Climate change and runoff from agricultural catchments in Norway Open Access

Agriculture contributes a significant portion of the nutrient load to the environment, being to a large degree responsible for the eutrophication of inland surface and coastal waters. Farming practices, soil types, and climatological conditions are important factors in nutrient and soil loss generation from agriculturally dominated catchments. Many studies have shown that runoff, hydrological pathways, and catchment scale also play an important role. Deelstra et al. (2010) showed that there are significant differences in hydrological behaviour when comparing small and large agricultural dominated catchments. Tiemeyer et al. (2006) observed that measurement scale influences nitrogen loss due to land use diversity and in‐stream processes leading to reduced NO₃‐N concentrations at higher scales. Vagstad et al. (2004) found that catchments having a large contribution of groundwater in the total runoff, in general had lower nitrogen losses. By taking into consideration different time resolutions, improved insight into the dynamics of agricultural hydrology can be obtained and Deelstra and Iital (2008) observed that the magnitude of the diurnal variation in discharge is important when considering nutrient loss processes.

Deelstra et al. (2010) observed that small agricultural catchments in Norway showed large diurnal variations in discharge compared to similar catchments in Baltic countries and that next to topography and soil types, subsurface drainage intensity played an important role in runoff and nutrient loss processes. A good understanding of hydrological processes is necessary to be able to effectively deal with the effects of climate change on nutrient and soil loss. Armstrong et al. (1992) in a study on the sensitivity of agricultural drainage systems to changes of climatic inputs concluded among others that the variation in required drain spacing due to changes in the rainfall inputs is small compared to the variation of drain spacing resulting from a more demanding performance required by changes in crop type. Several studies have been carried out using numerical models to obtain insight into the potential effects of climate change on runoff and nutrient loss. Arheimer et al. (2005), in a study on the impacts of climate change on water quality in Sweden, showed that the nitrogen load to the sea increased by almost 20 per cent, mainly due to a more pronounced winter runoff. A similar study, carried out by Rosberg and Arheimer (2007) showed that climate change is predicted to affect the phosphorus leaching from arable land.

Bouraoui et al. (2004) showed that climate change had been the main driver for the observed yearly increase in nutrient loss in a small agricultural catchment in Finland. A similar study was carried out in Sweden by Ulén and Johansson (2009), showing that the increase in temperature had been responsible for increased nitrogen mineralisation and leaching to subsurface drains. Falloon and Betts (2010), in a review and assessment of climate change impacts, state that the two main sources of uncertainty are the projections of future climate change and their resulting impacts on water and agriculture.

Bearing this in mind, the main objective of this study has been to investigate the effects of extreme weather conditions on soil and nutrient loss from agricultural dominated catchments at different locations in Norway. No numerical models have been used in this study but the potential effects of climate change are based on an analysis ands interpretation of runoff and nutrient loss on existing times series in small agricultural catchments.

Four catchments – Hotran, Time, Skas‐Heigre, and Skuterud – have been selected to study the potential effects of climate change on runoff and nutrient loss. The catchments are part of the Agricultural Environmental Monitoring Programme in Norway (JOVA) and represent different climatological conditions, agricultural production systems/practices, and soil types (Figure 1). The main characteristics of the four catchments are summarised in Table I.

Skuterud is located in the south‐eastern part of Norway with grain as the main crop. The long‐term annual precipitation is relatively low compared to the other catchments. Hotran is located in the central part of Norway, with a slightly lower mean annual temperature compared to Skuterud but a considerably higher long‐term annual precipitation. The dominating crops are grain and ley. Hotran has the lowest long‐term mean annual temperature. The Time and Skas‐Heigre catchments are located in the south‐western part of Norway with ley as main crop but also cereals are grown. They have the highest long‐term annual precipitation and mean annual temperature. The topography of the catchments varies from flat to hilly while soil types vary from marine silt clay loam deposits in addition to a marine sand and moraine deposits. The natural drainage condition in most of the Norwegian agricultural soils is poor and therefore artificial subsurface drainage systems have been installed, with main objectives being guaranteeing optimal cropping conditions during the summer season and facilitating early land preparation during spring and soil tillage practices in the autumn.

At the same time, artificial subsurface drainage systems also reduce surface runoff and hence erosion and phosphorus loss. Subsurface drainage systems are composed of lateral suction drains and collectors conveying access water to open streams. The lateral suction drains have an internal diameter of Ø=0.05 m, while drain spacings vary from 8 to 10 m and depths from 0.8 to 1.0 m below soil surface. Skuterud and Hotran typically represent areas in Norway where erosion is a major problem, especially during the period after the growing season from September to April. Major erosion and nutrient runoff problems are related to excess precipitation and/or snowmelt in combination with frozen soils and which can lead to surface runoff induced erosion and nutrient loss, causing deteriorating water quality in streams, rivers, and lakes (Øygarden et al., 2006).

The Time and Skas‐Heigre catchments typically represent the area in Norway with a high animal density. Excess nutrient application, both in the form of artificial fertiliser and animal manure, has for long been a problem and both through surface and subsurface runoff causing deteriorating water quality. Detailed information about farming practices at the level of the individual farmer field is collected on a yearly basis in the Time and Skuterud catchments. Similar information, that is less detailed due to the practical limitations because of catchment size, is obtained from Statistics Norway for Hotran and Skas‐Heigre.

Discharge is measured at the four catchments based on a continuously recording of the water level in combination with a data‐logger. At Skuterud and Hotran catchments, a crump weir is used with a known relation between the water level, measured upstream of the crest, and the discharge, depending on its dimension (Crump, 1952). At Time and Skas‐Heigre, a rating curve, i.e. a relation between the discharge and water level, had been established for a fixed cross‐section in the main stream. On the basis of the continuous measurement of the discharge at the catchment outlet, the annual runoff is calculated. Composite water samples are taken automatically on a volume proportional basis (Deelstra and Øygarden, 1998; Deelstra et al., 1998), collected every 14 days and analysed for total nitrogen (TN), nitrate (NO₃‐N), total phosphorus (TP), dissolved phosphate (PO₄‐P), suspended solids (SS), loss on ignition (SR), turbidity, electrical conductivity, and pH. The samples are analysed according to Norwegian (NS) and international standardised procedures (ISO, EPA). On the basis of the continuous discharge measurement and the component concentrations in the water samples, the annual nutrient and soil loss per unit area is calculated.

The highest average yearly temperatures during the observation period were recorded at the Time and Skas‐Heigre catchments located in the south‐western part of Norway. There has been a considerable variation in average yearly temperatures during the observation period (Table II). The mean actual yearly precipitation during the observation period has been considerably higher than the long‐term annual precipitation. The Time and Skas‐Heigre catchments had a significantly higher average annual precipitation compared to the Skuterud and Hotran catchments with the highest annual precipitation recorded at Skas‐Heigre and the lowest at Hotran. A large difference exists between the minimum and maximum precipitation. The highest mean annual runoff during the observation period was measured at the Time catchment while Skuterud catchment had the lowest runoff. There is a large variation in yearly runoff which is a reflection of the variation in precipitation. The precipitation at Skuterud was extreme in 2000‐2001 (1,305 mm), almost twice the long‐term annual precipitation. More than 50 per cent occurred during the period from September to December and, as a consequence, also more than 50 per cent of the yearly runoff (1,042 mm) was generated during those months. On the other hand, Hotran during that same year, received a record low precipitation (473 mm), seriously affecting yield and leading to an extreme low runoff.

The low loss of suspended sediments at the Time and Skas‐Heigre catchments in the western part of Norway (Table II) is a reflection of the dominating cropping system at those catchments, being ley. High loss of suspended sediments occurred at the Skuterud and Hotran catchment, with cereals as the dominating crop. Significant differences in phosphorus loss exist between Time/Skas‐Heigre and Skuterud/Hotran (Figures 2 and 3). Erosion, or the loss of suspended sediments, contributes more to the phosphorus loss from cereal‐dominated catchments than from the pasture‐dominated catchments, which can explain the difference (Bechmann et al., 2008). The mean annual nitrogen loss varied from 31 to 48 kg/ha in the Skuterud and Time catchment, respectively. In addition to the large differences between the catchments, there is also a considerable variation in the yearly loss. For Skuterud, Hotran, and Skas‐Heigre, and to a lesser degree Time, the variation in the yearly phosphorus loss corresponds well with the variation in runoff (Figures 2 and 3).

Processes generating phosphorus loss in the Skuterud and Hotran catchments are mainly related to surface runoff induced erosion (Øygarden et al., 2006). On the other hand, phosphorus runoff in the Time and Skas‐Heigre catchments is very much related to the spreading of animal manure. The yearly nitrogen loss corresponds well with the yearly runoff (Figure 4). Similar to the runoff in 2000‐2001, nutrient and soil loss reached its maximum value for Skuterud during 2000‐2001, being a direct result of the high precipitation and runoff during that year. On the other hand, runoff, nutrient, and soil loss was lowest during this year in the Hotran catchment, a direct consequence of the extreme low precipitation.

The effects of climate change on temperature and precipitation have been calculated for different regions in Norway by Hanssen‐Bauer et al. (2009). The calculations are based on different scenarios for CO₂ emissions, global climate change models, and different downscaling methods. Uncertainty about future greenhouse gas emissions together with limitations in climate change models affects the uncertainty of predictions of the future climate. Hanssen‐Bauer et al. (2009) have taken this uncertainty into consideration by indicating a lower, mean, and maximum expected value of change.

The projected changes in temperature and precipitation for the four selected catchments were obtained and are presented in Table III (mean values used in this paper). The projected increase in the annual temperature is approximately the same for the four catchments, varying from ΔT=3.1‐3.4°C, with the largest increase occurring during the winter season. The increase in yearly precipitation varies from 12.2 to 22.5 per cent. For the Hotran catchment, there is a significant increase in the precipitation during all seasons while for the Skuterud catchment there is decrease in precipitation during the summer and an increase in the other seasons. There is no appreciable increase in precipitation during the summer at Time and Skas‐Heigre while the remaining seasons show a considerable increase.

The climate change scenario forecasts a considerable reduction in the number of days with temperatures below freezing and a reduction in the number of freeze/thaw cycles. The climate change scenario also forecasts a considerable decrease in the amount of snow, especially in the low‐lying areas while, in addition, the number of days with snow cover will decrease. More episodes with high rainfall intensity are predicted while those intensities are predicted to increase compared to today's situation.

The following characteristics have been calculated to evaluate the effects of climate change on runoff and nutrient loss:

• 1.

  Seasonality, referring to the relative contribution of the different seasons in the yearly precipitation, runoff, and nutrient and soil loss.

• 2.

  Runoff generation, or the number of days it takes to discharge a certain percentage of the yearly runoff, nutrient and soil loss, obtained by:

  • sorting the daily losses in descending order; and

  • calculating the cumulative sum of the daily losses for runoff, nutrients and SS.

• 3.

  The coefficient of variation (CV) and skewness, reflecting, respectively, the variation in discharge and whether the distribution of discharges is determined by a relatively small number of large flows (positive skewness).

• 4.

  The flashiness index (FI), giving an indication of the intensity in changes in discharge. The FI is obtained by calculating the total path length of flow, i.e. the sum of the differences between the average daily discharge values, and dividing this by the sum of the average daily discharges (Baker et al., 2004) in our case calculated for one year periods. The total path length can be either the sum of the differences between the average daily or hourly discharge values.

• 5.

  The maximum specific discharge, providing information about runoff intensities.

In this analysis, emphasis has been given to the effects of a temperature increase but more specifically to the effects of the increase in extremes in precipitation on runoff, nutrient loss, and soil loss. Extremes in this case are events caused by high amounts of precipitation over both short‐ and long‐time periods.

Seasonality has been calculated on the basis of the existing measurements in the four catchments (Table IV). In general, for the four catchments, there is a high contribution of precipitation during the summer season while runoff contribution is lowest mainly due to the high evaporative demand, this also leading to relatively low nutrient and soil loss contributions. During the rest of the seasons, nutrient and soil loss is related to runoff generation, their seasonality therefore showing similarities.

A study on the influence of climate change on stream flow in Danish rivers showed that seasonality would change, the main reason being the change in precipitation (Thodsen, 2007). Similar findings were made by Loukas et al. (2002) when studying the climatic impacts on runoff generation in Canada. Andersen et al. (2006) also concluded that the seasonality in runoff changed, leading to higher runoff during winter while summer runoff was increasing or decreasing depending on the groundwater contribution.

In the catchments here, the groundwater contribution in runoff generation is small compared to subsurface and surface runoff, as the majority of soils have a poor natural drainage condition. Water is transporting nutrients and SS to the open water courses and therefore a change in the seasonality pattern of runoff will have a direct effect on the seasonality of nutrients and SS. A considerable increase in the yearly precipitation is predicted for the four catchments which will probably lead to an increase in yearly runoff. For the summer season, a decrease in precipitation is predicted for the Skuterud catchment while for Time/Skas‐Heigre no significant changes in precipitation are predicted. Combined with a temperature increase, leading to an increase in the potential evapotranspiration, for both Skuterud and Time/Skas‐Heigre, a reduction in runoff, nutrient loss, and soil loss can therefore be expected during the summer season. A considerable increase in precipitation is predicted for the period after the growing season from September to April for Skuterud and Time/Skas‐Heigre, probably leading to more runoff. Based on the similarity between runoff and nutrient loss (Figures 2‐4), this increase in runoff will lead to an increase in nutrient and soil loss, under the assumption of similar land use, cropping condition, and farming practices. Results obtained by Arheimer et al. (2005), Rosberg and Arheimer (2007), Bouraoui et al. (2004) and Ulén and Johansson (2009), when using numerical models to predict the effects of climate change, confirm the projected increase in nutrient loss.

For Skuterud and Time/Skas‐Heigre, future seasonality will see a lower contribution of the summer season in the runoff, nutrient loss, and soil loss and an increase during the rest of the seasons. No major differences in the predicted precipitation increase between the seasons for the Hotran catchment are expected, and therefore no major changes in the seasonality are expected to occur.

Seasonality, potentially caused by climate change, might have occurred already during 2000‐2001 at the Skuterud catchment. After the growing season from September to December, a period with extreme high precipitation led to very high runoff, nutrient loss, and soil loss, the highest measured since the start of the monitoring and a significant difference from the mean seasonality values occurred. The runoff contribution during the autumn increased to 0.5, compared to 0.3 for the observation period (Table IV), while the nutrient and SS contribution increased to 0.6. Øygarden et al. (2002) reported on this event and concluded that in addition to tillage methods, preferential flow, bank erosion, and erosion around hydrotechnical installations contributed to the very high soil and nutrient loss.

Seasonality showed that runoff and nutrient loss occurs mainly after the growing season. In practice, however, the monitoring results showed that during the observation period until 2010, on average, 50 per cent of the yearly runoff is discharged in 25‐55 days while 90 per cent is discharged in 126‐211 days (Table V). For the Skuterud/Hotran catchments, it takes significantly fewer days to discharge 50 and 90 per cent of the yearly phosphorus and SS compared to Time/Skas‐Heigre. The reason for this is due to erosion processes at Skuterud/Hotran mainly occurring during periods with high flow when excess energy is available to detach and transport sediments and particle‐bound phosphorus. At Time/Skas‐Heigre, having a year‐round crop cover, and therefore significantly less erosion, the phosphorus loss mainly occurs in the dissolved form. The number of days used to discharge 50 and 90 per cent of the yearly nitrogen is very much similar to the runoff generation, the main reason for this being nitrogen transported in the dissolved form. A study, carried out by Roald et al. (2002), on the predicted effects of climate change showed that both the yearly runoff and its seasonal distribution will change in Norway. An increase in the yearly runoff in general leads to an increase in the average yearly runoff intensities.

However, for Skuterud and Time/Skas‐Heigre, the increase in yearly runoff will occur mainly after the growing season and hence fewer days will be available for this increased runoff to occur, which might indicate that the number of days available to discharge 50 and 90 per cent of the yearly runoff will further decrease. This will additionally increase the runoff intensities which again can enhance soil and phosphorus loss processes under otherwise unchanged agricultural practices.

As the predicted increase in precipitation will be almost similar for the different seasons, it is expected that, contrary to Skuterud and Time/Skas‐Heigre, no additional increase in runoff intensities due to changes in seasonality are foreseen.

As the discharge at the Hotran, Skuterud, Time, and Skas‐Heigre is recorded at a time resolution of one hour, the specific discharge, CV, skewness, and FI have been calculated both on the basis of hourly as well as average daily discharge values (Table VI). For all catchments, a significant increase in the characteristic values occurs when calculated at a higher time resolution. For Skas‐Heigre, the values are lower while at the same time the increase is less pronounced as the catchment is operated as a “polder”, the discharge being controlled by a pumping station.

The FI is obtained by calculating the total path length of flow, i.e. the sum of the differences between the average daily discharge values, and dividing this by the sum of the average daily discharges, in this case taken over a period of one year (Baker et al., 2004). A significant increase in the FI can be observed when calculating the path length as the sum of the differences between hourly discharge values, confirming the increases in CV and skewness. This is an indication that large diurnal variation in discharge exists which also explains the large difference in specific discharge when calculated on the basis of a higher time resolution (Table VI). Figure 5 shows an example of this for the Skuterud catchment, showing the yearly maximum specific discharge values for both time resolutions. Especially during periods with high runoff, considerable amounts of energy are present in the runoff, leading to the detachment and transport of soil particles and nutrients. Deelstra and Iital (2008), when comparing Estonian and Norwegian catchments, concluded that high FI values could partly explain the high nutrient losses in Norwegian catchments.

In addition to an increase in precipitation, more episodes with high rainfall intensity are predicted to occur while those intensities will increase compared to today's situation (Hansen‐Bauer et al., 2009). Michael et al. (2005), when using a numerical model to simulate the potential effects of climate change, concluded that increased precipitation intensities lead to increased erosion under unchanged land use conditions. Uncertainty exists as to whether these high intensity rainfall episodes will occur:

• 1.

  during periods with high rainfall over longer time periods;

• 2.

  as single rainstorms during summer; or

• 3.

  during freeze/thaw periods in winter.

Occurrences as (2) will cause least damage as the soil is protected by a crop against the high energy in rainfall. If occurring during longer periods with high rainfall amounts, erosion and nutrient loss will increase, an example of which already might have occurred in the Skuterud catchment (September‐December 2000 period). Øygarden (2003) reported about extreme runoff conditions, caused by a combination of high intensity rainfall, frozen soils, and snowmelt leading to severe rill and ephemeral gully erosion in the winter of 1990, even in areas normally not considered susceptible to erosion, representing the third occurrence. Milder winters are predicted to lead to a reduced snow cover.

Bakken et al. (2004) concluded that, due to a reduced insulating effect of snow combined with an increase in the number of freeze/thaw cycles, the infiltration capacity of the soil would be reduced leading to more surface runoff. More frequent freeze/thaw cycles will also reduce the aggregate stability/shear (Kværnø and Øygarden, 2006). Combined with the predicted more frequent occurrence of high intensity rainfall episodes, this can lead to serious erosion and nutrient loss. Cropping conditions in the Hotran catchments are similar to those at the Skuterud catchment and therefore events as described by Øygarden (2003) also can occur at Hotran. However, these are unlikely to happen at Time/Skas‐Heigre as these catchments are little exposed to winter conditions as experienced in Skuterud and Hotran. Simultaneously, the agricultural practices are different, being mainly permanent grass cover protecting against erosion. Armstrong et al. (1992) concluded that, in principle, subsurface drainage systems did not need modification to cope with changes in rainfall inputs. However, the situation is different when extreme rainfall episodes and/or snowmelt in combination with frozen soils occur. A more detailed analysis, using numerical models, probably has to be carried out to obtain insight into those episodes and how these affect the different hydrological flow paths, i.e. surface/subsurface runoff and groundwater contribution, and their potential effects on nutrient and soil loss processes.

Based on the analysis of hydrology and nutrient and soil loss in four small agricultural dominated catchments, some conclusions can be drawn on the potential effects of climate change. In addition to an increase in temperature, an increase in precipitation is predicted which in all four catchments probably will lead to an increase in runoff. Under similar land use and tillage methods, this will most likely also lead to an increase in nutrient and soil loss.

For the Skuterud and Time/Skas‐Heigre catchments, climate change is predicted to change the seasonality, with a lower contribution of runoff, nutrient loss, and soil loss during the summer season and an increase during the other seasons. There will be no major changes in the seasonality for Hotran, with runoff and nutrient loss increasing during all seasons.

For Skuterud and Time/Skas‐Heigre catchments, the increase in yearly runoff will occur mainly after the growing season and hence fewer days will be available for this increased runoff to occur. This will further increase the runoff intensities which additionally will enhance soil and phosphorus loss processes under otherwise unchanged agricultural practices.

In addition, more episodes with increased high intensity rainfall are predicted to occur. If those episodes occur after the growing season, they can lead to extreme runoff conditions, potentially causing severe erosion and nutrient loss under similar agricultural practices in regions represented by the Skuterud and Hotran catchments.

This analysis showed that large diurnal variation in discharge exists and which partly can explain the magnitude of soil and nutrient losses in small agricultural catchments. These variations have to be taken into consideration in the future design of hydrotechnical implementations, to be able to effectively deal with the effects of the predicted climate change on runoff, nutrient loss, and soil loss. A more detailed analysis, using numerical models, should be carried out to obtain improved information about the potential effects of climate change on runoff generation, the magnitude of the different hydrological flow paths, i.e. surface/subsurface runoff and groundwater contribution, and their potential effects on nutrient and soil loss processes, thereby being able to choose the right mitigation measures.

Johannes Deelstra is a Researcher at the Norwegian Institute for Agricultural and Environmental Research (Bioforsk), working with environmental aspects related to agricultural activities in agriculture‐dominated catchments. His main interests are related to agro‐hydrology, nutrient and soil loss, mitigation measures, and how these are affected by climate change. Johannes Deelstra is the corresponding author and can be contacted at: Johannes.Deelstra@bioforsk.no

Lillian Øygarden is a Researcher at the Norwegian Institute for Agricultural and Environmental Research (Bioforsk), working with environmental aspects related to agricultural activities in agriculture‐dominated catchments. Her main interests are related to agro‐hydrology, nutrient and soil loss, mitigation measures, and how these are affected by climate change.

Anne‐Grete B. Blankenberg is a Researcher at the Norwegian Institute for Agricultural and Environmental Research (Bioforsk), working with environmental aspects related to agricultural activities in agriculture‐dominated catchments. Her main interests are related to agro‐hydrology, nutrient and soil loss, mitigation measures, and how these are affected by climate change.

Hans Olav Eggestad is a Researcher at the Norwegian Institute for Agricultural and Environmental Research (Bioforsk), working with environmental aspects related to agricultural activities in agriculture‐dominated catchments. His main interests are related to agro‐hydrology, nutrient and soil loss, mitigation measures, and how these are affected by climate change.