In the late 1970s, I was faced with the problem of running a small catchment study and interpreting the results. The aim was to clarify the influence of acid deposition on run off water, and to a lesser extent on the catchment itself. I got some way towards those aims, but not nearly so far as I wanted. Patterns could be seen in the runoff chemistry data: some of these could be interpreted in a qualitative way. I could take advantage of features of the catchment to quantify processes such as sulphate reduction or the effect of trees on soil properties. An overall quantitative understanding of catchment geochemistry remained elusive, however, and prediction of (for instance) the effect of changes in deposition on runoff water quality an impossibility. This was not very satisfying.
More than 20 years later, have things improved? I think they have, and in this overview I will try to review some of these improvements, mostly using examples from the geochemistry of acidification and recovery. There have been advances in both techniques and concepts. Analytical techniques have increased in speed, scope and affordability. Multi-element analyses on large numbers of samples are now possible with the development of techniques such as ion chromatography and plasma emission spectroscopy. Advances in mass spectroscopy have enabled stable isotope geochemistry to make a significant contribution to knowledge. Just as important, the information technology revolution has given us the tools to manage the torrent of data. And most significant of all, new concepts were developed (and old ones were recycled) which enabled us to make more sense of the patterns being observed in runoff water. The mobile anion concept is one example: the realisation of the importance of carbon dioxide equilibria; the quantification of ion exchange; and the clarification of aluminium solubility controls are others. The development of computer modelling techniques and technologies has allowed the development of acidification models which integrate these processes, and which can produce the quantitative predictions which were impossible in the 1970s.
So do we now understand enough about catchment responses, to acid deposition at least? I think not, and in the final part of this paper I will highlight some of the problems. Prediction of the responses of substances with strong biological interactions (e.g. nitrogen species and organic matter) is one area. Another is the interaction of geochemistry with hydrology - such as the prediction of chemical responses in episodes. And if we can attempt to predict the responses of small catchments, what can we say about large ones - which are probably more important to society in general? There seems to be plenty of work for catchment science over the next 20 years.
Knowledge of element speciation in the soil waters is essential to determine the nature of weathering processes at the soil/water interface. It is now well established that the solute fractions of samples separated using 0.2 mm membranes filters are composed by a mixture of true dissolved species and colloidal particles, and that colloids may strongly affect the speciation of element in the waters (Viers et al., 1997). The major aims of this study are: (1) to understand the spatial and temporal variability of trace-element speciation in groundwaters, (2) to unravel the role of this variability on the fate and transport of trace elemnts in small catchment.
The studied site is a small agricultural catchment located in temperate oceanic climate (Central Brittany, Western France). The soil distribution and the hydrology can be summarized by two contrasted domains: (1) a hillslope domain comprising organic-poor waters (2) a bottomland zone characterized by waterlogged soils, where shallow groundwaters are organic-rich and exhibit temporal variations of redox conditions. Water samples have been collected fornightly during the autumn-spring 1997 and 1998, in piezometers of the both domains and at the outlet of the catchment. The collected waters were immediately filtered on the field at 0.2 mm. The separation of colloids particles from 'true' dissolved species were performed by tangential ultrafiltration. Trace metal and Rare Earth Element (REE) concentrations were determined by ICP-MS.
Four important results stem from this study. (1) Organic-rich waters from the bottomland areas present higher trace-element concentrations than those from the hillslope domains. (2) Most of trace elements are controlled by colloids. However, the relative amount of colloidal trace elements and the nature of the colloids are somewhat different in the two domains. (3) In bottomland waters, a significant relationship between the Fe and REE concentrations is observed during reducing process. This trend indicates that, during this process, soil-forming iron oxi-hydroxides are destabilized and that trace elements, firstly released in solution, are complexed by organic colloids. (4) Trace-element and dissolved organic carbon contents at the outlet increases with increasing discharge. This agrees with previous hydrological studies, which showed that bottomland zones represents the major contribution of the stormflow (Durand and Torres, 1995).This study confirms the important control exerted by colloids in the fate and transport of trace element in catchment. Futhermore, a strong linkage is revealed between the hydrology of the system and the speciation of water-borne element. This could influence models dedicated to the estimation of fluxes from terrestrial to aquatic ecosystems.
Durand P.& Torres J.L.Journal of Hydrology, 181, 1-22, (1995).
Viers J., Dupré B., Polvé M., Schott J., Dandurand J-L., Braun J-J., Chemical Geology 140, 181-206, (1997).
The small Strengbach catchment (0.8 km2) ranging from 883 to 1146 m and located in the North-Eastern part of France, has been intensively monitored since 1985 in order to assess the consequences of acid atmospheric inputs on vegetation, soil and surface water quality. Mainly covered by spruce, this catchment lies on a granitic bedrock covered by acid brown to podzolic soils. Hydrochemical investigations were conducted simultaneously at plot and catchment scales. Atmospheric pollutants, open field precipitation, throughfall and soil solutions at various depths and in different stands, spring waters and streamwaters have been continuously monitored for 10 years. Hydrochemical budgets were calculated at the different stand and at the catchment scales.
Mid-term trends in water chemistry and associated fluxes of major elements during 1985-1995 period are presented in the different ecosystem compartments. The results are discussed according to hydrological and pollution climate changes. Over the study period, few changes in SO2 and NOx concentrations in local atmosphere are noticed, but the acidity of open field precipitation and throughfall decreases by almost 50%. This decrease is mainly associated to a decrease in sulphate, but also to an increase of ammonium. In a 90 year old spruce stand, proton concentration in the soil solutions decreases generally while total Al concentration remains constant. In a 35 year old spruce stand, proton concentration in soil solutions remains generally almost constant, base cation concentrations decrease while total Al generally increases. Hence, the ratio between base cation and Al concentrations decreases strongly. In spring and stream waters, divalent base cation concentrations significantly drop down simultaneously with sulfate. In spring waters, alkalinity drops down as well as the base cation to strong acid anion ratio.
Stand and catchment hydrochemical budgets show that acidity is mainly buffered in the regolith. At the catchment scale, net losses of silica, sulphate and base cation are strongly related to drainage intensity.
Two hypothesis can be drawn: (i) Because of less acid inputs, the catchment is in a recovery phase regard to an acidification trend; (ii) the exchangeable base cation pool of the soils and the regolith has decreased to such an extent that acidity is reaching the deep water compartment in the catchment. Unfortunately, both soil echangeable capacity depletion over the period and the decline in spring water buffering capacity indicate an acidification trend and let us think that the second hypothesis must however still be considered.
The different origins of streamwater in a catchment greatly influence the hydrological and geochemical behavior of the storm runoff. This behavior is controlled by the variable inputs of different reservoirs and/or contributing areas. In order to characterize as a geochemical point of view the main contributing areas which supply the streamwater in the small Strengbach catchment (80 ha) and to estimate their respective contributions to total streamflow, water samples have been collected regularly in time and space during a storm event (18th - 20th May 1994).
Among all the elements analyzed, DOC and dissolved silica appear to be the most efficient chemical tracers to characterize the different contributing areas because they are both conservative, they are not concentrated in atmospheric inputs and they present opposite patterns in the storm runoff. Consequently, they have a priori different origins and they trace different contributing areas. DOC is mainly supplied by the leaching of organic rich soil horizons of the saturated areas where the DOC concentration of the draining waters reaches 7.3-9.7 mg.l-1, whereas it is only 0.8-3 mg.l-1 in the waters draining the upper catchment. Silica is mainly released by the weathering of plagioclases and micas and originated from the deep horizons of the hillslopes where silica concentration in waters is three to four times higher (100 and 130 µmol.l-1) than in those from the saturated areas. The diagram silica versus DOC shows that the stream water is a mixing between two end-members: one is characterized by low silica and high DOC concentrations which corresponds to the saturated areas and the other by high silica and low DOC concentrations which corresponds to the deep layers of the hillslopes.
The hydrograph separation of stream flow components using DOC and silica concentrations in a mixing model indicates that the contribution of the saturated areas, which occupy only 2% of the total catchment area, reaches 25-29% of the total stream flow at the outlet, the remaining 71-75% being supplied by the hillslopes. Consequently the hillslope contribution to the major element fluxes exported by the Strengbach is very high (84 to 99%), except for DOC fluxes of which 56% originates from the saturated areas. Nevertheless the specific fluxes (kg.ha-1) of most of the elements are much more important in the saturated areas than in the rest of the catchment.
This study points out the important role of the saturated areas in such a catchment during storm events, both on hydrological and geochemical points of view. DOC, which is not classically used as a chemical tracer in such a study, appears to be very efficient to assess the contribution of saturated areas to storm flow generation and to stream fluxes of organic and inorganic elements.
The geochemical Model of Acidification of Groundwater in Catchments (MAGIC, Cosby et al., 1985) is a physically based, processed-oriented model of catchment soil and streamwater chemistry. MAGIC is based upon equilibrium and mass-balance equations for soil ion exchange, dissolution of Al(OH)3 and solution of CO2. Successful reconstructions of past long-term acidification of catchments, constrained by paleoecological studies, have been reported within Great Britain, Europe and the USA. On this basis MAGIC has been used to predict future acidification trends given various deposition-input scenarios to estimate the likely effect of environmental management strategies. Calibration of MAGIC parameters to the catchments under investigation has often been afforded by Rosenbrock Optimisation and Regionalised Sensitivity Analysis (RSA), Hornberger & Spear (1981). Many previous calibration methods have been extended using more traditional statistical techniques to estimate uncertainties in predictions.
This study explores the uncertainty within predictions made using the two-box version of MAGIC (Jenkins & Cosby, 1989) using data from the Plynlimon catchment, mid-Wales. The Generalised Likelihood Uncertainty Estimation (GLUE) procedure of Beven & Binley (1992) is used. The GLUE procedure explicitly takes into account the uncertainty in model predictions. GLUE is based upon the premise that, within the limitations of model structure and errors in boundary conditions and field observations, multiple parameter sets may be equally likely descriptors of a system. Monte Carlo simulations are run to randomly generate parameter values from realistic pre-specified parameter ranges. The likelihood of the numerous parameter sets being behavioural, or not, is conditioned using quantitative and/or qualitative information. Likelihood values are then used to estimate the uncertainties associated with model predictions.
Beven, KJ & Binley, A, Hydrological Processes, 6, 279-298, (1992).
Cosby, BJ, Hornberger, GM, Galloway, JN & Wright, RF, Water Resources Research, 21(1), 51-63, (1985).
Hornberger, GM & Spear, RC, J. of Environmental Management, 12, 7-18, (1981).
Jenkins, A & Cosby, BJ, In: Regional Acidification Models, Springer, 253-266, (1989).
During the last decade several forested catchments have been investigated in Scandinavia, with the aim to study sulfur dynamics. Effects of acidification reversal, acidification, elevated CO2(g) as well as 'pristine' catchments in the north have been studied. Samples have been analyzed for its 34S(sum)S and/or 34SSO4 values in soils, soil water, groundwater, runoff, throughfall, bulk deposition and snow. At low temperatures sulfur isotope fractionation is only significant if bacterial dissimilatory sulfate reduction (=BDSR) occurs in the saturated zone (Krouse and Grinenko, 1991). Processes in the unsaturated zone, like sulfate sorption, immobilization and mobilization give no or small isotope differences between source and end product (Krouse and Grinenko, 1991).Two catchments were investigated that concerns the roof project in the Lake Gårdsjön area, the roof catchment and its reference catchment. The reference catchment shows large variations in groundwater and runoff 34SSO4 values. These variations were anti correlated with the sulfate concentration. The roof catchment shows, however, very small 34SSO4 and sulfate concentration variations, even though the experiment includes addition of some sea water sulfate (which has a 34SSO4 value of +19.5‰ compared to about +6‰ in runoff). The roof also intercepts acid deposition, so that under the roof the total sulfur deposition reaches pre-industrial values. The difference between the catchments sulfur dynamics' is most likely caused by BDSR and oxidation of reduced sulfur (that was previously reduced) in the reference catchment and the lack of such processes in the roof catchment. Furthermore, in the roof catchment the lack of response to the isotope addition is suggested to be caused by mixing with a large soil sulfur pool and net sulfur mineralization from the soil horizons. This conclusion is supported by 34SSO4 measurements in soil water, modeling and soil sulfur pool changes with time.However, the roof catchment seems to be rather unique in Scandinavia, because the other forested Scandinavian catchment we have investigated (e.g. Risdalsheia, Lake Mjösjön, Svartberget, Coast land of Västerbotten and Kalix River) exhibits 34SSO4 and sulfate variations as observed in the reference catchment at Lake Gårdsjön.
Krouse HR & Grinenko VA, Natural and anthropogenic sulphur in the environment. SCOPE 43. John Wiley & Sons, New York., (1991).
The outlet of the Wadi Ahin catchment is located in the northern part of the Batinah coastal plain, which extends northwestward of Muscat and is one of the main agricultural areas of the Sultanate of Oman with the second highest population density in the country. B- and Sr-isotope ratios were measured in the catchment groundwaters in order to trace the degree of water-rock interaction and mixing processes in this small drainage basin, with special attention being paid to marine intrusion in the downstream catchment area. In such a coastal hydrosystem, the isotopic signature of the rain (11B = 38.9‰, 87Sr/86Sr = 0.70862) is close to that of seawater (11B = 42‰, 87Sr/86Sr = 0.70918). The upstream part of the Wadi Ahin basin (about 734 km2) corresponds to a mountainous area in which Wadi Ahin and its tributaries flow in narrow valleys. Despite the ophiolitic nature of the mountains, the surface waters (11B = 34.8‰, 87Sr/86Sr = 0.70748) show evidence of preferential dissolution of limestone olistoliths cast in the crust during the Cretaceous obduction. In its middle part (about 80 km2 and 30 km in length), the Wadi Ahin valley widens slightly and crosses limy, marly and sandstone outcrops attributed to the Eocene. Here the alluvium in the wadi bed is relatively thin, varying from 30 m to 50 m maximum. The Quaternary aquifer in these terraces exhibits a very homogeneous isotopic behaviour (11B = 34.1 ± 0.8‰, 87Sr/86Sr = 0.70849 ± 0.00005), but the basement of this aquifer, formed of Eocene mudstone with low permeability characteristics, contains waters whose isotopic composition (11B = 14.3‰, 87Sr/86Sr = 0.70792) indicates a strong influence of water-rock interaction processes. In the downstream basin area (about 120 km2 and 15 km in length), the wadi valley further widens down to the sea and runs through Quaternary alluvium of unknown thickness. Although intensive agricultural use of water in the lower basin has induced seawater intrusion, the Quaternary aquifer reveals no significant vertical variation of the isotopic composition that could indicate a stratification of different water qualities. Some wells located near the shore even suggest the existence of a confined aquifer between the Quaternary aquifer and the marine intrusion. The inferred geochemistry of this confined aquifer indicates a high altitude origin for the recharge waters.
Intensive agriculture involves 86.5% of the Coët-Dan catchment land (Brittany, France). Nitrate concentrations in the groundwater close to the water table in the weathered upper zone of the aquifer reach therefore levels of 200 mg/l resulting mainly of pig manure spreadings. The rapid decrease of NO3- concentrations with depth in the fissured part of the aquifer, is partly the result of dilution with older groundwater, probably never polluted, as deduced from isotopes of the water molecule (3H, D, 18O)and of sulphates (34S, 18O). The importance of denitrifying processes is evidenced by N2-concentrations as well as N-isotopes of nitrates. Several electron donors may be implied in the denitrification reactions: ferrous iron, pyrite and organic matter. At least the two last processes are mediated by the presence of bacteria. Mass balance calculations show that denitrification may contribute to a loss of at least 50-70 mg/l of NO3-. A small-scale tracer test has demonstrated a high rate of denitrification through pyrite and ferrous iron oxidation in the fissured part of the aquifer with a half life in the order of days (Pauwels et al., 1998). The autotrophic denitrification is expected to be the predominant process in the unweathered schist, whereas heterotrophic processes can occur in the overlying weathered zone. Metals contained in pig manure may catalyse chemical denitrification by iron (II). Autotrophic reactions tend to liberate SO4 but in the present case sulphate concentrations are probably controlled by the precipitation of sulphate minerals such as jarosite which are generally observed under acidic conditions.
Pauwels H, Kloppmann W, Foucher J-C, Martelat A & Fritsche V, Applied Geochemistry, 6, 767-778, (1998).
Riverine carbon occurs in different forms: dissolved organic (DOC) and particulate (POC), dissolved inorganic (DIC) and particulate (PIC). It can be both inherited from watershed inputs and affected by fluvial processes. 13C/12C ratios can help to understand carbon sources and cycling between the different compartments. Previous isotopic studies of riverine carbon have principally focussed either on DIC or on POC. We present, for carbon in the Rhône system, a coupled isotopic study of the organic and inorganic compartments.
River waters were sampled at 20 sites on the Rhône and its main tributaries. In the suspended sediments (collected on Whatman GF/F glass fibre filters), 13C values of POC indicate mixing of C3 terrestrial plants and phytoplankton organic matter, and the 13C values of PIC mixing of sedimentary rocks and soil/river carbonates. Significant contributions of phytoplankton organic matter and authigenic carbonates occur in the lowland tributaries. In the dissolved inorganic carbon, 13C values increase from the mouth to the headwater. This is consistent with decreased load of soil biogenic CO2 to weathering in Alpine catchments.
Elemental and isotopic mass-balances were calculated in order to test whether the observed carbon fluxes in the Rhône system can be solely accounted for by mixing between the Rhône and its main tributaries. The river was treated as a series of linked segments. Calculated and estimated fluxes out of each segment were compared. In the particulate compartments, attempts were made to estimate river carbonate fixation, photosynthetic carbon fixation and exchange between sediment and suspended sediments. In the dissolved inorganic compartments, mass-balances point to a minor but systematic input of CO2 from river respiration.
The St. Lawrence river ranks 16th among the world rivers (CSL, 1996), for its freshwater discharge. Its particulate fluxes are low, when compared with other rivers, due to the trapping of sediments in the Great Lakes. However, dissolution of the Paleozoic carbonate bedrock of the St. Lawrence lowland still results in significant fluxes of dissolved inorganic carbon (DIC) (Yang, 1996). They represent more than 1% of the total world terrestrial DIC-fluxes to the ocean. The present study was undertaken in order to determine their seasonal changes with special attention to the ratio between DIC-supplies from the Great Lakes vs those from the St. Lawrence tributaries. It started in the spring of 1997. Two sampling stations, at Montreal and Quebec city, provide information respectively on the incoming fluxes from the Great Lakes and the outgoing fluxes to the estuary. Two tributaries are also included in the monitoring program, one draining a sub-basin with Precambrian migmatites and marbles (Ottawa River), one draining a sub-basin with Paleozoic carbonates (Mascouche River). During the first sampling year, the annual water-flow was ~ 9800 m3/s at Montreal, with maximum/minimum values of 11900/8500 m3/s. The corresponding DIC fluxes were 13.6*109 (mean), 17.1*109 and 11.5*109 gC/day (minimum/maximum), respectively. The minimum values are observed during the late summer and fall. DIC-fluxes generally differ by less than 10% between Montreal and Quebec, indicating a prominent influence of the DIC supplied by the Great Lakes. DIC-13C values (vs PDB) as high as +0.9‰ are observed during the low-water fall period, and suggest isotopic equilibrium with atmospheric CO2, when the river drains almost exclusively the DIC from the Great Lakes. On the contrary, during high-water periods, and notably the spring flood, DIC-13C values as low as -3.8‰ at Montreal, and -6.8 at Quebec, indicate significant inputs from the soils of the drainage basin. The tributary supplies are indeed characterized by low DIC-13C values, ranging from -5 to -11‰ (Ottawa River) and -7 to -14‰ (Mascouche River), approximately. In this latter case, DIC-13C values suggest almost stoechiometric dissolution of Paleozoic carbonate matrices (~ +1‰) with soil CO2 (estimated at ~ -24‰), thus minimum subsequent re-equilibration with atmospheric CO2. It is concluded that the drainage basin of the St.Lawrence may be responsible for up to 30-50% of DIC-fluxes during the spring snow melt period, and for negligible supplies during dry-periods. The 13C-data set suggests minimum influence of in situ DIC-metabolism and/or re-equilibration with atmospheric CO2, although other studies show that some of the St. Lawrence riverine ecosystems may present a more complex carbon metabolism (Bart, 1997).
Centre Saint-Laurent, Rapport-synthèse sur l'état du Saint-Laurent, 1, 736, (1996).
Yang C, Telmer K & Veizer J, Geochimica et Cosmochimica Acta, 60, 851-866, (1996).
Bart J, PhD Thesis, (1997).
Hydrochemical investigations in catchments are commonly used to quantify the processes controlling surface water chemistry. In order to determine the origin of Ca, Sr was analysed as its analogue in a small catchment sensitive to acidification where Ca storage is low. The first step was to determine changes in behavior of Sr with soil, vegetation and hydrological conditions. The Strengbach catchment (80 ha) is covered by spruce (80%) and beech (20%) trees. Acid brown to ochreous podzolic soils have developed on a base poor granite. Several soil profiles were collected. Open field precipitation, throughfall and soil solutions (under beech and spruce stands in the upper part of the catchment), spring and stream water were sampled regularly during one hydrological year.
Initial results show that Sr isotopic ratios in two soil profiles range from 0.7494 to 0.8449. There is no correlation between Sr concentration and isotopic composition with depth. However, within a given horizon, the coarser are the soil fractions, the more radiogenic is the strontium, due to an abundance of radiogenic minerals. Both Sr isotopic ratios and Sr concentrations increase under the spruce stand from open field precipitation to soil solution at 60 cm depth (0.710-0.725 and 0.6-4.6 ppb respectively). A similar behavior is observed in the beech stand, however Sr is less concentrated there and more radiogenic. Throughfall and soil solutions plot on the same mixing line in the 87Sr/86Sr versus 1/Sr diagram. The 87Sr/86Sr ratio of stream water is close to that of the deepest soil solutions. This indicates that the isotopic composition of streamwater is already established in the upper part of the catchment. A weighted calculation using Sr isotopes indicates that the waters draining the beech stand would contribute about 28% of the streamflow at the outlet.
During the hydrological year, the Sr concentration of the streamwater varies between 10 and 14.5 ppb, and 87Sr/86Sr ratio shows a large seasonal variability. Waters are less radiogenic (0.7245-0.7248) and less Sr concentrated during the low water flow period than when the streamflow is much higher (0.7248-0.7253). The participation of more radiogenic deep waters stored during high water flow periods could explain these high values in streamwaters. This reflects the slight influence of incident rainwater to this catchment.At the catchment scale, surface waters plot on the same mixing line according to Sr-Ca diagram, except throughfall. However, in each type of water, a significant relationship between Sr and Ca could only be observed in throughfall.
Therefore Sr isotopes allow to recognize differences between all water types. As a result, it will be possible to quantify the proportions of atmospheric and lithologic strontium in streamwater and on soil exchangeable capacity.
7Be and 210Pb have been successfully used for tracing soil erosion and transport processes in the watershed, as well as sedimentation in lakes. Modelling of these processes requires a good knowledge of the input function from the atmosphere. In order to evaluate spatial and temporal variability of 7Be and 210Pb deposition over an entire annual cycle, atmospheric traps have been placed at four different sites around Lake Geneva (Versoix, Buchillon, Thônon, Montreux). The sampling was done with a monthly resolution, except at one site where one additional trap was also taken after each rain event. In parallel, the fluxes of major elements (Ca2+, Na+, Cl-) were determined, while the precipitation parameters were obtained from the Swiss Meteorological Network. The preliminary results, obtained over a period of eight months, show that the 7Be and 210Pb fluxes are strongly correlated by considering only the individual rains or by taking into account the monthly integration (r2 ~ 0.9 in both cases). Though it is less evident, the 7Be and 210Pb fluxes are positively correlated with the amount of precipitation too. That indicates that the fluxes of nuclides are mainly governed by the same scavenging processes. The examination of the specific 7Be and 210Pb concentrations (activity per litre) in individual event vs. the amount of precipitation illustrates another phenomenon. For very little rainfalls (< 4 - 5 mm), the specific concentrations are high, whereas when the rainfall is more important, they tend to decrease and then to stabilise at low values (210Pb~0.2 Bq/L, 7Be ~ 2.7 Bq/L). Actually, two different processes are operating in a rain: at the beginning, the scavenging of the radionuclides associated with the aerosols in the air level below the clouds is important (washout), and later the scavenging occurring inside the cloud during the nucleation process (rainout) becomes dominant. The high concentrations observed for the low rainfall seem to illustrate the preponderance of washout, or possibly a significant contribution of dry deposition. With heavy rainfall, the lower atmosphere is quickly washed and the relative importance of washout against rainout becomes negligible. That is why the concentrations tend to stabilise. For individual events, no correlation between radionuclide activities and major element concentrations is observed, except between concentrations of 210Pb and Ca, because of their common continental origin. Our preliminary results show that the radionuclide fluxes are most of the time similar over more than the 580 km2 of the lake. However, higher fluxes are locally recorded as at Montreux (+15%), likely because of a higher mean precipitation resulting from geomorphological situation.
Karstic areas near Montpellier (Southern France) have been investigated for 3 years (Schoen et al., Pinault et al., this issue). In order to understand groundwater processes in karstic hydrosystems, a chemical and isotopic characterisation of input signal is necessary. For this purpose, rainfall was sampled during the hydrological cycles 1996/1997 and 1997/1998 and analysed for major ions and stable water isotopes.
Mean chemical fluxes and variations are presented. The sampling sites are distributed along an altitudinal transect. The ionic contents analysed decreases with increasing distance from the sea and elevation. During the investigated period a sharp discrepancy was observed between the concentration in autumn rainfall events and the other periods, which indicates that chemical fluxes are much higher during these events. At all sites, the Na/Cl ratio is similar to the common bulk seawater ratio: 0.55. Thus, Sodium and Chloride find their unique origin in sea salt. Consequently, Cl can be considered as a conservative tracer and used to estimate sea contribution for other ions. Results show a mean contribution of sea-salt to the ionic composition of rainfall (about 10%), with higher values (about 40%) in autumn periods. In the study area, autumn rainfalls are mainly of Mediterranean origin, while during other seasons the precipitation can originate from the Atlantic ocean. Thus, the air masses coming from the Atlantic have undergone a long transfer before reaching the study area, which can partially explain the low marine contribution.
In the 18O-18H diagram, results are distributed between the Global and Mediterranean Meteoric Water Lines. Deuterium excess values ranged from 7,2 to 21,4‰, the maximum values were observed during autumn periods. This again suggests that autumn rainfall is mainly Mediterranean origin while during other seasons the precipitation can originate from the Atlantic.
As for rainwater, we analysed karstic springs in the study area stable for chemical composition and stable water isotopes. The weighted mean concentrations of the autumn rainfall were compared to mean concentrations of the karstic springs during baseflow periods. Mean chloride and sodium concentrations in groundwater match the autumn rainfall concentrations. The Karstic springs show high deuterium excess values, close to the autumn precipitation values. These results indicate that aquifer recharge mainly occurs during autumn limited rainfall events.
Increased weathering rates of natural stone in buildings and cultural heritage have become an acute problem as a response to acid deposition. The purpose of this study is to determine the weathering rate and the rate controlling mechanism of a granite rock surface in Tanum, SW Sweden, which is an area with numerous rock carvings from the Swedish Bronze Age of high preservation interest. The area is severely affected by acid rain, and the proximity to the North Sea expose the rock surfaces to high salt concentrations. As an extension of the study, a roof has been built over a rock surface to investigate the possibility to reduce weathering rates.
Chemical weathering of the rock surface is one source for the elements found in runoff waters. The chemical weathering can be calculated if other sources are identified and subtracted. To separate between the sources, wet- and bulk deposition are collected and also runoff from small confined ponds on the rock surface. Chemical analysis is made of all these waters for major- and trace elements and isotopes (Sr, S and Pb). The rock surface pond is sampled during rainfall, while the rock surface pond under roof is sampled during washing with deionized water. The mineralogical composition of the rock is determined by microscopic analysis and the specific surface area of the rock is determined through laser micro mapping and BET analysis of small surface rock pieces. Temperature, rainfall and humidity are measured at the rock surfaces using a computer-logged system. An attempt is also made to determine the physical weathering through collection of rock fragments in the ponds and collection of small rock particles in filtered samples from the runoff.
Comparisons of ICP-MS analysis between deposition and runoff samples show that the trace elements U, Nd, La, Ga, Ce, Mo, Pr, Sm, Sc, Sr, Ti, Y, and Zr are enriched in runoff. There is a strong correlation between Na and Mg versus Cl in the water samples, indicating a large influence of the coastal proximity. ICP-MS analysis of the filters shows that an important part of the weathering products are small rock fragments contributing to the total weathering rate. As a first approximation, chemical weathering rates have been estimated without correction for dust fall and specific surface area, and thus the maximum chemical weathering rate (release rate of Si) varies between 8*10-13 and 2*10-14 mol/cm2/s with a mean of 3*10-13 mol/cm2/s. The rate is positively correlated with the rock surface temperature. The rock surface under the roof show a mean maximum release rate of Si at 1*10-14 mol/cm2/s, indicating that a roof might be an alternative preservation technique against chemical weathering.
A large portion of runoff and associated hydrochemical transport in boreal regions occurs during the spring flood. The period is also of great significance for the aquatic biota. For these reasons the chemistry, and especially the pH decrease associated with the spring flood has been of considerable interest. Much of this interest has focused on the rapid release of pollutants from the snowpack. Less attention has been paid to the amount and character of organic carbon in the spring flood, as well as the problem of distinguishing the organic acidity from that generated directly by acid deposition. To address these issues, the 1997 and 1998 spring flood was studied on 12 catchments in northern Sweden in environments ranging from mountains, to inland forests and the coastal zone. Regular observations of runoff and sampling for anions, cations and total organic carbon (TOC) were supplemented by specialized isotope analyses to distinguish the recent snowmelt in runoff from older water displaced from the catchment. The potential for separating the organic contribution to acidity during spring flood from that contributed by acid deposition was explored. A key feature in this calculation is the estimate of how much the acid neutralizing capacity is reduced by natural dilution processes.
Investigations on springwater acidity were carried out in the Vosges mountains (north-eastern France). Acid or poorly buffered spring and streamwaters were detected in the same area. The proportion of acid springwaters (pH<5.6) is about 20% among 220 springs. The springwater pH on granite are equally spread between 5.0 and 6.8 whereas on sandstone a majority of springs is in the range 5.6 to 6.2.
As a whole, but mainly on sandstone, from the 1960's to 1990's, the shape of the pH distribution shifts toward greater acidity. In the sandstone area, trends in pH, alkalinity, total hardness (corresponding to divalent cations), sulfate and nitrate were considered over the 30 year period (1963-1996) in relation to the bedrock chemical composition. Kendall seasonnal tau coefficients indicated that decreasing trends were significant for the first three parameters. Linear regression on the smoothed mean value revealed 18% and 90% decrease for pH and alkalinity respectively, for springwaters draining poor-base cation sandstone whereas only 8% and 30% decrease respectively, was observed on clay-enriched sandstone. On silica-enriched sandstone, alkalinity began to decrease in the early 70's as well as pH. However, loss of alkalinity only occurred in the early 80's for springs draining clay enriched sandstone. This can be interpreted as a titration process by acid atmospheric inputs of the buffering capacity of weathering and exchange processes in the soils and the catchment bedrock. The nitrate presents an increasing step in the early seventies but possibly as a result of change in analytical technics and/or increase in atmospheric inputs mainly resulting from increase in fertiliser inputs in agricultural areas or in car traffic. Surprisingly no change in sulfate was noticed in any groups of springs probably as a result of the adsorption/mobilisation processes in the soils.
These long-term trends in spring waters confirm the soil and streamwater acidification trends already mentioned in this region, in relation to acid atmospheric inputs since no climate nor forestry practice changes have been detected over the period. This is consistent with similar observations of the literature for Nordic countries for example. Moreover, in spite of acid atmospheric input reductions, no recovery could so far be detected.
These trends in springwater chemistry can be associated to the different environmental factors of the drainage catchments using GIS in order to predict the extent of sensitive areas and to calibrate the critical loads of acidity calculated for streamwaters.
Monthly sediment-trap particules from Aydat lake (Massif-Central, France), a lake with seasonally anoxic bottom waters and strongly reducing sediments, were analysed for their neutral sugar content. Three sediment traps not treated with antibiotics were deployed at the depths of 5, 10 and 14 m in the deepest part of the lake during a whole year (from October 1995 to September 1996). Samples for neutral sugar analysis were successively hydrolysed with diluted and strong acid solutions. The resulting anomeric mixtures were equilibrated with lithium perchlorate in pyridine before analysis by gas chromatography as trimethylsilyl derivatives. Total sugars were calculated as the sum of individual peaks of sugars which were identified and quantified. During the whole investigation the total sugar fluxes remained in range of 0.01 to 0.12 g/m2/day for the three traps. On a weight basis, the sugar fluxes varied globally in phase with the total sediment flux, the biogenic silica flux and the organic carbon flux. The highest sugar fluxes appear after the autumnal overturn of the lake (November) and when the stratification of the water column takes place (June). These periods are caracterised by Fe-hydroxides precipitation resulting from the oxidation of ferrous ions in the water column and then by an increase of Fe fluxes. In both cases, the increase of the sugar flux can be explained by the flocculation of suspended (and/or even dissolved) material favoured by the ferric hydroxides precipitation. In all traps, general sugar distributions are dominated by rhamnose, glucose and fucose. But, xylose and mannose are also present in notable amounts. The three former compounds probably originate from the microflora while the two latter ones could be inherited from higher plants. Thus, sugar composition does not match that of the primary production in the lake but seems strongly related to the activity of the microflora and then to organic matter mineralization. Thus, sugars although present in relatively low amounts, appear useful indicators of the main biological (food chain) and physico-chemical (sedimentation) processes intervening in the lake.
In semi-arid and arid regions pedogenic carbonate commonly develops in soils formed on various parent material. The purpose of this work is to use carbon and oxygen isotopes to reconstruct the paleoclimatic conditions from soil carbonates developed on volcanic rocks from several areas in Morocco: (1) High Moulouya (1600 msl), (2) Middle Atlas (1300 msl) and (3) Mediterranean coast. Most soil profiles from the studied areas show an increase in carbonate content near the surface and are terminated by a hardpan layer (< 5 cm). Carbonate mineralogy is low Mg-calcite, which precipitated by replacement of host-rock minerals.
The isotopic composition of the bulk carbonate samples varies widely with values between -7.6 to -2‰ (vs PDB) and -10.2 to -6.4‰ (vs PDB) for 18O and 13C, respectively. Carbonate 18O values decrease with increasing altitude and distance from the coastline. This is in agreement with the effect of altitude and continentality on the isotopic composition of meteoric water at each area. However, the relatively low 13C values indicate that the carbonate formed in equilibrium with soil CO2 with a high C3/C4-plants ratio (0.5 to 0.87). This result is inconsistent with the present day dominant vegetation especially in the High Moulouya area, which is very arid and without C3-plants. This implies that carbonate precipitation occurred under more humid climate conditions than the present day but with sufficient seasonality to allow carbonate precipitation during the dry season. Furthermore, the isotopic composition of the thin layers (around the mm) comprising the hardpan indicate that the carbonate formation occurs under varying climatic conditions within the same area. Dating of the carbonates is in-progress in order to compare these climatic results with other climatic proxies, such as palynology.
A reconstruction of historical lake productivity was undertaken on sediment cores extracted from the Ponte Tresa Basin, a small sub-basin of Lake Lugano, Italy/Switzerland. The hydrological position of the Ponte Tresa Basin makes it a special "collecting pond" for inflowing particles. Lateral transport of suspended particles results in extremely high accumulation of organic matter (OM) leading to anoxic conditions in the bottom waters as far back as 250 years ago. Very high contents of OM in these older sediment samples, therefore, create the illusion that the Ponte Tresa Basin has always tended to have eutrophic conditions. This unique sedimentation pattern, however, enhances the geochemical signal of environmental change because of the better OM preservation under continuous anoxic conditions, even during oligotrophic times.
Fluctuations in the inorganic-biogenically precipitated calcite and OM contents were used as reliable indicators of relative changes in trophic state. These parameters reflect the recent history of eutrophication, as previously shown by a study of biogenic SiO2. With increasing anthropogenic nutrient loading during the mid 1960's, accumulation rates of all autochtonous components in the sediment increased dramatically. Primary productivity measured in the water column continued to increase until the end of the 1980's, despite decreasing phosphate concentrations with stricter water-quality regulations since 1983.
Although detrital carbonates strongly contribute to the 13C and 18O values of the bulk measurement, the authigenic isotope signal can be isolated in the upper portion of the core. The degree of apparent isotopic disequilibrium during calcite precipitation, correlates positively with the phosphate concentration. 13C and 15N values in bulk OM correspond basically with the trend of eutrophication in Lake Lugano. Occasionally, however, the geochemical results do not follow this general trend. Microbial degradation in the water column and sediments may alter the geochemical signal, making it difficult to assess the degree of change, as well as masking of the primary isotope signal which is closely coupled with biological activity in the epilimnion. In particular, enhanced decomposition due to denitrification probably contributes to a positive trend in the 15N value.
The relationships between concentrations of dissolved elements (C) and stream discharge (Q) have been commonly used in the literature to predict stormflow response to storm events. However, they have not been so often used to understand hydrological processes in the catchment during stormflow and particularly to assess the origin of storm flow components. Several C-Q relationships were considered for various storm events with different discharge intensities and pre-event soil moisture conditions in the small granitic Strengbach forested catchment (0.8 km2). Concentrations of dissolved solids like silica, sodium and sulfate always decrease with increasing discharge but the C-Q patterns present an hysteresis between the rising and the falling of the hydrograph: they are cyclical. These elements are concentrated in the deep layers of the hillslopes. On the contrary, when measured, the dissolved organic (DOC) concentration - discharge relationship is also cyclical but positive. DOC is mainly concentrated in the upper layers of a small saturated area. However, these hysteresis patterns are influenced by rainfall intensity and duration, but overall by antecedent moisture conditions.
During stormflow following relatively dry moisture conditions, left-hand hysteresis patterns are observed for Si, Na, SO4. At the same time, DOC (when available) and K present right-hand hysteresis cycles. This means that during the increasing discharge, water draining deep layers of the hillslope, enriched in elements originating from weathering (like Na, and Si) are diluted by water draining the saturated area characterized by low silica and high DOC contents. On the opposite, during the decreasing discharge period, the contribution of water draining the saturated area is lower, as indicated by higher Si and lower DOC concentrations in streamwater. However in such dry conditions, the C-differences between rising and falling discharge period are weak.
Following humid moisture conditions, the dilution patterns of silica, Na and SO4 describe right-hand cycles with important hysteresis. This indicates that during the stormflow rising, water draining the hillslope deep layers is mixed with water in the saturated area which have lain for a while in the catchment and was enriched in elements characteristics of deep layers. DOC-Q pattern shows a left-hand hysteresis cycle which confirms the relative important contribution of waters draining the saturated areas (DOC enriched and silica poor) when discharge decreases.
For other dissolved elements like Ca, Mg, NO3 and Cl, the C-Q relationship patterns and the hysteresis directions vary according to the season, the pre-event moisture conditions and the discharge intensity.
In this catchment, Si, Na, SO4 and DOC are the only elements which can be used to model the hydrochemical behaviour of water pathways according to moisture conditions.
This study intends to analyse and to determine the evolution of the chemical composition of drainage water in a changing rural environment of the Morvan region (Burgundy, France). In this medium mountainous area, the land cover is evolving from pasture and culture toward fallow lands and spontaneous or artificial afforestation. Thus, the apparent balance of the environment is disrupted, soil parameters change, and the dynamics of transformation, of storage and of release of soil elements is disturbed. This should lead to a change in the chemical and physical composition of the drainage water. The main objectives of the study are to estimate to what extend such composition, especially nitrogen and phosphorus, are affected and to determine the main controlling factors of the changes.
In order to answer these questions, the study focuses on a medium size catchment (442 km2) located in the Morvan region: the Cousin catchment. The upstream part of the basin lays on plutonic and metamorphic rocks and the land cover shows mainly forested areas, while the downstream region lays on sedimentary rocks (mainly calcareous and marly rocks) and is gradually covered by pastures and cultivated areas. The drainage water were sampled several times during the hydrological cycle (in low flow and high flow period) in 23 different points (7 on the main stream and 16 on the tributaries) covering all the basin area. In each sampling point, the major elemental composition of water, the sediment load and the discharge were determined. The environmental parameters (bedrock, soil, land cover, geomorphology) of the catchment were digitized and related to the chemical and physical properties of the water draining the 23 sub-basins using using GIS. The spatial distribution of the climatic parameters, especially the precipitations, was established using an original and efficient empirical modeling developped in the framework of this project.
The preliminary results show an usual spatial distribution of the major element composition which is controlled by the lithology (for Ca, Mg, Na, K, Cl, HCO3, SO4) and by human activities (for N and P). The ionic composition evolves from bicarbonate sodium-potassium water in the upstream part toward bicarbonate calcium water in the downstream part. Concerning N and P concentrations, the urban area in the downstream part of the catchment has a strong influence, resulting in a ten fold phosphate concentrations in the Cousin river.
Accurate knowledge of soil solution chemistry is highly important. Processes like mass transfer and ion exchange, dissolution and precipitation reactions depend on aqueous chemistry of dissolved species. Water quality and soil fertility management are two practical and important applications. Given its complexity, the use of thermodynamic computers codes to calculate the speciation of aqueous solutions is necessary. These modeling studies generally reveal undersaturated solutions in forest soils. Few others have obtained saturated solutions with respect to one or two mineral phases, mainly iron and aluminium hydroxides, low temperature silicates or sulfates. These phases are likely to control soil chemistry as long as arguments are provided against the role of ion exchange. This presentation is aimed to discuss thermodynamic calculations performed with the code EQ3NR (Wolery, 1992) on gravitational and capillary solutions from an acid brown soil (40 years Douglas stand, Vauxrenard, Beaujolais, France). The thermodynamic database has been updated and adapted to low temperature environments. Composition of gravitational and capillary solutions have been measured monthly for 6 years at different depth (15-30-60-120 cm). The uncertainty still prevailing for few minerals and aqueous complexes over the temperature dependence of the thermodynamic equilibrium constant are considered in the calculations. Modeling is carried out at 25°C and at the in situ temperature. Gravity-driven and capillary solutions are both oversaturated or saturated with respect to several secondary mineral phases: gibbsite, kaolinite, some clay minerals and sulfates. Linear regression is used to characterize thermodynamic equilibrium constants (K) of secondary phases and gives rise to the control of chemistry by precipitation. The solutions appear also to be saturated and oversaturated with quartz and muscovite respectively. A better estimation of K for the field muscovite is done with a solid solution model. Accordingly, the soil solutions become undersaturated with respect to this mineral. This is consistent with petrographic evidences (Ezzaïm, 1997). This result highlights the need to account for non-ideal mineral phases in geochemical modeling. The present study leads us to conclude on solute dynamic: solute transport is relatively homogeneous in this soil. This implies fast solute transfers between immobile and mobile water. Homogeneous solute transfer is slightly balanced by few noteworthy thermodynamic differences among the solutions. These differences are consistent with the theory assuming the larger proximity to equilibrium or greater oversaturation of the capillary solutions related to their larger residence time (e.g. Gérard, 1996; Ranger et al., 1993; Marques et al., 1996).
Ezzaim, A, Thesis University of Nancy I France, 197, (1997).
Gérard, F, Thesis ULP Strasbourg France, 250, (1996).
Marques R, Ranger J, Gelhaye D, Pollier B, Ponette Q & Goedert O, Europ. J. Soil Sci., 47, 407-417, (1996).
Ranger J, Discours D, Mohamed Ahamed D, Moares C, Dambrine E, Merlet D & Rouiller J, Ann. Sci. For, 50, 425-444, (1993).
Wolery T, LLNL Technical Report, Californie, USA, UCRL-MA-110662-PT-I, 230, (1992).
Varied sources of dissolved loads in river waters can be identified by appropriate studies of the isotopic- and elemental compositions of the constitutive streams in a watershed. The case study of the Cèze stream which is a confluent of the Rhône river in southeastern France, is of special interest because it drains successively highly differentiated rock lithologies forming large structures such as (1) the crystalline basement of the Massif Central, (2) the calcareous "cévenole" margin, (3) the marly gritty plain of Alès, and (4) the marly calcareous plateau of Garrigues.
The impact of these lithologies on the river waters was examined by determining Sr isotope- and elemental compositions which allowed differenciation between three main natural supplies into the Cèze catchment:
* weathering of the Hercynian silicic rocks and the Triassic evaporites in the upper part of the drainage basin, induced low dissolved loads (20-60 µS/cm) and highly radiogenic 87Sr/86Sr ratios (0.7127 - 0.7212),
* contribution of a first carbonate-rich reservoir consisting in Hettangian dolomites and Sinemurian limestones of the calcareous "cévenole" margin, induced higher conductivity (150 µS/cm) due to higher Ca-Mg-HCO3 contents, but low 87Sr/86Sr ratios (0.7112 - 0.7114),
* contribution of a second carbonate-rich reservoir consisting of the Oligocène plain of Alès which yields an increased conductivity (280 µS/cm) with higher Ca and Sr and lower Mg contents compared to those of the first carbonate-rich reservoir, involving even lower 87Sr/86Sr ratio (0.7091 - 0.7093). Some parameters, like the weak increase of conductivity (from 280 to 330 µS/cm) and the slight evolution of the isotopic 87Sr/86Sr ratio towards a lesser Sr-rich endmember in a mixture diagram, imply that the Albian-Aptian marly calcareous formation probably contributed partly to the dissolved load of the river waters. Furthermore, the influence of less mineralised waters of the "cévenole" margin originating probably from an underlying aquifer may also be considered, owing to the distinct decrease of the elemental compositions and a more radiogenic 87Sr/86Sr ratio.
In addition to these geogenic contributions, anthropogenic activities were also identified in three distinctive geographic sectors defined by the means of major- and trace-elemental compositions: (1) the crystalline basement where weathering of old mine dumps supplies the river waters with higher amounts of Sb, Zn, Cu, Ba, (2) the Alès plain and (3) the Cèze low-land, which are regions of extended wine-growing activities, where use of fertilizers induced higher amounts of Na, K and Cl in the waters. In the latter region, the behaviour of As was examined in the waters and the soils, to provide information about appropriate use of specific pesticides in wine-yard activities.
During almost thousand years the environment in the Falun area, central Sweden has been affected by anthropogenic S emissions. The dominating S source has been SO2; a product from oxidized sulfide ore during mining processes in the Falun copper mine. The S deposition from the mining activities has decreased considerably since the beginning of the 20th century and been replaced by long range transported S from fossil fuel burning. By data from previous studies good estimates of the 34S values of these two dominating S sources in the area exists. Therefore, by using stable sulfur isotopes as tracers, the soils in the Falun area provide an opportunity to study soil S dynamics and acidification reversals. Previously we have shown that soils close to the mining site still hold a signal from ancient deposited ore sulfur. In this study we have examined soil profiles at increased distance from the mining site in order to study the spatial effects of S deposition and on the soil sulfur cycling in the area.
Six podzol profiles at distances from 2 to 40 km in the prevailing wind direction from the mining site were sampled. The profiles were sampled with depth in O-, E- and B-horizons and analyzed for S isotope ratios and S concentrations of total sulfur (organic sulfur + inorganic sulfate) and inorganic (adsorbed + dissolved) sulfate. The 34S values ranged from +1.3‰ to +7.6‰ and showed increased 34S values with increased distance from the mining site for both organic sulfur (total sulfur - inorganic sulfate) and inorganic sulfate. The 34S values were nearly constant with depth, and no significant difference in 34S values between organic sulfur and inorganic sulfate was found. This could imply that no fractionation occur during soil sulfur cycling (immobilization and mineralization).
The higher 34S values with distance indicates a decreased influence from ore sulfur (34S 0‰) as the distance increases. At greater distances the higher 34S values from fossil fuel burning (34S +5‰) have contributed to a greater extent. However, the effects from the mining activities can be traced to a distance of at least 25-km from the mining site. At a distance greater than 40-km from the mining site no influence of deposited ore sulfur can be found.
From 1977-82 a catchment study was carried out in the Tillingbourne Catchment in Surrey, S. England with the original aim of assessing the influence of rain chemistry on stream chemistry. This was one of the earliest studies of the biogeochemical effects of acid deposition in the UK, and an extensive database was built up covering wet deposition; throughfall; soil and soil water chemistry; stream chemistry and groundwater chemistry. Numerous process studies were also carried out. Both deposition and streamwater proved to be very acidic, (mean rain pH 4.15; mean throughfall pH under conifers 2.82; mean outlet stream pH 4.02). The area is now thought to have a low critical load for acid deposition. Over the last 20 years, acid deposition in this area has been reduced considerably, and the catchment is being resampled to assess whether any recovery of the catchment soils or waters has taken place. Stream samples are being taken in identical positions. A survey of soil chemistry under 9 isolated trees has been repeated, allowing an assessment of soil change over the 20-year period. The catchment is very interesting hydrologically, consisting of permeable sandstone overlying impermeable clays, which gives rise to a mosaic of dry, acidic areas and wet, more basic areas which in turn produce sharply-contrasting water types. Understanding of acidification processes has increased a great deal since the original study, and modern acidification models and concepts will be applied both to the original and newly-collected data.. The results of the first year's re-sampling will be reported in this paper.
Inverse modeling has been carried out on the springs that have been hydraulically and chemically monitored for flow rates and water chemistry (Schoen et al., this issue). Inverse modeling characterizes the dynamic processes from which the properties of the karsts can be identified. Such fingerprints can be represented by the impulse responses (the response of the hydrogeological system after an isolated rainfall event) of flow and fluxes.
The model structure consists of the following steps: (1) The input of the system is defined as the quantity of rain which will induce a flow variation, considering that a part of the rain is lost out of the system. A threshold (omega), variable with time, separates the actual rain-fall (R) from the lost one. (2) The actual rain-fall is then separated into two components, the quick and the slow components. A parameter (alpha) defines the relative proportions of each components at each moment. (3) The impulse response of the outflow (gamma) to each component is computed through an iteration process.
The same functions are calculated for fluxes, i.e. the product of flow by the concentration of the chemical marker of a particular compartment of the hydrosystem, which is carried out in two steps. (1) The first one is similar to the process used for the flow. (2) A distinction is made between the pre-event water and the event water, i.e. the water originating from the last rainfall event, which allows the determination of water renewal.
The slow component of the hydrograph being reconstructed from a linear combination of impulse responses of fluxes, the weighting factors represent the contribution of every compartments of the hydrosystem. Such hydrograph decomposition allows the characterization of the dynamic processes, especially the more or less rapid transfer of the rain-water (or its influence on the water stored in the system). On the other hand, chemical tracers allow to characterize the origin of the water (from rain fall or from saturated zone) and not only the transfer-time. Comparison of hydrograph and chemiograph decomposition for the different springs emphasizes the characteristics of karst aquifers functioning from which objective criteria are defined for the identification of the degree of drainage organization inside the aquifer, the modes of infiltration as well as the mechanisms involved in karstification.
Schoen R, Ladouche B, Aquilin L, Bakalowicz M, Plagnes V, Cubizolles J, J. Conf. Abs., 4, (1999)
Three sites have been monitored in the karstic area north of Montpellier (Herault district, south of France) during a period of 2 hydrologic years: the major springs of Fontanilles, Cent-Fonts, and Hortus/Lamalou.
The flow rates, water temperatures and water electrical conductivity of these springs have been monitored continuously. The related time series were processed by correlation and spectral analyses, to outline hydrodynamic properties of the different karstic systems.
Spring water has been sampled monthly during baseflow periods, and up to daily during flood periods. Since the discharge of all these springs has been recorded, the exported chemical mass fluxes have been calculated. Sampling of spring water was also operated on other springs of the area, among which the Bueges spring. Chemical analysis included : major ions, trace elements, and oxygen 18, deuterium, Strontium 87, Carbon 13, Carbon 14 isotopic composition.
The Fontanilles and Cent-Fonts springs, yielding groundwater which has partly flown through dolomitic rocks, show a decrease of Mg and an increase of Ca during flood events. Inversely during the baseflow periods, a Mg increase is observed whilst Ca remains more or less constant. The Hortus/Lamalou spring, yielding groundwater which has mainly flown through limestone, shows the same variations of Ca content as in Fontanilles and Cent Fonts, whereas Mg content is low and shows very little variation. Variations in Cl content as well as Sr and C isotopic composition are also observed during the flood periods.
These variations are related to the behaviour of the different parts of the karstic system which are involved in the flows depending on the hydrological conditions. During flood periods, part of the flow originates from water that has undergone a significant residence time in the upper part of the system (epikarstic zone, locally saturated or unsaturated), where Cl is reconcentrated by evapotranspiration processes. During the baseflow periods, groundwater is characterized by natural tracers indicating a long residence time in a closed dolomitic carbonate system (saturated zone). Although the scheme is comparable for all the springs of this study, the amplitude and the characteristic times of the chemical variations strongly depend on the surveyed spring.
A concept for simulating nitrogen transport was recently developed, validated and applied in 3725 catchments in southern Sweden, to study the nitrogen contribution to the Baltic Sea. The model system includes GIS, the dynamic soil profile model SOIL-N (Johnsson et al., 1987), and the dynamic catchment model HBV-N (Arheimer and Brandt, 1998). Mapping of nitrogen retention and leakage pattern was performed, based on daily simulations during ten years. The large number of catchments involved allows detailed spatial analysis for the region, and it is obvious that the load pattern radically change when considering net load on the sea instead of local gross load. The retention in southern Sweden is low in a zone nearby the coast, however, in catchments with low lake percentages, this zone may sometimes reach far up in the river basin. Source apportionment shows that most of the nitrogen originates from non-point sources, and arable land is the single largest source.
The model system has been involved in several scenario analyses for predicting long term responses to environmental management strategies. Remedial measures to reduce the nitrogen leakage from arable land to the sea was undertaken in Sweden between 1985 and 1994, but the modelling shows that these measures have had minor influence on the total transport, which was only reduced by 7% (Arheimer and Brandt, 1999). Gross load reduction was higher than net load reduction. Thus, the reducing measures on arable land have not been located in the most effective way with aspects to the sea. To be effective, measures should be focused on the load that is a considerable portion of the net load and is little influenced by retention.
The model has also been used for the development of allocation strategies for constructed wetlands, and for estimations of the total effect of several combined measures in the landscape. It is then obvious that dynamic modelling is a well suited tool for system analysis when describing complex causalities and estimating the influence from several nestled processes. The latest model application includes the entire Baltic Sea drainage basin, divided into 37 subbasins. Also on this large-scale, the model manage to simulate dynamics in water discharge (Graham, 1999) and long-term average nitrogen concentrations to the sea. This application will be the basis for more detailed modelling of the region in the future. The poster presents the HBV-N model and some applications with relevance for environmental planning strategies.
Johnsson H, Bergström L, Jansson P-E & Paustian K, Agricultural ecosystems and environment, 18, 333-356, (1987).
Arheimer B & Brandt M, Ambio, 27, 471-480, (1998).
Arheimer B & Brandt M, Ecological engineering, (1999).
Graham LP, Ambio, (1999).
The Nsimi experimental catchment (0.60 km2, 3°10'N latitude), in South-Cameroon, has been choosen as a representative site for a study on the wet tropical area. The local climatic and hydrologic parameters, so that the chemistry of rain, soil and river waters have been observed along some annual cycles on this dense-forested basin (mean annual rainfall = 1700 mm, runoff =340 mm), where geological and pedological studies have also been performed. According to their sources, waters exhibit very different aspects and compositions: springs and deep groundwaters appear clear, with low suspended and dissolved contents (mean annual values: TDS = 9 mg/l including 10% DOC, TSS < 2 mg/l essentially mineral), whereas swamp and superficial soil waters, collected in the depressed zone, are coloured and more concentrated (TDS = 26 mg/l with DOC = 60%, TSS > 15 mg/l including more than 20% POC). Concerning their hydrogeochemical characteristics, the stream waters at the outlet of the catchment seem to be strongly influenced by the coloured swamp waters; however, the establishment of a budget is complicated by a recycling of most elements by plants and organic matter. From an hydrogeological study, the clear waters are those which have flowed at the bottom of the lateritic profile in the upper as in the lowest parts of the landscape. After a correction from the atmospheric inputs, it appears that their cation contents are very low, what means that the chemical weathering of the bedrock is practically ineffective under present conditions. In the coloured waters, the higher mineral concentrations could be related to the rather strong organic matter content of the depressed zone. The most probable origin of the cations analyzed in these waters, flowing in the superficial soil levels, is the intense leaching of sub-outcroping bedrock zones (few meters deep, against 20 to 40 meters in the general case), detected from geophysical studies. So, the dominant weathering processes according to the parts of this wet tropical catchment could be recognized from the compositions of different origin waters.
Index of EUG 10 Volume
Further EUG 10 Information
Index of the Journal of Conference Abstracts
Cambridge Publications Home Page
Last Updated on Wednesday, March 17, 1999.
© 1997 Cambridge Publications