Journal of Hydrology 497 (2013) 62–70
Contents lists available at SciVerse ScienceDirect
Journal of Hydrology journal homepage: www.elsevier.com/locate/jhydrol
Canopy enhanced chloride deposition in coastal South Australia and its application for the chloride mass balance method Zijuan Deng a,b,⇑, Stacey C. Priestley b, Huade Guan a,b, Andrew J. Love a,b, Craig T. Simmons a,b a b
Flinders Univ S Australia, Natl Ctr Groundwater Res & Training, Adelaide, SA 5001, Australia Flinders Univ S Australia, Sch Environm, Adelaide, SA 5001, Australia
a r t i c l e
i n f o
Article history: Received 28 June 2012 Received in revised form 29 April 2013 Accepted 24 May 2013 Available online 4 June 2013 This manuscript was handled by Laurent Charlet, Editor-in-Chief, with the assistance of M. Todd Walter, Associate Editor Keywords: Chloride deposition Chloride mass balance Throughfall Eucalyptus Pinus radiata South Australia
s u m m a r y Validity of the chloride mass balance (CMB) method relies on a good quantiﬁcation of the total chloride input to the land surface. The traditional way to quantify the chloride input, using bulk precipitation measured in the open ﬁeld is questionable to apply in forest catchments. Chloride deposition can be signiﬁcantly enhanced by the tree canopies as indicated by previous throughfall studies. However, this enhancement effect has not been examined or generally considered in the CMB method for groundwater recharge estimation and few chloride deposition data were available for the eucalyptus canopies. In this study, a throughfall experiment was carried out in the native eucalyptus forest and pine plantations in a coastal forest of South Australia. The results show that chloride depositions in canopy areas are signiﬁcantly higher than those in the adjacent open ﬁeld with 28% enhancement at the eucalyptus site and 89% at the pine site. These results indicate that a signiﬁcant underestimation of groundwater recharge can be introduced if this enhancement effect is neglected in the CMB application. A semi-empirical model is proposed to examine the controlling factors for the enhancement effect. The model will be useful to correct chloride input for the CMB application. Ó 2013 Elsevier B.V. All rights reserved.
1. Introduction The chloride mass balance (CMB) method has been widely used to estimate groundwater recharge, especially in arid and semi-arid areas (Allison et al., 1994; Scanlon et al., 2002, 2006). In Australia, more than half of the recharge estimation cases were conducted using the CMB method (Crosbie et al., 2010; Petheram et al., 2002). The method is based on a mass balance equation (see Eq. (1)), thus is applicable in conditions where the equilibrium between chloride input from atmospheric deposition and output to groundwater can be attained (Guan et al., 2010a; Wood, 1999). In practice, the chloride input is estimated from the bulk precipitation in the open ﬁeld, and the groundwater concentration is either derived from samples below the water table or inferred from the soil water below the root zone (Allison et al., 1994; Wood, 1999). The CMB method using chloride concentration of soil water in the unsaturated zone has been used to infer paleo-groundwater recharge and to evaluate the effect of land use change on groundwater recharge (Cook et al., 1992; Scanlon et al., 2003); while using
⇑ Corresponding author. Address: School of the Environment, Flinders University of South Australia, GPO Box 2100, Adelaide 5001, Australia. Tel.: +61 8 8201 2976. E-mail addresses: [email protected]
ﬂinders.edu.au (Z. Deng), [email protected]
ﬂinders.edu.au (S.C. Priestley), [email protected]
ﬂinders.edu.au (H. Guan), [email protected]
ﬂinders.edu.au (A.J. Love), [email protected]
ﬂinders.edu.au (C.T. Simmons). 0022-1694/$ - see front matter Ó 2013 Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.jhydrol.2013.05.040
groundwater chloride concentration is more suitable to quantify mountain block recharge (Guan et al., 2010a) and commonly applied in Australia due to its cost effectiveness (Crosbie et al., 2010). Proper quantiﬁcation of chloride input is important in the various applications of the CMB method. Here a simple one-dimensional chloride mass balance approach is used to demonstrate this signiﬁcance:
Cp P ¼ Cg R
where Cp (mg/L) is the chloride concentration of bulk precipitation integrating from both atmospheric wet and dry deposition, P is the average annual rainfall (mm/yr), Cg (mg/L) is the chloride concentration of groundwater or of the soil water below the root zone, R (mm/yr) is the mean annual recharge rate. The atmospheric deposition of chloride can vary both temporally (Scanlon, 2000; Crosbie et al., 2010) and spatially (Guan et al., 2010b; Keywood et al., 1997), thus proper methods or sampling designs are needed in order to account for the input variations. For example, 36Cl/Cl was employed to infer the long-term average chloride input rather than using several years’ chloride rain gauge measurement to avoid the inter-annual variation of the chloride input (Scanlon, 2000). Exponential regression ﬁtting and geo-statistical mapping were utilized to account for the spatial variation of the chloride input to continental Australia (Keywood et al., 1997) and coastal areas (Guan et al., 2010b), respectively. The uncertainty in the chloride
Z. Deng et al. / Journal of Hydrology 497 (2013) 62–70
input quantiﬁcation depends on the methods used and the sampling network density among other factors (Guan et al., 2010b; Scanlon, 2000). For example, an uncertainty of ±35% in the chloride input calculation was associated with the 36Cl/Cl method in the recharge estimation of Eagle Flat Basin (Scanlon, 2000). Vegetation canopy can also inﬂuence atmospheric chloride input (to be reviewed later). However, this factor has not been generally considered in the CMB calculations. It has been reported in various throughfall studies with multiple-year sampling that chloride deposition was much higher in a forest canopy than in the adjacent open ﬁelds regardless of vegetation species and climate types (Chang and Matzner, 2000; Lovett et al., 1996; Moreno et al., 2001; Neary and Gizyn, 1994; Staelens, 2006). The reported enhancement of chloride deposition in the canopy areas is typically around 50% but can be as high as 75% in a coastal area (see Fig. 1). This enhancement is acknowledged to be due to the dry deposition intercepted by forest canopies (Beier et al., 1993). Notes: all the enhancement magnitudes were calculated according to Eq. (3) neglecting the stemﬂow ﬂux. References are (from left to right) (Chang and Matzner, 2000; Neary and Gizyn, 1994; Moreno et al., 2001; Lovett et al., 1996; Staelens, 2006). Shown on top of the frame is the dominant tree species and ﬁgures shown above the column are either average long-term annual rainfall (labelled ‘‘1’’) or average rainfall during the sampling period (labelled ‘‘2’’). ‘‘a’’: there are four plots under canopies in a rainfall gradient from 526 mm to 1056 mm, the enhancement magnitude ranges from 110% to 50%, the column showing the smallest enhancement at plot with 1056 mm rainfall. Ulrich (1983) classiﬁed the atmospheric deposition processes of elements on a forest ecosystem into four categories: precipitation of rain or snow ‘‘particles’’, sedimentation of particles according to gravity, impaction of aerosols including fog and cloud droplets, and dissolution of gases on wet surfaces. The ﬁrst process is known as the wet deposition, which is not inﬂuenced by the surface characteristics, and the rest belongs to dry deposition, which can be inﬂuenced by surface characteristics. Thus, for inert elements of Na, Cl and S showing none or negligible exchange with leaves, the higher deposition rate to a canopy area compared to that of an open ﬁeld is attributed to dry deposition, which can be efﬁciently enhanced by canopies (Ulrich, 1983). For simplicity, the canopies’ higher efﬁciency for capturing the dry deposition is hereafter referred to as ‘‘enhancement effect’’. In sparsely vegetated areas the enhancement effect may not be signiﬁcant. However, in densely vegetated catchments, the total chloride input can be much higher than that measured in the open ﬁeld. For example, Fourcade et al. (2002) concluded after intensive mointoring and comparing chloride loads in stream water to those of bulk precipitation for 5 years that chloride input was underestimated by more than 100% in a small catchment in France. In addition, a 23-year long-term record in a reference watershed in
Hubbard Brook Experiment Forest, Northeast US, showed a consistently higher sulphate load in the stream water (the only discharge of the catchment) than the bulk precipitation, indicating a 37% underestimation of total sulphate input to the catchment by the bulk precipitation measurement in the open ﬁeld (Likens and Bormann, 1990). This underestimation can only be explained by the incapability of the open ﬁeld rain gauge to capture the dry deposition intercepted by the canopies (Fourcade et al., 2002). In Australia, most of groundwater recharge estimations using CMB method were conducted within 100 km to the coast (Crosbie et al., 2010) where the dry deposition portion is more signiﬁcant than the inland areas. Thus, the primary objective of this study is to examine the magnitude of enhancement effect of chloride in coastal areas. For this purpose a coastal site in South Australia including both native eucalyptus trees and the popular plantation trees Pinus radiata was selected to allow a comparison between species. There are few chloride deposition datasets available on eucalyptus trees, while pine trees have abundant records (to be reviewed later). A second objective is to establish a model to quantify the controlling factors on chloride deposition under the canopies. The model is aimed for chloride mass balance application by linking the enhancement effect to meteorological conditions and canopy characteristics. 2. Methodology 2.1. Site description The study was conducted in Kuitpo Forest (138.68 E, 35.18 S), 20 km east of Gulf St. Vincent and 30 km south of the central business district of Adelaide, South Australia. The climate of this area is semi-humid Mediterranean with 830 mm mean annual precipitation (1971–1998), from which around 60% occurs from June to October. Predominant wind directions are W, NW at 3 pm and E and SE at 9 am in long-term average (Kuitpo Forest HQ station, 2010). The forest consists of native eucalyptus trees, and commercial plantations most of which are pine trees. Three sites were selected including one bulk precipitation sampling site in the open ﬁeld (hereafter referred to as ‘‘Open site’’), and two throughfall sampling sites in the forest. The open site has no tree canopies within 100 m of the sampling device. The two throughfall sites include one sampling plot under native eucalyptus trees (hereafter referred to as ‘‘N site’’) and the other under pine trees (hereafter referred to as ‘‘P site’’) (see Fig. 2). The tree species at the N site are Eucalyptus camaldulensis, Eucalyptus leucoxylon and Eucalyptus baxteri with average diameter at breast height (1.3 m, DBH) of 18 cm. The eucalyptus trees are more than 100 years old (personal communication with the local forestry manager). The canopy height of the eucalyptus trees is averaged 13 m. The P site is located immediately west to the N site with only tree species of P. radiata. The pine trees have an almost uniform canopy height of 28 m and DBH of 45 cm. The plantations are 36 years old. 2.2. Field work and laboratory procedure
Fig. 1. Enhancement magnitude of chloride deposition in throughfall.
In each vegetation site, twenty collectors were placed randomly in a plot of 20 20 m2. The plots were set up at least 70 m away from the forest fringe to avoid the edge effect (Beier et al., 1992; Draaijers and Erisman, 1993). Each collector was made of 15 cm diameter funnel connected to a bottle of 1.5 L capacity. The funnel had coarse ﬁlters to stop leaf litter from falling into the bottle. The collector was held up in a plastic tube 55 cm above the ground to avoid dirt splashing into the funnels. In the open ﬁeld, four identical funnel collectors were placed to sample bulk precipitation. One automatic wet/dry collector sat amid the funnels to measure wet
Z. Deng et al. / Journal of Hydrology 497 (2013) 62–70
Fig. 2. Locations of study area and sampling site (base map is modiﬁed from Kuitpo Forest Reserve regional map of SA Forestry).
deposition and dry deposition separately. The wet/dry collector can record rainfall above 0.2 mm. To prevent water evaporation, each sampling bottle was ﬁlled with around 1 cm thick of liquid parafﬁn (Ajax Finechem Pty Ltd.) before being placed in the ﬁeld. Water samples were ﬁltered with 0.45 lm ﬁlter paper (GN-6 Metrice Grid, Pall Corporation) and kept in a cold room (4 °C) before the chemical analysis. Chloride content was measured using ion chromatography (DIONEX ICS-1500, MEP Instruments PTY Limited) with a measurement uncertainty of 1%. Sodium was measured with atomic absorption spectroscopy (GBC 933 plus, Scientiﬁc Equipment Pty. Ltd.) with a measurement uncertainty within 1%. To facilitate the comparison with the bulk precipitation in the open ﬁeld, the dry and wet depositions measured with the automatic collector were corrected by multiplying the ratio of the bulk precipitation over the sum of the dry and wet depositions. Leaf area index was obtained with the hemispherical photography method at an overcast day (Zhang et al., 2005). Each picture was taken vertically above each sampling point with a ﬁsh eye lens mounted to a Nikon D4000 at a stand height of 1.3 m. The pictures were then processed with Gap Light Analyser (GLA) (Frazer et al., 1997) and the leaf area index was calculated according to Chen et al. (1991) for each zenith ring. Similar work was done in Staelens et al. (2006).
can be rewritten in the form of Eqs. (3) and (4) to represent groundwater recharge in the open ﬁeld or the canopy areas respectively:
Do þ W o Cg
Dc þ W c Cg
where W (g/m2) and D (g/m2) are the average annual wet and dry deposition. Superscript ‘‘o’’ and ‘‘c’’ denotes ‘‘open ﬁeld’’ and ‘‘canopy’’ respectively. Cg (mg/L) is the chloride concentration of groundwater or of the soil water below the root zone. Wet deposition is assumed to be the same between the open ﬁeld and the canopy areas as not inﬂuenced by the surface characteristics. According to deﬁnitions, we have TF = Wc + Dc, BP = Wo + Do, and Wo = Wc. Thus with substitution of these into Eq. (2), the magnitude of enhancement can be represented as
TF BP 100% BP
where BP (g/m2) is the chloride deposition in bulk precipitation in the open ﬁeld, TF (g/m2) is the deposition in throughfall. Eq. (1)
Combining Eq. (3) through Eq. (5), we get Eq. (6) that directly reveals the possibility that signiﬁcant underestimation would occur if the bulk precipitation was used for recharge calculation in canopy areas.
2.3. Enhancement calculation The total deposition to a canopy was calculated with the throughfall ﬂux under a canopy, neglecting the contribution of stemﬂow in consideration that stemﬂow is a minor contributor to the total deposition for eucalyptus and pine canopies (Crockford et al., 1996b; Llorens and Domingo, 2007). In this case, the enhancement magnitude (E) can be quantiﬁed as:
Dc Do 100% BP
Rc Ro 100% Ro
2.4. Enhancement modelling Many studies have been devoted to quantifying deposition to forest canopies. In the forestry and ecology ﬁeld, linear regression method was commonly used to examine the inﬂuencing or controlling factors such as rainfall, dry antecedent period and leaf area index (Lovett et al., 1996; Puckett, 1990). However, the regression functions were purely empirical and difﬁcult to transfer between sites. In the atmospheric ﬁeld, the studies on dry deposition at can-
Z. Deng et al. / Journal of Hydrology 497 (2013) 62–70
opy areas focus on determining the deposition velocity as a function of the canopy surface characteristics (Draaijers et al., 1997; Slinn, 1977) which is usually complex. In this study, a simpliﬁed model is proposed to simulate the enhancement magnitude under the canopies. The enhancement magnitude is dependent on external micrometeorological conditions and the internal canopy characteristics (Knulst, 2004; Lovett et al., 1996; Staelens et al., 2006; Weathers et al., 2000). The throughfall and bulk precipitation are highly correlated in both the volume and chloride deposition over the sampling intervals in our sites (see Fig. 3). Thus, the method is ﬁrst to simulate chloride deposition in the open ﬁeld relating to the common meteorological conditions, then secondly, to estimate deposition under the canopies as a function of the canopy characteristics. The deposition in the open ﬁeld is modelled by adapting a model that was used to calculate sea salt sodium deposition in a Netherland coastal area (Ten Harkel, 1997). It is reasonable to assume that chloride in the rainfall predominantly originates from the sea water in our study area. In this model, the sea salt concentration is calculated with the empirical function following Erickson (1986). The proportion of chloride to the total sea salt in the air above the open sea is assumed to be the same as sea water (58% by mass). A ‘‘decay’’ factor (d < 1) denoting the ratio of chloride content in air at our study area to that of the open sea is added to the equations:
C air ¼ 0:58 d e0:16Uþ1:45
for U < 15 m=s
C air ¼ 0:58 d e0:13Uþ1:89
for U > 15 m=s
where Cair (lg/m3) is the chloride concentration in the air above the study area; U (m/s) is the wind speed at 15 m above the sea level, d is a dimensionless constant calculated by ﬁtting the modelled results to observed wet depositions. An open sea weather station overlooking the Indian Ocean (Neptune Island, 2010) was assumed
to be representative of the open sea condition, however, due to missing data of weather station, the nearest station to our study site (Kuitpo, 2010) is used instead and the wind speed was scaled to the open sea condition by comparing the existing data of the two stations. The 3 pm wind speed is used to characterise wind from the Gulf St. Vincent (Guan et al., 2010b). Dry deposition in the open ﬁeld is calculated according to Puckett (1990)
D ¼ C air V d L 108
where D (g/m2) is the total dry deposition during the sampling period; Vd (cm/s) is the deposition velocity of aerosols; L (s) is the length of the sampling interval, 108 is the unit conversion factor. Here, Vd is given the value of 0.43 cm/s representing the deposition velocity (Sehmel, 1980) of 2–5 lm diameter aerosols which comprise the majority of the chloride aerosols in 30 km from the coast (Ten Harkel, 1997). Wet deposition is calculated with an empirical function, established in a rain scavenging study in a coastal area of Virginia Key, in Miami, Florida, US (Savoie et al., 1987).
C air wr P 109
1240 P0:29 833
where W (g/m2) is the total wet deposition of all rainfall event during the sampling interval, wr is the washout ratio (dimensionless); P (mm) is the event rainfall amount, 109 is the unit conversion factor. The empirical function (Eq. (11)) can give unreasonably high chloride deposition results for heavy rain events; therefore, an upper limit of 10 mm is set for the precipitation amount in the calculation to best match the observed data. It is reasonable to treat the data in this way as the ending phase of rain contributes much lower chloride deposition compared to the early stage (Goncalves et al., 2000; Guan et al., 2010b). As mentioned in Section 1, the larger deposition in throughfall is due to the interception of dry deposition by canopies. Canopy characteristics including leaf area index, and tree height were observed to linearly correlate to the dry deposition (Erisman and Draaijers, 2003; Staelens et al., 2006). Therefore, the dry deposition onto canopies in our study area is depicted as
Dc ¼ Do ða LAI þ b H þ 1Þ
where a and b are the regression coefﬁcients determined by ﬁtting to the observed dry deposition data; H is the canopy height (m); other symbols are as deﬁned previously. The term a LAI + b H (>0) shows the efﬁciency of canopies enhancing the dry deposition compared to the open ﬁeld. Combing Eqs. (5) and (12) will give the enhancement magnitude in
Do ða LAI þ b HÞ 100% BP
Integrating Eqs. (9) and (10) into (13) will ﬁnally relate the enhancement magnitude to the meteorological and canopy structure factors:
V d L ða LAI þ b HÞ P 100% ðP wr Þ þ V d L
3. Results and discussion 3.1. Measurement results Fig. 3. Correlations between throughfall at canopy sites and rainfall in the open ﬁeld in water volume (upper) and chloride ﬂuxes (lower). Error bars show the one standard deviation.
Altogether, 16 batches of bulk precipitation and throughfall samples were collected from September 9, 2009 to September 6,
Z. Deng et al. / Journal of Hydrology 497 (2013) 62–70
2010. The total rainfall collected in the open ﬁeld was 888 mm. The throughfall collected under the eucalyptus canopies was averaged 804 mm, showing a 10% interception of the rainfall. Pine trees intercepted 25% of the rainfall with a throughfall volume of 668 mm. The total bulk precipitation of chloride was 6.2 g/m2 in the open ﬁeld, while, in canopy areas, the chloride depositions were signiﬁcantly higher. The throughfall showed a 28% enhancement of chloride deposition under the eucalyptus plot and 89% under the pine trees (see Table 1). On the assumption that sodium and chloride deposit at the same rate for marine aerosols (Beier et al., 1992), our results compare favourably with those of Crockford et al. (1996a). In their study, the enhancement of sodium deposition was 37% for the eucalyptus canopy and 118% for the pine trees (see Table 2). Baker and Attiwill (1987) also reported a range of 19–57% enhancement of sodium at different eucalyptus sites and 20–85% of enhancement at pine tree sites in a coastal area of Victoria, Australia. All three studies showed signiﬁcant enhancement of chloride or sodium under tree canopies, and consistently larger deposition rates under pine trees than eucalyptus trees. The higher enhancement of dry deposition on the pine tree canopy can be explained by its higher leaf area index (Stachurski and Zimka, 2000; Staelens et al., 2006) or larger canopy height (Erisman and Draaijers, 2003). Leaf area index can explain around 65% of the spatial variation of chloride deposition at the N site while only around 30% for the P site (see Fig. 4). This indicates that leaf area can be a major controlling factor on chloride deposition on the eucalyptus canopy; however, it may not be sufﬁcient to explain the spatial deposition pattern under the pine trees. The difference may be due to different canopy structure between the two kinds of trees. The native eucalyptus trees were randomly distributed with branches far extending and intersecting each other, while the pine trees were planted in lines and their branches gathered to form a dense umbrella-shape crown. However, the total deposition under the canopies shows good correlation with the total average leaf area index. Larger tree height also favours the deposition to canopies by increasing the stand-speciﬁc roughness length (Draaijers et al., 1997). Canopy height of pine trees is around
Table 1 Chloride deposition in the open ﬁeld and calculated enhancement magnitudes. No.
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 Sum
76 59 61 40 30 53 39 42 48 91 76 26 61 45 76 65 888
0.51 0.76 0.79 0.22 0.32 0.08 0.32 0.08 0.33 0.53 0.55 0.09 0.37 0.36 0.82 0.12 6.2 ± 0.1
0.51 0.72 / / / / / 0.11 0.33 0.53 0.55 0.12 0.45 / 0.90 0.13 /
0.08 0.09 / / / / / 0.11 0.09 0.07 0.06 0.03 0.03 / 0.04 0.03 /
25 29 34 61 64 71 23 77 23 30 5 21 16 39 17 58 28 ± 14⁄
47 53 67 253 270 331 124 91a 73 28 69 58 92 45 181 89 ± 18⁄⁄
Notes: No. refers to the sampling intervals. Rain is the accumulated rainfall during the sampling period. BP is bulk precipitation in the open ﬁeld; W and D are corrected wet and dry depositions measured with automatic wet/dry instrument in the open ﬁeld respectively; ‘‘/’’ no data was available due to the breakdown of the wet/ dry instrument. E% (N) and E% (P) display the enhancement magnitudes at N site and P site respectively. Superscript ‘a’: P site throughfall was sampled at different time from that of open site and N site thus was given total enhancement of interval 8 and 9; numbers after ‘‘±’’ are one standard deviation of the chloride deposition. ‘‘’’ Superscript shows that the median are signiﬁcantly different with Mann–Whitney test at <0.1 signiﬁcance level. ‘‘’’ Are signiﬁcantly different at <0.01 signiﬁcance level.
28 m, almost twice as high as the eucalyptus canopy. Leaching of chloride from coniferous leaves could also induce apparent enhancement (Puckett, 1990). But it is not the case in our study site according to the Na-reference method (Puckett, 1990). In addition, the Cl/Na ratios of the throughfall are not signiﬁcantly different from those of the bulk precipitation and sea water; neither shows an increasing trend with increasing leaf area index (see Fig. 5). It is worth to note that some sampling points at N site have a lower chloride deposition than that in the open ﬁeld (see Fig. 4) which is likely resulted from the shielding of rainfall by the tree canopies. Both shielded (for example under the dense canopy) and drip points (e.g. the edge of a canopy) can exist under a canopy (Robson et al., 1994). Sampling design is thus important to avoid biased measurement of the throughfall ﬂux (Levia and Frost, 2003). The sampling size required to limit relative error of event throughfall to a certain level depends on the rainfall event size, the forest type and the ion ﬂux to be measured (Levia and Frost, 2003; Puckett, 1991). Although, an unpractical high sample size is needed to restrict throughfall ﬂux to a certain accuracy (Zimmermann et al., 2010), the reader should bear in mind that the sampling requirements were intended for throughfall quantiﬁcation on an event or weekly sampling basis. For the calculation of annual chloride deposition, to reach within 10% relative error of 95% conﬁdence, a minimum of 20 collectors would sufﬁce (Houle et al., 1999). Therefore, a conclusion can be safely drawn from the observation of this study that the chloride depositions under eucalyptus and pine canopies are signiﬁcantly higher than that in the open ﬁeld. It can be further conﬁrmed from results of other throughfall studies on pine and eucalyptus canopies that have explored sampling methods with different sampling gauges and wide range of sampling sizes (see Table 2). 3.2. Modelling results By linking to meteorological conditions and canopy structures, the deposition model can help to identify or even quantify the controlling factors on the enhancement magnitude. The modelled wet depositions in the open ﬁeld match well with the observed results except for those intervals with very low depositions (see Fig. 6). Ten Harkel (1997) mentioned that low deposition periods were always overestimated, implying that the model may not be suitable for simulating large rainfall events. The overestimated sampling intervals all covered one to two events over 20 mm/day. The dry deposition model gives good approximation to the measurements (see Fig. 7). For the wet autumn and winter time (with 70% of the total rainfall in the study year), the sampling results showed 13% contribution of dry deposition to the bulk precipitation during the period (see Table 1), while the modelling, ﬁlling the gap of unmeasured periods, gave a 24% contribution of dry deposition. The modelled value is well comparable with the reported value of around 25% of chloride from dry deposition in the eastern Mediterranean basin in Turkey (Almomani et al., 1995). Coefﬁcients a and b (dimensionless, see Eq. (14)) were calibrated to match the observed enhancement magnitudes, resulting in 1.1 for a and 0.1 for b. The modelled results can explain more than 60% of the observed temporal variation (see Fig. 8). The term a LAI + b H as stated previously, demonstrates the efﬁciency of dry deposition enhanced by the canopies compared to the open ﬁeld (see Eq. (12)). The modelled value of a LAI + b H reveals that the dry deposition velocity under the canopies is 2.7 times and 6.9 times that of the open ﬁeld (0.43 cm/s), corresponding to dry deposition rates of 1.2 cm/s and 3 cm/s onto the eucalyptus and pine canopies respectively. These values are reasonable compared to direct measurement of dry deposition velocity. For example, dry deposition velocity of 1–2 cm/s was reported for aerosols in diameter (/) of 2–5 lm to a spruce forest canopy (sum-
Z. Deng et al. / Journal of Hydrology 497 (2013) 62–70 Table 2 Comparison between deposition rates under the Pinus radiata and eucalyptus canopies in different study areas. Sites
1. Kuitpo forest, South Australia 35°180 S, 138°410 E
20 Fixed funnels (/ 15 cm)
E. E. E. P.
camaldulensis leucoxylon baxteri radiata
2. Upper Yass Representative Basin, NSW, Australia 35°150 S, 149°200 E
2 Trough (22 cm 20 m) and 3 trough (22 cm 5 m)
E. E. E. E. P.
rossii macrorhyncha melliodora manngera radiata
3. Central Gippsland, Victoria, Australia 38°200 S, 146°150 E
1 Fixed funnel and 2 roving funnels (/ 11.5 cm)
E. E. E. P.
regnans oblique sieberi radiate
4. Basque County, Spain 43°110 N, 2°400 W 43°060 N, 2°390 W
12 Fixed funnels (/ 30 cm)
5. Basque County, Spain
5 Fixed funnels (/ 17 cm)
118 19–57 20–85
Notes: References are 1 this study, 2 (Crockford et al., 1996a), 3 (Baker and Attiwill, 1987), 4 (Gonzalez-Arias et al., 2006), and 5 (Gonzalez-Arias et al., 2000).
Fig. 4. Correlation between leaf area index and total enhancement magnitude for each sampling point.
Fig. 6. Modelled wet deposition (red squares) in the open ﬁeld compared to the measured bulk deposition at open site (green triangle) and wet deposition (purple diamonds). R2 shows the goodness of linear ﬁtting between the modelled and measured wet depositions and p is the signiﬁcance of the linear correlation. (For interpretation of the references to colour in this ﬁgure legend, the reader is referred to the web version of this article.)
Fig. 5. Cl/Na ratio of ﬁrst ﬁve sampling batches.
marized in Petroff et al. (2008)). In addition, their modelling results revealed that leaf area index was the major driving parameter in dry deposition to coniferous forest for aerosols smaller than / 2 lm and became much less important for aerosols over / 10 lm. Similarly, it was also demonstrated that, for aerosols larger than / 10 lm, only minor difference exists in deposition rates between forest and grassland (Zhang et al., 2001). Therefore, it is not surprising to see that negligible enhancement was observed for a pine tree plot within 1 km from the coast in Basque Country,
Fig. 7. Modelled (red squares) versus measured (green triangles) dry deposition at open site. (For interpretation of the references to colour in this ﬁgure legend, the reader is referred to the web version of this article.)
Spain (see Fig. 9) where the majority of the sea salt can exceed / 20 lm with a deposition velocity of over 3.5 cm/s in the open ﬁeld (Ten Harkel, 1997).
Z. Deng et al. / Journal of Hydrology 497 (2013) 62–70
methods (Levia and Frost, 2003) but also that the major controlling factors on enhancement magnitude may be site speciﬁc. 3.3. Enhancement magnitude in catchment scale
Fig. 8. Comparison magnitudes.
Fig. 9. Enhancement magnitudes versus the distances to the coast (see Table 2). Error bars were one standard deviation of throughfall deposition of chloride or sodium. Data points at 40 km are the averaged enhancement magnitude of the four plots in study of (Baker and Attiwill, 1987). Error bars at points of 80 km were one standard deviation of the sodium concentration in throughfall. The numbers beside the point correspond to the reference number in Table 2.
Dry deposition of sea salt is expected to decrease from coast to inland (Ten Harkel, 1997). At the same time, the differences between the deposition velocity to the forest canopy surface and the land surface in the open ﬁeld increase as the larger diameter aerosols preferentially precipitate along the path due to gravity. The former process results in a decreasing enhancement magnitude while the latter process leads to an increasing enhancement. Therefore, an increasing trend of enhancement magnitude with distance to coast should be expected until the two processes balance each other as can be seen from Fig. 9. However, the trend can only be viewed qualitatively not only because the different levels of uncertainties are associated with different sampling
Plot scale measurements of throughall may underestimate the total ﬂux to the whole stand by neglecting the edge effect (Beier and Gundersen, 1989; Erisman and Draaijers, 2003). For example, dry deposition of sea salt particles may be underestimated by 20–25% for the Netherland forests with more than 70% individual stands smaller than 1.5 ha (Erisman and Draaijers, 2003). In addition, absence of stemﬂow measurement will also add to the underestimation of the total chloride ﬂux as is the case in our study. However, to measure throughfall in a catchment scale is almost impossible, but can be inferred from catchment mass balance approach (Rustad et al., 1994). As a fully covered catchment, the reference watershed six in Hubbard Brook Experiment Forest provides a perfect example of the enhancement magnitude in a catchment scale with both bulk precipitation and stream ﬂow monitored since 1960s and even better, with throughfall ﬂux measured in growing seasons from 1989 to 1992. Since the stream ﬂow is the only discharge of the watershed and the tree biomass stops to accumulate since 1980s, the enrichment of chloride in stream ﬂow compared to the bulk precipitation should be majorly attributed to the dry deposition enhanced by canopies (Likens and Bormann, 1990). The chloride load in stream ﬂow is consistently higher than that of the bulk precipitation every year with an average enrichment of 54% from year 1980 to 2007 (data according to Likens, G.E., http://www.hubbardbrook.org/data/dataset.php?id=8, accessed May 2011), which is close to the upper range of 44–54% of chloride enhancement of the throughfall (Lovett et al., 1996). In their more recent paper (Lovett et al., 2005) summarising research on chlorine process in catchment over 36 years (1964–2000) in Hubbard Brook Experimental Forest, the authors believe that dry deposition can account for at least part of the extra chloride load in stream ﬂow compared to the bulk precipitation. For a partially covered catchment, the vegetation fraction Fi can be used to up-scale to the total chloride deposition as supported by the fact that dry deposition correlates linearly with the leaf area index. Therefore, the CMB calculation using the bulk precipitation input at the open site can be corrected by multiplying a factor of P (1 + Ei Fi). Suppose a catchment with 50% coverage of pine trees with the same 90% enhancement of chloride in this study, the correction factor should be 1.45. To give a real example, we compare the different recharge calculations of the CMB methods with other estimation methods for an experimental catchments in Japan (see Table 3) (Oda et al., 2009). The experimental catchments include a paired catchments with catchment A (0.8 ha in area) as the reference catchment and
Table 3 Comparison between recharge estimations with CMB methods with and without correction for enhancement effect and other methods (data after Oda et al., 2009). Method
Measurement Measurement Water balance CMB(BP)
758 959 1330–1430
730 1112 /
Measured discharge by weir for year 1994–1998 Measured discharge by weir for year 1999–2003 2230 mm of mean precipitation (1994–2003), 800–900 mm evapotranspiration
Measured bulk precipitation 66 ueq/l for 1998–2006, stream water concentration is 300 and 103 ueq/l for A and B in year 2006 Mean throughfall concentration 173 ueq/l for 1998–2006, combined stream water concentration is 190 ueq/l for two catchments for 2006
Notes: Water balance method give upper limit of recharge by neglecting the surface runoff. CMB(BP) display the recharge estimation without correction for the enhancement effect, that is only use the bulk precipitation as chloride input. CMB (E%) estimating recharge with chloride throughfall input.
Z. Deng et al. / Journal of Hydrology 497 (2013) 62–70
catchment B (1.1 ha) which is 100% cleared in 1998. The discharge of the catchments can be used to approximate the recharge with quick ﬂow as a minor contribution (inferred from stream water chemistry (Oda et al., 2009)). The CMB method with correction of the enhancement effect (see CMB(E%) in Table 3) estimates recharge close to the measured and the calculated results with the water balance method, while without considering the enhancement effect, the CMB calculations (see CMB(BP)) underestimate up to 100% of the recharge for the reference catchment A. If taking catchments A and B as a combined catchment with vegetation coverage of 42% (0.8/(0.8 + 1.1), the CMB(E%) still gives a good approximation to the recharge inferred from the measured results. It is meaningful to recall Eq. (14) quantifying the inﬂuencing factors on the enhancement magnitude. The equation suggests that larger leaf area index, taller canopy height and less annual rainfall all inﬂuence the enhancement magnitude positively. It implies that in Australia, forest catchments in southern South Australia and south-western West Australia with good canopy coverage and the Mediterranean-type climate are most likely to have problems of underestimation of groundwater recharge from the CMB calculation with bulk precipitation chloride deposition of the open area. 4. Conclusions In this study, chloride ﬂuxes under pine and eucalyptus canopies were compared with those in the adjacent open ﬁeld. The throughfall measurement results showed signiﬁcantly higher chloride deposition than the bulk precipitation with 28% enhancement at the eucalyptus tree plot and 89% enhancement at the pine tree site. The much higher enhancement magnitude of the pine tree canopy than the native eucalyptus canopy was attributed to the larger leaf area index and taller canopy height of the pine trees. The results indicate that signiﬁcant underestimation of groundwater recharge with the CMB method will be introduced if the canopy enhancement effect is not considered in a forest catchment. In addition, a model has been successfully utilized to account for both the temporal and spatial variations of the throughfall deposition by taking into account the meteorological conditions and canopy characteristics. The model highlights the situations in which a catchment can encounter signiﬁcant underestimation in groundwater recharge estimation if the open-site chloride deposition is used, such as high coverage of vegetation, large leaf area index, tall canopy height and low annual rainfall. Therefore, recharge studies conducted under these combined conditions should deal with the chloride input issue with caution especially for those areas experiencing vegetation alterations. The study suggests that canopies can complicate the evaluation of the chloride mass balance method by their enhancement effect. Acknowledgements This work was supported by National Centre for Groundwater Research and Training, the Department of Water, Land and Biodiversity Conservation (South Australia) as a part of the National Water Initiative project ‘‘Groundwater Flow Mechanisms in Fractured Rock Aquifers of the Mount Lofty Ranges and Kangaroo Island’’. The ﬁrst author is partly supported by China Scholarship Council. Special thanks to Jeff Drechsler from Forestry SA, who has facilitated the ﬁeld work. Thanks are due to VinodKumar and Meghan Daily who have assisted the ﬁeld work; Hugo A. Gutierrez-Jurado, Yulong Zhu and Yuting Yang who have provided valuable suggestions to this manuscript.
References Allison, G.B., Gee, G.W., Tyler, S.W., 1994. Vadose-zone techniques for estimating groundwater recharge in arid and semiarid regions. Soil Sci. Soc. Am. J. 58 (1), 6–14. Almomani, I.F. et al., 1995. Major ion composition of wet and dry deposition in the eastern Mediterranean basin. Sci. Total Environ. 164 (1), 75–85. Baker, T., Attiwill, P., 1987. Fluxes of elements in rain passing through forest canopies in southeastern Australia. Biogeochemistry 4 (1), 27–39. Beier, C., Gundersen, P., 1989. Atmospheric deposition to the edge of a spruce forest in Denmark. Environ. Pollut. 60 (3–4), 257–271. Beier, C., Gundersen, P., Rasmussen, L., 1992. A new method for estimation of dry deposition of particles based on throughfall measurements in a forest edge. Atmos. Environ. A – Gen. 26 (9), 1553–1559. Beier, C., Hansen, K., Gundersen, P., 1993. Spatial variability of throughfall ﬂuxes in a spruce forest. Environ. Pollut. 81 (3), 257–267. Chang, S.C., Matzner, E., 2000. The effect of beech stemﬂow on spatial patterns of soil solution chemistry and seepage ﬂuxes in a mixed beech/oak stand. Hydrol. Process. 14 (1), 135–144. Chen, J.M., Black, T.A., Adams, R.S., 1991. Evaluation of hemispherical photography for determining plant-area index and geometry of a forest stand. Agric. Forest Meteorol. 56 (1–2), 129–143. Cook, P.G., Edmunds, W.M., Gaye, C.B., 1992. Estimating paleorecharge and paleoclimate from unsaturated zone proﬁles. Water Resour. Res. 28 (10), 2721–2731. Crockford, R.H., Richardson, D.P., Sageman, R., 1996a. Chemistry of rainfall, throughfall and stemﬂow in a eucalypt forest and a pine plantation in southeastern Australia. 2. Throughfall. Hydrol. Process. 10 (1), 13–24. Crockford, R.H., Richardson, D.P., Sageman, R., 1996b. Chemistry of rainfall, throughfall and stemﬂow in a eucalypt forest and a pine plantation in southeastern Australia. 3. Stemﬂow and total inputs. Hydrol. Process. 10 (1), 25–42. Crosbie, R.S., Jolly, I.D., Leaney, F.W., Petheram, C., 2010. Can the dataset of ﬁeld based recharge estimates in Australia be used to predict recharge in data-poor areas? Hydrol. Earth Syst. Sci. 14 (10), 2023–2038. Draaijers, G.P.J., Erisman, J.W., 1993. Atmospheric sulfur deposition to forest stands – throughfall estimates compared to estimates from inference. Atmos. Environ. A – Gen. 27 (1), 43–55. Draaijers, G.P.J., VanLeeuwen, E.P., DeJong, P.G.H., Erisman, J.W., 1997. Base cation deposition in Europe. 1. Model description, results and uncertainties. Atmos. Environ. 31 (24), 4139–4157. Erickson, D.J., 1986. Seasonal estimates of global atmospheric sea-salt distributions. J. Geophys. Res. 91, 1067–1072. Erisman, J.W., Draaijers, G., 2003. Deposition to forests in Europe: most important factors inﬂuencing dry deposition and models used for generalisation. Environ. Pollut. 124 (3), 379–388. Fourcade, B., Coudrain-Ribstein, A., Martin, C., 2002. What can be deduced from chemical measurement in an open-ﬁeld raingauge? An example in the Maures Massif, southeastern France. Hydrol. Sci. J. – J. Des Sci. Hydrol. 47 (3), 469–478. Frazer, G.W., Trofymow, J.A., Lertzman, K.P., 1997. A method for estimating canopy openness, effective leaf area index, and photosynthetically active photon ﬂux density using hemispherical photography and computerized image analysis techniques. Information Report BC-X-373, Victoria, BC. Goncalves, F.L.T. et al., 2000. Modelling and measurements of below cloud scavenging processes in the highly industrialised region of Cubatao–Brazil. Atmos. Environ. 34 (24), 4113–4120. Gonzalez-Arias, A., Amezaga, I., Echeandia, A., Onaindia, M., 2000. Buffering capacity through cation leaching of Pinus radiata D. Don canopy. Plant Ecol. 149 (1), 23– 42. Gonzalez-Arias, A., de Arano, I., Barcena-Ruiz, M., Besga, G., Onaindia, M., 2006. Origin of atmospheric deposition and canopy buffering capacity in stands of radiata pine and pedunculate oak in the Basque Country. Forest Ecol. Manage. 229 (1–3), 268–284. Guan, H., Love, A.J., Simmons, C.T., Ding, Z., Hutson, J., 2010a. Catchment conceptualisation for examining applicability of chloride mass balance method in an area of historical forest clearance. Hydrol. Earth Syst. Sci. 14 (7), 1233–1245. Guan, H., Love, A.J., Simmons, C.T., Makhnin, O., Kayaalp, A.S., 2010b. Factors inﬂuencing chloride deposition in a coastal hilly area and application to chloride deposition mapping. Hydrol. Earth Syst. Sci. 14 (5), 801–813. Houle, D., Ouimet, R., Paquin, R., Laﬂamme, J.G., 1999. Determination of sample size for estimating ion throughfall deposition under a mixed hardwood forest at the Lake Clair Watershed (Duchesnay, Quebec). Can. J. Forest Res. – Revue Canadienne De Recherche Forestiere 29 (12), 1935–1943. Keywood, M.D., Chivas, A.R., Fiﬁeld, L.K., Cresswell, R.G., Ayers, G.P., 1997. The accession of chloride to the western half of the Australian continent. Aust. J. Soil Res. 35 (5), 1177–1189. Knulst, J., 2004. Ratio between throughfall and open-ﬁeld bulk precipitation used for quality control in deposition monitoring. Atmos. Environ. 38 (29), 4869– 4878. Kuitpo, 2010. South Australia, Daily Weather Observations. (accessed December 2010). Kuitpo Forest HQ station, 2010. Climate Statistics for Australian Locations. (accessed December 2010).
Z. Deng et al. / Journal of Hydrology 497 (2013) 62–70
Levia, D.F., Frost, E.E., 2003. A review and evaluation of stemﬂow literature in the hydrologic and biogeochemical cycles of forested and agricultural ecosystems. J. Hydrol. 274 (1–4), 1–29. Likens, G.D., 2011. Chemistry of Streamwater at HUBBARD BROOK EXPERIMENT FOREST WS-6. (accessed May 2011). Likens, G.E., Bormann, F.H., 1990. Dry deposition of sulfur: A 23-yr record for the Hubbard Brook Forest ecosystem. Tellus 71 (B), 319–329. Llorens, P., Domingo, F., 2007. Rainfall partitioning by vegetation under Mediterranean conditions. A review of studies in Europe. J. Hydrol. 335 (1–2), 37–54. Lovett, G.M., Nolan, S.S., Driscoll, C.T., Fahey, T.J., 1996. Factors regulating throughfall ﬂux in a new New-Hampshire forested landscape. Can. J. Forest Res. 26 (12), 2134–2144. Lovett, G., Likens, G., Buso, D., Driscoll, C., Bailey, S., 2005. The biogeochemistry of chlorine at Hubbard Brook, New Hampshire, USA. Biogeochemistry 72 (2), 191– 232. Moreno, G., Gallardo, J.F., Bussotti, F., 2001. Canopy modiﬁcation of atmospheric deposition in oligotrophic Quercus pyrenaica forests of an unpolluted region (central-western Spain). Forest Ecol. Manage. 149 (1–3), 47–60. Neary, A.J., Gizyn, W.I., 1994. Throughfall and stemﬂow chemistry under deciduous and coniferous forest canopies in south-central Ontario. Can. J. Forest Res. 24 (6), 1089–1100. Neptune Island, South Australia, 2010. Daily Weather Observations. (accessed December 2010). Oda, T., Asano, Y., Suzuki, M., 2009. Transit time evaluation using a chloride concentration input step shift after forest cutting in a Japanese headwater catchment. Hydrol. Process. 23 (19), 2705–2713. Petheram, C., Walker, G., Grayson, R., Thierfelder, T., Zhang, L., 2002. Towards a framework for predicting impacts of land-use on recharge: 1. A review of recharge studies in Australia. Aust. J. Soil Res. 40 (3), 397–417. Petroff, A., Mailliat, A., Amielh, M., Anselmet, F., 2008. Aerosol dry deposition on vegetative canopies: Part II. A new modelling approach and applications. Atmos. Environ. 42 (16), 3654–3683. Puckett, L.J., 1990. Estimates of ion sources in deciduous and coniferous throughfall. Atmos. Environ. A – Gen. 24 (3), 545–555. Puckett, L.J., 1991. Spatial variability and collector requirements for sampling throughfall volume and chemistry under a mixed-harwood canopy. Can. J. Forest Res. – Revue Canadienne De Recherche Forestiere 21 (11), 1581–1588. Robson, A.J., Neal, C., Ryland, G.P., Harrow, M., 1994. Spatial variations in throughfall chemistry at the small plot scale. J. Hydrol. 158 (1–2), 107–122. Rustad, L.E., Kahl, J.S., Norton, S.A., Fernandez, I.J., 1994. Underestimation of dry deposition by throughfall in mixed northern hardwood forests. J. Hydrol. 162 (3–4), 319–336.
Savoie, D.L., Prospero, J.M., Nees, R.T., 1987. Washout ratios of nitrate, non-sea-salt sulfate and sea-salt on Virginia Key, Florida and on American-Samoa. Atmos. Environ. 21 (1), 103–112. Scanlon, B.R., 2000. Uncertainties in estimating water ﬂuxes and residence times using environmental tracers in an arid unsaturated zone. Water Resour. Res. 36 (2), 395–409. Scanlon, B.R., Healy, R.W., Cook, P.G., 2002. Choosing appropriate techniques for quantifying groundwater recharge. Hydrogeol. J. 10 (1), 18–39. Scanlon, B.R., Keese, K., Reedy, R.C., Simunek, J., Andraski, B.J., 2003. Variations in ﬂow and transport in thick desert vadose zones in response to paleoclimatic forcing (0–90 kyr): ﬁeld measurements, modeling, and uncertainties. Water Resour. Res. 39 (7), 1179. Scanlon, B.R. et al., 2006. Global synthesis of groundwater recharge in semiarid and arid regions. Hydrol. Process. 20 (15), 3335–3370. Sehmel, 1980. Particle and gas deposition: a review. Atmos. Environ. 14, 983–1011. Slinn, W.G.N., 1977. Some approximations for wet and dry removal of particles and gases from atmosphere. Water Air Soil Pollut. 7 (4), 513–543. Stachurski, A., Zimka, J.R., 2000. Atmospheric input of elements to forest ecosystems: a method of estimation using artiﬁcial foliage placed above rain collectors. Environ. Pollut. 110 (2), 345–356. Staelens, J., 2006. Spatio-Temporal Patterns of Throughfall Water and Ion Deposition under a Dominant Beech Tree (Fagus sylvatica L.) in Relationship to Canopy Structure, Ghent University, Belgium. Staelens, J., De Schrijver, A., Verheyen, K., Verhoest, N.E.C., 2006. Spatial variability and temporal stability of throughfall water under a dominant beech (Fagus sylvatica L.) tree in relationship to canopy cover. J. Hydrol. 330 (3–4), 651–662. Ten Harkel, M.J., 1997. The effects of particle-size distribution and chloride depletion of sea-salt aerosols on estimating atmospheric deposition at a coastal site. Atmos. Environ. 31 (3), 417–427. Ulrich, B., 1983. Interaction of forest canopies with atmospheric constituents: SO2, alkali and earth alkali cations and chloride. In: Effects of Accumulation of Air Pollutants in Forest Ecosystems. D. Reidel Publishing Company. Weathers, K., Lovett, G., Likens, G., Lathrop, R., 2000. The effect of landscape features on deposition to Hunter Mountain, Catskill Mountains, New York. Ecol. Appl. 10 (2), 528–540. Wood, W.W., 1999. Use and misuse of the chloride-mass balance method in estimating ground water recharge. Ground Water 37 (1), 2–3. Zhang, L.M., Gong, S.L., Padro, J., Barrie, L., 2001. A size-segregated particle dry deposition scheme for an atmospheric aerosol module. Atmos. Environ. 35 (3), 549–560. Zhang, Y.Q., Chen, J.M., Miller, J.R., 2005. Determining digital hemispherical photograph exposure for leaf area index estimation. Agric. Forest Meteorol. 133 (1–4), 166–181. Zimmermann, B., Zimmermann, A., Lark, R.M., Elsenbeer, H., 2010. Sampling procedures for throughfall monitoring: a simulation study. Water Resour. Res., 46.