Reconstruction of water acidification in a forested catchment, western Harz Mountains, Germany

0883-2927/93 $6.00 + .00

Applied GeochmWtry, Suppl. Issue No.2, pp. 131-134, 1993 Printed in Great Britain

© 1992 Pergamon PressLtd

Reconstruction of water acidification in a forested catchment, western Harz Mountains, Germany ANKE LUKEWILLE

Federal Environmental Agency , Bismarckplatz 1, W-l000 Berlin 33, Germany

and JURGEN PRENZEL

Institute of Soil Science and Forest Nutrition, Biisgenweg 2, W-3400 Gottingen, Germany Abstract-A dynamic model based on the precipitation of aluminium hydroxosulfate minerals was used to reconstruct a hypothetical history of catchment acidification . The effects of 140a of increased atmospheric S deposition into the Lange Bramke valley, western Harz Mountains, was simulated. The model includes the following processes : complexation of several hydroxo- and sulfato-compounds; ion exchange based on selectivity constants for all uncomplexed cations; carbonic acid equilibria reactions system assuming a given partial pressure of C~ ; and the precipitation/dissolution equilibria for gibbsite (AI(OHh), alunite (KA h (OHM S0 4h ) and jurbanite (AlOHS04). The adequacy of the model was demonstrated by applying it to 4 a of deposition, soil water and streamwater chemistry data from the Lange Bramke area. The model indicated that the measured S retention in the catchment can be explained by precipitation of alunite and jurbanite.

but of changes in total component concentrations linked with input and output functions. Spatial resolution can only be achieved for one-dimensional transport. The hydrological approach is very simple: a constant water flux is assumed through all three compartments. The adequacy of models based on BEM was initially demonstrated by calibrating and applying it to batch experi ments with soil samples in the laboratory (PllENZEL, 1982). In the study reported in this paper the results of another calibration method are described, namely: reconstruction of past soil solution/streamwater chemistry for an entire catchment.

INTRODUCTION

ONE OF THE consequences of anthropogenically caused acid deposition is the accumulation of S in soil and bedrock (REUSS and JOHNSON , 1986; KHANNA et al., 1987). The nature of the basic mechanisms has been the subject of much debate (MAY and NORDSTROM, 1991; LUKEWILLE and VAN BREEMEN, 1992). In most watershed acidification models , S retention processes are based on S04 adsorption isotherms (REuss et al., 1986; CoSBY et al., 1986). This paper introduces an alternative approach: the Lange Bramke Model includes the precipitation of aluminium hydroxosulfate minerals which may control the concentrations of Al and S04 in the solution.

DESCRIPTION OF THE CATCHMENT

CONCEPTUAL BASIS OF THE LANGE BRAMKE MODEL The Lange Bramke Model (LB Model) is based on the computer program BEM (Batch Equilibrium Model ; FRENZEL, 1991). BEM was designed to create different models for the calculation of coupled chemical equilibria in aqueous solutions. The user can choose among several different components (ions, complexes, minerals) depending on the purpose. The LB Model includes the following processes: the complexation of several hydroxo- and sulfato -compounds ; ion exchange based on selectivity constants for all uncomplexed cations; carbonic acid equilibria reactions system assuming a given partial pressure of CO2 ; the precipitation/ dissolution equilibria for gibbsite (A1(OHh), alunite (KAl~(OHMS04h) and jurbanite (AIOHS04). The precipitation of these minerals is controlled by mass balances of all components, including AI and S04' The equilibrium constants, i.e. the solubility products and cation exchange selectivity coefficients, are treated as model parameters. The dynamics of the LB Model are not a function of time

Lange Bramke (LB) is a stream draining a 76 ha catchment in the western Harz Mountains, Germany. Precipitation averages 1300 mm1a. In 1984 the area was covered with 37 a old Norway spruce (Picea abies Karst.) The soils in the LB valley are underlain by almost impermeable Devonian sandstone (2-4 m thick). Thus , loss of percolating water to deep groundwater can be neglected. The soils are spododystic cambisols, well developed to 80 cm depth . Lange Bramke is an intensively-studied catchment. The area was clearcut in 1947 and replanted with spruce. Since then the precipitation and runoff rates have been continuously measured. Even the treeless soils showed only minor overland flow (DELFS et al., 1958). Thus it is reasonable to assume that infiltration capacity of the forested soils exceeds rainfall intensities (HAUHS, 1986, 1989). Data used to calibrate the LB Model were based on concentrations and quantified ion fluxes in seepage water, runoff and deposition at the LB catchment. The quantifications were derived from measurements at up to five monitoring plots in conjunction

131

A. Liikewille and J. Prenzel

132

Table 1. LB catchment, mean concentrations (1981-1984) of major elements and compounds in soil solution and streamwater compared with model outputs for 1984 Layer* Model output

pH Alkalinity (.umolJl) HzO+ (m1a) Na (.umolll) K (.umolll) Ca (.umolll) Mg (.umolll) Al (.umolll) Cl (.umolll) N03 (.umolll) S04 (.umolll) Base saturation ('Yo)

Stream Measured means

Model output

Measured means

1844

1984

1981-1984

1844

1984

1981-1984

5.0 +4

4.2 -282

4.2 -289

6.7 +45

6.2 +22

6.1 +43

0.69

0.69

0.69

0.69

0.69

0.69

72

77

76

72

85

75

7

15

14

8

19

14

10

29

28

22

88

88

7

21

19

15

67

69

0.2

62

77*

90

90

88

1

28

29

9

178

178

15.8

1.3

0.0

90

9

0.03

<0.7*

90

87

26

33

138

120

3.0§

*Lysimeter, 80 em depth (plots 1, 2, 3); tWater fluxes were estimated by using results from a model of saturated/unsaturated water flow (HAuHs, 1986); *Measured values: total Al (AAS analyses); model output: AI3+ concentrations; §Cation exchange capacity was estimated by N~Cl extraction.

with a physical model of catchment hydrology (HAUHS, 1986, 1989). The plots were established in 1977. They were arranged along a cross-section of the watershed (southern slope to northern slope). Wet deposition, throughfall, soil solution at 80 em depth, spring water and runoff have been sampled for chemical analyses weekly or bi-monthly (soil solution). Physical data such as air temperature, humidity, global radiation, pF curves, soil water potential and unsaturated hydraulic conductivities were available to calibrate the finite element hydrological model described in detail by HAUHS (1986). The model output indicates that flowis unsaturated at the hillslopes, and that seepage water passing the root zone percolates vertically. Table 1 summarizes mean concentrations (19811984) of major elements and compounds measured in soil solution and streamwater. These values were used as fixedpoints to calibrate the LB Model. Mean seepage water flow and runoff (1981-1984) were derived by using results from the hydrological model of HAUHS (1986).

RECONSTRUCTION OF PAST SOW STREAMWATER CHEMISTRY

The key processes incorporated in the LB Model were tested in an attempt to reconstruct the depo-

sition and acidificationhistory of the catchment. The model output in 1984 agrees with the measured annual averages in the LB-valley (Table 1). The main assumptions of the model were that the soil and subsoil were chemically homogeneous, that complex flow patterns such as macropore flow were negligible, and that contact between soil and aqueous solution was sufficient so that the reactions included in the model proceeded to equilibrium. Some reactions are "slow" and others are "quick" relative to the chosen time scale of years. The silicate weathering rate, for example, is treated as a zero order reaction and thus as an input flux. A three-compartment representation of the catchment was chosen: (1) 0-&0 em layer (soils), (2) 80250 em layer (bedrock), (3) stream (Lange Bramke). The model was run using average hydrological conditions, i.e, mean seepage water flow and runoff of 0.69 m/a (HAUHS, 1989). To obtain the model input for the first layer, atmospheric deposition and net biological uptake/release fluxes were estimated as inputs. An additional constant input of base cations through silicate weathering was included (rate = 900 molJha/a per meter depth; Ca and K). For the CO 2 partial pressure a mean annual value of 0.81 kPa was used (first and second layers). A PC02 of 0.10 kPa was chosen for the streamwater. The cation exchange capacity (CEC) for the first layer was set at 66 mmol.,lkg, and for the second at

Water acidification, Harz Mountains, Germany.

23 mmolzkg. Information on the cation exchange capacity was obtained by averaging measurements taken from 0-80 cm or 80-250 em depth at several locations throughout the catchment (HAUHs, 1989). The solubility constants of AI(OH)3 are in the range of possible p K; and measured pKiap values reported in the literature (iap = ion activity product). Different pK. values considered for the soil and bedrock layers imply a greater availability of soluble Al in the bedrock. This assumption may reflect a depletion of reactive amorphous Al(OH)3 down to 80 em depth. The constants for alunite (PK. = 84.3) and jurbanite (PK. = 17.95) were assumed to be identical in both layers.

HISTORICAL DEPOSmON SEQUENCE

A 140a deposition history for S, N and base cations from 1844 to 1984 was estimated. This long-term background sequence of input changes in the study area was based on historic emission records and deposition estimates (OECDIEMEP data; ALCAMO et al., 1990). During the 19th and early 20th centuries ore smelting was a very important regional source of emissions at the Harz Mountains. Data on heavy metal accumulation in non-acidified river and lake sediments (Srsvsss and ROOSTAl , 1990; MATSCHULLAT et al. , 1992)were used to help estimate additional S loads. It was assumed that S emissions developed in parallel to those of heavy metals ; and that S input followed these changes considered together with the slightly increasing background deposition levels. The local ore mines closed down between 1935and 1950. Therefore it was assumed that until 1935 the input of S04-S had increased to 42 kg/ha/a and then decreased linearly to the background load until 1950. Between 1950 and 1970the input again increased up to 45 kg/he/a. It remained constant until 1984. As mentioned above, for the period 1981-1984 average deposition fluxes at the LB catchment were estimated by using throughfall measurements in conjunction with a hydrological model. To achieve steady-state conditions at the beginning of the simulation , the model was run for -50 a with a constant background load of 2 kg S04-Slhala.

7.0 6.5 6.0

pH 5.5

...... ••.••••••

- ... - - -. -

133 - - 1. Layer - - 2. Layer ••••• Stream

.

.

- - - - - - - .... ,. __

5.0

4.5 4.0

~...--r...--r....,.......,.......,...

......,."T'T

-

1844 1864 1884 1904 1924 1944 1964 1984

Year FIG.

1. The pH of soil water (first and second layers) and

streamw ater in response to the historical deposition sequence.

reduction of percentage base saturation indicates a leaching of base cations from the soil/bedrock exchange sites (Table 1). The simulated changes in alkalinity are identical in the second layer and in the stream. The reconstructed trajectory of the Al concentration (first layer) is very similar to the alkalinity curve. Aluminium levels in the second layer and in the stream are <0.7,umol/l (Fig. 2). The shapes of the S04 concentration curves show clearly that S is accumulated in the catchment. In all three compartments the S04 levels are almost equal until 1908, following the input sequence assumed for S (Fig. 3). Sulfate is mainly retained as alunite . Precipitation in the bedrock layer starts in 1908, and in the soil layer in 1965. The rate and amount of precipitant formation are influenced by the availability of soluble Al(OHh and the weathering rate of K. Jurbanite builds up in small amounts in the first compartment (up to 125,umol/lof aqueous solution). Precipitation starts in 1974 at low pH levels (pH 4.2) and Al values of -60 ,umol/l. Although there is no change in S deposition rates between 1970and 1984, S04 concentrations in all three compartments slightly decrease (Fig. 3) whereas Al levels increase (Fig. 2). The S input to the Lange Bramke valley decreased between 1984 and 1990. It will be interesting to test the reaction of the model to these changes in input fluxes and to compare the model output with those of a model based on an adsorption isotherm for S04. 70 60

RESULTS

The pH values show a decreasing trend in all three compartments. If the percolating water is removed from contact with the bedrock matrix (second layer), e.g, when it discharges into the stream, the solution degasses causing an increase in streamwater pH (Fig. 1). Rising SO4levels in the solution lead to an increase in base cation concentration followed by a continuous reduction in alkalinity in all compartments. The

-~.=!:

;(

50

- - 1 . Layer - - 2. Layer • •••• Stream

40 30 20 10

o ~"''''''''-;=:::;::''---'~''''--'''-I''"--r--t'''''_ 1844 1864 1884 1904 1924 1944 1964 1984

Year FIG. 2. The Al concentration (umolll) in response to the historical deposition sequence.

A. Liikewilleand J. Prenzel

134 200

S0

150

E

~ 100

0


CJ)

50

1. Layer - - 2. Layer • •••• Stream

0"'T-...........-""T"'''"T"''''"T"''..........,.....,........,........,....,...T""T""r''""~ 1844 1864 1884 1904 1924 1944 1964 1984

Year FIG. 3. The S04 concentration Cumolll) in response to the historical deposition sequence.

CONCLUSIONS

The key processes included in the Lange Bramke Model are: input fluxes of base cations (especially through silicate weathering), cation exchange processes, the dissolution of AI-hydroxide, the precipitation of AI-hydroxysulfates and the dissociation and degassing of CO 2 • These are sufficient to reconstruct a hypothetical acidification history in the Lange Bramke catchment on the chosen time scale. The measured S retention in the catchment can be explained by the availability of reactive AI-hydroxide in the bedrock layer resulting in the precipitation of alunite and jurbanite. The precipitation of alunite implies a high input of K through silicate weathering. Acknowledgements-We are grateful to Michael Hauhs for the permission to use the Lange Bramke data . We also gratefullyacknowledgethe editorial handlingby Ron Fuge. Editorialhandling: Ron Fuge.

REFERENCES ALCAMO J., SHAW R. and HORDUK L. (eds) (1990) The

RAINS Model of Acidification: Science and Strategies in Europe. Kluwer Academic Publishers. CosBY B. J., HORNBBRGBR G. M., WRIGHT R. F. and GALLOWAY J. N. (1986) Modeling the effects of acid deposition: control of long-term sulfate dynamics by soil sulfate adsorption. Water Resour. Res. 22, 1283-1291. DELFS J., FRIEDRICH W., KISEKAMP H. and WAGENHOFF A. (1958)Der EinftuB des Waldes und des Kahlschlagesauf den AbfluBvorgang, den Wasserhaushalt und den Bodenabtrag. AIlSdem Walde 3, 1-212. HAUHS M. (1986) A model of ion transport through a forested catchment at Lange Bramke, West Germany. Geoderma 38, 97-113 . HAUHs M. (1989)Lange Bramke: an ecosystem study of a forested catchment. In Acidic Precipitation , Vol. 1, Case Studies (eds D. C. ADRIANo and M. HAVAS), pp. 275-305. Springer. KHANNA P. K., PRENZEL J., MEIWES K. J., ULRICH B. and MATZNER E. (1987)Dynamicsof sulfate retention by acid forest soils in an acidicdeposition environment. Soil Sci. Soc. Am . J . 51,446-452. LOKEWILLE A. and VAN BREEMEN N. (1992) Aluminium precipitates from groundwater of an aquifer affected by acid atmospheric deposition in the Senne, northern Germany. Water Air Soil Poll. 63, 411-416. MATSCHULLAT J., ANDREAE H., LESSMANN D., MALESSA V. and SIEVERS U. (1992)Catchment acidification-from the top down. Envir. Poll. (in press). MAy H. M. and NORDSTROM D. K. (1991) Assessing the solubilitiesand reaction kinetics of aluminous minerals in soils. SoilAcidity (eds B. ULRICH and M. E. SUMNER), pp. 125-147. Springer. FRENZEL J. (1982) Ein bodenchemisches Gleichgewichtsmodell mit Kationenaustausch und Aluminiumhydroxsulfat. Gottinger Bodenkundliche Berichte 72, 1-113 . FRENZEL J. (1991)Introduction to BEM (Batch Equilibrium Model). Berichte des Forschungszentrums Waldokosysteme 28, 1-51. REuss J. 0. , CHRISTOPHERSEN N. and SEIP H. M. (1986) A critique of models for freshwater and soil acidification. Water Air Soil Poll. 30, 909-930. REuss J. O. and JOHNSON D. W. (1986)Acid Deposition and the Acidification of Soils and Waters , Ecological Studies 59. Springer. SIEVERS U. and RooSTAI A. H. (1990) Verbundforschung Fallstudie Harz, Teilvorhaben 2: Schwermetallbilanzaus Immissionen und geogenem Anteil im Einzugsgebietder SOsetalsperre. Berichte des Forschungszentrums WaldokosystemeGottingen, Vol. 19.