HUMPHREYS, AARTS, WATSON, WACHENDORF, GALL, TAUBE, PFLIMLIN: Sustainable options for grassland-based dairy production in the northwest of Europe Tearmann: Irish journal of agri-environmental research, 7, 129-142, 2009 Variations in travel time for N loading to groundwaters in four case studies in Ireland: Implications for policy makers and regulators Owen Fenton1*, Catherine E. Coxon2, Atul H. Haria1, Brendan Horan3, James Humphreys3, Paul Johnson2, Paul Murphy3, Magdalena Necpalova3, Alina Premrov1, 2 and Karl G. Richards1 1Teagasc, Johnstown Castle, Environment Research Centre, Wexford, Ireland 2School of Natural Sciences, Geology, Trinity College, Dublin 2, Ireland 3Teagasc, Moorepark Research Centre, Cork, Ireland *Author for correspondence: E-mail: [email protected], Tel.: 00 353 (0)53 9171271 Abstract Mitigation measures to protect waterbodies must be implemented by 2012 to meet the requirements of the EU Water Framework Directive. The efficacy of these measures will be assessed in 2015. Whilst diffuse N pathways between source and receptor are generally long and complex, EU legislation does not account for differences in hydrological travel time distributions that may result in different water quality response times. The “lag time” between introducing mitigation measures and first improvements in water quality is likely to be different in different catchments; a process that should be considered by pol- icy makers and catchment managers. Many examples of travel time variations have been quoted in the lit- erature but no Irish specific examples are available. Lag times based on initial nutrient breakthrough at four contrasting sites were estimated to a receptor 500 m away from a source. Vertical travel times were esti- mated using a combination of depth of infiltration calculations based on effective rainfall and subsoil phys- ical parameters and existing hydrological tracer data. Horizontal travel times were estimated using a com- bination of Darcian linear velocity calculations and existing tracer migration data. Total travel times, assuming no biogeochemical processes, ranged from months to decades between the contrasting sites; the shortest times occurred under thin soil/subsoil on karst limestone and the longest times through thick low permeability soils/subsoils over poorly productive aquifers. Policy makers should consider hydrological lag times when assessing the efficacy of mitigation measures introduced under the Water Framework Directive. This lag time reflects complete flushing of a particular nutrient from source to receptor. Further research is required to assess the potential mitigation of nitrate through denitrification along the pathway from source to receptor. Key Index Words:Lag time, Water Framework Directive, mitigation, nitrate IInnttrroodduuccttiioonn ground and surface water pollution (Stark and Intensification of agriculture poses a challenge Richards, 2008; Humphreys et al.,2009). The to the sustainable management of soils, water Water Framework Directive (WFD; European resources and biodiversity. Nitrogen losses Parliament and Council, 2000) attempts to from agricultural areas can contribute to achieve at least “good ecological status” for all 129 Tearmann, the Irish journal of agri-environmental research, VOL 7, 2009 waterbodies by 2015 with programmes of a ‘groundwater body’ or aquifer, in the context measures in place by 2012. The Nitrates of the WFD, in Ireland is defined in terms of Directive (European Council, 1991) enacted the saturated zones in fractured bedrock or in in the Republic of Ireland (from now on significant bodies of sand and gravel. referred to as Ireland) in 2006 under Statutory Therefore, groundwater in subsoils, even Instrument (SI) 101, 2009 is currently the though it is the initial recipient of leached main legislative mitigation measure in place to nutrients,is not considered part of the aquifer. achieve the goals of the WFD. The Nitrates In a recent study investigating groundwater Directive sets limits on stocking rates on farms and surface water contributions to stream flow in terms of the quantity of N from livestock in Ireland, subsurface soil and subsoil water manure that can be applied mechanically or (with the exception of sand and gravel) were directly deposited by grazing livestock on agri- termed interflow, and shallow groundwater cultural land. A limit of 170 kg N ha-1year-1 were described as shallow bedrock groundwa- from livestock manure was set. However, the ter where permeability is higher and fracturing EU Nitrates Committee approved Ireland’s is more dominant (Jennings et al., 2007). application for a derogation of this limit to Saturated subsoils (interflow) and an allow grassland-based (mostly dairy) farmers underlying groundwater body (shallow and to operate at up to 250 kg N ha1 year1 from deep groundwater) may often form a livestock manures, under the understanding hydraulic continuum, but usually do not share that this derogation will not impinge on meet- the same hydrogeological characteristics. This ing the requirements of the Nitrates Directive. defining framework has clear implications for The current average stocking density on dairy nutrient migration pathways and their man- farms is 1.81 livestock units (LU) ha-1 agement. It is therefore important to investi- (Humphreys, 2008). gate all scenarios that could contribute to Nitrate (NO-) leaching pathways between water quality status. The components of lag 3 soils, groundwaters and rivers are generally time for nitrates are: vertical travel time, hori- long and complex (Collins and McGonigle, zontal travel time, flushing of an aquifer to 2008) and such pathways vary depending on below a threshold value and travel time soil/subsoil type, bedrock geolo- through the hyperoic zone. In this paper, ver- gy/hydrogeology and climatic factors such as tical and horizontal travel times will only be rainfall. The lag time between introducing considered. This travel time represents initial protection measures and first improvements in or fastest breakthrough of nitrate to a surface water quality is therefore likely to be different waterbody. Total lag time may be much longer. in different catchments comprising different soils and geologies and should be considered FFoouurr ccoonnttrraassttiinngg ccaassee ssttuuddiieess –– ssiittee cchhaarraacctteerr-- by policy makers and catchment managers iissttiiccss (Kronvang et al., 2008). Such hydrological The objective of this paper was to compare the travel times will be used in this study to indi- potential vertical and horizontal travel times of catethe nutrient breakthrough response times nitrate to a virtual surface water receptor on under different geologies and to categorise four contrasting study sites in Ireland. These each site as high risk or otherwise. hydrological travel times were then used to The saturated zone beneath a water table is indicate the likely first hydrological response technically defined as groundwater, which is a times to be expected in a virtual receptor 500 principle receptor of water and leached nutri- m from the site. Groundwater vulnerability ents from the unsaturated soil/subsoil. This categorisation is based on the thickness and definition applies to groundwater protection permeability of sub-soil overlying bedrock schemes. However, groundwater recognised as aquifers and the thickness of the unsaturated 130 FENTON, COXON, HARIA, HORAN, HUMPHREYS, JOHNSON, MURPHY, NECPALOVA, PREMROW, RICHARDS: Variations in travel time for N loading to groundwaters in four case studies in Ireland: Implications for policy makers and regulators zone in sand and gravel aquifers (Misstear and prises 1.69% of the land area of Ireland, and Brown, 2008). Each site was assigned a vul- soils with similar vulnerability extend to 4.6% nerability category: of the country (Ryan et al., 2006). The Oak Park, Co. Carlow – high vulnerabili- Teagasc research farm at Curtins, Moorepark ty (unsaturated zone sand and gravel thickness is located within a lowland limestone area. > 3 m); Moorepark, Co. Cork – extreme vul- The farm presently evaluates alternative man- nerability (moderate permeability soils agement systems for spring calving pasture- 0.5–4.0 m thick, karst); Solohead, Co. based milk production. Well-drained soils are Tipperary – moderate vulnerability (low per- present to a depth of 4.5 m. Due to the under- meability soils, 5–10 m thick); Johnstown lying karstified limestone aquifer, groundwa- Castle, Co. Wexford – moderate vulnerability ter flow direction is not uniform on site. (low permeability soils, 5–10 m thick). Depth to bedrock approximately is 2.5 m and depth to groundwater is approximately 29 m. Sand and gravel aquifer,Oak Park, Co. Carlow This case study was based on hydrogeological Poorly productive aquifer, Solohead, Co. investigations performed by Premrov et al. Tipperary. - (2007 and 2008), and on a bromide (Br ) The Teagasc research facility at Solohead is tracer study initiated by Hooker (2005). located on apoorly drained till to a depth of 5 Hydrogeological investigations were carried m, overlying a poorly productive Devonian out on a 10 ha area, underlain by a shallow sandstone and mudstone aquifer >10 m sand and gravel aquifer with inter-bedded depth. The site is undulating with occasional silt/clay lenses (Premrov et al., 2007 and emergence of sand and gravel at the surface. - 2008). The Br experiment was carried out on This is a heavy textured clay loam soil (25% an adjacent field approximately 200 to 300 m sand and 42% clay). Heavy textured soils clas- south-east of the case study site (Hooker, sified as clay, clay loam or silty clay loam rep- 2005). Sand and gravel aquifers underlie resent 32% of agricultural soils in Ireland approximately 2% of Ireland and are the only (Humphreys et al., 2008). The boundary aquifers with solely intergranular permeability. between the till and underlying bedrock is not Such aquifers are usually unconfined and are uniform and the sand and gravel aquifer is not generally thin; typically between 5 and 15 m wholly confined. Shallow groundwater flow saturated thickness. Premrov et al. (2007) direction in the till may deviate from general found mean concentrations at the start of the flow direction in the deeper confined aquifer experiment in November 2006 in the unsatu- (Daly and Teillard, 2001). Concentrations of rated zone, at 0.9 m below ground level (bgl), NO -N in shallow groundwater (1 m bgl, 3 of 59 mg NO -N L-1. Mean concentrations of below ground level) under four dairy grassland 3 35 mg NO -N L-1were observed in the satu- based systems from 2001–2002 were exam- 3 rated zone during the same period. ined by Humphreys et al.(2008). Losses of NO -N to shallow (1 m bgl) groundwaters 3 Karst limestone aquifer, Curtins farm, Fermoy, were low; mean groundwater NO -N concen- 3 Co. Cork trations were <3.0 mg L-1. NO -N losses were 3 Atypically in the EU, Ireland has predomi- largely independent of N inputs, N surpluses, nantly unconfined bedrock aquifers with fis- deposition of excreta-N at the soil surface and sure permeability only. Carboniferous lime- residual mineral N in the soil at the start of stone is the primary aquifer and occurs as out- drainage. crop or shallow bedrock covering approxi- mately 50% of Ireland. The free draining acid brown earth, which dominates the site, com- 131 149 The Teagasc research facility at Johnstown Castle, located on heterogeneous glacial 150 deposits on a 4.2 ha gently sloping (2%) field, comprised six non-grazed plots on a 151 beef farm. The overburden, morainic in nature, varies in thickness from 1–20 m. This 152 is underlain by Pre-Cambrian greywacke, schist and massive schistose quartzites that 153 have been subjected to low grade metamorphism (Diamond, 1988). The surface 154 elevation increases sharply at the head of the plots (>71 m Above Ordnance Datum) 155 forming a hill (Sandhill) comprising both sand and fine loamy till. Some of this sand 156 may have been soliflucted, resulting in stratification between sand and underlying low 157 permeability fine till. The Sandhill is well to excessively drained, whereas the plots 158 are moderately to poorly drained. Shallow groundwater NO-N concentrations ranged 3 159 from background levels <1 mg NO-N L-1 to >11.3 mg NO-N L-1 (Fenton et al., 3 3 160 2009). 161 162 Materials and methods 163 Calculation of vertical pathway travel time in soil/subsoil 164 Vertical water flow and leaching of nutrients through unsaturated soils (Fig. 1) are Tearmann, the Irish journal of agri-env1i6r5o nminefnluteanlc erde sbeya rncegha,t i vVeO Lpo 7re, 2w0a0te9r pressure. Flow is facilitated by matrix and 166 gravitational potential gradients. Depending on the matrix potential, which is related Poorly productive aquifer, Johnstown Castle, tribution or soil/subsoil type, and water con- 167 to the pore size distribution or soil/subsoil type, and water content in the unsaturated Co. Wexford tent in the unsaturated zone, the hydraulic The Teagasc research facility at Johnstow1n68 coznondeu, cthtiev hityyd rawuliilcl ccohndauncgtiev.i tyF woril l vcehratnigcea. lF otrr avveertlical travel time estimation, depth Castle, located on heterogeneous glacial time estimation, depth of infiltration calcula- 169 of infiltration calculations based on effective drainage (from January 2005 - January deposits on a 4.2 ha gently sloping (2%) field, tions based on effective rainfall (from January comprised six non-grazed plots on a beef farm. 2005 - January 2008) and subsoil physical 170 2008) and subsoil physical parameters versus watertable depth were used as follows The overburden, morainic in nature, varies in parameters versus watertable depth were used thickness from 1–20 m. This is underlain b1y71 asf ofor lallol wcass ef ostru dailels :c ase studies: Pre-Cambrian greywacke, schist and massive ED 172 RF [1] schistose quartzites that have been subjected = NED [1] to low grade metamorphism (Diamond, RF 173 PV RF [2] 1988). The surface elevation increases sharpl1y73 PV = [ 2 ] [2] = n at the head of the plots (>71 m Above PV Ordnance Datum) forming a hill (Sandhill1)74 TD PV /1000 [ 3 ] [3] 174 TD= NED/1000 [3] comprising both sand and fine loamy till. = Some of this sand may have been soliflucted11,7755 whwwehhreeerr eeR RRFFF isii ss rerreeccchhhaaarrrgggeee fffllluuuxxx (((mmmmmm dd aadyya--11y))-,,1 )EE,DD E Diiss eeiffsffeeccttiivvee ddrraaiinnaaggee ((rraaiinnffaallll--aaccttuuaall resulting in stratification between sand and effective rainfall (rainfall-actual evapotranspi- 176 evapotranspiration) for the study period (mm), NED is number of days recharge underlying low permeability fine till. Th1e76 raetivoanpo)t rfaonrsp itrhatei onst) ufdoyr tphee rsitouddy (pmermio)d, (NmmE)D, NiEsD is number of days recharge Sandhill is well to excessively drained, where1-77 nuomccburesr i no fth disa pyesr iroedc h(daarygse), oPVcc iusr pso irne vtehliosc iptye r(miomd day), n is effective porosity and 177 occurs in this period (days), PV is pore velocity (mm day), n is effective porosity and 593a s thFe ipgulortes Caraep mtioondse rately to poorly drained. (days), PV is pore velocity (mm day), nis 178 TD is total depth of infiltration during the study period (m). Depending on antecedent Shallow groundwater NO -N concentration1s78 effTeDct iisv teo taplo dreopsthit yof ainnfidlt raTtiDon dius ritnogt tahle dsteupdyth p eroiofd (m). Depending on antecedent 3 594r ange d from background levels <1 mg NO -N179 insfoilitl rcaotnidointio nds,u erfifnecgt ivteh dera instauged my ayp oerr imoady n(omt o)c.cur and can be determined daily 3 179 soil conditions, effective drainage may or may not occur and can be determined daily L-1 to >11.3 mg NO -N L-1 (Fenton et al., Depending on antecedent soil conditions, 5952 009F).igure 1. Generi3c diagram of vertica11l88 00a nedf fbbehyyco aatri visszppeoee ccrniiafftiiiaccnl fmm apoolaldd teemhll,,w aee..yagg y.. oSSsr cc thhomuu llttaaeey ee vttn iaarollt..tu ((a22ol00c 00cr55eu))c rff eooaprr ntggodrraarss ssllaanndd ssyysstteemmss iinn IIrreellaanndd uunnddeerr 181 cadni ffbeere ndte dtrearimnaignee cdo nddaitiiloyn sb (yM ao osrpepeacrikf,i cS omloohedaedl ,and Johnstown Castle). Premrov 596 500 m away. 181 different drainage conditions (Moorepark, Solohead and Johnstown Castle). Premrov MMaatteerriiaallss aanndd mmeetthhooddss e.g. Schulte et al.(2005)for grassland systems 182 et al. (2009) has developed an extension of this model for tillage systems, based on Calculation of vertical pathway travel time in182 ine It raell.a (n2d00 u9)n hdaesr ddeivfefleorpeendt adn reaxinteangsieo nc oonf dthiitsi omnosdel for tillage systems, based on 597 soil/subsoil 183 (Mreosuolrtse pfraormk ,t hSeo lOoahke Paadr ka ncads eJ osthundys tsoitwe.n M Cetaesotrloel)o.gical data was collected on sites 183 results from the Oak Park case study site. Meteorological data was collected on sites Vertical water flow and leaching of nutrients Premrov et al.(2009) has developed an exten- 598 184 in Oak Park, Moorepark and Johnstown Castle. Solohead weather data was compiled through unsaturated soils (Fig. 1) are influ1-84 sioinn Ooafk t hPaisrk m, Modooerle fpoarrk t ialnlad gJeo hsnyssttoewmn sC, absatlsee. dS oolnohead weather data was compiled 599e nced by negative pore water pressure. Flow i1s85 resuusilntgs sfirtoe msp ecthifeic Oraiankfa llP daarkta acnads et hset uredmya insiinteg. data from the Moorepark site. 185 using site specific rainfall data and the remaining data from the Moorepark site. facilitated by matrix and gravitational poten- Meteorological data was collected on sites in 186 Where available, tracer experiment breakthrough data in shallow ceramic 600t ial Fgirgaudriee n1.t s . Depending on the matri1x86 OWakh ePrea rka,v Mailaoboler,e ptarrakce ra nedx pJeorhimnesntot wbnre aCkathsrtoleu.gh data in shallow ceramic potential, which is related to the pore size dis1-87 Socluophs/epaidez owmeeattehrse rw dasa atals ow uasse dc o(Omapk iPleadrk uasnidn Mg osiotreepark). 187 cups/piezometers was also used (Oak Park and Moorepark). 188 Figure 1:Generic diagram of vertical and hor1i8z8o nta l pathways to a virtual receptor 500 m away. 500 m 189 Calculation of horizontal travel time in shallow groundwater 189 Calculation of horizontal travel time in shallow groundwater 190 Flows in the saturated zone (Fig 1.) are a function of the saturated hydraulic Shallow/perched 190 Flows in the saturated zone (Fig 1.) are a function of the saturated hydraulic groundwater Vertical pathway 191 conductivity (K ) and the po(tveanrtiiaabll eg rdaedpitehn)t, which in most cases is gravitational. In 191 conductivity (Ksat) and the potential gradient, which in most cases is gravitational. In sat Horizontal pathway 192 the saturated zone, K remains constant at a particular location but varies spatially 192 tRhiev esraturated zone, Ksat remains constant at a particular location but varies spatially sat Watertable 193 due to the heterogeneity of the aquifer and between different aquifers and geological Bedrock aquifer 193 due to the heterogeneity of theS uabqsuoifile br eadnrodc bk etween different aquifers and geological interface e.g.11 L99im44 estouunnne iiattqss..u ifKKersat mmaayy aallssoo vvaarryy dduuee ttoo aanniissoottrrooppiieess iinn tthhee aaqquuiiffeerr.. HHoorriizzoonnttaall ttrraavveell ttiimmee sat 195 estimation was calculated by effective Darcian linear velocity, v (m day-1). This can 195 estimation was calculated by effective Darcian linear velocity, v (m day-1). This can 196 be calculated from: 601 196 be calculated from: 6021 32 603 FENTON, COXON, HARIA, HORAN, HUMPHREYS, JOHNSON, MURPHY, NECPALOVA, PREMROW, RICHARDS: Variations in travel time for N loading to groundwaters in four case studies in Ireland: Implications for policy makers and regulators specific rainfall data and the remaining data RReessuullttss from the Moorepark site. Where available, The vertical and horizontal travel time results tracer experiment breakthrough data in shal- from the individual case studies are presented low ceramic cups/piezometers was also used below. (Oak Park and Moorepark). SSaanndd aanndd ggrraavveell aaqquuiiffeerr,, OOaakk PPaarrkk,, CCoo.. Calculation of horizontal travel time in shallow CCaarrllooww groundwater Vertical travel time Flows in the saturated zone (Fig 1.) are a func- Mean precipitation during the 2005–2008 tion of the saturated hydraulic conductivity period was 857.5 mm. The estimated vertical (K ) and the potential gradient, which in matrix travel time of approximately 1 to 2 sat most cases is gravitational. In the saturated years corresponded with mean total effective zone, K remains constant at a particular rainfall of 350.2 mm. Effective drainage was sat location but varies spatially due to the hetero- estimated to have occurred on 89 days year-1 geneity of the aquifer and between different (mean value for 2005 to 2008), which equates aquifers and geological units. K may also to an average vertical pore velocity of 15.7 sat vary due to anisotropies in the aquifer. mm day-1 (n=25%) through the unsaturated e Horizontal travel time estimation was calcu- zone and depth of infiltration in this period of lated by effective Darcian linear velocity, v (m 1.4 m. In the tracer experiment initiated by day-1). This can be calculated from: Hooker in 2004, the breakthrough of Br-trac- er occurred in suction cups at 0.9 m depth 1 dh 197 v K [ 4 ] ( n = 8 ) f r o m 2 7 t o 4 2 days after surface tracer =− satn dx e application (Hooker, 2005). This gives daily 198 w[4h] e re Ksatis estimated from slug injection vertical travel times of 0.02 to 0.03 m day-1 tests (Bouwer and Rice, 1976), hydraulic gra- (Hooker, 2005). In the saturated zone the 199 where K is estimated from slug injection tests (Bouwer and Rice, 1976), hydraulic dient is staht e change in the watertable head over time of first occurrence of the tracer was after 200 tghraed idenist tiasn tchee cbheantwgee einn t htew woa tewrtealbllse ihne adt hoev erf itehled dist2an2c7e bedtwayese,n twwoh wileel lst ihne peak concentration (dh/dx) and n is effective porosity calculated occurred 345 days after tracer application 201 the field (dh/dx)e and n is effective porosity calculated from soil cores. For the Oak from soil cores. Fore the Oak Park case study, (Premrov et al., 2007). The average water- 202 KPsaarkt cvaasleu estsu dwye, rKes att avkaelune sf rwoemre ttahkee nli tfreormat uthree liatnerdaturet aanbdl ea rde einpdtihc atiinv e tohfi sI ripshe riod was 2 to 5 m bgl are indicative of Irish fluvio-glacial sand and (Premrov et al.,2007 and 2008). 203 gflruavvieol- glaaqciuail fsearnsd. aKnd grvavaelul aeqsu iwfeersr.e K staht veanlu ecso wmer-e then combined with Premrov sat 204 beit nael.d ( 2w00i7th) dPatrae tmo rcoalvc uelta tae lt.h e( 2li0ne0a7r )vedloactiaty t aon dc atrla-vel Htimoersi.z Ionn tthael Storalovheela tdi mcaese culate the linear velocity and travel times. In Horizontal travel time must consider vertical 205 studies K and n values from the literature were used. Where these factors were not the Solosahtead cease studies K and n values variations and lateral changes in the sediment sat e 206 farvoamila bthlee, olirt ienr actirucruem wstaenrece us swehde.r We ashseersesm thenets oef f laince-ar vseeloqcuiteyn ics ev,e ryfr coomm plgerxa, vaes l to sand to clay/silt, tors were not available, or in circumstances which can occur over quite short distances, 207 with karst or fractured bedrock aquifers (Moorepark), actual horizontal travel time where assessment of linear velocity is very with domination of sand and gravel sediments 208 cuosimngp lteraxc,e ra sb rweaiktthh rokuagrhs ti no rw eflrlas cwtuarse dus ebde dtarkoecnk fromin B tahretl eayr e(a2 0(0P3r)e. mFroor vt heet al.,2007 and 2008). It aquifers (Moorepark), actual horizontal travel can be seen from Table 1 that a range of hori- 209 Johnstown Castle study, K data (slug injection tests) from shallow groundwater time using tracer breaksatthrough in wells was zontal travel times from 0.056 to 5.6 years is 210 upsieezdo mteatekresn ( Fefnrotomn etB aalr.,t l2e0y0 9() 2a0n0d 3n). dFatoa r( Ftehnteon eet xaple.,c 2te0d08 )f owre rae uflsuedv iotog lacial/mixed sand and e Johnstown Castle study, K data (slug injec- gravel aquifer. sat 211 estimate horizontal travel time. tion tests) from shallow groundwater 212 p iezometers (Fenton et al.,2009) and nedata Total travel time (Fenton et al.,2008) were used to estimate The total travel time of approximately 1 to 6.6 213 hRoersiuzlotsn tal travel time. years is the combination of vertical travel time 214 The vertical and horizontal travel time results from the individual case studies are 215 presented below. 133 216 217 Sand and gravel aquifer, Oak Park, Co. Carlow 218 Vertical travel time 219 Mean precipitation during the 2005–2008 period was 857.5 mm. The estimated 220 vertical matrix travel time of approximately 1 to 2 years corresponded with mean total 221 effective drainage of 350.2 mm. Effective drainage was estimated to have occurred on 604 Te armann, the Irish journal of agri-environmental research, VOL 7, 2009 605 TabTlaeb 1le: 1O. Oaka kP Paarrkk ccaassee ssttuuddyy a aqquuifiefre pr apraamraemterest earnsd a enstdim eastteimd haoteridz ohnotarli ztroanvetal lt itmraevs.e l times. 606 Aquifer type K dh/dx n v Horizontal travel time sat e m day-1 % m day-1 years Clayey gravel 0.1-1b 0.0122d 25g 4.9(cid:2)10-3 -0.488 f 280.7 - 2.81 f Fluvio-glacial sand and gravel 100 -500a, b, c 0.0122d 25g 4.88-24.4 f 0.28 -0.056 f Mixed sand and gravel h 5-100h 0.0122d 25g 0.244 -4.88 e 5.6 - 0.28 e Homogeneous gravel h 100-1000h 0.0122d 25g 4.88-48.8 f 0.28 -0.03 f 607 608 a Based on Allen and Milenic (2003); b Based on Misstear et al. (2009); c Based on McConville et al. (2001); d From 609 Premrov et al. (2007); e From Premrov et al. (2007) - computed values; f This study; g From Fetter (2001); h From 610 Kruseman and de Ridder (1992), given for comparative purposes. 611 612 (ap proximately 1 year) through the unsaturat- m on site, which means travel time at different ed zone (according to Hooker, 2005)and the locations will reflect this depth. Bartley (2003) 613 fas test horizontal travel times from Table 1 used a surface Br- tracer recharge experiment through the saturated zone to a virtual surface to quantify the vertical unsaturated travel time 614 rec eptor 500 m away from the chosen site to ceramic cups at 1 m bgl and to the screened (0.056 to 5.6 years - for fluvioglacial/mixed section of a well at 29 to 32 m bgl in ground- 615 sand and gravel). This reflects the best case water. Maximum vertical travel velocity due to scenario in terms of lag time between changed preferential flow paths through the unsaturat- 616 practices and improved water quality. ed zone was estimated as 1.66 m day-1, giving 617 a travel time of approximately 17 days. On a KKaarrsstt lliimmeessttoonnee aaqquuiiffeerr,, CCuurrttiinnss ffaarrmm,, neighbouring farm, also located on thin soils 618 FFeerr mmooyy,, CCoo.. CCoorrkk (depth to bedrock 1 m) underlying a karstified Vertical travel time limestone, first occurrence of tracer in a 619 Mean precipitation during the 2005–2008 screened interval at approximately 22 m bgl period was 953 mm. The vertical matrix trav- using the same methodology was 34 days 620 el time of approximately 1 to 2 years corre- (Richards et al.,2005). Maximum soil solu- 621 spo nded with mean total effective rainfall of tion concentrations were observed 34-65 days 437.2 mm. Mean effective rainfall was esti- after application, giving a vertical travel time 622 ma ted to have occurred on 158.5 days, which of 34 to 65 days (Richards et al.,2005). ). It equates to an average vertical pore velocity of was concluded that matrix flow through the 623 10 .03 mm day-1through the unsaturated zone overburden and preferential flow through the and depth of infiltration in this period of 1.2 remainder of the unsaturated zone (karst lime- 624 m. The depth to bedrock is approximately 1–4 stone) is quantitatively more important for 631 625 Table 2: Moorepark, horizontal travel time from K range and actual horizontal travel time 632 Table 2: Moorepark, horizontal travel time from K rasnagte and actual horizontal travel time from a 662363 from tra ac etrr aexcepre reimxpeenrt iimn tehnet kianr stth aeq ukifaerrs tt oa rqeuciefpetro rt o5 0rs0aet cme patwoary 5 f0ro0m m th ea wsoauyr cfer.o m the source. 634 627 Tracer Horizontal Travel time Tracer breakthrough Horizontal Travel K breakthrough at based on Tracer sat at point 4 time based on K 628 point 10 sat m day-1 days days time months 629 0.004 to 27a 44 51 Days to Years 2 to 3 635 663306 a Based on Bartley (2003), slug injection tests 637 638 134 639 640 641 642 643 644 645 646 647 648 649 650 651 652 653 654 655 656 FENTON, COXON, HARIA, HORAN, HUMPHREYS, JOHNSON, MURPHY, NECPALOVA, PREMROW, RICHARDS: Variations in travel time for N loading to groundwaters in four case studies in Ireland: Implications for policy makers and regulators transport of conservative tracers togroundwa- responded with mean total effective rainfallof ter. Over two monitoring years, the mean 481 mm. Mean effective rainfall was estimat- annual farm average NO -N concentrations ed to have occurred on 91 days, which equates 3 were 11.9 and 15.2 NO -N L-1 with consid- to an average vertical pore velocity of 52 mm 3 erable temporal variation observed within day-1through the unsaturated zone and depth years (Bartley, 2003). of infiltration in this period of 4.8 m. The mean (mean watertable depth ±standard devi- Horizontal travel time ation) perched shallow watertable on site dur- The horizontal travel time in the karst aquifer ing this time varied from 1.5±0.3 m bgl (sum- was difficult to estimate as considerable data mer) to 0.6±0.2m bgl (winter). Vertical trav- was required (ne, fracture density, fracture el time to the perched groundwater is achiev- truncation, fracture orientation, fracture trace able within one drainage season. Infiltration length, fracture spacing, mechanical aperture, may take several years to reach the boundary effective hydraulic aperture, and aperture of the sand and gravel confined layer. During opening size) to estimate K values. A K this time period effective rainfallranged from sat sat range was available from slug injection tests 312 to 605 mm, which means that in some (Bartley, 2003) but was not indicative or com- years vertical travel time in 1 year was not parable with actual horizontal flow travel achievable. times. Instead, tracer breakthrough in several wells originating from the vertical travel time Horizontal travel time study was used to calculate horizontal travel Horizontal travel times on the poorly produc- time. The horizontal travel time to the recep- tive aquifer were estimated to be 3424 years in tor 500 m away from the source following this the shallow perched groundwater to 152 years pathway was estimated as 2 to 3 months. in the confined aquifer (Table 3). Estimates of the horizontal travel time could be improved Total travel time further through investigation of K on the sat Vertical travel time was estimated at 1 to 2 site. years to allow for differences in soil depth. Ceramic cup and borehole tracer data indicat- Total travel time ed faster preferential flow paths on site. Once Combining vertical travel time of 3 years and water and nutrients pass through the soil layer, the horizontal travel time of 3424 years for the travel time increases rapidly. Horizontal travel shallow perched groundwater, the total travel time was estimated as 2 to 3 months. Total time is estimated to be 3427 years. For zones travel time from source to receptor following that are connected to the confined aquifer the groundwater flow direction was approximate- travel time may be shorter. ly 1.5 – 2.5 years and was dependent on soil thickness. In such a complex environment a PPoooorrllyy pprroodduuccttiivvee aaqquuiiffeerr,, JJoohhnnssttoowwnn CCaassttllee,, modelling exercise (e.g. global or distribution CCoo.. WWeexxffoorrdd methods) at the correct scale should be under- Vertical travel time taken. Mean precipitation during the 2005–2008 period was 1046 mm. The vertical matrix PPoooorrllyy pprroodduuccttiivvee aaqquuiiffeerr,, SSoolloohheeaadd,, CCoo.. travel time of approximately 1 year corre- TTiippppeerraarryy sponded with mean total effective rainfall of Vertical travel time 553 mm. Mean effective rainfall was estimat- Mean precipitation during the 2005–2008 ed to have occurred on 178 days, which period was 1059 mm. The mean vertical equates to an average vertical pore velocity of matrix travel time of approximately 1 year cor- 9.7 mm day-1 through the unsaturated zone 135 Tearmann, the Irish journal of agri-environmental research, VOL 7, 2009 657 658 TabTlaeb l3e: 3S: oSloolhoheeaadd,, hhoorriizzoonntatal lt rtarvaevl etli mtiem ine sihna llsohwal alonwd daenepde rd gereopuenrd wgraoteurn tdo wsuartfearc et ow astuerr fvaicretu awl ater 659 virrteucaelp rtoerc e5p0t0o mr 5a0w0a ym fr oamw athye f srooumrc et.h e source. 660 Groundwater Horizontal layer K dh/dx n v Travel time sat e m day-1 % m day-1 years A 0.01a 0.004 10c 0.0004 3424 B 0.224b 0.004 10c 0.00896 152 661 662 A Shallow perched groundwater 663 B Deeper Confined Aquifer 664 a, c Based on Fetter (2001) 665 b Based on Daly and Teillard (2001) 666 687 667 688 TTabablele 4 4:: JJoohhnnssttoowwnn C Casatsletl,e m, meane ahno rhizoorniztaoln trtaavl etlr ativmeel tinim siex imn osnixit omreodn piltootrse tdo pa lsoutrsf atcoe aw sauterrf ace water 68696 8 rerce ecpepttoorr 550000 mm a wawaya yfr ofrmo mth et hsoeu srcoeu. rce. 690 669 Horizontal Plot K dh/dx n a v Travel time sat e m day-1 % m day-1 Years 670 1 0.009a 0.39 32 0.011 125 671 2 0.008a 0.24 32 0.006 228 3 0.011a 0.35 32 0.012 114 4 0.011a 0.38 32 0.013 105 672 5 0.012a 0.32 32 0.012 114 6 0.008a 0.32 32 0.008 171 691 a From Fenton et al. (2008) 69627 3 and depth of infiltration in this period of 1.7 and therefore high residence times of water in 69637 4 m (Fenton et al.,2008). certain aquifers means that reversing this The depth to the median watertable during increasing trend may not be achievable within 69647 5 this period was 1.01 m (Fenton et al.,2009). the 2015 timeframe. A comparison of total 69657 6 travel times from all the contrasting case stud- Horizontal travel time ies is presented in Table 5. 69667 7 H o rizontal travel time ranged from 105 to For the purposes of illustration, a simple 228 years (Table 4.) model of transport has been used to show dif- 69677 8 ferences in first breakthrough travel time from Total travel time source to receptor in four contrasting sites. 69687 9 V e rtical travel time was estimated at less than This simple methodology shows the wide vari- 1 year and horizontal travel time was estimat- ability in travel times that can be expected 69698 0 ed from 105 to 228 years. This gives a total within catchments in Ireland (several months 70608 1 tr a vel time of approximately 106 to 229 years to years to reach a receptor 500 m away). In to the receptor 500 m away from the source. reality considerable heterogeneities and 70618 2 anisotropies in soil, sediment and bedrock DDiissccuussssiioonn hydrogeological properties make transport 70628 3 Ir is h case studies pathways and times highly variable. Examples This is the first attempt at calculating nitrate of this include discrepancies between travel 70638 4 lag times in Ireland and there are no direct time based on matrix flow (depth of infiltra- 70648 5 m e thodologies to calculate lag time across tion) or preferential flow (tracer breakthrough Europe. Groundwaters in Ireland are at risk of data). In some instances N transfer to shallow 70658 6 fa i ling to meet the water qualitytargets of the groundwater is rapid with little chance for WFD by 2015. Slow groundwater travel times attenuation. Zones of high conductivity asso- 706 707 1 36 708 709 710 711 FENTON, COXON, HARIA, HORAN, HUMPHREYS, JOHNSON, MURPHY, NECPALOVA, PREMROW, RICHARDS: 712 VaTraiabtlieo n5s.V inu ltnrearvaebli ltitiym ec aftoerg oNr yl,o aatdtienngu atoti ognr ocuapnadcwiatyt,e rtost ainl efsotuimr actaesde tsratuvdeli etsim ine sI raenlda nd: Implications for policy makers and regulators 713 Talbagle t5im:Ve uislnseureasb filoitry t hcaet efogourry c, oanttternaustaitniogn c acsaep asctiutdy,i etos.t a l estimated travel times and lag time issues for the four contrasting case studies. Case study Vulnerability Attenuation Total Lag time issues Case study Vulcnaeteragboirlyit y Atctaepnuacaittiyo n Ttroatvaell Lag time issues category capacity travel time time Indicative of initial breakthrough Contaminated shallow groundwater Months interaction with groundwater and Oak Park, Co High Moderate to surface water. Flushing utilising Carlow Decades specific yield of the aquifer should be incorporated into lag time. N mineralisation in soils, travel time will change depending on soil N Moorepark, status, thickness of soil and Extreme None Years Co. Cork preferential flow paths. Flushing utilising specific yield of the aquifer should be incorporated into lag time. Connectivity to semi confined aquifer. Solohead, Co. Phosphorus lag time to surface water Moderate High Decades Tipperary and groundwater. NO unlikely to be 3 problematic. Contaminated shallow groundwater Johnstown interaction with surface water in close Castle, Co. Moderate High Decades proximity to excess nutrients. N Wexford mineralisation in soils. ciated with gravel channel deposits or fracture groundwater (with the exception of sand and systems, may connect a source area with a gravel aquifers) under the WFD definition. receptor so that travel times may be signifi- Groundwater contribution to surface waters in cantly less than estimated. In this instance a poorly productive aquifers may be as low as small area of land may have a disproportionate 5% but can be as high as 30% when flow effect on water quality at the receptor. along the subsoil/bedrock interface is consid- Saturated soil and subsoil is not deemed ered; by contrast, in productive aquifers such 137 Tearmann, the Irish journal of agri-environmental research, VOL 7, 2009 as sand and gravel and karstified limestone, the coarser gravel lenses present at this site the contribution may be from 80% to 90% Irish bedrock aquifers are deemed to have (Misstear et al.,2009). Components of deep low attenuation potential due to their frac- groundwater, intermediate (interflow and tured and karstified nature and protection is shallow groundwater) and overland flow that mainly provided by the overlying glacial tills contribute to stream flow were assessed in (Daly and Warren, 1998). In karst environ- seven pilot catchments and used to constrain ments K measurements are only indicative sat and inform the numerical model NAM of how fast groundwater flows at a specific (Nedbør-Afstrømnings-Model) in Ireland point, which leads to vast ranges in K e.g. In sat (Jennings et al., 2007). This model was then Waulsortian limestone (a lithological sub- extended to regional catchments. The model group of Carboniferous Limestone), K val- sat estimated large amounts of intermediate flow ues range from 4 to 2500 m day-1 contributed to stream flow,e.g. 318 mm y-1in (Goldscheider and Drew, 2009). In the vicin- the Owenduff poorly productive aquifer. In ity of the case study site, K values such as a sat such a catchment deep groundwater con- public water supply, 873 m day-1 and nearby tributed 128 mm y-1 and overland flow con- water abstraction wells range from 10 to 200 tributed 1322 mm y-1. In the Suck pilot karst m day-1. From slug injection tests, bedrock catchment intermediate flows were 362, 171 K values in the case study area ranged from sat and 124 mm y-1 respectively. Such informa- 0.004 to 27 m day-1(Bartley, 2003). The soils tion, when combined with nitrate concentra- here have high total nitrogen levels and thus tions leaving the rooting zone can be used to elevated mineralisation rates. Even if fertiliser calculate loads of nutrients in intermediate inputs are reduced it will take a long time to and groundwater flow. This is important for prevent losses (Grimvall et al.,2000) from the calculating aquifer flushing times to certain unsaturated zone. In this respect groundwater EU threshold values. trend monitoring will not reflect mitigation For groundwater classification chemical measures in place at farm level but instead will and quantification status under the WFD sev- represent a mixture of bio-geochemical eral tests must be complied with. For example processes. “no significant diminution of surface water In the Oak Park and Moorepark case stud- chemistry and ecology” determines if ground- ies,shorter travel times and less denitrification water impacts on a surface water body achiev- potential in the soils and aquifer put receptors ing the goals of the WFD. The groundwater at risk. In terms of maximising the effective- threshold values used are surface water quality ness of resources applied to improve water standards with a dilution factor. This dilution quality, it is likely to be most successful if factor is determined by the shallow and deep problem receptors are identified and mitiga- groundwater contributions, but not the inter- tion measures targeted in areas that contribute flow components. Further differentiation of to that receptor. Implementing measures on intermediate flow is needed. sites for which there is no receptor water qual- Drilling records for the Oak Park site indi- ity problem or for which the travel time and N cate that the clay/silt lenses are not continuous remediation potential is such that changes in and may not represent a barrier for transport management are not likely to impact water of water soluble pollutants in this fluvio- quality (a blanket approach) is not an efficient glacial deposit aquifer (Premrov et al.,2007). use of resources. The time it takes to flush an Therefore, the estimated travel time of 1 to entire plume of nitrate through interflow, 6.6 years was shown as an indication of the shallow groundwater and deep groundwater fastest possible total travel time through the pathways should be considered as an impor- unsaturated and shallow groundwater through tant component of lag time. This of course 138