ELECTRONIC SUPPLEMENTARY MATERIAL 1 LCA FOR AGRICULTURE PestLCI 2.0: A second generation model for estimating emissions of pesticides from arable land in LCA Teunis J. Dijkman • Morten Birkved • Michael Z. Hauschild Received: 23 December 2011 / Accepted 4 May 2012 © Springer-Verlag 2012 Responsible editor: Ivan Muñoz T. J. Dijkman () • M. Birkved • M. Z. Hauschild Department of Management Engineering, Technical University of Denmark, 2800 Lyngby, Denmark e-mail: tedi@dtu.dk () Corresponding author: Teunis J. Dijkman Tel.: (+45) 45 25 48 86 e-mail: tedi@dtu.dk This document describes the equations used for the calculation of emissions of pesticides to air, surface water and ground water for the new and updated processes in PestLCI 2.0. For the processes and equations that are unaltered, compared to PestLCI 1.0, we refer to the paper by Birkved&Hauschild (2006). S.1.1 Wind drift The new wind drift curves generally take the following shape: f ( x) A0 e x / B0 A1e x / B1 (eq.1) Where f(x) the fraction lost at distance x (m). A and B are fitting parameters. For the various crop types, the parameters given in Table 1 are used (Holterman & Van de Zande, 2003). Table 1: Parameters used in spray loss curves (Holterman & Van de Zande, 2003) Spray boom Crop type A0 B0 A1 B1 Conventional boom Flower bulb 84 1.30 1.79 0.153 Cross flow Sugar beet 294 2.44 2.39 0.147 Cereals 39 2.28 0.90 0.147 Bare soil 25 1.54 1.50 0.133 Fruit trees, with leaves 48 2.7 0.45 0.091 Fruit trees, leafless 120 6.75 0.45 0.091 For potato crops, eq. 2 is used (Holterman & Van de Zande, 2003): f ( x) 114 x 1.29 (eq.2) S.2.1 Volatilization from plant leaves By combining 3 leaf volatilization datasets for pesticides having vapour pressures varying between 1.0∙10-4 Pa and 2.2∙104 Pa, Van Wesenbeeck et al. (2008) presented how the evaporation rate of pesticides can be regressed as ln [ER] = 12.2 + 0.933∙ln VP (eq. 3) Where ER is the evaporation rate (µg m-2h-1) and VP is the vapour pressure (Pa) at 25°C. Dividing the ER from eq. 3 with the applied pesticide dose, expressed in µg m-2, gives the rate constant of vaporization kTr (h-1) at 25°C. A correction for the ambient air temperature in the month of application is done using eq. 4 (Schwarzenbach et al., 2003): ln k Ta Ea 1 1 k Tr R Tr Ta (eq. 4) Where k(Ta) and k(Tr) are the rate constants of vaporization at actual and reference temperature (K), respectively, E a the activation energy (kJ mol-1) and R the gas constant (8.3145 J K-1 mol-1). For pesticides a typical Ea of 100 kJ mol-1 is assumed, based on the average value of 95 kJ mol-1 found by Leistra and Wolters (2004). The fraction of pesticide emitted by volatilization from leaves is hereafter calculated using the temperature-adjusted rate constant kTa , assuming first-order evaporation kinetics. S.2.2 Uptake in leaves Since most pesticides are lipophilic molecules, leaf uptake could be assumed to occur dominantly through cuticles (Korte et al., 2000). The first-order uptake of pesticides can then be calculated from the Arrhenius function (Baur and Schönherr, 1995): log k lu a bV ED R 2.3T 273.15 (eq. 5) Where klu is the first-order rate constant (s-1), a and b are plant-species dependent constants, V is the characteristic volume (cm3 mol-1), ED is the activation energy for diffusion (kJ mol -1). Following the same procedure as described by Birkved and Hauschild (2006), re-evaluating the data, the following 2 equations were derived for citrus- and pear-type leaves: Citrus: 78.875 e 3.470MV / 1000 3600 k lu 5.31 5.84 10 3 0.945 MV 2.772 R 2.303 Ta (eq. 6) Pear: 63.222 e 2.874MV / 1000 3 3600 k lu 4.93 5.30 10 0.945 MV 2.772 R 2 . 303 T a (eq. 7) With MV the molecular volume (cm3 mol-1). Citrus and pear leaves have cuticles with quite different thicknesses and compositions, with the citrus leaves being waxier and having a thicker cuticle. Each of the crop types in PestLCI 2.0’s database has been assigned to one of the 2 types, depending on the crop leave properties. S.2.3 Degradation on leaves The approach to model pesticide degradation on leaves has not been changed from PestLCI 1.0, but a new regression of the OH• radical concentration has been established as the previous regression was found to considerably underestimate the radical concentration at higher light intensities: log [OH•] = (4∙10-4 ∙Il) + 4.7783 (eq. 8) Where Il the monthly average light intensity (Wh m-2 day-1). The first-order rate constant for pesticide degradation on leaves hence takes the following form: k ld k OH 10 410 4 I l 4.7783 dl / 24 (eq. 9) Where kld is the first-order photodegradation rate constant (h-1), kOH is the overall OH• oxidation rate constant (cm3 molecules-1 h-1), and dl/24 term corrects eq. 9 for the latitude and day-of-year dependent daylight length. This equation hence works under the assumption that degradation does not take place in the absence of light. Dl is calculated as follows (Math forum, 2011): 0.8333 L sin sin sin P 24 180 180 dl arccos L cos cos P 180 (eq. 10) Where P arcsin 0.39795 cos0.2163180 2 arctan 0.9671396 tan 0.00860 DOY 186 Where L is the latitude (in degrees) and DOY is the day number of the application day in the year. (eq. 11) S.3.1 Soil volatilization The approach for calculation of pesticide volatilization from soil is based on a fugacity level 3 model by Mackay (2001). The rate constant for soil volatilization is calculated using equation 12: k s ,v DV VT Z T (eq. 12) Where DV is the total D value of the volatilization process (mol Pa-1 h-1) and VTZT is the product of the VxZx values of the phases in soil (x=air, water, or solid matter), with V x the volume of phase x (m3) and Zx the phase’s fugacity capacity (mol m-3 Pa-1). D-values are a transport parameter used for fugacity calculations. The overall D-value is calculated from the D values of diffusion through air (DA) and water (DW) and the boundary air layer (DBL). Diffusion through air and water are parallel processes, followed by diffusion trough the boundary air layer, resulting in the following overall transport parameter: 1 1 1 DV DBL D A DW (eq. 13) with DBL A Bia ZA BLT (eq. 14) In eq. 14, A is the surface area of the field under study (m2), Bia the diffusivity of organic molecules in air (cm2 s-1). BLT, the boundary layer thickness, is set to 0.00475 m (Mackay, 2001) but can be user-adjusted. Bia is calculated using a semi-empirical equation given by Schwarzenbach (2003): T 1.75 1 M air 1 / M i 1/ 2 Bia 10 7 1/ 3 p Vair Vi1 / 3 2 (eq. 15) Where T is the absolute temperature (K), Mair and Mi are the molar masses of air and active ingredient i (g mol-1), p the gas phase pressure (atm) and Vair and Vi are the molar volumes (cm3 mol-1) of air and the pesticide, respectively. DA and DW are calculated in a comparable way, using Dx BEx AZ x Y (eq. 16) Where B Ex /3 B x v 10 x v a v w 2 (eq. 17) In eq. 16 and 17 subscript x can be either A (air) or W (water), depending on whether D A or DW is calculated. BEx is the molecular diffusivity, calculated useing vx, the volume fractions of air and water in the total soil. Y is the diffusion path length measured as the vertical distance of the pesticide to the soil surface. Y by default is set to half the depth of the topsoil (0.005 m), thereby assuming the active ingredient is equally distributed over the topsoil. The Vx and Zx values are calculated separately for soil solids, water and air: Vx Vx Vs Vw Va ZS K OC , pH f OC b Z W 1 S (eq. 18) (eq. 19) ZW 1000 S VP MW (eq. 20) ZA 1 R T (eq. 21) In eq. 19-21, Vx the volume of phase x (m3), where s is the soil solid matter, w is soil water, and a is soil air, KOC,pH is the pH-dependent organic carbon-water partition coefficient (L kg-1), fOC is the fraction of organic carbon in the topsoil (dimensionless), ρb is the soil bulk density (kg m-3), ϕS is the soil porosity (dimensionless), calculated by summing the volume fractions of air and water in soil, S is the water solubility of the active ingredients (kg m -3), VP is the vapour pressure (Pa), MW is the molecular weight of the active ingredient (g mol-1). As is evident does the vapour pressure of an active ingredient depend on the ambient temperature. Based on the measurement reference temperature, the vapour pressure at the ambient temperature in the month of pesticide application is determined by assuming that the vapour pressure of the active ingredient increases exponentially with increasing temperature following a tenfold increase per 10 degrees Celsius. For many non-ionic pesticides, the organic carbon-water partitioning depends on the pH of the topsoil. This was not taken into account for the top soil layer in PestLCI 1.0. The pH-dependent KOC,pH is calculated as follows: K OC , pH K OC 10 pH pH 10 10 pKa (eq. 22) With pKa the first dissociation constant of the pesticide. S.3.2 Topsoil degradation In contrast to the approach applied in PestLCI 1.0, the new calculation approach relies on 1 equation for the entire temperature range. It is further assumed that the soil degradation rate will increase with a fixed factor per 10 degrees, resulting in correction factor Fs,T: Fs,t = Q(T-Tref)/10 (eq. 23) Where Q is the increase in degradation rate per 10 degrees, T is the ambient soil temperature, T ref is the reference temperature at which the soil degradation rate was determined. Q is set to 2.1 by default as recommended by the FOCUS soil modelling workgroup (Boesten et al., 1996), but can be adjusted by the model user if desired. The soil temperature T is assumed to be related to the air temperature T air through eq. 24, based on Zheng et al. (1993): Tref = 1.05∙Tair – 1.5 (eq. 24) Where both temperatures are expressed in degrees Celcius. S.3.3 Top soil runoff The runoff fraction is calculated using the approach presented by Berenzen et al. (2005): f sr Q 1 F f s ,t P 1 KD (eq. 25) Where fsr is the fraction of applied pesticide that is emitted to surface water, Q/P is the ratio of runoff amount (mm) and precipitation amount (mm). F is a slope dependent correction factor, f s,t is the fraction of applied pesticide that is present at the moment of precipitation. KD is the ratio of dissolved and sorbed pesticide, calculated by multiplying the pHdependent Koc and the fraction of organic matter in the top soil. In eq. 25, the latter 2 terms together determine the fraction of pesticide available for runoff, the first 2 terms determine which fraction will actually run off. It thus assumed that the rainfall and runoff occurs within such a short time span that the sorbed fraction of pesticide does not have time to desorb. The sorbed fraction of pesticide fraction is therefore assumed unavailable for runoff. In eq. 25, Q further depends on the soil type. For sandy soils (fsand > 0.50): Q = -0.016427 - 0.011377∙P + 0.0026284∙P2 – 5.8564∙10-6∙P3 (eq. 26) For all loamy soils, which were here assumed to be all soils with fsand < 0.50: Q = -0.061108 - 0.0041626∙P + 0.0040395∙P2 – 9.0361∙10-6∙P3 (eq. 27) The correction factor F in eq. 25 is the product of 2 terms, accounting for the slope of the field and the width of the nospray bufferzone: Fslope = 0.02153∙S + 0.001423∙S2 if S < 20% (eq. 28) Fslope = 1 if S ≥ 20% (eq. 29) Fbuffer = 0.83WBZ (eq. 30) Where S is the slope of the field (%), WBZ is the width of the bufferzone (m). In order for eq. 30 to be valid, it is assumed that the bufferzone is densely covered with vegetation (i.e. crop and/or wild flora) and hence acts a “run-off resistance” beyond the field borders. S.3.4 Macropore flow In the PestLCI 2.0 approach, the pore water is divided in ‘mobile’ and ‘immobile’ water, based on the approach from Hall (1993). The first class can flow, the second does not flow through the soil. The ‘mobile’ water is divided in a slowflowing (0.35 m day-1) and a fast-flowing domain. The fast-flowing domain represents the macropores. It is assumed that the pesticides dissolved in the water flowing through the macropores will quickly reach the groundwater and is hence considered to be emitted to ground water without undergoing degradation. The total pore volume is calculated as the sum of the volume fractions of air and water in soil. Based on Hall (1993), it is assumed that the mobile pore fraction depends on the composition of the solid soil fraction: fpore,mobile = 0.72∙fsand + 0.35∙fsilt + 0.14∙fclay (eq. 31) The macropore fraction of the total mobile pore volume is set to 0.30 by default, based on Hall (1993), but can be adjusted by the model user. Following Hall (1993), precipitation water will only flow into macropores when both the immobile pores in the topsoil and the slow-flowing pores are water filled. Using a flow rate of 0.35 m day-1 for the latter group of pores, the volume of water that can be taken up by the topsoil in 1 hour is calculated: 0.35 C f pore,immobile 0.01 1 0.30 f pore,mobile 24 (eq. 32) Where C is the water storage capacity of the soil (m3), fpore, immobile is the total pore fraction minus the mobile pore fraction. Since the soil water content prior to a precipitation event is known, the minimum hourly precipitation rate needed to initiate macropore can be calculated from the average rain intensity.The probability density function of rainfall can be approximated as an exponential function depending on the rainfall intensity (Eagleson, 1972): f i e i (eq. 33) In this equation, β is a constant, and i is the rain intensity (mm hour -1). The constant β is calculated from the mean of this distribution, and therefore equals the mean rain intensity: i 1 iday l precip (eq. 34) Where iday is the average precipitation on a rainy day, and lprecip the average length of a rain event which here is set at 5 hours by default. The constant β, and hence the shape of the probability density function depends on the local climatic circumstances. The average recommended rain event duration has been derived from precipitation data for Danish circumstances (Harremoës and Mikkelsen, 1995) and can in PestLCI 2.0 be adjusted according to user preferences if necessary. Integration of eq. 33 up to the rainfall intensity where no macropore flow occurs, yields the fraction of precipitation events without macropore flow. The fraction of precipitation events with macropore flow is then easily calculated. As the rainfall intensity distribution depends on local climatic circumstances, the rainfall intensity at which macropore flow will start, is also dependent on local climate. As LCA aims at calculating impacts from average scenarios, it can be assumed that the macropore rainfall fraction is the average fraction of rain in a rain event that flows into macropores. Knowing the fraction of pesticide molecules that is dissolved in the leaching water, the fraction of dissolved pesticide that enters macropores and flows to towards the ground water is calculated. References Baur P, Schönherr J (1995) Temperature dependence of organic compounds across plant cuticles. Chemosphere 30:1331-1340 Berenzen N, Lentzen-Godding A, Probst M, Schulz H, Schulz R, Liess M (2005) A comparison of predicted and measured levels of runoff-related pesticide concentrations in small lowland streams on a landscape level. Chemosphere 58:683-691 Birkved M, Hauschild MZ (2006) PestLCI: A model for estimating field emissions of pesticides in agricultural LCA. Ecol Model 198:433-451 Boesten J, Helweg A, Businelli M, Bergstrom L, Schäfer H, Delmas A, Kloskowski R, Walker A, Travis K, Smeets L, Jones R, Vanderbroeck V, Van der Linden A, Broerse S, Klein M, Layton R, Jacobsen O-S, Yon D (1996) FOCUS Report - Soil persistence and EU registration - EU document 7617/VI/96, EU, Brussels Eagleson PS (1972) Dynamics of flood frequency. Water Resour Res 8:878-899 Hall DGM (1993). An amended functional leaching model applicable to structured soils: I: Model description. J Soil Sci 44:579-588 Harremoes P, Mikkelsen PS (1995) Properties of extreme point rainfall I: Results from a rain gauge system in Denmark. Atmos Res 37:277-286 Holterman HJ, Van de Zande JC (2003) IMAG drift calculator v1.1: User manual Korte F, Kvesitadze G, Ugrehelidze D, Gordeziani M, Khatisashvili G, Buadze O, Zaalishvili G, Coulston F (2000) Organic toxicants and plants. Ecotoxicol Environ Saf 47:1-47 Leistra M, Wolters A (2004) Computations on the volatilization of the fungicide fenpropimorph from plants in a wind tunnel. Water Air Soil Poll 157:133-148 Mackay D (2001) Multimedia environmental models: The fugacity approach. Second edition. Taylor and Francis, Boca Raton Math forum (2011). Latitude and longitude and daylight hours. http://mathforum.org/library/drmath/view/56478.html. Accessed on 29 July, 2011 Schwarzenbach RP, Gschwend PM, Imboden DM (2003) Environmental organic chemistry. Second edition. John Wiley and Sons, Hoboken Van Wesenbeeck I, Driver J, Ross J (2008) Relationship between the evaporation rate and vapour pressure of moderately and highly volatile chemicals. Bull Environ Contam Toxicol 80:315-318 Zheng D, Hunt Jr ER, Running SW (1993) A daily soil temperature model based on air temperature and precipitation for continental applications. Clim Res 2:183-191 ELECTRONIC SUPPLEMENTARY MATERIAL 2 Table S1: Input data for PestLCI 2.0, SWASH 3.1 and PEARL 4.4 used for the validation study PestLCI 2.0 SWASH PEARL Pesticide x x x MCPA Molar mass (g mol-1) x x x 200.61 3 -1 Molecular volume (cm mol ) x 152.71 pKa x 3.731 Saturated vapour pressure (Pa) x x x 4.0∙10-4 1 -1 Solubility in water at 25°C (mg L ) x x x 293901 Molar enthalpy of vaporization at 25°C (J mol-1) x x 950002 -1 Molar enthalpy of dissolution (J mol ) x x 270002 2 -1 Diffusion coefficient in water (m d ) x x 5.27E-053 Diffusion coefficient in air (m2 d-1) x x 0.5383 Freundlich exponent (-) x x 0.92 -3 Reference concentration in liquid phase (g m ) x x 12 Factor for the uptake by plant in soil (-) x x 0.52 Wash-off factor from crop MACRO (mm-1) x 0.052 -1 Wash-off factor from crop PRZM (cm ) x 0.52 -1 Wash-off factor from crop (m ) x 504 t½ in soil at 20°C (d) x x x 241 t½ in water at 20°C (d) x 13.55 t½ in sediment at 20°C (d) x 175 t½ in crop at 20°C (d) x x 105 -1 Activation energy (J mol ) x x 654002 -1 Exponent (K ) x 0.09482 Q10fac (-) x 2.582 -1 Koc (L kg ) x 741 -1 Kom at 20°C (L kg ) x 376 log Kow x -0.811 Molar enthalpy of sorption (kJ mol-1) x -37 -1 Desorption rate coefficient at 20°C (d ) x 02 Factor relating CofFreNeq and CofFreEql x 02 Exponent for the effect of liquid x 0.72 Canopy process options x Lumped2 3 -1 -1 Atmospheric OH rate (days)(cm molecules s ) x 1.26∙10-11 1 Width no sprayzone (m) x 101 1: from PestLCI database; 2: model default values; 3: Schwarzenbach et al. (2003); 4: base don SWASH default; 5: PPDB (2011); 6: Kom = ½ Koc, Koc from PestLCI database; 7: Estimate based on Hiller et al. (2008) Table S2: Characteristics of soils used in validation (‘average’ soil) and case study (average, high sand and high clay) Soil # average high clay content high sand content layer Ap B Bt1 Bt2 Ah Bcs (B)Cg Cgx A1 Bw11 Bw12 Bw13 C thickness (m) 0.33 0.22 0.25 0.20 0.15 0.20 0.27 0.38 0.10 0.20 0.30 0.35 0.05 fclay 0.18 0.22 0.27 0.27 0.41 0.35 0.40 0.40 0.10 0.10 0.11 0.10 0.09 fsilt 0.36 0.30 0.28 0.34 0.33 0.41 0.49 0.53 0.20 0.20 0.21 0.21 0.21 fsand 0.46 0.48 0.45 0.39 0.26 0.24 0.11 0.07 0.70 0.70 0.68 0.69 0.70 fOC 1.54 0.25 0.16 0.11 2.8 0.7 0.0 0.0 3.8 1.1 0.4 0.2 0.3 pH 4.4 3.9 3.7 3.6 7.4 7.7 7.8 7.8 4.6 4.6 4.6 4.6 4.8 Table S3: Climate data used for case study (month: May) Location Latitude Longitude Elevation (m) Taverage (°C) Tmin (°C) Tmax (°C) Rainfall (mm) Rain days (>1mm) (mm) Average rainfall on rainy day (mm) Rain frequency (day-1) Annual potential evaporation (mm) Solar irradiation (Wh m-2 day-1) Temperate maritime Continental 2 Mediterranian 1 Tranebjerg (DK) Gyor (HU) Thessaloniki (GR) 55°50’ N 47°41’ N 40°38’ N 10°36’ E 17°38’ E 22°56’ E 11 115 32 13.5 16.8 19.6 8.2 10.9 14.3 16.5 22.6 25 49,4 67.2 43.6 8.6 7.9 5.3 5.7 8.5 8.3 3.6 3.9 5.9 1067 1466 1833 5040 5310 5580 Table S4: Results for the comparision of PEARL and PestLCI 2.0 Location Chateaudun Date of application 9 mar Application dose (kg/ha) 1 1 Annual emissions (kg/ha) 4.10E-06 2.40E-06 1.35E-05 2.47E-05 1.10E-04 8.97E-05 1.00E-04 5.60E-05 2.86E-05 3.23E-05 5.69E-05 7.51E-05 3.62E-05 5.22E-05 1.26E-05 0 0 0 0 1.20E-06 Average (kg/ha) 4.35E-05 Location Date of application Application dose (kg/ha) Annual emissions (kg/ha)1 Tours PEARL Hamburg Joikionen 31 mar 17 may 1 1 2.40E-04 1.12E-04 0 6.72E-05 3.16E-04 8.61E-05 1.08E-03 1.89E-04 1.30E-03 2.45E-04 5.96E-03 3.36E-04 1.24E-02 7.69E-04 1.34E-03 3.99E-04 1.07E-03 5.59E-04 9.55E-04 7.16E-04 7.89E-04 4.10E-04 5.99E-04 2.23E-04 1.31E-03 3.21E-05 2.47E-03 1.27E-05 2.16E-04 2.70E-06 0 7.80E-06 4.49E-04 4.80E-05 5.87E-04 2.25E-04 1.56E-03 3.25E-04 4.80E-03 1.68E-04 2.08E-03 2.47E-04 PestLCI 2.0 Tranebjerg Birzai Kremsmünster 31 mar 1 4.97E-04 6.93E-04 1.69E-03 4.60E-03 5.43E-03 1.71E-03 2.37E-03 2.17E-03 8.12E-04 5.19E-04 9.27E-04 1.27E-03 1.80E-03 1.48E-03 6.69E-04 0 0 2.14E-04 2.85E-04 4.16E-04 1.53E-03 Okehampton 31 mar 1 2.39E-03 3.48E-04 8.15E-04 1.40E-03 2.92E-03 3.41E-03 5.58E-03 2.96E-03 1.46E-03 1.82E-03 2.39E-03 2.48E-03 4.34E-03 6.14E-03 1.20E-03 8.29E-04 3.57E-03 5.40E-03 2.24E-03 2.65E-03 2.72E-03 Porto 9 mar 1 8.41E-05 1.01E-04 4.03E-03 8.74E-04 8.91E-04 5.65E-04 1.52E-04 2.64E-04 2.09E-04 7.67E-04 9.11E-04 3.52E-04 6.04E-04 1.35E-03 4.30E-05 9.47E-05 2.75E-04 2.16E-05 4.00E-07 1.65E-05 5.80E-04 Kremsmünster Gogerddan La Coruna March April May April April March 1 3.90E-03 1 2.10E-03 1 4.40E-03 1 6.10E-03 1 3.80E-03 1 6.10E-03 References Hiller E, Jurkovic L, Bartal M (2008) effect of temperature on the distribution of polycyclic aromatic hydrocarbons in soil and sediment. Soil Water Res 4:231-240 PPDB (2011) Pesticide properties database. http://sitem.herts.ac.uk/aeru/footprint/index2.htm. Accessed on 29 July, 2011 Schwarzenbach RP, Gschwend PM, Imboden DM (2003) Environmental organic chemistry. Second edition. John Wiley and Sons, Hoboken