HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS		QUICK SEARCH:   [advanced] Author: Keyword(s): Year:  Vol:  Page:

This Article Abstract Figures Only Full Text (PDF) Free Alert me when this article is cited Alert me if a correction is posted Services Similar articles in this journal Similar articles in Web of Science Alert me to new issues of the journal Download to citation manager Citing Articles Citing Articles via HighWire Citing Articles via Web of Science (1) Citing Articles via Google Scholar Google Scholar Articles by van Roosmalen, L. Articles by Sonnenborg, T. O. Search for Related Content PubMed Articles by van Roosmalen, L. Articles by Sonnenborg, T. O. GeoRef GeoRef Citation Agricola Articles by van Roosmalen, L. Articles by Sonnenborg, T. O. Related Collections Global Change Surface Hydrology

SPECIAL SECTION: GROUNDWATER RESOURCES ASSESSMENT UNDER THE PRESSURES OF HUMANITY AND CLIMATE CHANGE

Regional Differences in Climate Change Impacts on Groundwater and Stream Discharge in Denmark

^a Geological Institute, Univ. of Copenhagen, Øster Voldgade 10, 1350 Copenhagen K, Denmark
^b Geological Survey of Denmark and Greenland (GEUS), Øster Voldgade 10, 1350 Copenhagen K, Denmark
^* Corresponding author (lvr{at}geol.ku.dk).

All rights reserved. No part of this periodical may be reproduced or transmitted in any form or by any means, electronic or mechanical, including photocopying, recording, or any information storage and retrieval system, without permission in writing from the publisher.

Received 5 July 2006.

	ABSTRACT

TOP ABSTRACT INTRODUCTION Materials and Methods Results Discussion Conclusions REFERENCES

 
Regional impact studies of the effects of future climate change are necessary because projected changes in meteorological variables vary regionally, and different hydrological systems can react in various ways to the same changes. In this study the effects of climate change on groundwater recharge, storage, and discharge to streams are compared in two geologically and climatologically different regions in Denmark. Outputs are used for the periods 1961 to 1990 and 2071 to 2100 from a regional climate model representing the Intergovernmental Panel on Climate Change (IPCC) scenarios A2 and B2. A physically based, distributed hydrological model simulates changes in groundwater head, recharge, and discharge. Precipitation, temperature, and reference evapotranspiration increased for both the A2 and B2 scenarios. This results in a significant increase in mean annual net precipitation, but with decreased values in the summer months. The magnitude of the hydrological response to the simulated climate change is highly dependent on the geological setting of the model area. In the Jylland area, characterized by sandy top soils and large interconnected aquifers, groundwater recharge increased significantly, resulting in higher groundwater levels and increasing groundwater–river interaction. On Sjaelland, where the topsoil is dominated by low-permeability soils and the aquifers are protected by thick clay layers of regional extent, only minor changes in groundwater levels are predicted. The primary effect in this area is the change in stream discharge, caused by changes in drain flow and overland flow, with up to 50% increase in winter and 50% decrease in summer. This study shows the added value of studying different climate scenarios and hydrological systems, so that the simulated effects can be compared both qualitatively and quantitatively.

Abbreviations: GCM, general circulation model • GEUS, Geological Survey of Denmark and Greenland • IPCC, Intergovernmental Panel on Climate Change • masl, meters above sea level • ME, mean error • RCM, Regional Climate Model • RMS, root mean squared error

	INTRODUCTION

TOP ABSTRACT INTRODUCTION Materials and Methods Results Discussion Conclusions REFERENCES

 
Climate change is expected to have an impact on hydrological systems because of changes in precipitation, temperature, and reference evapotranspiration, which are the primary input variables for the terrestrial part of the hydrological cycle. Globally averaged surface air temperature is projected to increase between 1.4 and 5.8°C by the year 2100 relative to 1990 for the range of scenarios described in the IPCC Special Report on Emissions Scenarios (IPCC, 2000, 2001). Regional variations in the predicted temperature changes occur, with higher latitudes showing greater relative warming (IPCC, 2001). Changes in precipitation also show a large regional variation (IPCC, 2001). The most important sources and driving forces of climate change are at the regional scale. Considering that the political and technical measures to adapt to climate change also take place at a regional level (river basin level), regional-scale studies on global change are a matter of great concern (Varis et al., 2004).

General circulation models (GCMs) have been used to generate a number of projections of climate change based on greenhouse-gas and aerosol concentrations described in emissions scenarios. Statistical or dynamical downscaling of the climate change signals projected by the GCM is necessary since the horizontal resolution of the GCMs is too coarse to be used in regional hydrological models (Wood et al., 2004). Physically based regional climate models (RCMs) are used to dynamically downscale the signals by accounting for sub-GCM gridscale forcing (e.g., complex topographical features and land-cover inhomogeneity) and by enhancing the simulation of atmospheric circulations and climatic variables at fine spatial scales (Xu et al., 2005). The RCM is driven by sea-surface temperatures and driving atmospheric lateral boundary values from the forcing GCM (Déqué et al., 2005).

The use of several simulations from the same RCM with varying emissions scenarios and GCM forcings better reflects the uncertainty in the range of possible future climate change (Salathé, 2005). Unfortunately, the number of simulations from an RCM is often limited and only available for certain time slices because climate model simulations are computationally demanding. A number of studies have compared climate outputs for several RCMs, emissions scenarios, or GCM forcings (Rummukainen et al., 2001; Räisänen et al., 2004; Salathé, 2005; Van den Hurk et al., 2005).

Most hydrological impact studies that investigate the effects of climate change focus on surface water (Mimikou et al., 2000; Andréasson et al., 2004; Christensen et al., 2004; Graham, 2004; Kunstmann et al., 2004; Payne et al., 2004; Vanrheenen et al., 2004; Kleinn et al., 2005; Krysanova et al., 2005; Yuan et al., 2005; Kay et al., 2006). Only a few studies have shown the impact of climate change on groundwater. Scibek and Allen (2006) compared the potential impacts on two unconfined aquifers, one dominated by recharge and the other dominated by river–aquifer interaction. In the river-dominated aquifer, a 0.5-m increase in head near the river was predicted, while a 0.2-m increase in head was predicted away from the river. These results were generally consistent with the simplified steady-state approach used for the same aquifer in a previous study by Allen et al. (2004). The predicted changes were largely due to climate-change impacts on river discharge, which were incorporated into the model. In the recharge-dominated aquifer, groundwater levels decreased by 0.05 to 0.25 m. Thus, the responses of the two geologically similar aquifers differed substantially due to different predicted climate change and degree of interaction with surface water.

The effects of future climate change on groundwater recharge in the Rhône Valley, France, was studied using a stochastic weather generator to downscale GCM output in combination with a distributed, physically based hydrological model (Bouraoui et al., 1999). The study concluded that the doubling of atmospheric CO₂ would have negative impacts on groundwater resources because of an increase in the soil water deficit in the root zone, followed by larger groundwater withdrawal. Brouyère et al. (2004) found that future climate changes could result in a decrease in groundwater levels in a relatively small watershed in Belgium, whereas the seasonal variation did not change. The effects of climate change on a groundwater system in Texas resulted in a reduction of the aquifer's groundwater resources under climate scenarios with a doubling of the atmospheric CO₂ concentration (Loáiciga et al., 2000).

Thomsen (1989) investigated the effects of historical climate variability and change on groundwater recharge on the Danish island Samsø for the period 1865 to 1983. He noted that groundwater in Denmark responds to climatic variation and concluded that the groundwater recharge in most of western Denmark at the end of the nineteenth century was only half of the groundwater recharge during the period 1964 to 1983, due to much lower winter rainfall during the former period. Varanou et al. (2002) studied the changes in surface runoff, lateral inflow, and groundwater flow contributions to the total river flow using a conceptual model. A range of climate change scenarios was used, and the mean monthly groundwater flow showed a pronounced decrease for most months and all of the scenarios.

The objective of this study is to compare the effects of future climate change on groundwater recharge, storage, and discharge to streams for two regions in Denmark, representing climatologically and geologically varying parts of the country. The main reason to focus on groundwater resources is that 99% of the water supply in Denmark is based on groundwater. The advantage of using two study areas is that the effects of the same climate-change signals on varying climatological and hydrogeological settings can be compared. A transfer method was applied to link the climate-change signals predicted by the RCM to the meteorological input for the two study areas. This method gives meteorological data sets for the hydrological scenario simulations. The changes in input variables will alter the water balance and result in changes in groundwater recharge.

In this study the vadose zone was modeled in a spatially distributed and temporally varying way to take into account regional variations. Sharma (1989) performed a sensitivity analysis of groundwater recharge beneath a grassland and a pine plantation in Western Australia to changes in precipitation and evapotranspiration. The simulations showed that recharge was modified by a much larger proportion than rainfall, and this was greatly influenced by land use. Many climate scenario predictions for northern Europe and Scandinavia show the largest increases in precipitation during the winter months (Christensen and Christensen, 2007). An increase in recharge to groundwater was therefore expected because the soil and subsoil are already filled to field capacity during winter months in the current climate.

Determining how changes in available water affect the deep percolation and groundwater storage is not straightforward. It requires a detailed characterization of the subsurface, taking into account the spatial heterogeneity in the two study areas. A deterministic, distributed, physically based model called the DK model (Henriksen et al., 2003; Henriksen and Sonnenborg, 2003; Sonnenborg et al., 2003) was used to describe groundwater flow with a high spatial resolution. A distributed model provides the possibility to study the spatial distribution of the changes in hydrological variables within each region as well as the differences between regions.

The DK model used in this study is an integrated surface water–groundwater model, allowing us to study changes in the groundwater system and its discharge to rivers and drains. The baseflow is a large component of the flows in rivers and streams, which is especially important during the low-flow period when surface runoff and drain flow are limited. The DK model has documented predictive capabilities at a regional and catchment scale (Henriksen et al., 2003), making it a suitable tool for studying general trends in the effects of climate change on the hydrological system. The high-resolution (horizontal grid size 1 x 1 km), three-dimensional groundwater-flow model makes it possible to study the projected effects of climate change in various geological layers. Compared with most of the other studies mentioned above that concentrated on the effects of climate on groundwater, this study focuses on relatively large study areas of a few thousand square kilometers covering a large spatial and geological variation. This study focuses on both the water-table aquifer and the change in hydraulic head in deeper aquifers. These changes can alter the flow patterns in the groundwater system, affecting, for example, the flow path to groundwater abstractions, which is especially relevant for water supply.

We present in this paper the changes in meteorological input to the hydrological simulations for the two study areas and for both the IPCC A2 and B2 scenario simulations, as well as the resulting effects on groundwater recharge. The spatial and temporal changes in hydraulic head are shown for a number of model layers in the two study areas. Additionally, changes in stream flow are presented, and the contributions of the flow components, including baseflow, drain discharge, and overland flow to the rivers and streams, are studied.

	Materials and Methods

TOP ABSTRACT INTRODUCTION Materials and Methods Results Discussion Conclusions REFERENCES

 
Study Areas
Two areas with different meteorological and geological settings were selected for our study. One area is on the west coast of the peninsula Jylland, with an area of 5459 km². The other consists of the entire Sjaelland Island, covering an area of 7226 km², located in the eastern part of the country (Fig. 1 ). Both areas are relatively flat, with maximum elevations around 150 meters above sea level (masl). In Jylland the topography slopes gently from east to west. The central and northern part of the area is drained by the Skjern River system, while the southern part is drained by smaller stream systems. Land use is 68% agriculture, 22% wetlands, 8% forest, and 2% urban areas. The Jylland area is bounded by the North Sea to the west, while the water divide at the Jutland Ridge serves as the boundary to the east. The northern and southern boundaries are delineated on the basis of local water divides. Sjaelland is characterized by a glaciomorphological, hilly topography and is drained by numerous smaller streams. Land use is composed of 70% agriculture, 11% forest, 9% wetlands, and 10% urban areas. The island is bounded by the Baltic Sea along the entire periphery.

View larger version (51K): [in this window] [in a new window]   FIG. 1. Map of Denmark showing the study areas Jylland and Sjaelland and the transects AA' and BB'.

View larger version (41K): [in this window] [in a new window]   FIG. 2. (Upper) Geology of the Jylland study area along the transect AA' (Fig. 1). (Lower) Geology of the Sjaelland study area along the transect BB' (Fig. 1).

Climate and Hydrology
Denmark has a temperate coastal climate, and the weather is dominated by fronts caused by polar cyclones above the North Atlantic because of its location in the belt of prevailing westerlies. Maximum precipitation is observed in autumn, while minimum precipitation occurs in spring. The weather is highly dependent on the wind direction because of the proximity to both the ocean and the European continent. The dominant westerly wind results in mild winters and chilly summers with variable weather and often with rain and showers. Winds from the south and east are influenced by Continental weather systems characterized by low temperatures in winter and high temperatures in summer.

The present climate in the two areas was evaluated on the basis of observed data from the climate grid for Denmark, where daily values for temperature, precipitation, and reference evapotranspiration (potential evapotranspiration for a well-watered grass of uniform height) at a 40-km resolution have been used (Scharling, 1999). Precipitation was corrected for wetting and aerodynamic effects using the standard correction values of Allerup et al. (1998). Reference evapotranspiration was calculated using the Makkink (1957) formulas. Temperature and reference evapotranspiration are slightly higher on Sjaelland compared with Jylland (Fig. 3 ). However, the most significant climatic difference between the two areas is the amount of precipitation. Annual precipitation in Jylland is almost 300 mm higher than on Sjaelland. The difference is greatest in the winter period, from October to March, when monthly precipitation is more than 50% higher in Jylland than in Sjaelland.

View larger version (47K): [in this window] [in a new window]   FIG. 3. Annual mean temperature T (°C), annual mean precipitation P (mm), and annual mean reference evapotranspiration ETref (mm) for the model areas Jylland and Sjaelland and the period 1990–2004. These values are annual means of the meteorological input used for the current climate hydrological simulations.

The DK model is a transient, spatially distributed groundwater–surface water model based on the MIKE SHE code (Abbott et al., 1986; Refsgaard and Storm, 1995). Three-dimensional groundwater flow is solved based on Darcy's law using detailed input on the distribution of the geology and associated hydraulic properties. The flow from the saturated groundwater zone to the rivers, here referred to as baseflow, is one of the three possible contributions to the total discharge in the rivers (Fig. 4 ). Coupled exchange of water between the groundwater zone and the rivers is described using a Darcy type relationship between flux and head difference, where a leakage coefficient that depends on the river bottom permeability acts as the controlling parameter. Drain flow is conceptualized as a linear reservoir and simulates flow through small streams, ditches, and drain pipes not described explicitly by other modules of the model. The drain flow is calculated on the basis of distributed information about drain level and drainage coefficient. When the groundwater level rises above the level specified for the drains, water is routed to the nearest river at a rate controlled by the drainage coefficient and the difference between the groundwater level and the drainage level. The last possible inflow to rivers consists of overland flow. Overland flow is generated when water is ponded on the ground surface either as a consequence of high groundwater level or because the rainfall rate exceeds the infiltration capacity. Overland flow is described using the diffusive wave approximation of the Saint Venant equations and requires information about roughness of the surface expressed through a Manning coefficient.

View larger version (12K): [in this window] [in a new window]   FIG. 4. Schematic representation of inflows to a river from the groundwater zone (baseflow, QBF), drains (drain flow, QDF), and flow on the ground surface (overland flow, QOF).

The DK model is divided into several submodels, including the Jylland and Sjaelland models. All major streams and rivers are described by the models. In both models, the subsurface was resolved using a horizontal discretization of 1 km. In the vertical plane Jylland is discretized using 16 model layers, whereas 9 model layers are used in Sjaelland. To avoid overparameterization of the models, uniform parameter values are used for the geological units defined in the hydrostratigraphical model (Fig. 2). Table 1 shows the horizontal hydraulic conductivities and the parameters controlling overland flow, drain flow, and baseflow for the Sjaelland and Jylland models. The initial best estimates of hydraulic parameter values and expected ranges were assessed using data from previous fieldwork (unpublished results from pump tests), previous modeling results, and literature (Henriksen et al., 2003). In the Sjaelland model four hydrofacies were defined, including an upper high permeable till, interchanging layers of Quaternary clayey till and sand, and pre-Quaternary chalk. The pre-Quaternary chalk aquifer was attributed distributed values for conductivity, of which the spatial distribution was based on measured values of transmissivity. Calibration of the chalk was performed by adjusting a multiplication factor that acts on all the chalk conductivity values. In Jylland five geological units were defined (Quaternary clay, Quaternary sand, Miocene quartz sand, Miocene mica sand, and pre-Quaternary clay and silt).

The models were calibrated in two steps. First, a steady-state version of the model was optimized using the automatic parameter estimation procedure UCODE (Poeter and Hill, 1999). Here the hydraulic conductivities of the groundwater zone and the leakage coefficient controlling the magnitude of baseflow were estimated. Subsequently, a transient version of the model was calibrated by trial and error against observations of hydraulic head and stream discharge from the calibration period. The parameters found in the steady-state automatic optimization were transferred to the transient model (Sonnenborg et al., 2003), while the parameters controlling the dynamics of the model response, such as storage coefficients of the saturated zone and drainage coefficient, were estimated. The calibration procedure was followed by a split-sample validation test against measurements of hydraulic head and stream discharge from the subsequent validation period.

Recorded groundwater heads from GEUS's well log archive from 1970 to 1999 were used for calibration and validation purposes. In the Sjaelland model the calibration period included the 3-yr period 1988 to 1990, and the validation tests were performed for the 6-yr period 1991 to 1996. Head data from 4439 wells (one observation per well), with screens distributed over nine computational layers, were used as the measure of observed steady-state hydraulic heads. The density of head data was highest in the bottom layer, representing the chalk. Head values ranged from mean sea level to approximately 100 masl in the central part of the island. Daily stream flow data from 26 river gauging stations for the period 1988 to 1996 were used for calibration and validation (Fig. 5 ). In the Jylland model the calibration period was 1991 to 1995, while the validation period was 1996 to 1999. Hydraulic head data from 1715 wells with screens distributed over the upper 15 model layers were used in the steady-state automatic calibration. The density of head data decreased from several hundred in the top layers to only four in layer 15. The hydraulic heads varied from about 100 masl at the eastern part of the area to mean sea level at the western boundary. Additionally, discharge values from 26 gauging stations (Fig. 5) were included as targets.

View larger version (48K): [in this window] [in a new window]   FIG. 5. Location of discharge stations used for calibration and validation in Jylland and Sjaelland. The two stations marked in red indicate the stations used in the Results section on stream flow.

[1]

[2]

where n is the number of observations, and H_obs,i and H_sim,i are the observed and simulated heads, respectively. The Nash-Sutcliffe model efficiency coefficient (E) (Nash and Sutcliffe, 1970) was used to assess the river discharge simulation:

[3]

where Q_obs^t Q_sim^t and are the observed and simulated river discharges at time t, respectively. The RMS values were calculated for each of the model layers, while E was calculated for all gauging stations. This resulted in a large number of performance criteria, so global values based on averages over the calculation layers (weighted by the number of observations per layer) and the river gauging stations (simple arithmetic mean) were used in the calibration and validation process. A criterion F_bal was used to express the ability of the model to simulate average discharge for a particular river gauging station:

[4]

where and are the average observed and simulated discharges, respectively.

The RMS values for the groundwater heads should be compared to uncertainty of the observed head data. Henriksen and Sonnenborg (2005) estimated the aggregated uncertainty of the observed head data relative to model simulations at a 1-km scale to have a value of 3.1 m, corresponding to RMS values of 6.2 m at the 95% confidence levels. This aggregated uncertainty includes factors such as measurement errors, errors in assessing the elevation of the well, scaling errors due to the 1-km model grid size and to geological heterogeneity, and nonstationarity in the observed data. Additionally, if high-quality synoptic head observations had been available instead of the rather crude data from the historical well log archive, the RMS performance criteria used for the validation would have been reduced accordingly. Smaller deviations could then have been expected because the observed data were less uncertain, but the model result would not necessarily be closer to the unknown reality (Henriksen et al., 2003). Thus, simulated RMS values of 6.2 or less are considered satisfactory.

The validation results for the two models are presented in Table 2. The average groundwater level, expressed through the ME, is reproduced satisfactorily by both models with a maximum deviation in the Sjaelland model of 0.5 m. The RMS value of the Sjaelland model on the other hand is relatively high, indicating that the ability of the model to reproduce point measurements of hydraulic head is associated with uncertainty. The RMS value for Jylland of 3.8 m is considered to be satisfactory. The relatively high RMS value of the Sjaelland model compared with the Jylland model partially reflects the more complex geology on Sjaelland, dominated by heterogeneous deposits of tills and glacial outwash sand from the last ice age of the Quaternary period. Such formations are difficult to represent in detail in a regional-scale model. The geology in Jylland is dominated by Miocene marine deposits correlated over large distances and characterized by less small-scale heterogeneity. Therefore, these formations are better captured by a regional model.

The validation results for river discharge are generally excellent, with E values of approximately 0.8. Also the overall water balance expressed by F_bal indicates that the models can capture the average flow dynamics adequately (see Table 2).

The calibration period used for Sjaelland was relatively dry, with precipitation values in 1988 and 1989 of 630 and 508 mm, respectively, compared with the annual precipitation during the period 1990 to 2004 (Fig. 3). The validation test was performed for 1991 to 1996, including the very wet years 1993 (840 mm) and 1994 (904 mm). Hence, the Sjaelland model was calibrated on a relatively dry period and then validated against much wetter years in which the annual precipitation was 78% higher for the wettest year compared to the driest year.

The Jylland model was calibrated for the period 1991 to 1995, of which 1994 was a very wet year (precipitation of 1255 mm) and 1991 a relatively dry year (895 mm) (Fig. 3). The Jylland model was validated for the period 1996 to 1999, of which 1996 and 1997 were dry years (674 and 839 mm, respectively) while 1998 and 1999 were very wet years (1239 and 1315 mm, respectively). Also for Jylland a considerable range in average annual precipitation occurred with an extremely dry year and a very wet year during the validation period. The validation shows the ability of the models to simulate the hydrological responses for both very wet years and very dry years, spanning the range in precipitation for the current climate.

Regional Climate Model
The climate change signals from a GCM were downscaled dynamically using the RCM HIRHAM developed by the Danish Meteorological Institute (Christensen et al., 1996, 1998). The RCM output can be downloaded from the PRUDENCE project (Christensen and Christensen, in press) website (http://prudence.dmi.dk/). In this experiment HIRHAM was nested in the Hadley Centre GCM HadAM3H, simulating the future emissions scenarios IPCC Special Report on Emissions Scenarios A2 and B2 (IPCC, 2000). In the A2 scenario, atmospheric CO₂ concentrations reach 850 ppm in the year 2100 (IPCC, 2001) in a world characterized by high population growth, medium gross domestic product growth, high energy use, medium-to-high land-use changes, low resource availability, and slow introduction of new and efficient technologies (IPCC, 2000). In the B2 scenario, atmospheric CO₂ concentrations reach 600 ppm in the year 2100 (IPCC, 2001) in a world with continuously increasing global population at a rate lower than A2, intermediate levels of economic development, moderate energy use, moderate land-use changes, and more rapid and more diverse technological change than in the A2 storyline (IPCC, 2000).

Daily precipitation and temperature data are direct HIRHAM output. Daily reference evapotranspiration was estimated using the FAO Penman–Monteith equation (Allen et al., 1998) and RCM output such as incoming and outgoing, short- and long-wave radiation, temperature, water vapor pressure, and wind speed:

[5]

where ET₀ is reference evapotranspiration (mm d^–1), R_n is net radiation at the crop surface (MJ m^–2 d^–1), G is soil heat flux density (MJ m^–2 d^–1), T is mean daily air temperature at 2 m height (°C), u₂ is wind speed at 2 m height (m s^–1), e_s – e_a is saturation vapor pressure deficit (kPa), is the slope of the curve showing the saturated vapor pressure as a function of temperature (kPa °C^–1), and is psychrometric constant (kPa °C^–1).

The reference evapotranspiration described in Eq. [5] is the potential evapotranspiration for a hypothetical grass reference crop with an assumed crop height of 0.12 m, a fixed surface resistance of 70 s m^–1, and an albedo of 0.23 (Allen et al., 1998). Reference evapotranspiration can be converted to a potential evapotranspiration by multiplying with a surface coefficient (crop coefficient). In this study the relative change in evapotranspiration between the current and future climate was used (see following section), so it is unnecessary to include the crop coefficient because it is divided by itself.

HIRHAM output for the A2 scenario is available at a horizontal grid resolution of 12 km and for the B2 scenario at a 50-km resolution. Transient simulations using the RCM were not run for the entire period 1961 to 2100 because these simulations are computationally very demanding. Instead, two 30-yr time slices are available, one representing the climate in the period 1961 to 1990 (control) and the other representing future climate in 2071 to 2100 (scenario). There is much uncertainty as to how humans will adapt to climate change and perhaps change the emissions scenarios. For this reason it should be noted that the A2 and B2 scenarios used in this study are only two plausible descriptions of how future emissions might develop and are not any more likely than any other scenario. The developments described for the second half of the twenty-first century are even more uncertain than for the first part of the century, due to the longer time span. There were two main reasons for choosing a period so far in the future (2071–2100): (i) the change signals from the scenarios are generally less strong for the first half of the century, and it is therefore difficult to distinguish a clear climate change signal for earlier periods due to natural variability, and (ii) the change signals from the individual scenarios are difficult to distinguish from each other before the end of the century. Although interpolation also has its limitations, it is more reliable to interpolate the results for the period 2071 to 2100 to an earlier period in the century than it is to extrapolate from this earlier period to the period 2071 to 2100, due to the nonlinearity of the changes and the changes being larger at higher greenhouse-gas concentrations.

Delta Change Method
Meteorological output from RCMs cannot be used directly as input for hydrological simulations due to biases between the simulated meteorological variables for the current climate and observed values (Jones et al., 2004). Precipitation output is especially subject to these biases. The delta change approach is a commonly used method to bypass these biases by transferring the signal of climate change derived from a climate model simulation to an observed database (Hay et al., 2000). The climate-change signals are derived from the difference in mean meteorological variables for the periods representing the control and scenario climate. This method is more robust than a simple sensitivity analysis of the meteorological variables, where the changes in the variables are selected independently from the others, because the factors result from a climate model simulation where the changes in the meteorological variables have been simulated in a physically based system, where feedback processes are included.

A drawback of this method is that the baseline for the future simulations is an observed database, so information on the changes in variability and extremes in the future climate, as simulated by the climate model, is lost. Also, the cross-correlations between climate variables remain the same as for the baseline climate, so that a change in temperature, for example, will result in the same change in precipitation in the future climate as it did in the current climate, whereas this relationship might be altered for future climate change. An uncertainty is introduced in the climate-change impact study because the observed database has a large influence on the way the climate-change signals are transferred to the hydrological scenario simulation. On the other hand, the use of an observed database has the advantage that a consistent set of input data is used, whereas the use of climate model output directly could result in unrealistic dynamics in input variables due to climate model variance.

In this study the means were calculated on a monthly basis for each 30-yr period of climate output (current climate, A2 scenario, and B2 scenario). The 12 delta change factors for each meteorological variable, precipitation, temperature, and reference evapotranspiration were used to perturb the observed database, which is used in the hydrological simulation for the current climate. This method is described by the following equation:

[6]

where P is the precipitation input to the hydrological model for the A2 and B2 scenario runs, P_obs is the observed precipitation representing the current climate, and the suffixes i and j stand for the i^th day of the j^th month. _P is the delta change factor, which is calculated using the expression

[7]

where (j) is the precipitation in month j averaged for the 30-yr control or scenario period as simulated by the RCM. The index scen stands for the scenario period (2071–2100), and the index cont indicates the control period (1961–1990). For reference evapotranspiration the same method is applied to calculate the delta change factors and to generate the input for the scenario simulations. The delta change factors for both precipitation and reference evapotranspiration represent relative changes.

For temperature the absolute change is used for the delta factor, as follows:

[8]

where _T is given by

[9]

The delta change factors in Table 3 represent spatial averages for the whole of Denmark, corresponding to 256 cells covering Danish land surface for the A2 scenario output with a 12-km horizontal resolution. The B2 scenario output is at a 50-km resolution, resulting in an average of 16 cells. The delta change factors are used for both study areas, so no regional differences in scenario input occur as a result of the use of varying climate change signals.

Delta change factors used in hydrological impact studies vary due to the use of different forcing GCMs, climate scenarios, timescales, downscaling methods, and areas. Déqué et al. (2005) compared the climate change signal for temperature and precipitation for four GCMs and nine RCMs, forced by the same GCM, with model areas over Europe. For temperature, the GCMs and RCMs behaved similarly, and it was concluded that the results for seasonal mean temperature over Europe could be trusted. For precipitation, however, the bias with observed current climate was twice as large as the signal produced by the models. Although the models behaved differently for the current climate for precipitation, the authors still expressed confidence in the PRUDENCE results because the models offered similar climate change signals (delta change factors) to IPCC A2 scenario forcing.

A study focusing on RCM simulations over the land area of Norway, Sweden, Finland, and Denmark for the B2 scenario in 2050, but using other forcing GCMs than used in this study, showed an increase in average annual temperature of 1.5°C and a relative increase in average annual precipitation of 5% for Denmark (Rummukainen et al., 2003). An increase in annual evapotranspiration of 4% was given for the whole of Scandinavia, although it was not explained whether this was actual evapotranspiration as calculated by the RCMs or reference evapotranspiration and how this value was calculated. A study focusing on river runoff in Denmark also used data from the RCM HIRHAM for the A2 scenario and the scenario period 2071 to 2100, but with ECHAM/OPYC as the forcing GCM (Andersen et al., 2006). An increase in mean annual temperature of 3.2°C and an increase in mean annual precipitation of 5% were used. A complete discussion on the ranges in climate predictions for Denmark is not within the scope of this article, but one must keep in mind that this scenario study presents only a limited representation of possible climate changes. For this reason the outcomes of this study should be seen as an indication of a trend and not as precise predictions of future water availability.

	Results

TOP ABSTRACT INTRODUCTION Materials and Methods Results Discussion Conclusions REFERENCES

 
The observed annual meteorological input data used as the baseline for the hydrological simulations are presented in Fig. 3. First, the meteorological input data, generated by perturbing the observed climate data by applying the delta change method, are presented. Then the output from hydrological simulations using the same model setup for each study area and either the observed meteorological input for the current climate or the perturbed meteorological data for the climate scenarios A2 and B2 for the time period 2071 to 2100 are presented. The results were extracted from transient simulations for the period 1990 to 2004 after a warm-up period from 1971 to 1989, to minimize the effect of initial values on the results. The results presented are the average for the 15-yr simulation period for that specific variable. To isolate the effect of climate changes in the natural hydrological system, groundwater abstraction for water supply and irrigation was excluded from the model simulations.

Meteorological Input
Temperature
In the hydrological simulations temperature directly influenced a number of processes, the most important of which are whether precipitation falls as snow, sleet, or rain; the calculation of snow accumulation and melting; and the magnitude of estimated potential evapotranspiration. Table 4 presents the average annual temperature spatially averaged over the model areas for Jylland and Sjaelland. The A2 scenario resulted in an increase of 3.2 and 3.1°C for Jylland and Sjaelland, respectively, and an increase of 2.2°C for the B2 scenario for both areas.

View larger version (24K): [in this window] [in a new window]   FIG. 6. Average monthly temperature for the current climate and the IPCC A2 and B2 scenarios (2071–2100) for the Jylland and Sjaelland study areas.

The seasonal changes in precipitation are most pronounced for the winter months December to February, in which an increase of about 30 to 50 mm mo^–1 was simulated for Jylland and 20 to 30 mm mo^–1 for Sjaelland (Fig. 7 ). In the current climate the maximum precipitation is observed in autumn, whereas a shift of the maximum to the winter months appears in the future climate scenarios. The changes throughout the rest of the year are relatively small, although both scenarios show a decrease in summer precipitation. The largest decrease in precipitation occurs in August, with values about 20 to 30 mm below those for the current climate for both regions.

View larger version (22K): [in this window] [in a new window]   FIG. 7. Average monthly precipitation for the current climate and the IPCC A2 and B2 scenarios (2071–2100) for the Jylland and Sjaelland study areas.

The monthly values for reference evapotranspiration show that the evapotranspiration for both climate scenarios is higher than in the current climate for all months (Fig. 8 ). Large absolute increases occur during the months May to October, whereas the low evapotranspiration in the winter results in small absolute changes during these months. Especially in August, a large increase of about 25 and 18 mm occurs for the A2 and B2 scenarios, respectively.

View larger version (21K): [in this window] [in a new window]   FIG. 8. Average monthly reference evapotranspiration (ETref) for the current climate and the IPCC A2 and B2 scenarios (2071–2100) for the Jylland and Sjaelland study areas.

The seasonal changes in net precipitation are shown in Fig. 9 . The future net precipitation is notably higher during the period December to March for both areas. From April to June the differences between the net precipitation for the climate scenario simulations and the current climate are relatively small, while a decrease in net precipitation was simulated for July to October for both scenarios. Especially in September a large reduction in net precipitation occurs, where the A2 scenario results in a value equal to the August level for the current climate. In November the B2 scenario shows a relatively large increase in net precipitation, whereas the A2 value is close to that for the current climate.

View larger version (22K): [in this window] [in a new window]   FIG. 9. Average monthly net precipitation from the root zone for the current climate and the IPCC A2 and B2 scenarios (2071–2100) for the Jylland and Sjaelland study areas.

View larger version (64K): [in this window] [in a new window]   FIG. 10. Average change in mean annual groundwater head relative to the 1990–2004 period in model layer 1 and 5 of the Jylland model area when comparing the IPCC A2 or B2 scenario (2071–2100) to the current climate.

Seasonal Changes in Groundwater Head
The changes in mean annual groundwater head in layer 1 for the summer and winter show that the seasonal differences are not very large compared with the changes in mean annual groundwater head (results not shown). For example, the spatially averaged increase in groundwater level for the A2 scenario is 0.18 m during summer and 0.27 m during winter compared with the annual average increase of 0.20 m. For the B2 scenario the average increases are 0.32 m (summer) and 0.45 m (winter) compared with the annual average increase of 0.36 m.

Groundwater Storage in Sjaelland
For the Sjaelland model, layers 3 and 9 were used to analyze the effects of future climate change on groundwater. Layer 3 represents the uppermost groundwater reservoir, which constitutes a mostly confined aquifer. Layer 9 is the lowest layer of the model and the primary groundwater reservoir in the area, consisting of pre-Quatenary carbonate formations. The depth of layer 9 varies considerably throughout the model area as can be seen in the cross-section in Fig. 2b.

Absolute Changes in Mean Annual Groundwater Head
The spatially averaged change in mean annual groundwater head for the A2 scenario is negligible, with values of –0.01 m for layer 3 and 0.02 m for layer 9. On the other hand, the B2 scenario results show a moderate increase in mean annual groundwater head of 0.20 m for both layers. The absolute changes in mean annual groundwater head are smaller than the absolute value of the mean error of 0.53 m, estimated through the validation of the Sjaelland model. Even though insignificant changes in spatially averaged, mean, annual groundwater head are predicted for the A2 scenario, a geographical variation throughout the area still occurs. For example, 15% of the area shows an increase >0.25 m in layer 1 and 11% in layer 9 (Table 5). Almost no areas show an increase >1.0 m for the A2 scenario. In the B2 scenario approximately 27% of the area shows increases >0.25 m and 5% shows >1.0 m.

Spatial Distribution of Changes in Groundwater Heads
The geographical distribution of the change in mean annual groundwater head is shown for layer 3 and 9 for both scenarios in Fig. 11 . In the northeastern part of Sjaelland, the mean annual groundwater head increases for both scenarios, and the B2 scenario shows increases >0.50 m for a large part of northeast Sjaelland. For the A2 scenario the results show a decrease in mean annual groundwater head in many places in the southwestern part of Sjaelland, with decreases most apparent for layer 3. For the B2 scenario the mean annual groundwater head in this area is affected only slightly by the change in climate input.

View larger version (79K): [in this window] [in a new window]   FIG. 11. Average change in mean annual groundwater head relative to the 1990–2004 period in model layer 3 and 9 of the Sjaelland model area when comparing the IPCC A2 or B2 scenario (2071–2100) to the current climate.

View larger version (46K): [in this window] [in a new window]   FIG. 12. Average change in groundwater recharge relative to the 1990–2004 period for the Sjaelland model area when comparing the IPCC A2 or B2 scenario (2071–2100) to the current climate.

Jylland
Absolute and Relative Changes in Total Stream Flow
The average stream discharge increases for both climate scenarios. The average current climate discharge is 24.0 m³ s^–1, for the A2 scenario it is 27.0 m³ s^–1, and for the B2 scenario it is 28.8 m³ s^–1. The relative increases are 13% (A2) and 20% (B2).

Seasonal Changes in Total Stream Flow
The seasonal variation in stream discharge is shown in Fig. 13 . For the A2 scenario all the months except September and October show an increase in discharge, with the period January to March showing the largest increases. The maximum absolute increase of 11 m³ s^–1 occurs in February, and the largest absolute decrease of 2 m³ s^–1 occurs in September. Almost the same seasonal pattern can be seen for the B2 scenario, but with overall higher discharges.

View larger version (20K): [in this window] [in a new window]   FIG. 13. Average monthly stream discharge and the relative change in stream discharge at the station shown in Fig. 5 in the Jylland model area when comparing the IPCC A2 or B2 scenario (2071–2100) to the current climate (1990–2004).

View larger version (38K): [in this window] [in a new window]   FIG. 14. Average monthly discharge and the relative change in discharge of the three flow components, baseflow, drain flow, and overland flow, at the station shown in Fig. 5 in the Jylland model area when comparing the IPCC A2 or B2 scenario (2071–2100) to the current climate (1990–2004).

Drain Flow
The peak in stream discharge for the current climate coincides with the peak in drain flow, and the drain flow shows relatively large seasonal variation, with the maximum discharge occurring during the winter months. The A2 and B2 scenarios show that the drain flow increases about 7.5 m³ s^–1 in January and February, which is caused by groundwater table rising above the drain level (located at a half-meter below the surface) in a larger part of the study area compared with the current climate simulation.

Overland Flow
A relatively large increase in overland flow occurs, but the absolute value of the contribution to total flow is small. With respect to summer discharge, overland flow shows the largest decreases from July to September in the A2 and B2 simulations. The largest reduction in drain flow and baseflow takes place in September and October. In September, when the largest decrease in total flow occurs, the baseflow, drain flow, and overland flow for the A2 scenario decrease by 0.87 m³ s^–1, 1.27 m³ s^–1, and 0.94 m³ s^–1, respectively. This shows that all three components contribute significantly to the decrease in discharge in this period.

Sjaelland
Absolute and Relative Changes in Total Stream Flow
Sjaelland also shows an increase in discharge for both scenarios, with the B2 scenario resulting in the largest increases (Fig. 15 ). The average discharge for the current climate, A2 scenario, and B2 scenario is 6.2 m³ s^–1, 7.1 m³ s^–1, and 7.9 m³ s^–1, respectively. The relative increases for the A2 and B2 scenario are 16 and 28%, respectively. The largest absolute increases occur from December to February.

View larger version (15K): [in this window] [in a new window]   FIG. 15. Average monthly stream discharge and the relative change in stream discharge at the station shown in Fig. 5 in the Sjaelland model area when comparing the IPCC A2 or B2 scenario (2071–2100) to the current climate (1990–2004).

View larger version (38K): [in this window] [in a new window]   FIG. 16. Average monthly discharge and the relative change in discharge of the three flow components baseflow, drain flow, and overland flow at the station shown in Fig. 5 in the Sjaelland model area when comparing the IPCC A2 or B2 scenario (2071–2100) to the current climate (1990–2004).

Drain Flow
The changes in drain flow for Sjaelland show the same temporal distribution as for Jylland, but the relative increases in drain flow during winter for Sjaelland are smaller. The drain discharge on Sjaelland also shows decreases earlier in the year, and the decreases during summer are larger, especially for the A2 scenario.

Overland Flow
Overland flow on Sjaelland shows larger increases in winter than in Jylland, reflecting that a larger part of the increase in winter precipitation for the future scenarios is lost to runoff in Sjaelland and contributes less to groundwater recharge than in Jylland.

	Discussion

TOP ABSTRACT INTRODUCTION Materials and Methods Results Discussion Conclusions REFERENCES

 
Uncertainties
This study gives insight into the effects of future climate change on groundwater and river discharge in Denmark and indicates the trends that can be expected as a result of the changes in meteorological input. The results presented should be interpreted as trends and not as an accurate quantitative prediction of hydrological changes. The uncertainties in the predictions are a function of numerous sources, including future greenhouse-gas emissions, the GCM used, the downscaling performed by the RCM, the transfer method applied to convert the output from the climate model to input to the hydrological model, and the hydrological model.

The uncertainty related to future greenhouse gas emissions is a result of lack of knowledge about the future human development and the sensitivity of the climate system to all the feedback mechanisms influenced by increasing emissions. Two climate scenarios were used in this study to compare the effects on the hydrological system for different emissions concentrations, but these two scenarios do not cover the full range of climate scenarios. The A2 and B2 scenarios represent relative high and low emissions, respectively, and it may therefore be reasonable to assume that the range in possible hydrological impacts of climate change is captured.

Another category of uncertainties relates to the signals from the climate model. In this study the output from one GCM was used to establish the boundary conditions to one RCM. The forcing GCM has a large influence on the resulting climate output (Salathé, 2005) and the downscaling method. In this case, dynamical downscaling using one RCM also influences the climate signal derived from the GCM. If a thorough analysis of the combined effects of global and regional climate model uncertainty is desired, then an ensemble of results produced by combining different GCMs and RCMs should be performed. The resulting climate predictions could subsequently be used as input in the hydrological analysis. At the moment, however, this kind of climate modeling is not available, but such outputs are expected to be generated in the near future through the EU project ENSEMBLES (contract number GOCE-CT-2003-505539; http://ensembles-eu.metoffice.com/).

Most climate-change impact studies use an uncoupled simulation of the hydrological model, requiring a transfer method to transfer the climate-change signal from the climate model to the hydrological model. This study used the delta change method, in which the observed current climate is the baseline for the future climate input. Hence, this method does not transfer information on changed dynamics of the meteorological variables. The delta change method is assumed appropriate for this type of impact assessment that focuses on groundwater and mean stream discharge because extremes may have relatively little influence on groundwater flow and baseflow; the shortest timescale at which the changes for river discharge are studied here is monthly, limiting the influence of extremes. The use of only one transfer method brings along uncertainties in the climate-change signal transfer, and together with the combined uncertainties of the climate model predictions, the uncertainties on the simulated changes are expected to be significant. Quantification of these uncertainties, however, is beyond the scope of the present work.

The uncertainties of the hydrological model are a complex function of the accuracy of the different elements of the model. The driving variables (precipitation, reference evapotranspiration, and temperature), the model parameters found by calibration against uncertain observation data, and the structural model, including, for example, the process descriptions applied and the distribution of inputs and parameters in the model, are associated with errors. These uncertainties in input and model setup are transferred to uncertainties in the model results. Butts et al. (2004) compared the magnitude of model structure uncertainty to estimates of the uncertainty in measurement data, model parameters, and rainfall input. Ten different model structures using a combination of model codes, including MIKE SHE, were compared. The results showed that the sensitivity to variations in acceptable model structure was of the same magnitude as uncertainties arising from the other evaluated sources. The study showed that model structure can cause significant uncertainty in the simulated results and should therefore be considered when assessing the results.

In the present study, the uncertainty of the absolute magnitude of the hydrological model predictions was assessed by a validation test on present climate data. The test was performed to obtain an indication of the accuracy of the predictions of the hydrological model for periods different from the calibration period. The calibration and validation periods included very dry and wet years, and the DK model was robust for a relatively large range in average annual precipitation. The test also revealed that the precision of point predictions of hydraulic head was quite low, as indicated by relatively high RMS values. This is partially caused by the low quality in the observation data and scale problems. Nonetheless, the models were capable of producing results that satisfactorily captured the average response (low ME and F_bal values) and the stream discharge response at larger gauging stations. Hence, the ability of the model to reproduce aggregated quantities of the system is better than for small-scale variables, and the uncertainty in the aggregated model results are therefore estimated to be smaller. It is, however, interesting to examine how large the signal from the models should be to exceed the noise in the model response.

Because the hydrological model is unchanged when simulating the response of the present climate and the climate change scenarios, a high degree of correlation in uncertainty in results for the two situations can be expected. If, for example, the model simulates too-high groundwater levels in a particular region with the present climate, one expects that it will also do that in the future climate, since it contains the same errors in both situations. Therefore, the uncertainty of the difference or change in model predictions for the present and the future climate, such as change in groundwater level, should not be calculated assuming that the two variables are independent, resulting in summation of the individual variances. On the contrary, because of the high correlation between the errors of the present and the future model predictions, the uncertainty of the differences is expected to be much smaller than the uncertainty of the absolute values of climate change predictions (Reichert and Borsuk, 2005). It is therefore assumed that the confidence in the changes presented in this work is higher than indicated by the validation results. Further, by using the averages for a 15-yr simulation period and averaging the spatially distributed results for the whole model area, the relative strength of the signal-to-noise ratio is increased. For these reasons the simulated results are considered to represent a significant trend, even though the precision of the absolute predictions of the future hydrological situation are within the realm of model uncertainty.

	Conclusions

TOP ABSTRACT INTRODUCTION Materials and Methods Results Discussion Conclusions REFERENCES

 
The work undertaken here is a first estimate of the effects of future climate change on groundwater and river discharge in hydrological systems with different climatology, geomorphology, and geology. The results indicate a plausible trend in the hydrological response to future climate changes. The magnitude of the mean annual net precipitation was estimated to increase by up to 30%, and the impact of this additional input of water was found to be highly dependent on the geological settings of the model areas. In the Jylland area, characterized by sandy top soils and large interconnected aquifers, groundwater recharge increased significantly, resulting in an increase in mean annual groundwater heads up to 0.5 m and increasing groundwater–river interaction. Significant increases in winter stream discharge were found, whereas the summer discharge was found to be only slightly affected. On Sjaelland, where the topsoil is dominated by low-permeability soils and the aquifers are protected by thick clay layers of regional extent, only minor increases up to 0.2 m in mean annual groundwater heads were predicted. Here, the primary effect was realized for stream discharge, caused by changes in drain flow and overland flow, with up to a 50% increase in winter and a 50% decrease in summer.

The comprehensive three-dimensional representation of the flow in the groundwater system made it possible to transfer the climate change signals from the surface to the groundwater system in a physically based manner and to explain the differences in response through an improved understanding of the hydrological system. The two hydrological systems studied here showed different responses, even though the imposed climate change signal was the same for both study areas. This demonstrates the importance of using site-specific models that capture the physical characteristics of the area when studying the impacts of climate change.

The many uncertainties related to the climate models and emissions scenarios, the downscaling and transfer method, and the adequacy of the hydrological model complicate the evaluation of the usefulness of the simulated results. The quantitative predictions of the hydrological changes therefore have a large uncertainty margin and indicate a trend more than they indicate a precise value. The use of two emissions scenarios and two study areas had a large added value for determining the trends because changes could be compared quantitatively and qualitatively. To determine the full range in the simulated trends, however, a study would have to include many more climate change models and scenarios, downscaling methods and various hydrological model structures and parameterizations. Hydrological climate impact studies would benefit significantly from using a framework that includes ensembles of climate models and scenarios, which is commonly applied in climate research, as well as ensembles of downscaling methods and an evaluation of the sensitivity to variations in acceptable model structure and parameterizations. The use of such a framework severely increases the complexity and work load of climate impact studies on hydrological systems, making it necessary to carefully select what is to be included in the ensembles, if these kinds of frameworks are to be used in practice.

In the present study only the hydrological consequences of the direct climatic effects of increasing greenhouse-gas concentrations (temperature, precipitation, and reference evapotranspiration) were assessed, allowing us to separate the response of the natural system from more indirect effects, which otherwise would have blurred the response of the changes in climate only. However, to obtain more realistic predictions of the hydrological response to the future climate, the impact of indirect consequences of climate changes such as sea-level rise, changes in agricultural practice and land use, and the development in water demand for domestic and irrigation purposes also need to be addressed. This will most likely involve the definition of different scenarios describing plausible human developments and require the use of a physically based model, as presented here, because these system changes need to be studied in an integrated way. More realistic effects of climate change on water resources in Denmark would be obtained by such an approach and would serve as the basis for possible adaptation plans.

	ACKNOWLEDGMENTS

 
The RCM data was provided through the PRUDENCE data archive (http://prudence.dmi.dk/), funded by the EU through contract EVK2-CT2001-00132. Special thanks go to the staff at the Danish Climate Centre for their help with processing the climate model data. We also gratefully appreciate the constructive discussion points contributed by Jens Christian Refsgaard and the review comments of Karsten H. Jensen. Diana M. Allen, Zhiming Lu, Maria A. Mimikou, an anonymous reviewer, and one of three guest editors of this special issue, Timothy R. Green, are thanked for their constructive comments in their reviews, which have led to significant improvements in the paper. The project was partly funded by the Danish EPA.

	REFERENCES

TOP ABSTRACT INTRODUCTION Materials and Methods Results Discussion Conclusions REFERENCES

 

• Abbott, M.B., J.C. Bathurst, J.A. Cunge, P.E. O'Connel, and J. Rasmussen. 1986. An introduction to the European Hydrological System–Systeme Hydrologique Europeen, SHE, 2: Structure of a physically-based, distributed modeling system. J. Hydrol. (Amsterdam) 87(1–2):61–77.
• Allen, D.M., D.C. Mackie, and M. Wei. 2004. Groundwater and climate change: A sensitivity analysis for the Grand Forks aquifer, southern British Columbia, Canada. Hydrogeol. J. 12(3):270–290.
• Allen, R.G., L.S. Pereira, D. Raes, and M. Smith. 1998. Crop evapotranspiration: Guidelines for computing crop water requirements. FAO Irrigation and Drainage Paper 56. FAO, Rome.
• Allerup, P., H. Madsen, and F. Vejen. 1998. Standard values (1961–90) for precipitation correction. (In Danish.) Tech. Rep. 98-10. Danish Meteorological Inst., Copenhagen.
• Andersen, H.E., B. Kronvang, S.E. Larsen, C.C. Hoffmann, T.S. Jensen, and E.K. Rasmussen. 2006. Climate-change impacts on hydrology and nutrients in a Danish lowland river basin. Sci. Total Environ. 33(4–5):228–234.
• Anderson, M.P., and W.W. Woessner. 1992. Applied groundwater modeling. Academic Press, San Diego, CA.
• Andréasson, J., S. Bergström, B. Carlsson, L.P. Graham, and G. Lindström. 2004. Hydrological change: Climate change impact simulations for Sweden. Ambio 33(4–5):228–234.
• Bouraoui, F., G. Vachaud, L.Z.X. Li, H. Le Treut, and T. Chen. 1999. Evaluation of the impact of climate changes on water storage and groundwater recharge at the watershed scale. Clim. Dyn. 15(2):153–161.
• Brouyère, S., G. Carabin, and A. Dassargues. 2004. Climate change impacts on groundwater resources: Modeled deficits in a chalky aquifer, Geer basin, Belgium. Hydrogeol. J. 12(2):123–134.
• Butts, B., J.T. Payne, M. Kristensen, and H. Madsen. 2004. An evaluation of the impact of model structure on hydrological modelling uncertainty for streamflow simulation. J. Hydrol. (Amsterdam) 298(1–4):242–266.
• Chow, V.T., D.R. Maidment, and L.W. Mays. 1988. Applied Hydrology. McGraw-Hill, New York.
• Christensen, B.S.B., and T.O. Sonnenborg. 2006. Groundwater level trend on Zealand 1989–2001. (In Danish.) GEUS Rep. 2006/59. Geol. Survey of Denmark and Greenland, Copenhagen.
• Christensen, J.H., and O.B. Christensen. 2007. A summary of the PRUDENCE model projections of changes in European climate by the end of this century. Clim. Change (in press), doi:10.1007/s10584-006-9210-7.
• Christensen, J.H., O.B. Christensen, P. Lopez, E. van Meijgaard, and M. Botzet. 1996. The HIRHAM4 regional atmospheric climate model. SC Rep. 96-104. Danish Meteorological Inst., Copenhagen.
• Christensen, O.B., and J.H. Christensen. 2004. Intensification of extreme European summer precipitation in a warmer climate. Global Planet. Change 44(1–4):107–117.
• Christensen, O.B., J.H. Christensen, B. Machenhauer, and M. Botzet. 1998. Very high-resolution regional climate simulations over Scandinavia: Present climate. J. Clim. 11(12):3204–3229.
• Christensen, N.S., A.W. Wood, N. Voisin, D.P. Lettenmaier, and R.N. Palmer. 2004. The effects of climate change on the hydrology and water resources of the Colorado River basin. Clim. Change 62(1–3):337–363.
• Déqué, M., R.G. Jones, M. Wild, F. Giorgi, J.H. Christensen, D.C. Hassell, P.L. Vidale, B. Rocke, D. Jacob, E. Kjellström, M. de. Castro, F. Kucharski, and B. van den Hurk. 2005. Global high resolution versus limited area model climate change projections over Europe: Quantifying confidence level from PRUDENCE results. Clim. Dyn. 25(6):653–670.
• Graham, L.P. 2004. Climate change effects on river flow to the Baltic Sea. Ambio 33(4–5):235–241.
• Havnø, K., M.N. Madsen, and J. Dørge. 1995. MIKE 11: A generalized river modeling package. p. 733–782. In V.P. Singh (ed.) Computer models of watershed hydrology. Water Resources, Highlands Ranch, CO.
• Hay, L.E., R.L. Wilby, and G.H. Leavesley. 2000. A comparison of delta change and downscaled GCM scenarios for three mountainous basins in the United States. J. Am. Water Res. Assoc. 36:387–398.[CrossRef]
• Henriksen, H.J., and A.B. Sonnenborg. 2003. The freshwater cycle. (In Danish.) NOVA Temarapport. Geol. Survey of Denmark and Greenland, Copenhagen.
• Henriksen, H.J., and T.O. Sonnenborg. 2005. Accuracy criteria. p. 12-1–12-21. In T.O. Sonnenborg and H.J. Henriksen (ed.) Handbook of groundwater modeling. (In Danish.) GEUS report 2005/80. Geological Survey of Denmark and Greenland, Copenhagen.
• Henriksen, H.J., L. Troldborg, P. Nyegaard, T.O. Sonnenborg, J.C. Refsgaard, and B. Madsen. 2003. Methodology for construction, calibration and validation of a national hydrological model for Denmark. J. Hydrol. (Amsterdam) 280(1–4):52–71.
• IPCC. 2000. Special report on emissions scenarios (SRES): A special report of working group III of the intergovernmental panel on climate change. Cambridge Univ. Press, Cambridge, UK.
• IPCC. 2001. Climate Change 2001: The scientific basis. Cambridge Univ. Press, Cambridge, UK.
• Jones, C.G., U. Willén, A. Ullerstig, and U. Hannsson. 2004. The Rossby centre regional atmospheric climate model: Part I. Model climatology and performance for the present climate over Europe. Ambio 33(4–5):211–220.
• Kay, A.L., R.G. Jones, and N.S. Reynard. 2006. RCM rainfall for UK flood frequency estimation: II. Climate change results. J. Hydrol. (Amsterdam) 318(1–4):163–172.
• Kinzelbach, W. 1986. Groundwater modeling: An introduction with sample programs in BASIC Dev. Elsevier, Amsterdam, the Netherlands.
• Kleinn, J., C. Frei, J. Gurtz, D. Luthi, P.L. Vidale, and C. Schar. 2005. Hydrologic simulations in the Rhine basin driven by a regional climate model. J. Geophys. Res. 110:D04102, doi:10.1029/2004JD005143.
• Krysanova, V., F. Hattermann, and A. Habeck. 2005. Expected changes in water resources availability and water quality with respect to climate change in the Elbe River basin (Germany). Nord. Hydrol. 36(4–5):321–333.
• Kunstmann, H., K. Schneider, R. Forkel, and R. Knoche. 2004. Impact analysis of climate change for an Alpine catchment using high resolution dynamic downscaling of ECHAM4 time slices. Hydrol. Earth Sys. Sci. 8(6):1030–1044.
• Loáiciga, H.A., D.R. Maidment, and J.B. Valdes. 2000. Climate-change impacts in a regional karst aquifer, Texas, USA. J. Hydrol. (Amsterdam) 227(1–4):173–194.
• Makkink, G.F. 1957. Examination of the Penman formula. (In Spanish.). Neth. J. Agric. Sci. 5:290–305.
• Mimikou, M.A., E. Baltas, E. Varanou, and K. Pantazis. 2000. Regional impacts of climate change on water resources quantity and quality indicators. J. Hydrol. (Amsterdam) 234:95–109.
• Nash, J.E., and J.V. Sutcliffe. 1970. Riverflow forecasting through conceptual models: Part 1. A discussion of principles. J. Hydrol. (Amsterdam) 10:282–290.
• Payne, J.T., A.W. Wood, A.F. Hamlet, R.N. Palmer, and D.P. Lettenmaier. 2004. Mitigating the effects of climate change on the water resources of the Columbia River basin. Clim. Change 62(1–3):233–256.
• Poeter, E., and M.C. Hill. 1999. UCODE, a computer code for universal inverse modeling. Comput. Geosci. 25(4):457–462.
• Räisänen, J., U. Hansson, A. Ullerstig, R. Doscher, L.P. Graham, C. Jones, H.E.M. Meier, P. Samuelsson, and U. Willén. 2004. European climate in the late twenty-first century: Regional simulations with two driving global models and two forcing scenarios. Clim. Dyn. 22(1):13–31.
• Refsgaard, J.C., and B. Storm. 1995. MIKE SHE. p. 809–846. In V.P. Singh (ed.) Computer models of watershed hydrology. Water Resources, Highlands Ranch, CO.
• Reichert, P., and M.E. Borsuk. 2005. Does high forecast uncertainty preclude effective decision support? Environ. Model. Softw. 20(8):991–1001.
• Rummukainen, M., J. Räisänen, D. Bjørge, J.H. Christensen, O.B. Christensen, T. Iversen, K. Jylhä, H. Ólafsson, and H. Tuomenvirta. 2003. Regional climate scenarios for use in Nordic water resources studies. Nord. Hydrol. 34(5):399–412.
• Rummukainen, M., J. Räisänen, B. Bringfelt, A. Ullerstig, A. Omstedt, U. Willen, U. Hansson, and C. Jones. 2001. A regional climate model for northern Europe: Model description and results from the downscaling of two GCM control simulations. Clim. Dyn. 17(5–6):339–359.
• Salathé, E.P. 2005. Downscaling simulations of future global climate with application to hydrologic modeling. Int. J. Clim. 25(4):419–436.
• Scharling, M. 1999. Climate grid Denmark, precipitation, temperature, and potential evapotranspiration 20*20 km and 40*40 km. (In Danish.) Tech. Rep. 99-12. Danish Meteorological Inst., Copenhagen.
• Scibek, J., and D.M. Allen. 2006. Comparing modeled responses of two high-permeability, unconfined aquifers to predicted climate change. Global Planet. Change 50(1–2):50–62.
• Sharma, M.L. 1989. Impact of climate change on groundwater recharge. p. 511–519. In Conference on climate and water, Helsinki, Finland. 11–15 Sept. Vol. 1. Publications of the Academy of Finland.
• Sonnenborg, T.O., B.S.B. Christensen, P. Nyegaard, H.J. Henriksen, and J.C. Refsgaard. 2003. Transient modeling of regional groundwater flow using parameter estimates from steady-state automatic calibration. J. Hydrol. (Amsterdam) 273(1–4):188–204.
• Thomsen, R. 1989. The effects of climate variability and change on groundwater in Europe. 489–500. In Conference on climate and water, Helsinki, Finland. 11–15 Sept. Vol. 1. Publications of the Academy of Finland.
• Van den Hurk, B., M. Hirschi, C. Schaer, G. Lenderink, E. Van Meijgaard, A. Van Ulden, B. Rockel, S. Hagemann, L.P. Graham, and E. Kjellstroem. 2005. Soil control on runoff response to climate change in regional climate model simulations. J. Clim. 18(17):3536–3551.,
• Vanrheenen, N.T., A.W. Wood, R.N. Palmer, and D.P. Lettenmaier. 2004. Potential implications of PCM climate change scenarios for Sacramento–San Joaquin River basin hydrology and water resources. Clim. Change 62(1–3):257–281.
• Varanou, E., E. Gkouvatsou, E. Baltas, and M. Mimikou. 2002. Quantity and quality integrated catchment modeling under climate change with use of soil and water assessment tool model. J. Hydrol. Eng. 7(3):228–244.
• Varis, O., T. Kajander, and R. Lemmelä. 2004. Climate and water: From climate models to water resources management and vice versa. Clim. Change 66(3):321–344.
• Wood, A.W., L.R. Leung, V. Sridhar, and D.P. Lettenmaier. 2004. Hydrologic implications of dynamical and statistical approaches to downscaling climate model outputs. Clim. Change 62(1–3):189–216.
• Xu, C.Y., E. Widen, and S. Halldin. 2005. Modeling hydrological consequences of climate change: Progress and challenges. Adv. Atmospheric Sci. 22(6):789–797.
• Yuan, F., Z.H. Xie, Q. Liu, and J. Xia. 2005. Simulating hydrologic changes with climate change scenarios in the Haihe River basin. Pedosphere 15(5):595–600.

This article has been cited by other articles:

				T. R. Green, M. Taniguchi, and H. Kooi Potential Impacts of Climate Change and Human Activity on Subsurface Water Resources Vadose Zone J., August 23, 2007; 6(3): 531 - 532. [Full Text] [PDF]

This Article Abstract Figures Only Full Text (PDF) Free Alert me when this article is cited Alert me if a correction is posted Services Similar articles in this journal Similar articles in Web of Science Alert me to new issues of the journal Download to citation manager Citing Articles Citing Articles via HighWire Citing Articles via Web of Science (1) Citing Articles via Google Scholar Google Scholar Articles by van Roosmalen, L. Articles by Sonnenborg, T. O. Search for Related Content PubMed Articles by van Roosmalen, L. Articles by Sonnenborg, T. O. GeoRef GeoRef Citation Agricola Articles by van Roosmalen, L. Articles by Sonnenborg, T. O. Related Collections Global Change Surface Hydrology