Simulation of future stream alkalinity under changing deposition and climate scenarios

Models of soil and stream water acidification have typically been applied under scenarios of changing acidic deposition, however, climate change is usually ignored. Soil air CO2 concentrations have potential to increase as climate warms and becomes wetter, thus affecting soil and stream water chemistry by initially increasing stream alkalinity at the expense of reducing base saturation levels on soil exchange sites. We simulate this change by applying a series of physically based coupled models capable of predicting soil air CO2 and stream water chemistry. We predict daily stream water alkalinity for a small catchment in the Virginia Blue Ridge for 60 years into the future given stochastically generated daily climate values. This is done for nine different combinations of climate and deposition. The scenarios for both climate and deposition include a static scenario, a scenario of gradual change, and a scenario of abrupt change. We find that stream water alkalinity continues to decline for all scenarios (average decrease of 14.4 microeq L-1) except where climate is gradually warming and becoming more moist (average increase of 13 microeq L-1). In all other scenarios, base cation removal from catchment soils is responsible for limited alkalinity increase resulting from climate change. This has implications given the extent that acidification models are used to establish policy and legislation concerning deposition and emissions.     

Riparian zone influence on stream water chemistry at different spatial scales: a GIS-based modelling approach, an example for the Dee, NE Scotland

A geographical information system (GIS-ARC/INFO) was used to collate existing spatial data sets on catchment characteristics to predict stream water quality using simple empirical models. The study, based on the river Dee catchment in NE Scotland, found that geological maps and associated geochemical information provided a suitable framework for predicting chemical parameters associated with acidification sensitivity (including alkalinity and base cation concentrations). In particular, it was found that in relatively undisturbed catchments, the parent material and geochemistry of the riparian zone, when combined with a simple hydrological flow path model, could be used to accurately predict stream water chemistry at a range of flows (Q95 to > Q5) and spatial scales (1-1000 km2). This probably reflects the importance of the riparian zone as an area where hydrological inputs to stream systems occur via flow paths in the soil and groundwater zones. Thus, evolution of drainage water chemistry appears to retain the geochemical characteristics of the riparian area as it enters the channel network. In more intensively managed catchments, riparian land use is a further influential factor, which can be incorporated into models to improve predictions for certain base cations. The utility in providing simple hydrochemical models, based on readily available data sets, to assist environmental managers in planning land use in catchment systems is discussed.     

Impacts of warm winters and extreme rainstorms on the base consumption in a limed lake,southern Norway

The chemical composition of a limed lake, the two main inlets and the outlet was monitored during a period of 3 years. The winters of 1991-1992 and 1992-1993 were unusually warm while the winter of 1993-1994 was more normal. The lake surface water was wind exposed in the warm winters and as a consequence of frequent turnovers the acid input from the catchment mixed with the whole lake water body. In the winter of 1993-1994, the lake was ice-covered for approximately 4 months. During this period the drainage water from the catchment flowed to the outlet of the lake in the upper 2-3 m of the water column and only some of the acid input was neutralised. This is compared to a complete neutralisation in the winter of 1992-1993. The in-lake loss of alkalinity during this warm winter was approximately 29 microeq/l (November-June) compared to approximately 7 microeq/l lakewater in 1993-1994. Acid drainage from the catchment induced by an extraordinary rainstorm with heavy sea-salt deposition contributed to the in-lake alkalinity consumption in spring 1993. As winter temperatures above 0 degrees C and more frequent rainstorms may be common due to expected global warming, future increased lime consumption in-lakes may be projected in acidified areas as southern Norway.     

Partitioning the variation within the acid neutralizing capacity of surface waters in Scotland in relation to land cover, soil and atmospheric depositional factors

A method of decomposing the variation in the acid neutralizing capacity (ANC) of surface waters in Scotland is described. Using national datasets, a series of variables relating to 703 catchments across Scotland is divided into three components representing (i) land cover, (ii) soil and (iii) atmospheric deposition/altitude. Redundancy analysis (RDA) and (partial) redundancy analysis are used to quantify the amount of variation in ANC uniquely attributable to each of these components, independent of the effects of the others. The variation accounted for by covarying combinations of these components is also determined. Approximately 55% of the total variation in ANC across the 703 sites is explained by the variables representing catchment characteristics and atmospheric deposition. Of this, 8.5%, 2.4% and 6.9% are uniquely attributable to the land cover, soil and deposition/altitude components, respectively. A further 38% of ANC variation is associated with the covariation between components, with 18% accounted for by the combination of all three. Approximately 45% of the variation in ANC remains unexplained. The results reflect the integrated nature of catchment processes and demonstrate, for these data, that it is a combination of land cover, soil and deposition and altitude factors which most explain variation in freshwater ANC level. The approach offers a tool with which to assess the sensitivity of surface waters to acid deposition at a regional scale and provides a way of identifying regional differences in catchment response to acid loading.     

Modeling and monitoring of long-term acidification in an upland catchment of the Western Ore Mountains,SE Germany

Modeling and monitoring of acidification patterns in a limed forest catchment in the Ore Mountains, SE Germany are presented. A lumped-parameter model, MAGIC, satisfactorily reproduced the main parameters of stream water chemistry. Despite repeated whole-catchment liming, between 1993 and 1999 stream water pH increased from 4.3 to only 4.6, while calcium concentrations declined slightly. Stream water sulfate concentration declined from 687 to 396 microeq l(-1), and the pronounced effect of deposition decline during the 1990s was probably more important than liming.     

Catchment characteristics controlling the mobilization and potential toxicity of aluminium fractions in the catchment of the River Dee, northeast Scotland.

Elevated streamwater concentrations of aluminium have been associated with the onset of acidification, both by natural and anthropogenic means. This has important implications for river water quality. Concentrations of total, labile-inorganic and non-labile-organic fractions of aluminium were determined across the River Dee catchment, northeast Scotland. Fifty-nine subcatchments, chosen to reflect the variety of soils, parent materials and land use patterns across this major river system were sampled bi-weekly for 1 year. The distribution of aluminium was closely linked to factors of parent material and organic soil cover. Strong spatial and temporal relationships were observed between pH and all fractions of aluminium. Significant episodic peaks in aluminium occurred, these being especially pronounced when a storm event followed a period of dry weather. A weathering rate index utilizing the Na dominance of base cations was found to be a predictor of potential streamwater toxicity implied through Ca/inorganic aluminium ratios. It was demonstrated that Al was mobilized from acid headwater streams, whilst concentrations in the main stem remained much lower.     

Linked models to assess the impacts of climate change on nitrogen in a Norwegian river basin and FJORD system

Dynamically downscaled data from two Atmosphere-Ocean General Circulation Models (AOGCMs), ECHAM4 from the Max-Planck Institute (MPI), Germany and HadAm3H from the Hadley Centre (HAD), UK, driven with two scenarios of greenhouse gas emissions (IS92a and A2, respectively) were used to make climate change projections. These projections were then used to drive four effect models linked to assess the effects on hydrology, and nitrogen (N) concentrations and fluxes, in the Bjerkreim river basin (685-km(2)) and its coastal fjord, southwestern Norway. The four effect models were the hydrological model HBV, the water quality models MAGIC, INCA-N and the NIVA FJORD model. The downscaled climate scenarios project a general temperature increase in the study region of approximately 1 degrees C by 2030-2049 (MPI IS92a) and approximately 3 degrees C by 2071-2100 (HAD A2). Both scenarios imply increased winter precipitation, whereas the projections of summer and autumn precipitation are quite different, with the MPI scenario projecting a slight increase and the HAD scenario a significant decrease. As a response to increased winter temperature, the HBV model simulates a dramatic reduction of snow accumulation in the upper parts of the catchment, which in turn lead to higher runoff during winter and lower runoff during snowmelt in the spring. With the HAD scenario, runoff in summer and early autumn is substantially reduced as a result of reduced precipitation, increased temperatures and thereby increased evapotranspiration. The water quality models, MAGIC and INCA-N project no major changes in nitrate (NO(3)(-)) concentrations and fluxes within the MPI scenario, but a significant increase in concentrations and a 40-50% increase in fluxes in the HAD scenario. As a consequence, the acidification of the river could increase, thus offsetting ongoing recovery from acidification due to reductions in acid deposition. Additionally, the increased N loading may stimulate growth of N-limited benthic algae and macrophytes along the river channels and lead to undesirable eutrophication effects in the estuarine area. Simulations made by the FJORD model and the HAD scenario indicate that primary production in the estuary might increase up to 15-20%, based on the climate-induced changes in river flow and nitrate concentrations alone.     

The discriminatory power of two biomonitors of trace metal bioavailabilities in freshwater streams

The relative discriminatory powers of two trace metal biomonitors in European streams (the amphipod crustacean Gammarus fossarum and mayfly larvae of the genus Baetis-B. rhodani and B. vernus) are compared by discriminant function analysis using data from streams draining a zinc and lead mining area of Upper Silesia, Poland. The mayfly larvae (whether or not distinguished into the two separate species) had the better discriminatory power to distinguish between sites on the basis of local bioavailabilities of cadmium, copper, iron, lead and zinc. The bioavailabilities of the two metals lead and cadmium were the major local contributing factors to the stronger discrimination shown by the Baetis larvae.     

Analysis of catchment evapotranspiration at different scales using bottom-up and top-down approaches

Physically-based hydrological models are used to predict catchment water balance through detailed simulation of hydrological processes at small temporal and spatial scales. However, annual catchment water balance can also be easily and simply predicted using lumped conceptual model. Comparison between physically-based hydrological models and lumped conceptual models can help us understand the dominant factors on catchment water balance at different scales. In this paper, a distributed physically-based hydrological model (i.e., bottom-up approach) and a simple water-energy balance model (i.e., top-down approach) are used to predict actual evapotranspiration in nine sub-catchments, and the whole basin of the Luan River in northern China. Both simulations give very close values of annual evapotranspiration and show the same complementary relationship between actual and potential evapotranspiration at annual time scale. From the analysis at different time scales through comparison of the top-down and the bottom-up methods, it is shown that the annual catchment evapotranspiration is controlled mainly by annual precipitation and potential evapotranspiration, and the variability of soil water and vegetation becomes more important at a smaller time scale in the study areas. It is also known that the relationship between potential and actual evapotranspiration shows a highly nonlinear relationship at the annual and catchment scale but can be simplified to a linear relationship at hourly temporal and hillslope scales, which is commonly used in the physically-based hydrological models.     

River flow and associated transport of sediments and solutes through a highly urbanised catchment, Bradford, West Yorkshire

The hydrological characteristics of catchments become drastically modified in response to urbanisation. The total contributions and dynamics of runoff, suspended sediment and solutes may change significantly and have important implications downstream where they may affect flooding, instream ecological habitat, water quality and siltation of river channels and lakes. Although an appreciation of the likely hydrological changes is crucial for effective catchment management they are still poorly understood. In this paper we present data from a network of river monitoring stations throughout the heavily urbanised Bradford catchment, West Yorkshire. Sites are upstream, within and downstream of the highly urbanised central part of the catchment. Flow, turbidity (calibrated to suspended sediment concentration) and specific conductance (surrogate for solute concentration), logged at 15-min intervals, are presented for a 12-month period (June 2000 to June 2001). The total amounts and dynamics of flow, solute and suspended sediment transport were investigated. Estimated total flow and suspended sediment transport for the monitoring period were found to be high in response to the high total rainfall. Flow and sediment transport regimes were extremely 'flashy' throughout the catchment and became increasingly flashy in a downstream direction. Suspended sediment discharged from the Bradford subcatchment makes an important contribution to downstream sediment transport on the river Aire at Beal. Data suggest that the urbanised part of the Bradford catchment is extremely important in contributing solutes to the Beck (river). Although flow and sediment are also contributed to the Bradford Beck in the urbanised part of the catchment the data suggest that significant amounts may enter the combined sewer system and bypass the river. Understanding the spatial and temporal variations of flow and the transport of suspended sediment and solutes in rivers in urbanized subcatchments is crucial to their effective management and monitoring. Furthermore, this knowledge may be extremely important to the management and monitoring of downstream rivers in large scale mixed catchments.     

Analysis of catchment evapotranspiration at different scales using bottom-up and top-down approaches

Physically-based hydrological models are used to predict catchment water balance through detailed simulation of hydrological processes at small temporal and spatial scales. However, annual catchment water balance can also be easily and simply predicted using lumped conceptual model. Comparison between physically-based hydrological models and lumped conceptual models can help us understand the dominant factors on catchment water balance at different scales. In this paper, a distributed physically-based hydrological model (i.e., bottom-up approach) and a simple water-energy balance model (i.e., top-down approach) are used to predict actual evapotranspiration in nine sub-catchments, and the whole basin of the Luan River in northern China. Both simulations give very close values of annual evapotranspiration and show the same complementary relationship between actual and potential evapotranspiration at annual time scale. From the analysis at different time scales through comparison of the top-down and the bottom-up methods, it is shown that the annual catchment evapotranspiration is controlled mainly by annual precipitation and potential evapotranspiration, and the variability of soil water and vegetation becomes more important at a smaller time scale in the study areas. It is also known that the relationship between potential and actual evapotranspiration shows a highly nonlinear relationship at the annual and catchment scale but can be simplified to a linear relationship at hourly temporal and hillslope scales, which is commonly used in the physically-based hydrological models.     

Alternative sources for acidification of river water in Norway

In a recent monograph, in Norwegian¹, the author discussed various reasons for acidification of river water and the geochemistry of a proton budget for southern Norway. A short resumé will be given in this article. It is demonstrated that the acid of precipitation during the last decades is only causing a minor part of the acidity of the brook, river and lake waters. The greater part of the acidity in the fresh waters is the result of ion-exchange reactions in raw humus in the soil profiles of the catchment areas. The increased amount of biomass and sub-fossil humus due to changes in agriculture, cattle farming and forestry is discussed.     

Spatial heterogeneity of the spring flood acid pulse in a boreal stream network

Spatial and temporal patterns in streamwater acidity are ecologically important, but difficult to measure in parallel. Here we present the spatial distribution of streamwater chemistry relevant to acidity from 60 stream sites distributed throughout a 67 km² boreal catchment, sampled during a period of winter baseflow (high pH) and during a spring flood episode (low pH). Sites were grouped based on pH level and pH change from winter baseflow to spring flood. The site attributes of each pH group were then assessed in terms of both stream chemistry and subcatchment landscape characteristics. Winter baseflow pH was high throughout most of the stream network (median pH 6.4), but during the spring flood episode stream sites experienced declines in pH ranging from 0-1.6 pH units, resulting in pH ranging from 4.3-6.3. Spring flood pH was highest in larger, lower altitude catchments underlain by fine sorted sediments, and lowest in small, higher altitude catchments with a mixture of peat wetlands and forested till. Wetland-dominated headwater catchments had low but stable pH, while the spring flood pH drop was largest in a group of catchments of intermediate size which contained well-developed coniferous forest and a moderate proportion of peat wetlands. There was a trend with distance downstream of higher pH, acid neutralizing capacity (ANC) and base cation concentrations together with lower dissolved organic carbon (DOC, strongly negatively correlated with pH). This apparent scale-dependence of stream chemistry could be explained by a number of environmental factors which vary predictably with altitude, catchment area and distance downstream—most notably, a shift in surficial sediment type from unsorted till and peat wetlands to fine sorted sediments at lower altitudes in this catchment. As a result of the combination of spatial heterogeneity in landscape characteristics and scale-related processes, boreal catchments like this one can be expected to experience high spatial variability both in terms of chemistry at any given point in time, and in the change experienced during high discharge episodes. Although chemistry patterns showed associations with landscape characteristics, considerable additional variability remained, suggesting that the modeling of dynamic stream chemistry from map parameters will continue to present a challenge.     

Use of continuous water quality records for hydrograph separation and to assess short-term variability and extremes in acidity and dissolved carbon dioxide for the River Dee, Scotland

A combination of continuous (15-min) pH, conductivity and temperature measurements and fortnightly spot-sampled water quality data were used to examine temporal variability and extremes in river water quality in an upland Scottish river: the River Dee at Mar Lodge. An empirical relationship was established for Gran-alkalinity by multiple regression against flow and conductivity for the fortnightly data. Applying this relationship to the continuous data, an estimate of continuous Gran-alkalinity was calculated. The continuous Gran-alkalinity record was used as (1) a conservative tracer in a simple two-component mixing model to determine the relative proportions of near-surface runoff and deeper groundwater contributing to stream flow; (2) to deconvolute the contribution of weathering and sea-salt contributions to stream conductivity; and (3) to calculate the excess partial pressure of carbon dioxide in stream water. The episodic variations in pH, weathering and sea-salt conductivity and excess partial pressures of carbon dioxide (EpCO2) associated with high flow events in the River Dee suggest that hydrological pathways play an important role in determining stream chemistry. The results of the hydrograph separation indicate that groundwater provides an important contribution to stream flow, and that there are large and hydrologically active stores of groundwater within the upper River Dee catchment. Sea-salts have an important influence on stream conductivity, particularly with the onset of storm runoff following summer drought periods. This suggests that sea-salts are concentrated in the upper soil horizons by dry deposition and/or evapotranspiration. EpCO2 behaves non-conservatively and shows marked diurnal variability under low-flow conditions during summer, inducing diurnal pH variations, and indicating the importance of within-river biological processes. This study emphasises the very intermittent nature of water quality extremes with stream spates and the diurnal nature of biologically-induced responses. Fortnightly sampling programmes do not capture the range of high flow extremes, and with sampling undertaken during the working day, biological extremes occurring at night are also missed, introducing bias. This study shows the value of continuous measurements for infilling aspects of these intermittent extremes. However, it is also noted that the relatively simple patterns of response observed from the continuous measurements may well belie a much more varied response at the sub-catchment and hillslope scales, as local chemical and hydrological heterogeneities do occur. Thus, the two-component mixing model used is, in practice, based on catchment integrated values for a range of soil water and groundwater endmembers.     

Controls on the nutrient content of suspended sediment transported by British rivers

Recent studies of nutrient fluxes from river basins have emphasised the importance of sediment-associated transport. In order to develop an improved understanding of sediment-associated nutrient transport in UK rivers and of contrasts between individual rivers, attention has been directed to spatial and temporal variations in the nutrient (N and P) content of suspended sediment transported by four rivers, which embrace a representative range of British conditions. Bulk samples of suspended sediment were collected during storm events and these were analysed for both total N and P content and for individual fractions of these nutrients. Significant temporal and spatial variability in these various measures of the nutrient content of suspended sediment has been documented. Spatial variability has been linked to catchment characteristics, which in turn influence sediment sources. Patterns of temporal variability, which are reasonably consistent among the different rivers, have been related both to variations in other sediment properties and to hydrometeorological conditions.     

Climate variability and forecasting surface water recovery from acidification: modelling drought-induced sulphate release from wetlands

Climate-induced drought events have been shown to have a significant influence on sulphate (SO(4)(2-)) export from forested catchments in central Ontario, subsequently delaying recovery of surface waters from acidification. Field and modelling studies have demonstrated that water table drawdown during drought periods promotes oxidation of previously stored (reduced) sulphur (S) compounds in wetlands, with subsequent efflux of SO(4)(2-) upon re-wetting. Although climate-induced changes in processes are generally not integrated into soil-acidification models, MAGIC (Model of Acidification of Groundwater in Catchments) includes a wetland compartment that incorporates redox processes driven by drought events. The potential confounding influence of climate-induced drought events on acidification recovery at Plastic Lake, south-central Ontario (under proposed future S emission reductions) was investigated using MAGIC and two climate scenarios: monthly precipitation and runoff based on long-term means (average-climate scenario), and variable precipitation and runoff based on the past 20 years of observed monthly data (variable-climate scenario). The variable-climate scenario included several periods of summer drought owing to lower than average rainfall and higher then average temperature. Nonetheless, long-term regional trends in precipitation and temperature suggest that the variable-climate scenario may be a conservative estimate of future climate. The average-climate scenario indicated good recovery potential with acid neutralising capacity (ANC) reaching approximately 40 micromol(c)L(-1) by 2020 and 50 micromol(c)L(-1) by 2080. In contrast, the forecasted recovery potential under the variable-climate scenario was very much reduced. By 2080, ANC was forecasted to increase to 2.6 micromol(c)L(-1) from -10.0 micromol(c)L(-1) in 2000. Elevated SO(4)(2-) efflux following drought events (introduced under the variable-climate scenario) has a dramatic impact on simulated future surface water chemistry. The results clearly demonstrate that prediction of future water quality, using models such as MAGIC, should take into account changes or variability in climate as well as acid deposition.     

The Role of Geographical Information Systems in Urban Hydrological Modelling

'Geographical information systems'(GIS) technology holds considerable benefits to urban hydrology because of the spatial nature of catchment analysis. The availability of a spatio-temporal database of catchment characteristics can eliminate the adoption of simplifying assumptions, which reduce the quality of a research study. This research study was undertaken to demonstrate the need for accurate spatial and temporal accounting of land-use in an urbanizing catchment. The definition of the extent of GIS use, the approach to its integration into a study, and spatial database development, are crucial for eventual success.     

Modelling of phosphorus inputs to rivers from diffuse and point sources

The difference in timing of point and diffuse phosphorus (P) delivery to a river produces clear differences in the P concentration-flow relationship. Point inputs decrease in concentration with increasing river flow, due to dilution of a relatively constant input, whereas diffuse (non-point) load usually increases with river flow. This study developed a simple model, based on this fundamental difference, which allowed point and diffuse inputs to be quantified by modelling their contribution to river P concentration as a power-law function of flow. The relationships between total phosphorus (TP) concentration and river flow were investigated for three contrasting UK river catchments; the Swale (Yorkshire), the Frome (Dorset) and the Avon (Warwickshire). A load apportionment model was fitted to this empirical data to give estimates of point and diffuse load inputs at each monitoring site, at high temporal resolution. The model produced TP source apportionments that were similar to those derived from an export coefficient approach. For many diffuse-dominated sites within this study (with up to 75% of the annual TP load derived from diffuse sources), the model showed that reductions of point inputs would be most effective in order to reduce eutrophication risk, due to point source dominance during the plant and algae growing period. This modelling approach should provide simple, robust and rapid TP source apportionment from most concentration-flow datasets. It does not require GIS, information on land use, catchment size, population or livestock density, and could provide a valuable and versatile tool to catchment managers for determining suitable river mitigation options.     

Linking catchment characteristics and water chemistry with the ecological status of Irish rivers

Requirements of the EU Water Framework Directive for the introduction of ecological quality objectives for surface waters and the stipulation that all surface waters in the EU must be of 'good' ecological status by 2015 necessitate a quantitative understanding of the linkages among catchment attributes, water chemistry and the ecological status of aquatic ecosystems. Analysis of lotic ecological status, as indicated by an established biotic index based primarily on benthic macroinvertebrate community structure, of 797 hydrologically independent river sites located throughout Ireland showed highly significant inverse associations between the ecological status of rivers and measures of catchment urbanisation and agricultural intensity, densities of humans and cattle and chemical indicators of water quality. Stepwise logistic regression suggested that urbanisation, arable farming and extent of pasturelands are the principal factors impacting on the ecological status of streams and rivers in Ireland and that the likelihood of a river site complying with the demands of the EU Water Framework Directive, and be of 'good' ecological status, can be predicted with reasonable accuracy using simple models that utilise either widely available landcover data or chemical monitoring data. Non-linear landcover and chemical 'thresholds' derived from these models provide a useful tool in the management of risk in catchments, and suggest strongly that more careful planning of land use in Ireland is essential in order to restore and maintain water quality as required by the Directive.     

A model for predicting chloride concentrations in river water in a relatively unpolluted catchment in north-east Scotland

The River Dee is an oligotrophic soft water system, in the NE of Scotland, with a catchment area of approximately 2100 km2. The river rises in the Cairngorm Mountains and enters the North Sea at Aberdeen, approximately 140 km from its source. Water chemical quality data was collected every 2 weeks over 12 months for 59 sites distributed throughout the catchment. River water chloride concentrations increased significantly from west to east. In depth investigation of the relationship with distance from the coast revealed the significant difference in spatial distribution of river water chloride concentrations between upland and lowland/agricultural areas, suggesting the possible importance of agricultural practices to streamwater chloride concentrations. Thirty of the sample sites are independent and have been used to develop a simple model for prediction of streamwater Cl- concentration throughout the catchment. The model has been validated using data from the remaining sub-catchments. The model shows that mean Cl- concentration may be reliably predicted from distance from the coast and the percentage of improved grassland and arable land cover in each sub-catchment (r2 = 0.98). It is postulated that the land use effects may be partly due to the evolved link between landuse and catchment altitude characteristics, rather than just the direct effect of applied potassium chloride fertiliser on agricultural land. It was noted that there was insufficient forestry within the River Dee Catchment to reliably include % forest cover in the model.