We hypothesized that rivers draining large loughs in the Corrib and Shannon surface water systems would have enriched isotopic composition from evaporative enrichment.
Methods
Sample conditions
Summer 2018.
In mid-May two compounding atmospheric circulation patterns (1) positive summer North Atlantic Oscillation and (2) northern hemisphere high amplitude Rossby Waves, called Wave–7 pattern, formed anticyclonic conditions over Ireland [28, 29], causing prolonged dry weather. Few storms passed over the island but the timing of precipitation was not evenly distributed throughout the island. June consistently had the lowest monthly cumulative amount of precipitation measured at stations across the island (Fig 1). Ireland received rainfall again towards the end of July that resulted in the termination of the summer drought [17].
Precipitation amounts from May, June, and July 2018 are represented as pie charts for Irish synoptic weather stations and Northern Ireland weather stations [19, 26]. Numbers in boxes next to each pie chart show the total precipitation amount (mm) from May, June, and July 2018. The color of the box corresponds to the month when most precipitation was measured, as shown on the pie chart. The table shows the percent of summer (May, June, July) precipitation measured at each station as the percent long–term average. Long–term average is classified as 30-year average (1981–2010) by Met Éireann Meteorological Services. LTA for Northern Irish weather stations was calculated over the same period with monthly total precipitation (mm). All projected spatial data, including surface waters, catchments, and island boundary, were sourced from the Irish Environmental Protection Agency (EPA) and downloaded from EPA Geoportal (gis.epa.ie/GetData/Download) (CC-BY 4.0; https://creativecommons.org/licenses/by/4.0/legalcode).
Met Éireann (the Irish Meteorological Service) has identified three classes of drought: (1) a dry spell is a period of +15 consecutive days with less than 1 mm of precipitation; (2) an absolute drought is a period of +15 consecutive days with less than 0.2 mm precipitation; and (3) a partial drought is a period of +29 consecutive days with <0.2 mm precipitation per day [30]. In summer 2018 heat wave conditions were recorded at 15 weather stations between June 24 and July 4. From May 22 to July 14, 21 stations recorded absolute droughts conditions. From May 28 to July 25, ten stations recorded partial drought conditions and five stations recorded dry spell conditions between June 18 and July 14 [17].
October 2019.
Extratropical Storm Lorenzo passed over Ireland on October 3 and 4, 2019. Precipitation from the storm accounted for approximately 2 to 10% of October long–term averages long–term average at synoptic weather stations. The storm caused river flows ≥ 50 percentile values (Q50) [31].
March 2020.
February 2020 was one of the wettest winter months on record [19]. Three extratropical cyclones pass over Ireland, and precipitation totals were 155 to 332% of 1981–2010 year long–term average [32]. In response to a high volume of precipitation in February, river flows were between 10 percentile (Q10) and 1–percentile (Q1) values and flooding was widespread across the island [31].
Sample collection
Surface water samples (n = 148) were collected from 40 rivers during 2018–2020 (Fig 2). Rivers across Ireland were sampled in the summer (May–August) of 2018, while additional sampling was conducted in the Corrib and Shannon catchments in March 2018, March and October 2019, and March 2020 (S1 Table). Detail of Corrib and Shannon river systems and catchment characteristics can be found in S1 Text [32–36]. Not all locations were sampled in each sampling campaign. Samples were collected by hand in 20mL HDPE scintillation vials with urea caps or with a high-density polyethylene (HDPE) sampler. Samples collected in the sampler were transferred to 20 mL HDPE scintillation vials for transport and storage. Samples had no headspace to prevent evaporation and were shipped back to The Ohio State University, Columbus Ohio for analysis and analysed within a week of arrival. River samples were taken from public access points and no permits or approval were required for this research. Additional information regarding the ethical, cultural, and scientific considerations specific to inclusivity in global research is included in the supporting information (S1 Checklist).
River sample locations from this study and groundwater δ18O sample locations from Regan et al. [24]. Samples collected in the Corrib and Shannon catchments are displayed as yellow circles. Ireland synoptic weather stations and Northern Ireland weather stations with historical data are shown as green diamonds [19, 26]. GNIP δ18O and δ2H data were collected and measured at Valentia and Armagh, which are labelled on the map [27]. Lough (lake) names in the Corrib and Shannon catchments are noted next to the loughs in blue. All projected spatial data, including surface waters, catchments, and island boundary, were sourced from the Irish Environmental Protection Agency (EPA) and downloaded from EPA Geoportal (gis.epa.ie/GetData/Download) (CC-BY 4.0; https://creativecommons.org/licenses/by/4.0/legalcode).
δ18O and δ2H analysis
Samples were analysed on a Picarro Wavelength Scanned-Cavity Ring Down Spectroscopy Analyzer Model L1102-I for δ18O and δ2H. A sample aliquot of 2 mL was used for analysis, where seven injections of 2.3 μL were made per sample. The first three injections were discarded to avoid memory effects and the last four injections were averaged to obtain the sample δ18O and δ2H raw values. The sample δ18O and δ2H were corrected by internal laboratory standards that had been calibrated to VSMOW (δ18O = 0‰, δ2H = 0‰) at the Institute of Arctic and Alpine Research (INSTAAR) at University of Colorado at Boulder by a dual inlet mass spectrometer. Internal laboratory standards were Colorado (δ18O = -16.53‰; δ2H = -126.3‰), Nevada (δ18O = -14.20‰; δ2H = 104.80‰), Ohio (δ18O = -8.99‰; δ2H = -61.80‰), and Florida (δ18O = -2.09‰; δ2H = -9.69‰). Two internal standards were run at the beginning and end of each sample run and after every fifth sample. Sample duplicates were run for every tenth sample. Instrument precision was determined by Picarro precision tests, where 150 deionized (DI) water samples were run to calculate instrument precision. The precision was 0.016‰ for δ18O and 0.15‰ δ2H. Accuracy for samples was determined with sample duplicates and values were ≤0.83‰ and ≤2.5‰ δ18O and δ2H, apart from one run where the δ2H precision of March 2019 samples is ≤3.9‰. Deuterium excess (d-excess) values (d-excess = δ2 H − 8 × δ18 O) were calculated for all samples. Local evaporation line was calculated for all river sample locations draining major loughs in the Corrib and Shannon catchments [37] (Fig 2). Sample populations from each sampling campaign were not normally distributed and a Kruskal-Wallis test [38] was used in MATLAB 2019b to identify statistical differences among sample collection groups. All data are reported in S1 Table.
Spatial analyses
Sample point drainage area, defined as the within-catchment area above the sample point, was delineated in ArcMap 10.6.1. All projected spatial data, including Irish Environmental Protection Agency (EPA) Water Catchment shapefiles were downloaded from EPA Geoportal (gis.epa.ie). The EU-DEM v1.1 was used to delineate the within-catchment drainage area [39]. Delineated drainage basin areas are reported in S2 Table and methods are detailed in S2 Text [33, 40, 41].
Isoscape generation
Maps showing the distribution of river water δ18O and δ2H composition (isoscapes) were generated for Ireland in May and August 2018. Isoscapes were created for the Corrib catchment in March 2019, October 2019, and March 2020 and for the Shannon catchment in March 2018, June 2018, October 2019, and March 2020 were also constructed. Isoscapes were generated at 100 m2 resolution with a minimum-curvature spline methods with barriers with a smoothing factor of 0.1 in ArcMap 10.6.1 [42].
Evaporation fraction calculation
An evaporative fraction , also referred to as the E/I values, was calculated for all rivers sampled in both May and August 2018. Calculations were modelled after Diamond and Jack [43], which are founded on the fundamentals of Craig and Gordon [44]. The fraction of water evaporated compared to groundwater and precipitation inflow water (E/I) was calculated with equations detailed in S3 Text [43–48]. The analytical propagated error for E/I calculations was 8.3% of reported E/I values. Additional uncertainties were included to calculate a total propagated uncertainty for May and August E/I values with Valentia and Armagh precipitation: relative humidity, temperature, groundwater δ18O values [48], and GNIP 2018 precipitation δ18O values. Uncertainty for May E/I values was 32% of the E/I reported values for Valentia and Armagh precipitation inputs. Uncertainty for August E/I values was 81% (Valentia) and 55% (Armagh) of the reported E/I values. Propagated error for calculated August E/I values was greater because the July 2018 precipitation δ18O values were more enriched or depleted compared to July averages at Valentia (-4.05‰) and Armagh (-6.79‰) [27].
Weather and discharge data
Precipitation, temperature, and relative humidity values were obtained from Met Éireann Weather Observing Stations (synoptic stations) and Rainfall Observation Stations in the Republic of Ireland and UK Met Office Climate Stations [19, 26]. Long–term averages for monthly and annual precipitation amounts are reported for the period 1981–2010. Long–term average values are reported by Met Éireann for each synoptic station and were used to classify summer 2018 precipitation deficits [17].
Discharge data were available for 24 of the sampled rivers. Data were obtained from the Office of Public Works (OPW) Hydro-data Archive [31] and the Irish Environmental Protection Agency (EPA) Hydronet [49]. Discharge data for Athlone were calculated from a rating curve data generated from OPW discharge and gauge height data. Fitted rating curve flows may deviate from flow measurement by 10–20% and high values are determined by extrapolation of the curve (personal communication, OPW Hydrometric Section). Discharge percentiles were calculated and reported for the 24 rivers with discharge measurements provided by the Office of Public Works, with the exception of the river Shannon at Dowra [31].
Results
Summer 2018 river discharge conditions
River flows began to decline in April 2018 and continued to fall their lowest point in late June and July (Fig 3) [31]. All rivers had at least 26 days of average daily discharge rates below 75 percentile (Q75) flow values. Thirteen rivers had more than 90 days of average daily discharge below Q75. All the rivers had more than 10 days of average daily discharge lower than Q90, except for Rivers Suck and Slaney. The greatest number of days with average daily discharge values less than Q90 was the River Boyne. Rivers on the western side of the island, in the Corrib catchment and River Boyle, had the fewest days when discharge <Q90. River discharge of <Q75 is denoted as low flow rather than baseflow, which is defined as slow, consistent water input from multiple sources that sustain river flow between water–input events [50]. Flow values <Q75 are referred to as low flows instead of baseflow because several precipitation events occurred across the island between May and August. Rivers had notable low flows throughout the drought period, but events with small precipitation volumes may have contributed to river flow (Fig 3).
Data were obtained from the Office of Public Works [31] and EPA Hydronet [49]. (a) Discharge data for 2018 presented for major Irish rivers outside of the Corrib and Shannon catchment with low percentile flow Q99, Q90 and Q75 values displayed in red, blue, and yellow lines. Vertical orange bars show dates of sampling campaigns. (b) Discharge data for 2018–2020 for Corrib and Shannon catchment rivers with Q10 and Q1 flow values displayed in purple and pink in addition to the Q99, Q90, Q75 discharge thresholds. River Athlone only shows Q1 flows. Vertical orange bars show dates of campaigns when rivers were sampled. Gaps in hydrograph indicate missing data. Discharge measurements displayed as a grey line indicate discharge data was estimated.
River δ18O and δ2H values
Local meteoric water lines (LMWLs), which define precipitation isotopic composition in relation to water sources, local geographic, and topographic variables have been published by the IAEA at Armagh [δ2H = 7.49(±0.14) × δ18O + 5.38(±1.10)] (R2 = 0.97) and Valentia [δ2H = 7.01(±0.09) × δ18O + 2.99(±0.5)] (R2 = 0.91) [27] (Figs 2 and 4). Both LMWLs were calculated with a least squares regression methods [27]. The Armagh LMWL was generated from 51 samples collected monthly since 2012 to 2019 and has a standard error of 2.6‰. The Valentia LMLW has a standard error of 3.4‰, and it is generated from 542 samples collected monthly from 1960 to 2018, with consistent monthly collection since 1977 [27].
(a) Samples collected 2018–2020 plotted with the global meteoric water line (GMWL) [δ2H = 8 × δ18O + 10] GNIP Armagh [δ2H = 7.49(±0.14) × δ18O + 5.38(±1.10)] and Valentia [δ2H = 7.01(±0.09) × δ18O + 2.99(±0.5)] local meteoric water lines (LMWL) in Northern Ireland and the Republic of Ireland, respectively [27]. A calculated local evaporation line [δ2H = 5.58 × δ18O ‒ 4.47] from sample locations draining major loughs in the Corrib and Shannon catchments is presented. (b) River Corrib samples collected 2018–2020 plotted with GMWL and LMWLs. Samples collected in 2018 are shown as circles, sample collected in 2019 are shown as squares, and samples collected in 2020 are shown as diamonds.
River samples fall on or above the LMWLs. The distribution of river sample δ18O values and groundwater δ18O were significantly different (p<0.05) than GNIP precipitation data [27]. River sample δ18O values and groundwater δ18O were not significantly different. River data plotted in isospace show that river samples reflect historical precipitation isotopic composition despite statistical differences in the datasets (Fig 4). The surface water line is δ2H = 6.06 × δ18O – 1.42 and the local evaporation line is δ2H = 5.58 × δ18O – 4.47 (Fig 4). Statistical differences of δ18O and δ2H among sample groups were inconsistent (S3 Table). The range of δ18O and δ2H data for most sampling campaign datasets overlapped, particularly for campaigns with a greater number of samples collected (S2 Fig). The majority of data from summer months plot below the GMWL and fell into the same range of summer 2003 data (δ18O = -7.4 to -2.4‰ δ2H = -53 to -17‰) [23], but the most enriched sample values in summer 2018 were -5.08‰, -30.27‰ in May and -3.56‰, -25.26‰ in August (S1 Table, Fig 5). The most depleted δ18O and δ2H values of rivers were measured in March and October (Fig 4, S2 Fig).
The color gradient of red to blue indicates a shift from enriched to depleted δ18O and δ2H values (‰). Dashed contour lines and corresponding numbers show contour intervals of river water δ18O (1‰), δ2H (5‰), and d-excess (1‰) values. All projected spatial data, including surface waters, catchments, and island boundary, were sourced from the Irish Environmental Protection Agency (EPA) and downloaded from EPA Geoportal (gis.epa.ie) (CC-BY 4.0).
The isoscapes generated from May and August sampling campaigns showed a pattern of δ18O and δ2H depletion from west to east (Fig 5). The samples collected along each river system reflect the mixed signature of all water upstream of the sample point. Samples collected from rivers in the north and northeast reflected May 2018 GNIP precipitation δ18O and δ2H values that were reported at Armagh (-7.16‰, -50.9‰), while rivers in the south and southwest reflected δ18O and δ2H values recorded at Valentia (-4.65‰, -32.00‰) [27]. These data follow the general pattern of prevailing winds and decreasing precipitation amount across the island (Fig 5). In August, the same pattern was observed but southwest rivers were depleted compared to July GNIP precipitation (-2.70‰, -13.60‰; Valentia) and northeast rivers were enriched (-8.21‰, -56.1‰; Armagh). River water isotopic composition showed inconsistent insignificant relationships with aquifer vulnerability and soil drainage (S4 Text).
There was less spatial variation in river δ18O and δ2H in May than August. In May, the most enriched δ18O and δ2H values were -5.08, 30.27 ‰ (southwest) and the most depleted values were -7.14‰, -47.35‰ (east) (Fig 5). The range of southwest river isotopic composition (δ18O = -5 to -6‰; δ2H = -33 to -40‰) was similar to the May 2018 GNIP precipitation sample that was reported at Valentia (-4.65‰, -32.00‰) [27] (Fig 5). In May, rivers in central Ireland had δ18O values of -6‰ to -7‰ and δ2H values of -40‰ to -45‰ and shifted further east in August (Fig 5). River d-excess values ranged from 7.5‰ to 12.1‰ with the majority of river d-excess values between 9‰ to 11‰ (S1 Table).
In August, River Corrib had the most enriched δ18O and δ2H values (-3.56‰, -25.26‰) and the lowest d-excess value (3.2‰). The upper Shannon also had enriched δ18O and δ2H values (-4.31‰, -29.42‰) and a low d-excess (5.0‰) (Fig 5). The River Bann, located in northeast Ireland, had the most depleted δ18O and δ2H values (-7.34‰, -49.60‰) (Fig 5). Overall, there was enrichment in river δ18O and δ2H values from May to August (Fig 5). The d-excess values for August ranged from 3.2–11.8‰ (S1 Table), where lower d-excess values were correlated with the enriched δ18O and δ2H sample values.
Evaporation fraction (E/I)
The calculated evaporation fraction (evaporation/inflow, E/I) of rivers increased between the onset of the drought in late May and the termination of the drought in August (Fig 6A and 6B). The E/I values were greatest when the minimum (depleted) groundwater δ18O compositions were used as the river inflow value (S3 Text, Eq. 2). Conversely, E/I values were lowest when the maximum (enriched) groundwater δ18O composition were used for inflow value (S1 Fig, S4 Table) [24]. Rivers with E/I ≤ 0.01 were considered to have a negligible evaporative fraction of water. The E/I values groups calculated from minimum and maximum groundwater δ18O values varied in statistical difference (S5 Table). The E/I values calculated for May and August with average groundwater δ18O inflow values were significantly different (p<0.05) (S5 Table). Average groundwater δ18O values were used to calculate E/I value shown in Fig 6. Ranges of maximum, minimum, and average groundwater δ18O inflow values [24] used to calculate E/I values are shown in S2 Fig. Hereon, results presented in Fig 6 are discussed.
Maps show interpolated Irish river E/I values [43] of river water in (a) May and (b) August 2018. Average drainage basin groundwater δ18O values were used as inflow values for surface water prior to evaporation [24]. Values are reported in S4 Table. The color gradient of red to blue show a shift from a greater E/I values (more evaporation) to lower E/I values (less evaporation). Dashed lines are contour intervals for E/I intervals of 0.01. E/I values ≤0.01 were considered negligible. All projected spatial data, including surface waters, catchments, and island boundary, were sourced from the Irish Environmental Protection Agency (EPA) and downloaded from EPA Geoportal (gis.epa.ie/GetData/Download) (CC-BY 4.0; https://creativecommons.org/licenses/by/4.0/legalcode).
In May, the rivers Corrib and Annascaul (southwest) had the greatest evaporative fraction (0.05). May E/I values were negligible across the rest of southern Ireland. The E/I values in the Corrib catchment rivers (0.02–0.05), Shannon tributaries (0.02), and River Bann (0.03) resulted in a band of elevated E/I values from west to northeast (Fig 6A). The band of elevated E/I values from west to northeast was not observed in August, but E/I values along the west coast increased and extended toward the southeast of Ireland (Fig 6B). Rivers in western Ireland and in the Shannon catchment had greater E/I values in August than in May. In August, the River Corrib had the greatest calculated E/I value (0.10), and negligible E/I values were calculated for locations in the north and southeast (≤0.01) (Fig 6B, S4 Table). The There was a notable E/I value increase in the southwest from May to August (Fig 6A and 6B). The E/I value was greatest for the River Corrib and the River Shannon above Lough Ree in both May and August. A significant (p<0.05) negative relationship between d-excess and E/I values calculated from average groundwater δ18O values (r = -0.7) from May to August provided robust support for E/I value calculations despite uncertainty associated with the E/I calculations (S3 Text).
Lough and river δ18O and δ2H
Isoscapes were also generated for the Corrib and Shannon catchments for additional sampling campaigns in 2018, 2019, and 2020 (Fig 7, S3 Fig) to show the extent of change in 2018 and identify the role that loughs play in low and high flow events. In early March 2018, Shannon catchment rivers were sampled after unusual snow fall in February [51]. All sampled rivers, including locations below loughs, had δ18O and δ2H values within 0.18‰ and 5.46‰ of February precipitation mean δ18O and δ2H values (6.61‰, -36.4‰) (Fig 7, S3 Fig), showing that surface water systems reflected the isotopic composition precipitation [27].
Isoscapes were created for March 2018 (n = 10), May 2018 (n = 9), June 2018 (n = 9), August 2018 (n = 13), March 2019 (n = 6), October 2019 (n = 23), and March 2020 (n = 23). The color gradient is the same as Fig 5 and indicates a shift from enriched to depleted δ18O values. All corresponding δ2H values (‰) are shown in S3 Fig. Dashed lines and corresponding numbers show contour intervals of river water δ18O (1‰). Corrib and Shannon catchment rivers are displayed in black. All projected spatial data, including surface waters, catchments, and island boundary, were sourced from the Irish Environmental Protection Agency (EPA) and downloaded from EPA Geoportal (gis.epa.ie/GetData/Download) (CC-BY 4.0; https://creativecommons.org/licenses/by/4.0/legalcode).
Patterns from May to August follow island-wide patterns previously reported, but there was no significant difference from June and May 2018 sampling campaigns in the Shannon (S3 Table). Shannon catchment rivers showed a shift from depleted to enriched δ18O and δ2H values between March and June. Rivers sampled during a period when 19 synoptic weather stations recorded absolute drought conditions in June [17] (Fig 7, S3 Fig).
Additional rivers were sampled in the Corrib (n = 2) and Shannon (n = 3) catchments in 2019 and 2020 (S1 Table), which provided more information for isoscape generation. Samples were collected below two large loughs in the Corrib catchment and 3 large loughs in the Shannon catchment (Fig 2). River Corrib consistently had the most enriched δ18O and δ2H values in the Corrib catchment (Fig 7, S3 Fig). October 2019 sampling in the Corrib and Shannon catchments occurred after extratropical Storm Lorenzo, and elevated river discharge ranged between Q50 and Q25 (Fig 3). Rivers had the largest range in isotopic composition during this sampling campaign. River δ18O and δ2H values varied from -4.60‰, -28.38‰ (River Owenriff) to -7.61‰, -50.53‰ (Shannon Pot) (Fig 7, S1 Table). The western Corrib catchment rivers had enriched δ18O and δ2H values compared to east catchment rivers. River Shannon also had enriched δ18O and δ2H values below the three loughs compared to tributary δ18O and δ2H values (Figs 2 and 7, S3 Fig). These isotopic patterns were consistent in samples collected in March 2020 during extremely high flows (Q10–Q1) (Fig 3) and after one of the wettest winter months since 1850 [19, 52].
Discussion
Island-wide river δ18O and δ2H patterns
River mean δ18O and δ2H values for 2018–2020 were similar to average precipitation values in global models [25] but plotted above the LMWLs for precipitation samples collected at Armagh and Valentia (Fig 4). These GNIP LMWLs were constructed based on monthly samples that were collected from the precipitation collector, rather than sample collection on an event-by-event basis [27]. Single monthly sample collection might not represent seasonal events or daily fluctuations. Furthermore, the δ18O-δ2H relationship of the LMWLs that were derived from precipitation samples collected at these stations incorporated the effect of meteorologic, geographic, and topographic variables that are specific to Ireland [53, 54], but they do not encapsulate the δ18O-δ2H relationship of all locations across the island. Spatial presentation of the river δ18O and δ2H data across Ireland show isotopic depletion from west to east that have been presented in previous Irish surface and groundwater studies [23, 24]. Precipitation patterns and the depletion of δ18O and δ2H values in precipitation are due to the prevailing westerly winds and the Atlantic as the predominant precipitation source on the island of Ireland [8, 9]. As airmasses travel from west to east, progressive depletion of the heavier isotopologues results in depleted δ18O and δ2H values along the east coast. The amount effect and relative humidity at location of moisture uptake also control precipitation isotopic composition [8], so localized meteorological differences may not be fully captured by the two GNIP LMWLs. This stratified pattern of isotopic depletion in surface waters has also been shown to occur on a continental scale [55], country wide scales [56, 57] and catchment scale [58]. Furthermore, contribution of event water or groundwater may cause samples to plot off of the LMLWs [59, 60]. The aforementioned reasons may have caused the river samples to plot along and above the LMWLs but within the range of reported precipitation samples [27].
Groundwater and evaporation influence
The increase in the riverine δ18O and δ2H values in May to August could be interpreted as (1) a change in precipitation signature, (2) shift to predominant groundwater source, (3) evaporative influence, or a combination of these factors. Since Regan et al. [24] established a direct relationship between precipitation and groundwater δ18O values there is no differentiation between these two inputs in this work, and while the precipitation isotopic signature is dampened by moderately and poorly productive aquifers [24], for the purposes of this work it is assumed that groundwater δ18O values are still representative of precipitation input. The inconsistent relationships among river isotopic data reported in this study and catchment aquifer type support the use of groundwater δ18O [24] as inflow values for the E/I calculation despite varying recharge rates (S4 Text). Almost all rivers had +25 days of daily average discharge values <Q90 water inflow in July (Fig 3), where weather stations recorded between 20 and 27 days without effective rainfall across Republic of Ireland [19]. These data indicate that river flow in early August sampling was predominantly input from groundwater or water stored in loughs. Conversely, river water inflow was likely a mixture of precipitation and groundwater input in May (Fig 3) [61].
The evaporative fractions calculated with average groundwater values δ18O for May were negligible (E/I ≤0.01) except for a band traveling from the west of the lower Corrib catchment (53.2792, -9.0563) to the northeast of the Bann catchment (54.4203, -6.4548) (Fig 6A, S4 Table). This E/I value interpolated in the northeast is interpreted with caution, since this interpolation relies only on the point of the River Bann (Fig 6A). Nevertheless, there is evidence of evaporation in the Bann catchment. A low order stream in the southwest, also had an elevated E/I fraction (Fig 6A). The lowest river d-excess values were calculated for locations with the greatest E/I values in the Corrib catchment and northeast Ireland in May (S1 Table). These data support the E/I value calculation because of the preferential mass-dependent fractionation of lighter isotopologues 2H216O and 1H216O over 1H218O or 2H218O in the evaporative process [53]. Other rivers had E/I values ≤0.01 because the river δ18O values closely match the inflow value of average groundwater δ18O (S2 Fig). Negligible E/I values do not imply that no evaporation is occurring, however that the replacement of source water is sufficient for isotopic evaporative enrichment to go undetected. When May E/I values were calculated with maximum δ18O end members, E/I values were <0.02 across the island (S1 Fig, S4 Table). The E/I values calculated with maximum and minimum groundwater δ18O inflow values differed up to values 0.06 (6%) at the River Corrib and River Robe. The median difference between E/I values calculated with maximum and minimum groundwater δ18O values was 0.02 (2%) for all rivers in May (S4 Table, S1 Fig). In June E/I values were ≤0.01–0.03 for rivers Shannon and Maigue in the Shannon catchment (S4 Table). The E/I values of rivers that were sampled in May and June did not significantly differ from each other and few rainfall events probably produced event flow contributions at the time of sampling and minimized, or at least diluted, an evaporative effect (Fig 3B) [17].
For all evaporation fraction calculations, except the River Bann, values in August E/I ≥ May E/I values. The calculated E/I values showed more spatial variation in early August, after the summer dry spells, absolute drought, and partial drought conditions in different locations of the island (Fig 6). Some rivers had an E/I value of 0.01, suggesting groundwater contribution to rivers during drought resulted in little to no evidence of isotopic enrichment induced by dry conditions [58]. Overall, the river d-excess values in August support the evaporation calculations, as lower d-excess values corresponded greater E/I values (Fig 5), and there was an inverse relationship between E/I values and d-excess from May and August. Increases in E/I values were particularly noticeable in the Corrib and Shannon catchments, east and southwest of the island (Figs 2 and 6). The greatest difference in August E/I values calculated with maximum and minimum groundwater δ18O inflow values was 0.08 (8%) at River Corrib. The median difference among all rivers was 0.02 (2%) (S4 Table, S1 Fig). Despite uncertainties associated with these E/I data (S3 Text), they show seasonal and geographic shifts of evaporative water deficit across Ireland (S5 Table), and they are supported by additional meteorologic and hydrologic datasets.
Precipitation disparities from May through July 2018 support the E/I values results. Lower total precipitation amounts likely increased the E/I value in southwest rivers throughout the summer (Fig 1). The River Bann in the northeast was the only river that showed a detectable decrease in E/I fraction. This E/I value shift of 0.03 in May to 0.01 in August can be explained by the timing of local precipitation (Fig 1). Pronounced river isotopic enrichment in the northeast in May was likely the result of the larger precipitation deficit, as there was less rainfall to replace evaporative loss (Fig 1).
Notably, the upper Shannon above Lough Ree, had the greatest E/I value in both May and August for the River Shannon (Fig 6). The nearest weather station, Mullingar, (Fig 1), recorded 44% of the summer long–term average precipitation amount. This section of the river also has a series of small, shallow, loughs that can elongated water residence time, increase evaporation and induce isotopic enrichment. This pattern of enrichment after loughs opposes the pattern of depletion below the larger loughs, Ree and Derg. However, the presence of small lakes and reservoirs has been shown to result in evaporitic enrichment of water isotopes, particularly in drought years [62–64].
River Corrib also had high E/I values. Lough levels on Lough Corrib, measured in the southern portion of the lough at Angligham (station #30089) at the beginning of May were at ~50% level (8.7 mOD; meters above ordnance datum), and dropped until hitting of minimum height of 8.4 (mOD), the 99% gauge height [31]. Low water levels can be an indication of evaporation from water body in addition to reduced water delivery by surface and groundwater sources and extraction for domestic water use [65]. Lough water height and lough extent were reduced in multiple surface water systems across Ireland in the summer 2018 [18].
Despite the anomalous summer drought conditions, the calculated Irish river E/I values were low on the global scale (≤0.12). The mean E/I value for global lakes is 0.2 [66]. Evaporative isotopic enrichment patterns in lake systems have been shown to vary by local climate and hydrology around the world [67] but the study of river evaporative enrichment is less common. The elongated residence time and reduced flow in lake systems can facilitate isotopic enrichment via increased evaporation. Brooks et al. [68] found that E/I values for lakes varies widely (<0.1 to 1.0) by climatic region over the contiguous United States, and E/I variations were better described by annual precipitation amount and relative humidity rather than temperature. In addition to climatic variation, seasonally driven E/I fluctuations with a range of 0.57 to 1.04 have been documented in a temperate lake [47]. Anthropogenic modifications, like dam building [62, 64] and water abstraction [43] can increase evaporation from river systems. Diamond and Jack [43] found an overall 0.215 E/I for the Gariep River in South Africa, which has been dammed and used as a water source. Major points of evaporation were in and below the reservoir along the river. Even at high latitudes, evaporation fractions ranging from <0.05 to >0.5 have been calculated for Canadian lake catchments, however these calculations also included transpiration water losses [69].
The isotopic composition and calculated E/I values of river sample locations draining loughs, particularly River Corrib (Fig 4), indicate that evaporative enrichment was occurring in some of the Irish loughs. Previous work in Ireland [23] and recent work in Scotland [56] showed that river systems expressed enriched isotopic composition in summer and autumn, which support the presented results. However most Irish rivers were not sampled below loughs or reservoirs, and so the overall low E/I values that were calculated in this study can be attributed to hydrologic setting, climate of Ireland, and the series of assumptions made in the E/I calculation.
The role of loughs
Sample locations draining loughs in the Corrib and Shannon catchments showed a local evaporation line slope of 5.58 (Fig 4) similar to expected slope value (5.6) for humid, high latitude locations [70] and the local evaporation line for lakes in all climates (-5.5) [66]. It is also very similar to the slope calculated by Diefendorf and Patterson [23] for Irish rivers downstream of lakes (5.9). Interestingly, the major lough systems in the Corrib and Shannon catchments did not consistently act the same isotopically (S4 Fig). Water isotopic composition of lakes can demonstrate the sensitivity of these system to larger regional hydroclimate and localized catchment changes [66]. The inlet, mid, and outlet sample points of the loughs in the Corrib catchment showed continual enrichment in 2018, and enrichment from further upstream to downstream point for all seasons (S4 Fig). The progressive enrichment of River Corrib sample isotopic compositions throughout summer 2018 suggested evaporative enrichment influenced by Lough Corrib (Figs 2, 4B). The two River Corrib samples collected in May showed enrichment in δ18O and δ2H in the span of 25 days (Fig 4B, S1 Table). River Corrib 2018 summer samples plotted below the LMWLs and had low (<8) d-excess values, which indicated increased evaporation (Fig 4B, S1 Table) [71]. The enriched δ18O and δ2H values and E/I value were likely the result of increased evaporation in the lower basin of Lough Corrib. The lower basin of Lough Corrib is smaller and shallower than the upper basin, and it experiences reduced stratification and quicker temperature changes than the upper basin [35]. The uncommon seasonal heat and greater than average sunshine days probably generated ideal conditions for increased evaporation in Lough Corrib [17] as sensitive heat fluxes are exert large control over isotopic enrichment in lakes in temperate environments [66].
River Corrib δ18O and δ2H values were also related to lower basin residence time. The lough was assumed to be a mixed system, however this could be an underestimation of residence time if a stratification regime was dominant. Residence times were calculated as
(1)
where the lough volume of lower Lough Corrib (1.09 x 108 m3) was adjusted based on water level gauge height, with the assumption that 50 percentile gauge height readings represent the reported lough volume [31, 35]. In 2018 the residence time in lower Lough Corrib increased from <8 days in January and February to a peak of 122 days on July 25. Longer calculated residence time in Lough Corrib coincided with enriched δ18O and δ2H values and elevated E/I values. We hypothesized that evaporation in the shallow lower lough basin was the cause of enriched δ18O values and greater E/I values in the dataset, which coincides with results from global wide research that shows evaporation is greatest in summer months at high latitudes (≥40°N) and that evaporation volume of lakes is up to 0.5 km3 yr-1 [72].
The summer isotopic enrichment pattern was not consistent for Lough Ree and Lough Derg in the Shannon (Fig 7, S4 Fig). The inlet to outlet points of Lough Ree displayed a pattern of depletion in summer 2018 and enrichment in October 2019 and March 2020 (S4 Fig). This depletion pattern also occurred at the inlet and outlet of Lough Derg, expect for March 2020. The isotopic composition of these lakes outlets demonstrate seasonal fluctuation from changing precipitation signatures but show less variation than inlet sample locations. These data suggest that tributary mixing could be affecting the isotopic composition of Loughs Ree and Derg [64] but also that these reservoirs are well mixed and elongated residence times dampens the seasonality and potential evaporative enrichment.
Water residence time in Lough Ree (5.44 x 108 m3), along River Shannon, increased throughout 2018 to a peak of 374 days in July based on discharge outflow values. During peak residence time, gauge height was <90 percentile value. Despite low flows, δ18O and δ2H values below Lough Ree did not reflect a distinct evaporative signature in June or August 2018 (Fig 7, S4 Fig), instead, the samples collected at inlets of both loughs Ree and Derg were enriched compared to outlets (Fig 7, S4 Fig). As mentioned, the upper Shannon above Lough Ree had the most enriched δ18O and δ2H value and a greatest E/I value in the Shannon (Figs 6B and 7). An evaporative signature may have been damped downstream from Lough Ree because it is a deep lough (max depth: 36m; average depth: 6.2 m), and the ratio of evaporated water to inflow has been shown to decrease with lake increasing lake depth [65]. Overtime, a deeper, less evaporated water source could mix with surface water.
In October 2019, the Corrib and Shannon catchment rivers displayed differences in isotopic signature when rivers were sampled after Storm Lorenzo (Fig 7). Most notable about the October 2019 isoscape is the enriched δ18O and δ2H values below loughs compared to sample locations not draining loughs in the Corrib and Shannon catchments. We do not propose that these enriched values were due to evaporation of the lough because precipitation amount was above average most location across the island and temperature was ±0.6°C of the long-term average [73]. Instead, we suggest that the isotopic signature may be reflective of non-event water that had been stored in the loughs. The isotopic composition below loughs more closely reflected the long-term weighted mean values at Valentia (-5.48‰ ± 0.56; -35.4‰ ± 3.5) compared to more depleted river samples, which indicated surface water mixing [64]. Loughs may have been mixed by winds, especially from a storm that produce a more homogenizing environments along this river system.
Rivers showed an increase in discharge response following the October storm event, which corresponded with sample collection (Fig 3B). The depleted river water isotopic composition was likely from the storm event. However, precipitation samples were not collected from Storm Lorenzo. The isotopic composition of cyclonic events at subtropical latitudes can be relatively enriched or depleted in relation to bulk precipitation measurements depending on storm evolution and track [60, 74, 75]. Therefore, we can only hypothesize that larger water volumes stored along the Shannon and its loughs resulted in δ18O and δ2H values that represented seasonal, not event, δ18O and δ2H values.
Sampling in March 2020 followed one of the wettest winter months on record [19, 52], and river water isotopic compositions fell above the GMWL (Fig 4), and this may be reflective of a seasonal signal. However, flooding was widespread throughout the Shannon catchment and overland flow was dominant. These samples were collected after three extratropical cyclonic storms passed over Ireland in February 2020, and hydrographs (Fig 3B) show that samples were collected near the peak of the hydrograph limb. The hydrograph data suggests that it was almost certainly event water that dominated river flow during sample collection. In support of this assertion, most river d-excess values in March 2020 were elevated (>10) (S1 Table). Previous work has tied strong large-scale ocean evaporation to extratropical cyclone events [76] to elevated d-excess [77], which have been shown to affect the isotopic composition of precipitation in Iceland [78]. It is not possible to tie a direct connection between river water and precipitation isotopic composition because precipitation was not sampled. However, it is probable that the Irish surface waters in March 2020 resemble the isotopic fingerprint of recent extratropical cyclonic precipitation. Even with evidence of event flow, the isotopic compositions of sample locations below loughs were enriched in March, which supports the previously mentioned hypothesis that elongated residence time in loughs results in mixing and an isotopic composition weighted towards a seasonal, or annual, δ18O and δ2H signature (Fig 7, S4 Fig).
Conclusions
River water δ18O and δ2H values transformed from the onset of to the termination of the 2018 drought. The pattern of west to east δ18O and δ2H depletion in surface waters was apparent in both May and August. River water δ18O and δ2H values showed enrichment at most locations from the onset to the termination of the drought, and calculated E/I values showed the same pattern. The d-excess August values agree with E/I values to suggest that the locations of greatest evaporation were along the Rivers Corrib and Shannon. Overall, the E/I values ranged from <0.01 to 0.12 and were low on a global scale. The E/I values also did not consistently follow expected evaporation enrichment patterns below the loughs due to isotopic depletion at lough outlets. Large loughs along the Corrib system demonstrated an expected enrichment pattern and seemed to promote evaporative enrichment, while large loughs along the Shannon facilitated mixing of seasonal (combination of precipitation and groundwater) and event water. Lough systems have strong influence over the isotopic patterns of surface water systems, but differences in lough depth and mixing alter the isotopic enrichment and depletion patterns. Findings in this work support continued investigation of surface waters better understand hydrologic response in the context of increased hydrologic extremes due to climate change.
Supporting information
S4 Text. Aquifer vulnerability relationships.
https://doi.org/10.1371/journal.pwat.0000197.s004
(DOCX)
S1 Fig. Interpolated evaporation fraction (E/I) of Irish rivers for maximum and minimum groundwater values.
Maps show interpolated Irish river E/I values [43] calculated for May and August 2018 were with maximum and minimum groundwater δ18O values used as inflow values for surface water prior to evaporation [24]. Values are reported in S4 Table. The color gradient of red to blue demonstrates a shift from a greater E/I value that shows greater evaporation to lower E/I value that shows less evaporation. E/I values ≤0.01 are considered negligible. All projected spatial data, including surface waters, catchments, and island boundary, were sourced from the Irish Environmental Protection Agency (EPA) and downloaded from EPA Geoportal (gis.epa.ie/GetData/Download) (CC-BY 4.0; https://creativecommons.org/licenses/by/4.0/legalcode).
https://doi.org/10.1371/journal.pwat.0000197.s005
(TIF)
S2 Fig. Box plot of groundwater and river δ18O values.
The box plot shows average, maximum, and minimum groundwater δ18O values [24] used for E/I value calculation and river water δ18O values from each sample campaign. The δ18O values of precipitation collected Valentia and Armagh [27] are shown as “v” and “a” for May 2018 (black) and July 2018 (red).
https://doi.org/10.1371/journal.pwat.0000197.s006
(TIF)
S3 Fig. δ2H isoscapes of the Corrib and Shannon catchments.
Isoscapes of the Corrib and Shannon catchments in March 2018 (n = 10), May 2018 (n = 9), June 2018 (n = 9), August 2018 (n = 13), March 2019 (n = 6), October 2019 (n = 23), and March 2020 (n = 23). The color gradient is the same as Fig 5 and indicates a shift from enriched to depleted δ2H values (‰). These data correspond to δ18O shown in Fig 7. Dashed lines and corresponding numbers show contour intervals of river water δ2H values (5‰). Corrib and Shannon catchment rivers displayed in black. All projected spatial data, including surface waters and catchments, were sourced from the Irish Environmental Protection Agency (EPA) and downloaded from EPA Geoportal (gis.epa.ie/GetData/Download) (CC-BY 4.0; https://creativecommons.org/licenses/by/4.0/legalcode).
https://doi.org/10.1371/journal.pwat.0000197.s007
(TIF)
S4 Fig. Lough inlet and outlet δ18O values.
The δ18O values of river inlet and outlet sample locations of (a) Lough Corrib and Lough Mask in the Corrib catchment, (b) Lough Ree and (c) Lough Derg in the Shannon catchment from 2018–2020. Names of sample locations are in parenthesis.
https://doi.org/10.1371/journal.pwat.0000197.s008
(TIF)
S1 Table. River water δ18O, δ2H, and d-excess values.
Samples collected from 2018–2020, and sample ID numbers correspond to sample ID numbers in Fig 2.
https://doi.org/10.1371/journal.pwat.0000197.s009
(XLSX)
S2 Table. Delineated drainage basin area for sample locations.
https://doi.org/10.1371/journal.pwat.0000197.s010
(XLSX)
S3 Table. Kruskal-Wallis test results to compare the δ18O and δ2H groups for the 2018–2020 sample campaigns.
The null hypothesis states that the groups are not significantly different.
https://doi.org/10.1371/journal.pwat.0000197.s011
(XLSX)
S4 Table. Evaporation fractions (E/I) for May, June, and August 2018.
E/I values are presented for average, maximum, and minimum groundwater δ18O (‰) inflow values [24]. The GNIP Station and precipitation δ18O values are shown for each sample.
https://doi.org/10.1371/journal.pwat.0000197.s012
(XLSX)
S5 Table. Kruskal-Wallis test results to compare the calculated evaporation fractions for May and August 2018.
Calculated evaporation fractions (E/I) with average, maximum, and minimum groundwater δ18O inflow values were included [24]. The null hypothesis states that the groups are not significantly different.
https://doi.org/10.1371/journal.pwat.0000197.s013
(XLSX)
S1 Checklist. Inclusivity in global research.
https://doi.org/10.1371/journal.pwat.0000197.s014
(DOCX)
Acknowledgments
Sample collection in 2018 was conducted while WBL was on a Fulbright-GSI Fellowship at University of Galway. We would like to thank Peter Croot for hosting WBL during the Fulbright-GSI Fellowship and Department of Earth & Ocean Sciences at the National University of Ireland Galway for their hospitality and assistance in sample collection. We would like to thank Sean Wheeler for assistance in sample collection in 2019 and 2020.
References
- 1.
IPCC. Climate Change 2021: The Physical Science Basis. Contribution of Working Group I to the Sixth Assessment Report of the Intergovernmental Panel on Climate Change. Cambridge Univ Press. 2021; 3949.
- 2. Vicente-Serrano SM, Quiring SM, Peña-Gallardo M, Yuan S, Domínguez-Castro F. A review of environmental droughts: Increased risk under global warming? Earth-Science Rev. 2020;201: 102953.
- 3. Van Loon AF. Hydrological drought explained. WIREs Water. 2015;2: 359–392.
- 4. Parry S, Prudhomme C, Wilby RL, Wood PJ. Drought termination: Concept and characterisation. Prog Phys Geogr. 2016;40: 743–767.
- 5. Gosling SN, Arnell NW. A global assessment of the impact of climate change on water scarcity. Clim Change. 2016;134: 371–385.
- 6. Yuan X, Wang Y, Ji P, Wu P, Sheffield J, Otkin JA. A global transition to flash droughts under climate change. Science (80-). 2023;380: 187–191. pmid:37053316
- 7. Vicente-Serrano SM, McVicar TR, Miralles DG, Yang Y, Tomas-Burguera M. Unraveling the influence of atmospheric evaporative demand on drought and its response to climate change. Wiley Interdiscip Rev Clim Chang. 2019;11: 1–32.
- 8. Baldini LM, McDermott F, Baldini JUL, Fischer MJ, Möllhoff M. An investigation of the controls on Irish precipitation δ18O values on monthly and event timescales. Clim Dyn. 2010;35: 977–993.
- 9. Sweeney J. Regional weather and climates of the British Isles—Part 6: Ireland. Weather. 2014;69: 275–281.
- 10. Noone S, Murphy C, Coll J, Matthews T, Mullan D, Wilby RL, et al. Homogenization and analysis of an expanded long-term monthly rainfall network for the Island of Ireland (1850–2010). Int J Climatol. 2016;36: 2837–2853.
- 11. Noone S, Broderick C, Duffy C, Matthews T, Wilby RL, Murphy C. A 250-year drought catalogue for the island of Ireland (1765–2015). Int J Climatol. 2017;37: 239–254.
- 12. Wilby RL, Noone S, Murphy C, Matthews T, Harrigan S, Broderick C. An evaluation of persistent meteorological drought using a homogeneous Island of Ireland precipitation network. Int J Climatol. 2016;36: 2854–2865.
- 13. Murphy C, Broderick C, Burt TP, Curley M, Duffy C, Hall J, et al. A 305-year continuous monthly rainfall series for the island of Ireland (1711–2016). Clim Past. 2018;14: 413–440.
- 14. McCarthy M, Christidis N, Dunstone N, Fereday D, Kay G, Klein-Tank A, et al. Drivers of the UK summer heatwave of 2018. Weather. 2019;74: 390–396.
- 15.
Murphy C, Broderick C, Matthews T, Noone S, Ryan C. Irish Climate Futures: Data for Decision-making. 2019.
- 16. Peters W, Bastos A, Ciais P, Vermeulen A. A historical, geographical and ecological perspective on the 2018 European summer drought. Philos Trans R Soc B. 2020;375: 20190505. pmid:32892723
- 17.
Moore P, Murphy A, Downes P, Watters V. Summer 2018: An analysis of the heatwaves and droughts that affected Ireland and Europe in the summer of 2018. Dublin; 2018. met.ie
- 18. Falzoi S, Gleeson E, Lambkin K, Zimmermann J, Marwaha R, O’Hara R, et al. Analysis of the severe drought in Ireland in 2018. Weather. 2019;74: 368–373.
- 19.
Met Éireann. Historical Data. In: Met Éireann [Internet]. 2022. met.ie
- 20. Turner S, Barker LJ, Hannaford J, Muchan K, Parry S, Sefton C. The 2018/2019 drought in the UK: a hydrological appraisal. Weather. 2021;76: 248–253.
- 21.
EPA. Drinking Water Quality in Public Supplies 2018. Environmental Protection Agency; 2019.
- 22. Baldini LM, McDermott F, Foley AM, Baldani JUL. Spatial variability in the European winter precipitation δ 18O-NAO relationship: Implications for reconstructing NAO-mode climate variability in the Holocene. Geophys Res Lett. 2008;35.
- 23. Diefendorf AF, Patterson WP. Survey of stable isotope values in Irish surface waters. J Paleolimnol. 2005;34: 257–269.
- 24. Regan S, Goodhue R, Naughton O, Hynds P. Geospatial drivers of the groundwater δ18O isoscape in a temperate maritime climate (Republic of Ireland). J Hydrol. 2017;554: 173–186.
- 25. Bowen GJ, Revenaugh J. Interpolating the isotopic composition of modern meteoric precipitation. Water Resour Res. 2003;39: 1–13.
- 26.
Met Office. Historic Station Data. United Kingdom; 2022. metoffice.gov.uk
- 27.
IAEA/WMO. Global Network of Isotopes in Precipitation, International Atomic Energy Agency. Vienna: International Atomic Energy Agency; 2022 [cited 1 Nov 2022]. iaea.org/services/networks/gnip
- 28. Dirmeyer PA, Balsamo G, Blyth EM, Morrison R, Cooper HM. Land‐atmosphere interactions exacerbated the drought and heatwave over northern Europe during summer 2018. AGU Adv. 2021;2: 1–16.
- 29. Kornhuber K, Osprey S, Coumou D, Petri S, Petoukhov V, Rahmstorf S, et al. Extreme weather events in early summer 2018 connected by a recurrent hemispheric wave-7 pattern. Environ Res Lett. 2019;14.
- 30.
Murphy A. Drought Summary. In: Met Éireann [Internet]. 2022 [cited 4 Oct 2022]. https://www.met.ie/drought-summary
- 31.
OPW. Hydro-Data. In: Office of Public Works [Internet]. 2021. http://waterlevel.ie/hydro-data/
- 32.
Met Éireann. February 2020. 2020. met.ie
- 33.
GSI. Data and Maps. In: Geological Survey of Ireland [Internet]. 2022. gsi.ie
- 34.
Lydon K, Smith G. Corine landcover 2012 Ireland Final Report. Environ Prot Agency. 2014; 1–54.
- 35. Mooney EP. A study of Lough Corrib, western Ireland and its phytoplankton. Hydrobiologia. 1989;175: 195–212.
- 36. Dabrowski T, Berry A. Use of numerical models for determination of best sampling locations for monitoring of large lakes. Sci Total Environ. 2009;407: 4207–4219. pmid:19406455
- 37. Hughes CE, Crawford J. A new precipitation weighted method for determining the meteoric water line for hydrological applications demonstrated using Australian and global GNIP data. J Hydrol. 2012;464–465: 344–351.
- 38. Kruskal WH, Wallis WA. Use of Ranks in One-Criterion Variance Analysis. J Am Stat Assoc. 1952;47: 583–621.
- 39.
Copernicus Programme. European Digital Elevation Model (EU-DEM), version 1.1. European Environment Agency (EEA); 2016. copernicus@eea.europa.eu
- 40.
Creamer R, Simo I, Fealy R, Hallett S, Hannam J, Holden A, et al. National Soil Map of Ireland. Environmental Protection Agency; 2014.
- 41. Misstear BDR, Brown L, Daly D. A methodology for making initial estimates of groundwater recharge from groundwater vulnerability mapping. Hydrogeol J. 2009;17: 275–285.
- 42. Terzopoulos D. The Computation of Visible-Surface Representations. IEEE Trans Pattern Anal Mach Intell. 1988;10: 417–438.
- 43. Diamond RE, Jack S. Evaporation and abstraction determined from stable isotopes during normal flow on the Gariep River, South Africa. J Hydrol. 2018;559: 569–584.
- 44.
Craig H, Gordon LI. Deuterium and oxygen 18 variations in the ocean and the marine atmosphere. In: Tongiorgi E, editor. Stable Isotopes in Oceanographic Studies and Paleotemperatures. Pisa; 1965. pp. 9–130.
- 45.
Gonfiantini R. Environmental Isotopes in Lake Studies. In: Fritz P, Fontes JC, editors. Handbook of Environmental Isotope Geochemistry. Amsterdam: Elsevier; 1986. pp. 113–168.
- 46. Horita J, Rozanski K, Cohen S. Isotope effects in the evaporation of water: A status report of the Craig-Gordon model. Isotopes Environ Health Stud. 2008;44: 23–49. pmid:18320426
- 47. Wu H, Li XY, He B, Li J, Xiao X, Liu L, et al. Characterizing the Qinghai Lake watershed using oxygen-18 and deuterium stable isotopes. J Great Lakes Res. 2017;43: 33–42.
- 48. O’Brien RJ, Misstear BD, Gill LW, Deakin JL, Flynn R. Developing an integrated hydrograph separation and lumped modelling approach to quantifying hydrological pathways in Irish river catchments. J Hydrol. 2013;486: 259–270.
- 49.
EPA. Hydronet. In: Environmental Protection Agency [Internet]. 2021. https://epawebapp.epa.ie/hydronet
- 50.
Dingman SL. Physical Hydrology. 2nd ed. Lynch P, Hall S, editors. Upper Saddle River: Prentice-Hall, Inc.; 2002.
- 51. Vero SE, Mcdonald NT, Mcgrath G, Mellander PE. The Beast from the East: impact of an atypical cold weather event on hydrology and nutrient dynamics in two Irish catchments. Irish J Agric Food Res. 2020;59: 113–122.
- 52. Matthews T, Mullan D, Wilby RL, Broderick C, Murphy C. Past and future climate change in the context of memorable seasonal extremes. Clim Risk Manag. 2016;11: 37–52.
- 53. Dansgaard W. Stable isotopes in precipitation. Tellus. 1964;16: 436–468.
- 54. Rozanski K, Araguás-Araguás L, Gonfiantini R. Isotopic Patterns in Modern Global Precipitation. Clim Chang Continetnal Isot Rec. 1993;78: 1–36.
- 55. Kendall C, Coplen TB. Distribution of oxygen-18 and deuterium in river waters across the United States. Hydrol Process. 2001;15: 1363–1393.
- 56. Birkel C, Helliwell R, Thornton B, Gibbs S, Cooper P, Soulsby C, et al. Characterization of surface water isotope spatial patterns of Scotland. J Geochemical Explor. 2018;194: 71–80.
- 57. Reckerth A, Stichler W, Schmidt A, Stumpp C. Long-term data set analysis of stable isotopic composition in German rivers. J Hydrol. 2017;552: 718–731.
- 58. Chiogna G, Skrobanek P, Narany TS, Ludwig R, Stumpp C. Effects of the 2017 drought on isotopic and geochemical gradients in the Adige catchment, Italy. Sci Total Environ. 2018;645: 924–936. pmid:30032088
- 59. Cole A, Boutt DF. Spatially-Resolved Integrated Precipitation-Surface-Groundwater Water Isotope Mapping From Crowd Sourcing: Toward Understanding Water Cycling Across a Post-glacial Landscape. Front Water. 2021;3: 1–22.
- 60. Boutt DF, Mabee SB, Yu Q. Multiyear Increase in the Stable Isotopic Composition of Stream Water From Groundwater Recharge Due to Extreme Precipitation. Geophys Res Lett. 2019;46: 5323–5330.
- 61.
MacCárthaigh M. Parameters of low flow and data on low flows in selected Irish rivers. National Hydrology Seminar. Environmental Protection Agency; 2002. pp. 64–70.
- 62. Wu H, Li J, Song F, Zhang Y, Zhang H, Zhang C, et al. Spatial and temporal patterns of stable water isotopes along the Yangtze River during two drought years. Hydrol Process. 2018;32: 4–16.
- 63. Jiang D, Li Z, Luo Y, Xia Y. River damming and drought affect water cycle dynamics in an ephemeral river based on stable isotopes: The Dagu River of North China. Sci Total Environ. 2021;758: 143682. pmid:33288252
- 64. Wu H, Song F, Li J, Zhou Y, Zhang J, Fu C. Surface water isoscapes (δ18O and δ2H) reveal dual effects of damming and drought on the Yangtze River water cycles. J Hydrol. 2022;610: 127847.
- 65. Fergus CE, Brooks JR, Kaufmann PR, Pollard AI, Mitchell R, Geldhof GJ, et al. Natural and anthropogenic controls on lake water-level decline and evaporation-to-inflow ratio in the conterminous United States. Limnol Oceanogr. 2022;67: 1484–1501. pmid:36212524
- 66. Vystavna Y, Harjung A, Monteiro LR, Matiatos I, Wassenaar LI. Stable isotopes in global lakes integrate catchment and climatic controls on evaporation. Nat Commun. 2021;12: 1–7. pmid:34893644
- 67. Jasechko S, Sharp ZD, Gibson JJ, Birks SJ, Yi Y, Fawcett PJ. Terrestrial water fluxes dominated by transpiration. Nature. 2013;496: 347–350. pmid:23552893
- 68. Brooks JR, Gibson JJ, Jean Birks S, Weber MH, Rodecap KD, Stoddard JL. Stable isotope estimates of evaporation: Inflow and water residence time for lakes across the United States as a tool for national lake water quality assessments. Limnol Oceanogr. 2014;59: 2150–2165.
- 69. Gibson JJ, Edwards TWD. Regional water balance trends and evaporation-transpiration partitioning from a stable isotope survey of lakes in northern Canada. Global Biogeochem Cycles. 2002;16: 10-1–10–14.
- 70. Gibson JJ, Birks SJ, Yi Y. Stable isotope mass balance of lakes: A contemporary perspective. Quat Sci Rev. 2016;131: 316–328.
- 71. Gat JR, Bowser CJ, Kendall C. The contribution of evaporation from the Great Lakes to the continental atmosphere: estimate based on stable isotope data. Geophys Res Lett. 1994;21: 557–560.
- 72. Zhao G, Li Y, Zhou L, Gao H. Evaporative water loss of 1.42 million global lakes. Nat Commun. 2022;13: 1–10. pmid:35764629
- 73.
Met Éireann. September 2019. Dublin; 2019. met.ie/climate-statement-for-2019
- 74. Good SP, Mallia DV., Lin JC, Bowen GJ. Stable isotope analysis of precipitation samples obtained via crowdsourcing reveals the spatiotemporal evolution of superstorm sandy. PLoS One. 2014;9: e91117. pmid:24618882
- 75. Smith DF, Saelens E, Leslie DL, Carey AE. Local meteoric water lines describe extratropical precipitation. Hydrol Process. 2021;35: e14059.
- 76. Aemisegger F, Papritz L. A climatology of strong large-scale ocean evaporation events. Part I: Identification, global distribution, and associated climate conditions. J Clim. 2018;31: 7287–7312.
- 77. Aemisegger F, Sjolte J. A climatology of strong large-scale ocean evaporation events. Part II: Relevance for the deuterium excess signature of the evaporation flux. J Clim. 2018;31: 7313–7336.
- 78. Aemisegger F. On the link between the North Atlantic storm track and precipitation deuterium excess in Reykjavik. Atmos Sci Lett. 2018;19: 1–9.