2 K. Granlund et al. Agriculture, Ecosystems and Environment 80 2000 1–13
fields or intensive livestock production and fur farms are located adjacent to groundwater formation areas.
Dissolved reactive
phosphorus, being
totally bioavailable, has been considered to be of crucial im-
portance in controlling the eutrophication of surface waters in Finland. In recent years, however, the role
of nitrogen as a limiting factor in primary production in lakes and rivers has become more clear. Pietiläi-
nen 1997 carried out a comprehensive survey of nutrient status in Finnish lakes and rivers, based on
the ratio of dissolved inorganic nitrogen to dissolved inorganic phosphorus. The results showed that 27
of the lakes and 25 of the rivers were nitrogen lim- ited. Furthermore, nitrogen is the limiting factor for
algal growth in the Baltic Sea during the major part of the year Granéli et al., 1990; Kivi et al., 1993. As
Finnish agriculture is concentrated in coastal areas or in river valleys discharging to the Baltic Sea, the need
to reduce nitrogen losses is greatest in these areas Valpasvuo-Jaatinen et al., 1997.
According to a Decision-in-Principle by the Finnish government in 1998 Ministry of the Environment,
1998, nitrogen loading from agriculture should be re- duced by 50 from the level in the beginning of the
1990s by 2005. Additionally, international agreements within the HELCOM framework aim at similar reduc-
tions in nitrogen loads to the Baltic Sea HELCOM, 1994.
In order to reduce the agricultural nutrient losses, the Finnish Agri-Environmental Support Scheme
was implemented when Finland joined the European Union in 1995, in keeping with Common Agriculture
Policy EEC, 1992. The goal of the Scheme is to en- sure the development of agricultural practices towards
higher sustainability. In a priori assessments it has been estimated that a 25–40 reduction in nutrient
losses may be achieved by full implementation of the Scheme Valpasvuo-Jaatinen et al., 1997. Accord-
ing to this programme farmers are paid for adopting environmentally sound management practices. A de-
tailed description of the programme was presented by Valpasvuo-Jaatinen et al. 1997. In 1998, more
than 90 of Finnish farmers participated in the pro- gramme. During the first 3 years 1995–1997 of the
programme, its implementation and changes in agri- cultural practices have been monitored by interview-
ing farmers in four study areas in Finland Grönroos et al., 1998.
A monitoring network of water quality is needed to evaluate the actual impacts of altered practices. How-
ever, due to slow responses of water quality to changes in practices and high fluctuations in annual nutrient
losses caused by meteorological variation, monitoring programmes cannot detect changes occurring within
short time periods. The objective of this study was to assess the potential impacts of changes in cultivation
practices on nitrate leaching in the study areas by using a mathematical simulation model. The SOIL-N
nitrogen model Johnsson et al., 1987 was used as a tool to estimate the potential changes in nitrate losses
on the basis of information about crops, soils, fertilisa- tion rates and management practices. This paper gives
an overview of the regional changes in nitrogen fer- tilisation rate resulting from the Agri-Environmental
Support Scheme and presents the calculated potential changes in nitrate leaching caused by changes in
fertilisation rate.
2. Materials and methods
2.1. Data collection Detailed management data based on field-parcels
were collected from farms in four study areas in differ- ent parts of the country Fig. 1, representing different
natural conditions and agricultural structures. Alto- gether about 400 farmers were interviewed. The study
areas varied from 35 to 1373 km
2
, and the agricultural land use from 10 to 50. The dominating particle size
in soil is clay in Lepsämänjoki, clay and fine sand in Yläneenjoki, silt in the most eastern area Taipaleen-
joki and fine sand in the northern area Lestijoki Fig. 2. The clay soils are most commonly Vertic
Cambisols, loam and silt soils Eutric Cambisols and the coarsest silt loam and sandy soils with less than 8
of clay are Dystric Regosols or Haplic Podzols Food and Agriculture Organization, 1988; Rasmussen et al.,
1991. In the two southern areas, cereal production is predominant, whereas in the two others dairy produc-
tion, and thus also grass cultivation is common. Most of the livestock farms in the study areas are dairy and
beef cattle farms, except in Yläneenjoki, where pig and poultry husbandry is also common. In Lepsämän-
joki, only 20 of the farms are livestock farms. The proportion of livestock farms of all farms and the total
K. Granlund et al. Agriculture, Ecosystems and Environment 80 2000 1–13 3
Fig. 1. Location of the Lepsämänjoki, Yläneenjoki, Taipaleenjoki and Lestijoki study areas.
amount of animals expressed as livestock units in each study area are presented in Table 1. According to the
Support Scheme 1 ha per 1.5 livestock units is needed for manure application. For cattle, the livestock units
vary from 0.18 to 1, for pigs from 0.14 to 0.5 and for poultry from 0.0028 to 0.011. The total number
of livestock units for cattle is highest in Lestijoki. In Yläneenjoki the number of pigs and poultry in terms
of livestock units is much higher than that of cattle.
The interview was conducted twice: in 1996 cov- ering the years 1994 and 1995, and in 1998 covering
the years 1996 and 1997. The years 1994 and 1995 represent agricultural practice before adopting the
Agri-Environmental Support Scheme. The interview was concentrated on those practices to which the
Scheme sets restrictions or limits. The monitored variables were: fertilisation, storage and spreading of
Table 1 The proportion of livestock farms of all farms and total number of livestock units in the study areas in 1997
Area Livestock farms per cent of all farms
Total number of livestock units Cattle
Pigs Poultry
Cattle Pigs
Poultry Lepsämänjoki
17 3
– 337
118 7
Yläneenjoki 19
15 15
631 1062
1347 Taipaleenjoki
66 –
– 612
– –
Lestijoki 75
2 –
1565 67
– Fig. 2. Distribution of soil types in the study areas per cent of
cultivated area.
manure, winter green coverage, animal density, use of pesticides, soil tillage methods, management of
set-aside and filter strips, and management of silage effluents. The main interest in this study was to eval-
uate the potential impacts of changes in nitrogen fertilisation rates and manure application.
2.2. Modelling The soil profile model SOILSOIL-N Jansson and
Halldin, 1979; Johnsson et al., 1987 was used to estimate nitrate leaching for different crops and soil
types under varying fertilisation regimes. The model has been widely used and tested e.g., in Sweden and
Finland for studying hydrological processes in the soil profile and leaching of nitrate from agricultural land
e.g., Johnsson et al., 1987; Borg et al., 1990; Johns- son, 1990; Johnsson and Jansson, 1991; Kätterer,
1995; Rekolainen and Leek, 1996; Kallio et al., 1997. The results of the previous model applications have
shown that the model can describe mineral nitrogen in soil and nitrogen leaching satisfactorily. Thus, it was
4 K. Granlund et al. Agriculture, Ecosystems and Environment 80 2000 1–13
assumed that the model is applicable to conditions of the present study.
Mean nitrate leaching kg ha
− 1
per year was calcu- lated for different crops and soils in four study areas
in different parts of the country. For each combina- tion, 10-year weather records were used to calculate
the mean annual leaching value in order to eliminate annual variations caused by climatic fluctuations. In
order to make regional assessments, the results of these model calculations were combined with digital
spatial data about soils, crops and fertilisation using GIS-software.
2.2.1. Description of the model The model is applied in two phases. First, the soil
water content and heat conditions are calculated with a one-dimensional water and heat model SOIL, Jans-
son and Halldin, 1979 for each soil layer. The results are then used as driving variables in a nitrogen model
SOIL-N, Johnsson et al., 1987 for calculation of ni- trate leaching from the soil profile.
Fig. 3. Structure of the SOIL-N nitrogen model showing state variables boxes and flows arrows. The structure is replicated for each soil layer Johnsson et al., 1987.
The SOIL model has a one-dimensional vertical structure, with the profile divided into layers ac-
cording to site-specific physical and biological soil properties. The model is based on two coupled partial
differential equations describing heat and water dy- namics in a soil profile derived from Fourier’s and
Darcy’s laws, respectively. Snow dynamics, frost, evapotranspiration, infiltration, surface runoff and
drainage flows are included. Using meteorological data as input, the model predicts the driving variables
for the SOIL-N model i.e., daily values of surface runoff and infiltration, water flow between soil layers
and flow to drainage tiles and to deeper groundwater, unfrozen soil water content and soil temperature in
different soil layers. Information about soil character- istics, such as pF-curves and hydraulic conductivities
is needed as input data, as well as parameters related to the calculation of evapotranspiration.
The SOIL-N nitrogen model Fig. 3 includes the major processes determining inputs, transformations
and outputs of nitrogen in agricultural soils. Inor-
K. Granlund et al. Agriculture, Ecosystems and Environment 80 2000 1–13 5
ganic fertiliser, manure and atmospheric deposition are inputs to the uppermost soil layer. Losses through
denitrification and leaching of nitrate to drainage tiles can occur from each soil layer. Nitrate in solution can
be transported between soil layers or to drainage tiles. Inorganic and organic nitrogen pools are replicated
for each soil layer. Mineral nitrogen pools include ammonium and nitrate. Organic nitrogen pools are lit-
ter, manure-derived faeces and humus. Carbon pools for litter and faeces are included for controlling ni-
trogen mineralisation and immobilisation rates. The litter component includes undecomposed material,
whereas stabilised decomposition products make up the humus component. The single plant component
includes nitrogen in both above- and below-ground biomass. Root distribution within the soil profile is
also taken into account. Ammonium is considered to be immobile in the soil profile.
Mineralisation of humus nitrogen is calculated as a first-order rate process controlled by a specific min-
eralisation constant and response functions for soil temperature and moisture. Decomposition of the litter
and faeces carbon pools is calculated in the same way. The net mineralisation of litter nitrogen is determined
by the balance between the release of nitrogen during decomposition and the nitrogen immobilised during
microbial synthesis and humification. Mineralisation of faeces is handled in the same way. Plant uptake of
nitrogen is calculated using a simple time-dependent empirical function requiring parameter values specific
for the crop and site considered. Total crop nitrogen uptake is used as input for the simulations, and can be
measured or estimated from harvest yield. The total uptake is distributed as N uptake from the soil during
the whole growing season. At harvest and ploughing the roots and harvest residues are incorporated into
the soil litter pool. Denitrification is calculated as a first order rate process controlled by potential rate and
response functions accounting for the effects of soil temperature, soil oxygen status and soil nitrate content.
2.2.2. Input data and parametrisation of the model The nitrate leaching was simulated for a 10-year
period 1981–1990 using meteorological data sepa- rately for each of the four study areas. The daily values
of temperature, humidity, wind speed, precipitation and cloudiness measured at weather stations located
within the study areas were used as input driving vari-
Fig. 4. Mean annual precipitation a and temperature b in the study areas 1981–1990.
ables to the model. The mean annual precipitation and temperature in the study areas are shown in Fig. 4.
Nitrate leaching was calculated for eight texturally most common soil types in the study areas. The tex-
tures of the soils varied from clay to silt loam. The water retention characteristics estimated for each sim-
ulated soil were considered to represent typical soils falling into these eight classes. The particle size dis-
tribution and physical characteristics of the soils are shown in Table 2. The soil profiles were divided into
five layers 4×0.25 m and 1 m down to a depth of 2 m, where free drainage was assumed. The physical
soil characteristics were assumed to be uniform for the whole profile. For all soils the organic matter content
was set to a constant value of 30 g kg
− 1
in the topsoil and 3 g kg
− 1
beneath the topsoil, which is slightly less than the average contents of organic matter in soils of
Finland Sippola, 1974; Sippola and Tares, 1978.
6 K. Granlund et al. Agriculture, Ecosystems and Environment 80 2000 1–13
Table 2 Textural composition and physical characteristics of the soils used in the simulations
Soil Textural class
Composition per cent Porosity
Water content Saturated hydraulic
at pF 4.2 conductivity
Food and Agriculture Organization, 1990
Clay
a
Silt
b
Sand
c
vol per cent vol per cent
cm h
− 1
1 Clay
72 25
3 54
41 0.6
2 Silty clay
45 49
6 50
31 1.2
3 Clay loam
37 27
33 50
26 1.5
4 Silty clay loam
33 50
17 44
18 3
5 Loam
19 38
43 41
12 9
6 Silt loam
12 79
6 60
9 9.7
7 Silt loam
8 70
15 45
4 3.3
8 Silt loam
6 54
40 42
7 9
a
0.002 mm.
b
0.002–0.06 mm.
c
0.06–2 mm.
Estimation of hydrological parameters was based on information on soil textures collected from farmers.
The calculation of unsaturated hydraulic conductivity and water retention characteristics was based on the
methods described by Mualem 1976 and Brooks and Corey 1964. Soil water content and soil temperature
were calculated for each soil and area for six differ- ent crop groups. Rape Brassica rapa oleifera DC.
Metzg. and spring cereals barley, Hordeum vulgare L. and spring wheat, Triticum aestivum L. were
assumed to have similar properties regarding evapo- transpiration. One common set of parameters was also
used to simulate the water balance of the winter cere- als winter rye, Secale cereale L. and winter wheat.
Separate parameter sets were developed for sugarbeet Beta vulgaris var. altissima Döll, potato Solanum
tuberosum L., silage and hay e.g., Timothy, Phleum pratense L.. Despite the fact that the same species are
cultivated for silage and hay, different sets of param- eters were used for these crops because their manage-
ment differs in terms of harvesting times. The length of the growth period for each crop was kept equal for
the two southern Lepsämänjoki and Yläneenjoki and the two northern areas Taipaleenjoki and Lestijoki.
Estimation of parameters related to the nitrogen cy- cle was mainly based on previous applications of the
model e.g., Johnsson et al., 1987; Borg et al., 1990; Johnsson, 1990; Rekolainen and Leek, 1996; Kallio
et al., 1997. The values of potential crop nitrogen up- take were given as input values. These values were
based on information on average yields in the study areas and assumed relationships between grain, straw
and root biomass. Data on nitrogen concentration of grain and straw was mainly derived from literature
e.g., Haak, 1993. For each crop the potential N up- take was independent of the fertilisation scheme or
rate and it was kept constant for each simulated year. Same initials values of litter-N and mineral-N were
used in all simulations and they were estimated based on normal amounts of these pools in previous ap-
plications. The model was not calibrated in different areas with respect to mineralisation constants or abi-
otic response functions for temperature and moisture.
Information about management practices e.g., sowing and harvest dates and use of fertilisers was
collected from the interview data. At Yläneenjoki and Lepsämänjoki simulations were made for barley,
spring wheat, rape, sugarbeet, potato, winter rye, win- ter wheat, silage and hay. At Taipaleenjoki and Lesti-
joki only barley, winter rye, potato, silage and hay were modelled. For silage and hay the parameters re-
lated to management practices were identical in all the study areas, whereas for the other crops the two south-
ern and the two northern areas were treated separately.
Several N fertilisation schemes and rates were used in the simulations to estimate the effects on nitrate
leaching Table 3. These schemes were the same for all the study areas. A fertilisation scheme based only
on various rates of mineral fertiliser Scheme 1, Table 3 was simulated for all the selected crops. In addition,
combined fertilisation schemes mineral+organic, Schemes 2–5, Table 3 were applied for cereals and
K. Granlund et al. Agriculture, Ecosystems and Environment 80 2000 1–13 7
Table 3 The amounts of inorganic and organic nitrogen fertilisers as soluble N for different crops used as input data in the model calculations
Crop Fertilisation
Inorganic N Slurry
Farmyard manure scheme
kg ha
− 1
per year N kg ha
− 1
per year N kg ha
− 1
per year N Spring
Harvest Autumn
Spring Autumn
Spring Autumn
Barley 1
60–140 –
– –
– –
– 2
10–100 –
– 40–110
– –
– 3
35–115 –
– –
50 –
– 4
60–100 –
– –
– 40–105
– 5
60–100 –
– –
– –
40–105 Spring wheat
1 70–150
– –
– –
– –
2 10–100
– –
40–110 –
– –
3 35–115
– –
– 50
– –
Winter wheat, rye 1
30–110 –
30–110 –
– –
– 2
90–130 –
– –
30 –
– Rape
1 60–140
– –
– –
– –
2 10–100
– –
40–110 –
– –
3 35–115
– –
– 50
– –
Sugarbeets 1
120–200 –
– –
– –
– Potato
1 20–100
– –
– –
– –
Grass for silage
a
1 100
60–90 –
– –
– –
Grass for hay
a
1 60
36–54 –
– –
– –
a
After the first harvest.
rape. A simplified three phase fertilisation scheme was applied for perennial silage and hay. When es-
tablishing the grass in autumn first year, a constant amount of slurry was applied. Inorganic fertiliser was
applied twice: first in spring, and then in summer af- ter the first harvest. No fertiliser was applied after the
second harvest. The amount of inorganic fertilisers varied depending on crop type and time of appli-
cation between 10 and 200 kg ha
− 1
per year N, and that of manure between 20 and 60 m
3
ha
− 1
per year, producing 40–110 kg ha
− 1
per year of soluble N. The range of fertilisation rates chosen for each crop
was based on crop-specific information on actual use of fertilisers in the study areas. Averaged values for
manure application rates and nitrogen storages in ma- nure were used as input. The estimation of nitrogen
content in different fractions of manure soluble N, organic N and N in bedding was based on litterature
e.g., Kemira Agro Oy, 1997. The nitrogen content of precipitation and dry deposition were kept constant for
all areas 1 mg l
− 1
and 0.001 g m
− 2
per day N, respec- tively. The estimation of these values was based on air
quality measurements carried out by the Finnish Me- teorological Institute Leinonen, 1996 and on model
calculations by EMEP Long-Range Transport LRT models Tuovinen et al., 1990.
Instead of loss prediction in absolute terms, in- formation on relative differences of losses is useful
in decision-making. Taking this into account, it was considered that earlier model validations in similar
environmental conditions could provide justification for the use of the model for this purpose see e.g.,
Johnsson et al., 1987; Borg et al., 1990; Bergström and Jarvis, 1991. In addition, general information on
nitrate losses in Finland and Sweden was collected and used for evaluating the model results. The cal-
culated nitrate losses were compared with reported measured nitrate losses for several cropping systems
Jaakkola, 1984; Yläranta et al., 1993, for several soil types Yläranta et al., 1993, for different fertilisation
rates Jaakkola, 1984 and for different manure appli- cation schemes Niinioja, 1993; Kemppainen, 1995.
Generally, the model results were in agreement with these observed loss estimates. Comparison of model
results with measured data is discussed in Section 4.
2.3. Upscaling In order to estimate regional losses the model
results for each climate-soil-crop-fertilisation combi- nation were connected to the spatial data for actual
field parcels in the study areas using GIS-software.
8 K. Granlund et al. Agriculture, Ecosystems and Environment 80 2000 1–13
The spatial database contained the corner coordi- nates and the soil type and management data for the
field parcels. To calculate the regional losses and their changes, the loss estimates for each field parcel
were summed up separately for all farms in the four study areas for the years 1995 and 1997. Average
regional nitrate leaching in kg ha
− 1
per year was then calculated for both years and for all the study
areas.
3. Results