![Full access]({"type":"image" , "src":"/pb-assets/3rdPartyLogos/accessFull-1452596451717.png"}){kind=small}
![CrossMark Logo]({"type":"image" , "src":"/templates/jsp/images/CROSSMARK_Color_horizontal.svg"})
Abstract
Abstract
The Soil and Water Integrated Model (SWIM) was used to assess potential climate and land-use change impacts in the Central European catchments of Schwarze Elster, Spree and Lusatian Neisse which are heavily influenced by opencast lignite mining. To account for potential climate change, scenarios of two statistical regional climate models, STAR and WettReg, were used. Regional anthropogenic change was considered in terms of increasing cultivation of energy crops (oilseed rape, silage maize, sunflower and sorghum) and decreasing mining activities (decreasing groundwater depression cone). In the climate scenarios, decreased natural discharge, by up to 60% in the long-term average, was simulated. In simulations with climate scenarios and oilseed rape, this effect is halved; the other energy crops have a small additional impact on discharge. The decreasing groundwater depression cone slightly compensates for climate change impacts. Overall, potential impacts of regional anthropogenic activities are secondary to those of climate change.
Editor Z.W. Kundzewicz; Guest editor M. White
Résumé
Le modèle intégré sol et eau (SWIM) a été utilisé pour évaluer les impacts potentiels des changements climatiques et d’utilisation du sol dans les bassins versants de l’Elster Noire, de la Spree et de la Neisse de Lusace, qui sont fortement influencés par l’exploitation minière à ciel ouvert de lignite. Pour tenir compte des changements climatiques potentiels, on a utilisé les scénarios de deux modèles climatiques statistiques régionaux, STAR et WettReg. Les changements anthropiques régionaux ont été pris en compte en considérant une augmentation des cultures énergétiques (colza, tournesol, maïs ensilage et sorgho) et une diminution des activités minières (diminution du cône de dépression de la nappe). Dans les scénarios climatiques, une diminution du débit moyen naturel allant jusqu’à 60% a été simulée sur le long terme. Dans les simulations avec des scénarios climatiques et le colza, cet effet est réduit de moitié, les autres cultures énergétiques ayant peu d’impact supplémentaire sur les débits. La diminution du cône de dépression des eaux souterraines compense légèrement les impacts du changement climatique. Globalement, les impacts potentiels des activités anthropiques régionales sont secondaires par rapport à ceux des changements climatiques.
INTRODUCTION
The relationship between climate and land-use change is complex at different spatial and temporal scales (Chase et al. 2000, Pielke 2005, Rounsevell and Reay 2009). Regional climate change occurs as a consequence of global climate change and land-use changes, especially at the meso- and microscales (e.g. Mölders 2000, Conradt et al. 2007), whereas anthropogenic land-use changes occur at local or regional scales, driven by local, national and global drivers (Foley et al. 2005). The quantification of both climate and land-use change impacts is a major challenge of hydrology (Sivapalan et al. 2003). As climate is the major driver of the water cycle, even small changes in meteorological parameters affect water balance components directly (Bronstert et al. 2007). Also land-use changes can result in strong impacts on hydrological processes (Calder 1992, DeFries and Eshleman 2004). For example, urbanization results in higher flood peaks (Moscrip and Montgomery 1997), afforestation increases evapotranspiration and decreases runoff (e.g. Fohrer et al. 2001), while deforestation, in particular, increases peak discharges (Seibert and McDonnell 2010, Birkel et al. 2012). In contrast, changes between different types of annual vegetation, such as agricultural fields, meadows or fallow land, generally have only a minor impact on discharge (Wechsung et al. 2000, Lahmer et al. 2001, Klöcking and Haberlandt 2002, Niehoff et al. 2002, Bormann et al. 2007). The extents of climate and land-use change impacts on the water balance depend on catchment properties, such as climate conditions, predominating land use, and storage capacity (Sahin and Hall 1996, Capell et al. 2014). Studies aimed at both climate and land-use change impacts often find that climate change dominates at larger scales, while land-use change impacts can be more important at local up to regional scales. For example, Lahmer et al. (2001) found that potential climate change impacts outweigh the effects of moderate conversion of cropland into dry pasture, meadow, forest or fallow land, in mesoscale catchments in northern Germany. Nathkin et al. (2012) identified climate change as the main driver of decreased groundwater recharge of headwater catchments in northeast Germany. However, land-use change in terms of changes in the ground vegetation of pine forest and changes in the tree age distribution also contributed to this decrease.
As climate is a major driver of crop growth, climate change impacts on crop growth are also of fundamental concern (Rötter and Van De Geijn 1999, Tuck et al. 2006, Bellarby et al. 2010, Haberl et al. 2011, Wenkel et al. 2013). However, the quantification of these impacts is subject to huge uncertainties (Godfray et al. 2010, Bassu et al. 2012, Asseng et al. 2013). Since climate change has impacts on both crop growth and water balance components, these impacts should be assessed simultaneously.
In the Central European river catchments of Schwarze Elster, Dahme, Spree and Lusatian Neisse, which are vulnerable to changing climate conditions due to their continental climate with relatively high summer temperatures and low annual precipitation, potential climate change impacts are of particular interest. So far, increasing temperature and decreasing precipitation have been observed in Germany in the past decades (Schönwiese et al. 2006). By 2100, Central European mean temperature increase of up to 5.5 K is expected, which is most likely associated with a decrease in precipitation in Central Europe (Houghton et al. 2001, Eisenreich 2005, Solomon et al. 2007). Consequently, studies aimed at potential climate change impacts in eastern Germany mostly show decreases in discharge (Hattermann et al. 2008b, Pohle et al. 2012). Increasing temperatures and decreasing precipitation in the study region are also reflected in the scenarios of the regional climate models STAR and WettReg (see section on ‘Climate scenario development’). Other studies, however, show a temperature increase of the same magnitude, but a slight increase in annual precipitation in Central Europe (Jacob et al. 2014). Generally, the simulated changes in discharge strongly depend on the regional climate model used, as Gädeke et al. (2014) discovered for a sub-catchment of the Spree River.
Apart from climate change, increasing cultivation of energy crops in the region—desired for renewable energy production (CEC 2005, BMELV and BMU 2009, Kenkmann 2010)—is also expected to affect discharge. Furthermore, due to lignite mining activities, the water balance in the region is seriously affected: large-scale opencast lignite mining has been conducted for more than 100 years (Koch et al. 2008). To extract lignite in opencast mining, the groundwater table has to be lowered prior to and during the mining operation. To extract one tonne of lignite, in the study region, more than six tonnes of groundwater are pumped and released into the river system. Consequently, a wide groundwater depression cone has been formed around the opencast mines, covering more than 2100 km2 in 1989, whereas river discharges have been increased (Koch et al. 2005, Grünewald et al. 2008). After German re-unification in 1990, most of the lignite mines were closed and, consequently, less groundwater was pumped. Therefore, groundwater started to rise again and the area of the groundwater depression cone was reduced.
Due to the various influences on natural discharges in the region, the potential impacts of climate change and regional anthropogenic activities (increasing cultivation of energy crops and decreasing mining activities) should be assessed simultaneously.
Hence, the ecohydrological Soil and Water Integrated Model (SWIM; Krysanova et al. 1998, 2000) was applied to study the river catchments of Schwarze Elster, Dahme, Spree and Lusatian Neisse. SWIM had been developed to simulate the impacts of climate and land-use change on the water balance, water quality and crop growth. As such, it has been applied in various regional studies, particularly in Central Europe (Post et al. 2008, Hattermann et al. 2008b, Holsten et al. 2009, Huang et al. 2010). Wechsung et al. (2000) applied SWIM in the German Federal State of Brandenburg and simulated a small increase in runoff for the conversion of cropland to temporary fallow land, and also for the conversion to meadows within river corridors. In a study using SWIM for the Elbe River basin, Conradt et al. (2012b) found that discharge decreases with temperature increase and that climate change impacts outweigh moderate land-use change impacts.
The main objectives of the current study were to apply SWIM for the catchments of Schwarze Elster, Dahme, Spree and Lusatian Neisse in Central Europe in order to assess potential future natural discharges under climate change conditions, whilst also considering the effects of regional anthropogenic activities in terms of both increasing cultivation of energy crops and decreasing mining activities. The results presented here provide a means to assess potential impacts of climate change and regional anthropogenic activities on water management and water quality issues in the region. The results of the study can form the basis for planning adaptation measures in close cooperation with regional stakeholders.
DATA AND METHODS
Study region
The study was conducted for the river catchments of Schwarze Elster, Dahme, Spree and Lusatian Neisse (Fig. 1). The catchments are located in the region of Lusatia in Central Europe, mainly in the German Federal States Brandenburg and the Free State of Saxony. The Schwarze Elster River is a tributary of the River Elbe. The Dahme flows into the River Spree, which joins the Havel as another tributary of the River Elbe. The headwater catchment of the Lusatian Neisse River is located in the Czech Republic; the river forms the Polish–German border until it joins the River Oder. In this study, the entire catchments of the Schwarze Elster (5700 km2) and the Dahme (2000 km2) are considered, whereas the catchment of the Spree River is included up to the Große Tränke gauging station (6200 km2) and that of the Lusatian Neisse River up to the Steinbach gauging station (2100 km2). The Schwarze Elster, Dahme and Spree are all lowland catchments, with average altitudes of 120, 54 and 110 m a.s.l., respectively. The Lusatian Neisse catchment is mountainous to hilly, with an average altitude of 330 m a.s.l. and a maximum altitude in excess of 1000 m a.s.l.
Impacts potentiels des changements climatiques et des activités anthropiques régionales dans les bassins versants de méso-échelle d’Europe centrale
Published online:
18 May 2015Fig. 1 Map of the study region including gauging stations referred to and the Status Quo scenario (2005) of the groundwater depression cone.
![]({"type":"image","src":"/na101/home/literatum/publisher/tandf/journals/content/thsj20/2015/thsj20.v060.i05/02626667.2014.968571/20150710/images/medium/thsj_a_968571_f0001_b.gif"})
Fig. 1 Map of the study region including gauging stations referred to and the Status Quo scenario (2005) of the groundwater depression cone.
Lusatia is characterized by a continental climate with relatively high summer temperatures and low annual precipitation (monthly mean temperature in July and mean annual precipitation at the Cottbus climate station during the reference period 1961–1990 were 18.4°C and 563 mm, respectively). As a result, the climatic water balance (precipitation minus potential evapotranspiration) is negative in many parts of the region.
Cropland, forest and wetlands are important forms of land use in the study region (Fig. 2). Smaller catchments, for example those of two headwaters of the River Spree, the Löbauer Wasser (up to Gröditz gauging station) and the Weißer Schöps (up to Särichen gauging station) are dominated by cropland. The data shown are based on CORINE land cover (Bossard et al. 2000), with areas of riparian soils regarded as wetland according to Hattermann et al. (2008a).
Impacts potentiels des changements climatiques et des activités anthropiques régionales dans les bassins versants de méso-échelle d’Europe centrale
Published online:
18 May 2015![]({"type":"image","src":"/na101/home/literatum/publisher/tandf/journals/content/thsj20/2015/thsj20.v060.i05/02626667.2014.968571/20150710/images/medium/thsj_a_968571_f0002_b.gif"})
Fig. 2 Land-use types in the river catchments.
Hydrological modelling with SWIM
The SWIM model is a process-based, semi-distributed ecohydrological model. It was developed based on the models SWAT (Arnold et al. 1994) and MATSALU (Krysanova and Luik 1989). In this study, a version of SWIM with two adaptations by Conradt et al. (2012b) is used: (a) groundwater is represented by two linear storages and (b) riparian vegetation is simulated with access to river discharge thus allowing a negative discharge contribution from sub-catchments.
For the study area, four separate models were set up for each of the catchments of the rivers Schwarze Elster, Dahme, Spree and Lusatian Neisse. The catchments were subdivided into sub-catchments and then into homogeneous landscape units (hydrotopes) based on soil (BGR 1998, CGS 2005, IUNG 2012) and land-use (Bossard et al. 2000) maps.
Due to the strong anthropogenic impact on discharge by lignite mining and water management, the traditional approach of calibrating hydrological models based on time series of observed discharges is constrained in the region. Also, the approach of Conradt et al. (2012b) for the Elbe River basin, a combination of a global calibration at the lowest gauging station and a sub-basin scale calibration of sensitive parameters, was not applicable to the study area. Therefore, SWIM was first calibrated for sub-catchments without significant influence of mining and water management on discharge in such a way that all model parameters were calibrated separately for selected sub-catchments. In accordance with the availability of climate data and observed discharges, 1998–2001 was used as the calibration period, and 2002–2006 as the validation period. For the Lusatian Neisse catchment, the years 1998, 2002 and 2003 were used as the calibration period, and 1999–2002 as the validation period (observed climate data for the Czech Republic were only available to the authors up to 2003). In a second step, some parameters were adjusted globally for all catchments and the most sensitive parameters in the region were identified and calibrated for further sub-catchments. Consistent with the findings of Conradt et al. (2012b), these sensitive model parameters are: (a) a correction for potential evapotranspiration and (b) a factor for groundwater recession rates. Additionally, (c) a correction coefficient for hydraulic soil conductivity and (d) a riparian zone parameter were adjusted for the sub-catchments individually. For the first parameter (a), Conradt et al. (2012a) found that, in order to reduce volume differences between simulation and observation, a small correction of simulated potential evapotranspiration is usually preferable to that of precipitation. Concerning the second parameter (b), Schmalz et al. (2008) emphasized that, in lowland catchments, where hydrological processes are dominated by groundwater dynamics and storage, the groundwater recession rates are very sensitive. These most sensitive model parameters in the region were therefore regionalized by spatial proximity (Egbuniwe and Todd 1976) and physical similarity (Burn and Boorman 1993). Additional data, including long-term mean monthly observed discharges for the Schwarze Elster River for 1974–2008, as well as observed pre-mining annual discharges of the N-A-U map for 1921–1940 (IfWW 1959), were used as a comparison to verify the simulation results. The regionalization and also the comparison with additional data is described in more detail in Pohle et al. (2013).
In the scenario study, impacts of climate change and regional anthropogenic activities on the water balance components were quantified using the commonly used approach introduced by Gleick (1986). This compares simulation results of hydrological models driven for ‘status quo’ conditions and scenarios with projected climate change, as well as scenarios of regional anthropogenic activities. SWIM was driven with daily climate input for the observation period (1951–2003) and climate scenarios (2004–2055). For ‘status quo’ conditions of land use, cropland is parameterized as winter wheat. The simulated natural water balance components do not include influences of mining discharges and water management.
Climate scenario development
The spatial resolution of global climate model (GCM) outputs is too coarse for regional impact assessment (Gleick 1987, Kundzewicz and Stakhiv 2010). Moreover, the use of climate data on a regional scale requires the downscaling of data delivered by GCMs using a regional climate model (RCM). For this analysis, climate data were regionalized using the statistical RCMs, STAR (Werner and Gerstengarbe 1997, Orlowsky et al. 2008) and WettReg (Enke et al. 2005a, 2005b). Both RCMs have been used in various climate impact studies in Central Europe (e.g. Huang et al. 2010, Bormann 2012, Conradt et al. 2012b, Gädeke et al. 2014). Bronstert et al. (2007) evaluated both to be more suitable for hydrological climate impact studies in terms of mean seasonal runoff and evapotranspiration than the direct use of the GCM or the dynamic regional climate model REMO (Jacob and Podzun 1997, Tomassini and Jacob 2009).
The statistical analogue resampling scheme STAR generates time series of climate parameters by resampling segments of daily observations at climate and precipitation stations. The simulated time series are forced by only the linear air temperature trend of the future period. The STAR 0K scenario assumes no further climate change in the study region. However, recent years have already been warmer than the reference period (1961–1990), so that a temperature increase of 0.9K compared to the reference period is included in the STAR 0K scenario. The other STAR scenarios used in this analysis assume a further temperature increase of 2K (STAR 2K) and 3K (STAR 3K) in the region until 2055. For each STAR scenario, 100 stochastically generated time series (realizations) of climate data at climate and precipitation stations in the Czech Republic (three stations) and Germany (65 stations) were used.
The statistical–empirical RCM WettReg (Wetterlagenbasierte Regionalisierungsmethode: weather type based regionalization method) uses the statistical relationships between large-scale atmospheric conditions and local climate. Driven by GCM results, scenarios of future local climate conditions are generated based on climatic variables of the original time series, including corrections concerning geopotential height and the precipitation regime. Ten realizations at 149 German climate and precipitation stations using the WettReg 2010 version (Spekat et al. 2010) were used for simulations of the past (1960–2000) and the A1B scenario of WettReg was used for the future.
The standard reference period 1961–1990 is also used as the reference period in this study. The scenario results are presented for the last 10 years for which STAR data for both Germany and the Czech Republic were available (2046–2055). Here, the effect of inter-decadal variability is minimized by considering daily climate variables of 100 realizations of each STAR scenario and of 10 realizations of the A1B scenario of WettReg.
Scenarios of regional anthropogenic activities
Regional anthropogenic activities were considered in terms of scenarios of both an increased cultivation of arable energy crops and the potential development of mining. In the standard version of SWIM, only one crop is considered for the whole time period and the whole area. Therefore, in order to get a first idea about the direction and magnitude of potential land-use change impacts on the water balance, ‘extreme’ scenarios with a complete change of the dominating crop were considered. Here, winter oilseed rape and the summer crops silage maize, sunflower and sorghum were included. Nutrient parameters were not adjusted in the calibration of SWIM, but optimal fertilization is assumed in the study area. Furthermore, irrigation and the effects of elevated CO2 levels are disregarded in the simulation of crop yields. Therefore, potential changes in crop yield are analysed considering changes in climate variables only. Simulated crop yield for winter wheat, silage maize, oilseed rape and sunflower were compared to average observed crop yields of Germany as a whole and of the Federal States of Saxony and Brandenburg in the years 1997–2012 (DESTATIS 2007, 2013, StLA 2007, 2009, AfS 2012). Observed crop yield data could not be obtained for the spatial extent and discretization of the catchments. Furthermore, a direct comparison between observed and simulated yields is limited to a few years for which both crop yield and climate data were available. Nevertheless, a rough comparison was made between absolute values of mean observed crop yields of Germany as a whole and of the Federal States of Saxony and Brandenburg on the one hand, and mean simulated crop yields of the catchments (reference period and scenarios) on the other. Additionally, the correlations (Pearson’s r) between crop yield and summer temperature, summer precipitation, and the aridity index, Ai (De Martonne 1926) were compared. For years where observed yields of a certain crop were available, mean summer temperature, summer precipitation and the aridity index were also considered. For the simulations, medium climate variables and simulated crop yields for the reference period (observed climate and WettReg) and the climate scenarios STAR 0K, STAR 2K and WettReg A1B in the scenario period were considered.
Three mining scenarios in terms of the area of the groundwater depression cone, which does not contribute to subsurface runoff, were taken into account: one scenario without groundwater depression cone (No Mining), one that assumes the size of the groundwater depression cone to remain constant at the level of 2005 (Status Quo) and one that assumes the reduction of the groundwater depression cone (Phasing Out) according to a scenario described in Lienhoop et al. (2011) (Fig. 3). In the mining scenarios, it is assumed that any water that percolates within the area of the groundwater depression cone is used to fill the pore space to reduce the groundwater deficit, and does not contribute to subsurface flow and river discharge. That way, the entire area of the groundwater depression cone around both the active and former mines, and not only the area of the active mines themselves, impacts on river discharges in the region.
Impacts potentiels des changements climatiques et des activités anthropiques régionales dans les bassins versants de méso-échelle d’Europe centrale
Published online:
18 May 2015Fig. 3 Development of the spatial extent of the groundwater depression cone as percentage of the catchment area in the mining scenario Phasing Out.
![]({"type":"image","src":"/na101/home/literatum/publisher/tandf/journals/content/thsj20/2015/thsj20.v060.i05/02626667.2014.968571/20150710/images/medium/thsj_a_968571_f0003_b.gif"})
Fig. 3 Development of the spatial extent of the groundwater depression cone as percentage of the catchment area in the mining scenario Phasing Out.
RESULTS AND DISCUSSION
Parameterization of SWIM
The parameter values that were calibrated for several sub-catchments, and then taken as global parameter values for all sub-catchments, are shown in Table 1. Also, the ranges of the parameters that were adjusted for sub-catchments by calibration, and later by regionalization, are listed. A relatively high correction for potential evapotranspiration was used in some hilly sub-catchments, where climate heterogeneity is highest, and observed climate variables at the stations differ most from mean climate variables of the sub-catchments. Low values for groundwater recession rates were calibrated in hilly areas, whereas higher recession rates were used in shallow groundwater-dominated sub-catchments. The groundwater recession rates for the fast and the slow groundwater storage were set in the ratio of 20:1. The correction coefficient for the saturated hydraulic soil conductivity is lower in hilly areas and higher in shallow sandy catchments. The riparian zone parameter was set to 1 for catchments with a slope higher than 3% and to 0.9 for catchments with a lower slope. For the wetland area of the Spreewald Biosphere Reserve this parameter was set to 0.6 to account for the numerous channels. Figure 4 shows comparisons between observed and simulated discharge for five catchments with little anthropogenic impact on discharge, where model parameter calibration was performed. The five catchments are: the Pulsnitz River, a tributary to the Schwarze Elster River, at Ortrand (OR) gauging station; the Spree River at Bautzen (BA) gauging station; the Löbauer Wasser River at Gröditz 1 (GR) gauging station; the Weißer Schöps River at the Särichen (SR) gauging station; and the Lusatian Neisse River at the Zittau 1 (ZI) gauging station. The mean simulated discharges in the calibration and validation period correspond well to the observed discharges. Also, the goodness of fit, as expressed by the Nash-Sutcliffe efficiency (NSE; Nash and Sutcliffe 1970), the Nash-Sutcliffe efficiency with logarithmic discharges (NSElog), the coefficient of determination (r2) and the volume efficiency (VE; Criss and Winston 2008) are satisfactory for those catchments. Such a comparison is not possible for the entire catchment discharges, which are heavily influenced by lignite mining and water management. However, for the Schwarze Elster River at Löben gauging station, a comparison was made between long-term means of observed (1974–2008) and simulated monthly discharge (1974–2006). As the observed discharges were only available in aggregated form, while the observed climate data used to drive SWIM were only available up to 2006, a comparison at a similar time span was not possible. However, for the mean of the years considered, monthly observed and simulated discharges show a good overall agreement (Fig. 5(a)). Simulated discharges are slightly higher in winter and lower in autumn compared to the observed values. This can be explained by (a) high discharges in winter being used to fill the reservoirs and fish ponds, and (b) water being released from fish ponds in the region in autumn. For catchments where nowadays the discharge is heavily impacted by lignite mining and water management, pre-mining long-term discharges from the N-A-U-map were used to verify simulation results (Fig. 5(b)). For the gauging stations considered, the observed discharges for the period 1921–1940 are slightly higher than the simulated discharges for the period 1961–1990. This might be explained by different climate conditions in the two periods.
Impacts potentiels des changements climatiques et des activités anthropiques régionales dans les bassins versants de méso-échelle d’Europe centrale
Published online:
18 May 2015Fig. 4 (a) Mean annual discharge (Q) for the calibration (C) and validation (V) periods, and (b) quality criteria (NSE: Nash-Sutcliffe efficiency; NSElog: Nash-Sutcliffe efficiency with logarithmic discharge; r2: coefficient of determination; VE: volume efficiency) for the gauging stations Ortrand (OR), Bautzen (BA), Gröditz (GR), Särichen (SR) and Zittau (ZI).
![]({"type":"image","src":"/na101/home/literatum/publisher/tandf/journals/content/thsj20/2015/thsj20.v060.i05/02626667.2014.968571/20150710/images/medium/thsj_a_968571_f0004_b.gif"})
Fig. 4 (a) Mean annual discharge (Q) for the calibration (C) and validation (V) periods, and (b) quality criteria (NSE: Nash-Sutcliffe efficiency; NSElog: Nash-Sutcliffe efficiency with logarithmic discharge; r2: coefficient of determination; VE: volume efficiency) for the gauging stations Ortrand (OR), Bautzen (BA), Gröditz (GR), Särichen (SR) and Zittau (ZI).
Impacts potentiels des changements climatiques et des activités anthropiques régionales dans les bassins versants de méso-échelle d’Europe centrale
Published online:
18 May 2015Fig. 5 (a) Observed and simulated mean monthly discharge (Q) at the Löben gauging station, and (b) mean annual discharge at the gauging stations Lauchhammer (LH), Bad Liebenwerda (BL), Löben (LO), outlet of the Schwarze Elster (SE), outlet of the Dahme (DA), Cottbus (CO), Lübben (LU) and Große Tränke (GT).
![]({"type":"image","src":"/na101/home/literatum/publisher/tandf/journals/content/thsj20/2015/thsj20.v060.i05/02626667.2014.968571/20150710/images/medium/thsj_a_968571_f0005_b.gif"})
Fig. 5 (a) Observed and simulated mean monthly discharge (Q) at the Löben gauging station, and (b) mean annual discharge at the gauging stations Lauchhammer (LH), Bad Liebenwerda (BL), Löben (LO), outlet of the Schwarze Elster (SE), outlet of the Dahme (DA), Cottbus (CO), Lübben (LU) and Große Tränke (GT).
Table 1 Model parameters.
In general, simulated discharges correspond well with observed discharges, so that the parameterization of SWIM is considered to be credible. As such, the model provides a suitable tool for climate and land-use change impact assessment.
Potential climate change
Annual precipitation (P), temperature (T) and potential evapotranspiration (Ep; Turc-Ivanov) (Turc 1961, Wendling and Schellin 1986) for both the reference based on observations (Ref) and the simulation of WettReg in the reference period 1961–1990 (W Ref) are shown in Fig. 6. The results for the climate scenarios STAR 0K, STAR 2K, STAR 3K and WettReg A1B in the period 2046–2055 are also shown.
Impacts potentiels des changements climatiques et des activités anthropiques régionales dans les bassins versants de méso-échelle d’Europe centrale
Published online:
18 May 2015Fig. 6 Annual climate input: temperature (T), potential evapotranspiration (Ep), and precipitation (P) for the reference period 1961–1990 based on observation (Ref) and WettReg (W Ref) and for the scenario period 2046–2055 for STAR 0K (S 0K), STAR 2K (S 2K), STAR 3K (S 3K) and WettReg A1B (W A1B). Uncertainty is expressed by error bars showing 25% and 75% quantiles based on 30 years of measurement (Ref) or 10 (WettReg) and 100 (STAR) realizations.
![]({"type":"image","src":"/na101/home/literatum/publisher/tandf/journals/content/thsj20/2015/thsj20.v060.i05/02626667.2014.968571/20150710/images/medium/thsj_a_968571_f0006_b.gif"})
Fig. 6 Annual climate input: temperature (T), potential evapotranspiration (Ep), and precipitation (P) for the reference period 1961–1990 based on observation (Ref) and WettReg (W Ref) and for the scenario period 2046–2055 for STAR 0K (S 0K), STAR 2K (S 2K), STAR 3K (S 3K) and WettReg A1B (W A1B). Uncertainty is expressed by error bars showing 25% and 75% quantiles based on 30 years of measurement (Ref) or 10 (WettReg) and 100 (STAR) realizations.
Impacts potentiels des changements climatiques et des activités anthropiques régionales dans les bassins versants de méso-échelle d’Europe centrale
Published online:
18 May 2015Fig. 7 Medians of mean annual discharge simulated in the reference period 1961–1990 based on observed climate (Ref) and WettReg (W Ref) and winter wheat, and in the scenario period 2046–2055 based on the climate scenarios STAR 0K (S 0K), STAR 2K (S 2K), STAR 3K (S 3K), WettReg A1B (W A1B) and the crop scenarios winter wheat, silage maize, oilseed rape, sunflower and sorghum.
![]({"type":"image","src":"/na101/home/literatum/publisher/tandf/journals/content/thsj20/2015/thsj20.v060.i05/02626667.2014.968571/20150710/images/medium/thsj_a_968571_f0007_b.gif"})
Fig. 7 Medians of mean annual discharge simulated in the reference period 1961–1990 based on observed climate (Ref) and WettReg (W Ref) and winter wheat, and in the scenario period 2046–2055 based on the climate scenarios STAR 0K (S 0K), STAR 2K (S 2K), STAR 3K (S 3K), WettReg A1B (W A1B) and the crop scenarios winter wheat, silage maize, oilseed rape, sunflower and sorghum.
Long-term climate variables of the lowland catchments of Schwarze Elster, Dahme and Spree in the reference period are comparable. For these catchments, the climate variables in the reference period generated by WettReg also correspond to the reference based on observations. Due to increasing temperature compared to the reference period, potential evapotranspiration increases in the climate scenarios. Together with decreasing precipitation, this causes a decline in the climatic water balance. The strongest increase in temperature and potential evapotranspiration is projected in STAR 3K, the strongest decrease in precipitation in WettReg A1B.
In the Lusatian Neisse catchment, the observed temperature is about 1.5 K lower and the observed precipitation more than 250 mm year-1 higher than in the lowland catchments. As the Czech stations included in the interpolation of observed climate in the Lusatian Neisse catchment are located at higher altitudes, lower temperature and higher precipitation are observed compared with the German stations. These stations are not included in the WettReg data, so that, in the WettReg reference period, higher temperature and lower precipitation are considered for the Lusatian Neisse catchment than on the basis of both Czech and German stations in the observed reference. In comparison with observed data in the reference period, precipitation increases for the 0K scenario of STAR, but decreases for all other scenarios. Due to increasing temperature and decreasing precipitation in the scenarios, the climatic water balance, which is positive in the reference period, turns negative for the scenarios STAR 3K and WettReg A1B.
For all catchments, bandwidths expressed by the 25% and the 75% quantiles are higher for precipitation than for potential evapotranspiration and also higher for STAR than for WettReg.
Potential impacts of climate change and energy crop scenarios on discharge
Due to increasing temperature and potential evapotranspiration, actual evapotranspiration (Ea) increases unless limited by water availability. Together with decreasing precipitation, discharge (Q) is reduced. Land-use change, in terms of energy crop scenarios, additionally influences evapotranspiration and discharge. Figure 7 shows mean annual discharge for the major catchments in the study region, as well as two headwater catchments dominated by agriculture, the Löbauer Wasser (Gröditz) and the Weißer Schöps (Särichen), for both the reference period (based on observed climate and winter wheat) and for the period considering climate and energy crop scenarios. In the reference period, the simulated discharge based on WettReg is slightly lower than that based on observed climate in the lowland catchments, but slightly higher for the headwater catchments. Consistent with the differences in the climate input of the observation and reference period of WettReg in the Lusatian Neisse catchment, simulated discharge based on the observed climate in the reference period is higher than that based on the reference period of WettReg. Compared to simulations with observed climate in the reference period, slightly lower discharge is simulated assuming STAR 0K and winter wheat in the lowland catchments (up to 17% decrease in discharge), while for the headwater catchments and for the Lusatian Neisse catchment these differences are negligible. For the other climate scenarios, discharges decline in the order Q(STAR 2K) > Q(STAR 3K) > Q(WettReg A1B). This decline is stronger for the lowland catchments (−45% for STAR 2K, −50% for STAR 3K and up to −60% for WettReg A1B) than for the headwater catchments of the Spree and for the Lusatian Neisse catchment. Notably, despite a similar temperature increase, the impacts on discharge are more pronounced in WettReg A1B than in STAR 2K, which is mainly caused by higher actual evapotranspiration simulated for the WettReg A1B scenario (not shown here). This emphasizes the effects of the RCM used in climate impact studies focussing on hydrology.
Impacts potentiels des changements climatiques et des activités anthropiques régionales dans les bassins versants de méso-échelle d’Europe centrale
Published online:
18 May 2015Fig. 8 Mean annual simulated discharge for 100 realizations each of the climate scenarios STAR 0K, STAR 2K and STAR 3K and the crop scenarios winter wheat (WW), silage maize (SM), oilseed rape (OR), sunflower (SF), sorghum (SO) (in the scenario period 2046–2055). The 25%, 50% and 75% quantiles of the simulation results driven by the WettReg A1B scenario (not shown here) are comparable to those by STAR 3K.
![]({"type":"image","src":"/na101/home/literatum/publisher/tandf/journals/content/thsj20/2015/thsj20.v060.i05/02626667.2014.968571/20150710/images/medium/thsj_a_968571_f0008_b.gif"})
Fig. 8 Mean annual simulated discharge for 100 realizations each of the climate scenarios STAR 0K, STAR 2K and STAR 3K and the crop scenarios winter wheat (WW), silage maize (SM), oilseed rape (OR), sunflower (SF), sorghum (SO) (in the scenario period 2046–2055). The 25%, 50% and 75% quantiles of the simulation results driven by the WettReg A1B scenario (not shown here) are comparable to those by STAR 3K.
Simulation results with oilseed rape show notably higher discharges than those with winter wheat for all catchments and climate scenarios. In the STAR 0K scenario, simulated discharge for silage maize and sorghum is slightly higher than for winter wheat, whereas for the other climate scenarios it is slightly lower. Simulation results with sunflower are somewhat higher than those with winter wheat. Except for oilseed rape, the effects of the climate scenarios dominate over those of an increased cultivation of energy crops. In the headwater catchments dominated by agriculture, the relative impacts of an increased cultivation of energy crops on discharge are more pronounced compared to the entire catchment where the proportion of agricultural land is much smaller.
Regarding climate and land-use change impacts, not only medium conditions, but also uncertainty associated with the number of realizations of the regional climate models, are of major interest. Therefore, Fig. 8 illustrates the bandwidths of simulated discharge for the Spree catchment (Große Tränke gauging station) and its headwater catchments in the form of box plots. Each box plot represents 100 realizations of the combination of one STAR and one energy crop scenario. In general, the uncertainties associated with these bandwidths are notable. Due to the generation of STAR realizations, the bandwidths are highest for the STAR 0K and lowest for the STAR 3K scenario. The 25%, 50% and 75% quantiles of the simulation results driven by the WettReg A1B scenario (not shown here) are comparable to those driven by STAR 3K.
Impacts potentiels des changements climatiques et des activités anthropiques régionales dans les bassins versants de méso-échelle d’Europe centrale
Published online:
18 May 2015Fig. 9 Monthly water balance components for realizations 46 and 19 of STAR 3K: precipitation (P), potential evapotranspiration (Ep) and actual transpiration (Ea) for the Spree catchment (left), and discharge (Q) for the Spree, the Löbauer Wasser and the Weißer Schöps.
![]({"type":"image","src":"/na101/home/literatum/publisher/tandf/journals/content/thsj20/2015/thsj20.v060.i05/02626667.2014.968571/20150710/images/medium/thsj_a_968571_f0009_b.gif"})
Fig. 9 Monthly water balance components for realizations 46 and 19 of STAR 3K: precipitation (P), potential evapotranspiration (Ep) and actual transpiration (Ea) for the Spree catchment (left), and discharge (Q) for the Spree, the Löbauer Wasser and the Weißer Schöps.
Two realizations of STAR 3K in 2055 were chosen to visualize the feedbacks between water balance components for different crops: Fig. 9 shows monthly water balance components in the Spree River catchment for the realization with the lowest summer precipitation and climatic water balance (Realization 46) and the realization with the highest potential evapotranspiration in summer (Realization 19) of STAR 3K in the Spree River catchment in 2055. In both realizations, the potential evapotranspiration exceeds precipitation in all months of the vegetation period, so that actual evapotranspiration is limited by water availability throughout the vegetation period. The precipitation in Realization 46 is lower than in Realization 19, hence less water is available resulting in lower actual evapotranspiration and discharge. In both realizations, actual evapotranspiration in April and May is highest for winter wheat due to its earlier growth, but in June to August it is higher for the summer crops. For all crops, the lowest discharge is simulated in early summer. The relative differences in discharge between the crops are strongest in the Löbauer Wasser catchment and lowest in the entire Spree catchment. In all months, the highest discharge is simulated for oilseed rape. The impacts of the other crops on discharge can best be seen in the Löbauer Wasser catchment in Realization 19. Compared to winter wheat, slightly higher discharge is simulated for sunflower throughout the vegetation period. Discharge simulated with silage maize is to some extent lower than discharge simulated with winter wheat. In April and May, simulations with sorghum show slightly higher discharge than those with winter wheat. For Realization 46, and in the catchments of the Weißer Schöps and the entire Spree catchment, the differences in discharge simulated with winter wheat, silage maize, sorghum and sunflower are very small. The fact that an increased cultivation of energy crops is related to small effects on discharge corresponds with other studies that found changes between different types of annual vegetation having only minor impacts on discharge, for example Klöcking and Haberlandt (2002) and Niehoff et al. (2002).
Potential impacts of climate change and energy crop scenarios on crop yield
Potential changes in crop yields are of interest, firstly, due to the interlinkages between crop yield, leaf area index and potential evapotranspiration, and secondly, to find out whether cultivation of energy crops might still be an option under a changing climate. Figure 10 shows observed and simulated crop yields, as well as the correlation between crop yield and summer temperature, summer precipitation and the aridity index, Ai (De Martonne 1926). For winter wheat and silage maize, simulated crop yields correspond well with the observed values. Crop yields of oilseed rape seem to be underestimated by the simulation, especially in the lowland catchments, while crop yields of sunflower seem to be overestimated. For sunflower, this can be explained by the fact that this crop is usually grown in less favourable areas, while in the scenario it is assumed to be grown on the entire cropland. Furthermore, the comparison is limited to no more than 3 or 4 years of observed crop yields. For both observation and simulation, crop yields of winter wheat correlate positively with precipitation, as well as the aridity index, yet negatively with temperature. These correlations are less pronounced for the colder and wetter Lusatian Neisse catchment than for the lowland catchments. The positive correlations between silage maize yield and temperature and the negative correlation between silage maize yield and both precipitation and aridity index of the simulation correspond well with those based on observed values for Germany as a whole. For oilseed rape, the observations mostly show a positive correlation between crop yield and precipitation, as well as the aridity index, and a negative correlation with temperature, which is similar to the simulations. Crop yield of sunflower is correlated positively to temperature and negatively to precipitation for both the observed and simulation data. Simulated crop yield of sorghum is correlated negatively with precipitation and the aridity index, but is strongly positively correlated with temperature.
Impacts potentiels des changements climatiques et des activités anthropiques régionales dans les bassins versants de méso-échelle d’Europe centrale
Published online:
18 May 2015Fig. 10 Observed (Obs) crop yields for Germany (D), Saxony (SN) and Brandenburg (BB), and simulated (Sim) crop yields for the catchments of Schwarze Elster (SE), Dahme (DA), Spree (SP) and Lusatian Neisse (LN). Correlations r with climate variables: temperature (T), precipitation (P) and aridity index (Ai), as well as numbers of data used (n), are shown.
![]({"type":"image","src":"/na101/home/literatum/publisher/tandf/journals/content/thsj20/2015/thsj20.v060.i05/02626667.2014.968571/20150710/images/medium/thsj_a_968571_f0010_b.gif"})
Fig. 10 Observed (Obs) crop yields for Germany (D), Saxony (SN) and Brandenburg (BB), and simulated (Sim) crop yields for the catchments of Schwarze Elster (SE), Dahme (DA), Spree (SP) and Lusatian Neisse (LN). Correlations r with climate variables: temperature (T), precipitation (P) and aridity index (Ai), as well as numbers of data used (n), are shown.
Despite the absolute crop yields being hard to evaluate due to limited data availability, the correlations between crop yields and climate variables are well reproduced. This provides the basis for assessing potential changes in crop yield associated with climate change. In accordance with the correlations shown, changes in temperature and precipitation also influence simulated crop yields in the scenarios (Fig. 11). In scenarios with increasing temperature and decreasing precipitation, decreasing crop yields of the winter crops, winter wheat and oilseed rape, but increasing crop yields of the summer crops, silage maize, sunflower and sorghum, are simulated. For winter wheat, the results match the simulation results of Wessolek and Asseng (2006) for Brandenburg, and of Hattermann et al. (2007) for the Elbe River catchment, but disagree with the observations of increasing crop yields for wheat in the Czech Republic by Chloupek et al. (2004). Increasing crop yields for silage maize correspond with the findings of Chloupek et al. (2004), but differ from simulation results obtained by Bassu et al. (2012). Besides, a simulated decrease in crop yield of oilseed rape is contrary to the findings of Chloupek et al. (2004). The changes in crop yield are stronger for the colder and wetter Lusatian Neisse catchment than for the lowland catchments.
Concerning crop yields, it should be noted that, for example, Asseng et al. (2013) and Bassu et al. (2012) found high uncertainties in yield simulations of different crop models. Other sources of uncertainty include: (a) that climate change impacts might be secondary to impacts of management and fertilization (Olesen and Bindi 2002), and (b) it is still challenging to predict the effects of elevated CO2-levels on plant growth (Ewert et al. 2005, Körner et al. 2007, Leakey et al. 2009) that are not included in the simulations of this study. Furthermore, comparisons with observations (e.g. Chloupek et al. 2004) might sometimes be misleading, as climate change assumed in the scenarios STAR 2K, STAR 3K and WettReg A1B goes beyond historical observations. Due to interactions between crop growth and water consumption, uncertainties in the simulation of crop growth also cause uncertainties in the simulation of water balance components. Due to declining simulated crop yields of oilseed rape and, additionally, as oilseed rape should not be grown 2 years in a row on the same field, an extremely high cultivation of oilseed rape is not to be expected. Therefore, it can be assumed that climate change impacts on discharge are of higher importance than the impacts of increased cultivation of energy crops over the entire catchment.
Impacts potentiels des changements climatiques et des activités anthropiques régionales dans les bassins versants de méso-échelle d’Europe centrale
Published online:
18 May 2015Fig. 11 Changes in simulated crop yield of the scenario period 2046–2055 of STAR 0K (S 0K), STAR 2K (S 2K), STAR 3K (S 3K) and WettReg A1B (W A1B) compared to the reference period (1961–1990, observed climate).
![]({"type":"image","src":"/na101/home/literatum/publisher/tandf/journals/content/thsj20/2015/thsj20.v060.i05/02626667.2014.968571/20150710/images/medium/thsj_a_968571_f0011_b.gif"})
Fig. 11 Changes in simulated crop yield of the scenario period 2046–2055 of STAR 0K (S 0K), STAR 2K (S 2K), STAR 3K (S 3K) and WettReg A1B (W A1B) compared to the reference period (1961–1990, observed climate).
Potential impacts of mining scenarios on discharge
The impacts of mining scenarios, in terms of the area of the groundwater depression cone, are illustrated in Fig. 12. Compared to the reference based on observed climate and mining, assumed to be similar to the No Mining scenario, decreasing discharges are simulated for all combinations of climate and mining scenarios. For the same climate scenario, discharge is generally highest for the No Mining scenario and lowest for the scenario Status Quo. For the STAR 0K scenario, the differences between the mining scenarios are stronger than the decrease in discharge of the No Mining scenario compared to the reference. For the other climate scenarios, the decrease in discharge compared to the reference is higher than the differences between the mining scenarios. However, except for the STAR 0K scenario, the differences between discharges simulated in different climate scenarios are lower than those between the No Mining scenario and the Status Quo scenario. For the gauging station Lauchhammer, higher discharge is simulated in the Phasing Out scenario in both STAR 3K and WettReg A1B, in comparison to the results in Status Quo and STAR 2K. For the other gauging stations, lower discharge is simulated in the Phasing Out and WettReg A1B scenarios compared to Status Quo and STAR 2K. In general, except for catchments with a high proportion of area affected by the groundwater depression cone, the impacts of climate change in terms of increasing temperature and decreasing precipitation associated with the scenarios STAR 2K, STAR 3K and WettReg A1B are more dominant than those associated with a decreasing groundwater depression cone.
Impacts potentiels des changements climatiques et des activités anthropiques régionales dans les bassins versants de méso-échelle d’Europe centrale
Published online:
18 May 2015Fig. 12 Mean annual discharge (Q) for the reference period (1961–1990) based on observed climate (Ref) and WettReg (W Ref) and for the scenario period (2046–2055) under the climate scenarios STAR 0K (S 0K), STAR 2K (S 2K), STAR 3K (S 3K) and WettReg A1B (W A1B) and the mining scenarios No Mining, Status Quo and Phasing Out.
![]({"type":"image","src":"/na101/home/literatum/publisher/tandf/journals/content/thsj20/2015/thsj20.v060.i05/02626667.2014.968571/20150710/images/medium/thsj_a_968571_f0012_b.gif"})
Fig. 12 Mean annual discharge (Q) for the reference period (1961–1990) based on observed climate (Ref) and WettReg (W Ref) and for the scenario period (2046–2055) under the climate scenarios STAR 0K (S 0K), STAR 2K (S 2K), STAR 3K (S 3K) and WettReg A1B (W A1B) and the mining scenarios No Mining, Status Quo and Phasing Out.
SUMMARY AND CONCLUSION
The model SWIM was successfully calibrated for sub-catchments and parameterized for the heavily anthropogenically-impacted catchments of the rivers Schwarze Elster, Dahme, Spree and Lusatian Neisse by model parameter regionalization. Based on the model, a scenario analysis was performed including regional climate scenarios, scenarios assuming an increasing cultivation of energy crops, as well as scenarios about the potential development of the groundwater depression cone related to lignite mining activities. The results suggest that climate change impacts on natural discharge are substantial and exceed the effects of an increasing cultivation of energy crops, as well as a decreasing groundwater depression cone. This is consistent with the results of other studies in the wider region, namely the Elbe River basin and catchments in northeastern Germany, which show that climate change impacts have a greater influence than land-use changes on discharge (e.g. Wechsung et al. 2000, Lahmer et al. 2001, Nathkin et al. 2012, Conradt et al. 2012b) and that changes in different annual crops only have minor impacts on discharge (Klöcking and Haberlandt 2002, Niehoff et al. 2002).
Also, decreasing crop yields were found with increasing temperature and decreasing water availability for wheat and oilseed rape, grown as winter crops in the region. In contrast, for the summer crops, silage maize, sunflower and sorghum, temperature-induced increases in crop yield were simulated.
While climate change impacts on discharge are stronger in the lowland catchments than in the Lusatian Neisse catchment, climate change impacts on crop yield are more pronounced for the Lusatian Neisse catchment. The former can be explained by the fact that drier catchments are more vulnerable to climate changes (Mengistu et al. 2013), the latter by temperature limitation of crop growth, which is more important in the colder Lusatian Neisse catchment.
Uncertainties relating to potential climate change have been considered by using daily climate variables of the regional climate models STAR (three scenarios with 100 realizations each) and WettReg (one scenario with 10 realizations). The scenarios, assuming an increasing cultivation of energy crops, are to be understood as extreme scenarios to estimate the highest potential impacts of arable energy crop cultivation on discharge in the study region. In addition, two extreme and one medium scenario of the potential development of the groundwater depression cone were considered. Hence, the results obtained in this study represent a bandwidth of potential future natural discharges in the study region. They provide the basis for long-term water management modelling of managed discharges in the region, where mining discharges, water users, reservoirs and water transfers are included (Pohle et al. 2014). Furthermore, potential climate change impacts on water quality are assessed by water quality modelling focusing on the mining-related water quality problems of the region, low pH and high concentrations of iron and sulphate (Zimmermann et al. 2014).
Disclosure statement
No potential conflict of interest was reported by the author(s).
| Initial values | Adjusted values | |
|---|---|---|
| Global model parameters | ||
| Baseflow factor | 1 | 0.3 |
| Initial groundwater flow contribution to streamflow | 0.2 mm d-1 | 0.02 mm d-1 |
| Curve numbers for dry/intermediate/wet conditions | 50/55/80 | 50/55/55 |
| Routing coefficients for surface flow/subsurface flow | 3/13 | 1/1 |
| Snowfall temperature/snowmelt temperature | 0.0°C/0.0°C | 0.3°C/1.2°C |
| Snowmelt rate | 4.0 mm d-1 | 2.2 mm d-1 |
| Catchment-specific model parameters | ||
| Correction for potential evapotranspiration | 1.0 | 0.9–1.3 |
| Groundwater recession rate: fast groundwater storage | 0.5 | 0.01–0.1 |
| Groundwater recession rate: slow groundwater storage | 0.5 | 0.0005–0.005 |
| Correction coefficient for saturated hydraulic soil conductivity | 1 | 1–13 |
| Riparian zone parameter | 1 | 0.6–1.0 |
Acknowledgements
The authors thank the Saxon State Office for the Environment, Agriculture and Geology, and the Landesamt für Umwelt, Gesundheit und Verbraucherschutz for data provision. The authors gratefully thank Mike White, Hamid Solaymani and one anonymous referee for their constructive comments and recommendations.