Assessment of the intrinsic vulnerability of agricultural land to water and nitrogen losses : case studies in Italy and Greece

LOS indices (abbr. of Losses) can be used for the assessment of the intrinsic vulnerability of agricultural land to water and nitrogen losses through percolation and runoff. The indices were applied on the lowland region of Ferrara Province (FP) in Italy and the upland region of Sarigkiol Basin (SB) in Greece. The most vulnerable zones in FP were the coastal areas consisting of high permeability sandy dunes and the areas close to riverbanks and palaeochannels, and in SB were the areas characterized by high slopes and high permeability soils at high altitude and areas belonging to the upper part of the alluvial plain close to the boundaries between agricultural land and mountainous regions. The application of LOS indices highlighted the specific features of both lowland and upland regions that contribute to water and nitrogen losses and showed their ability for use as tools in designing environmental management plans.


INTRODUCTION
Nitrate leaching from agricultural land is a worldwide concern, and especially in Europe where a large part of the farmed areas has been affected by nitrate pollution for decades (Howden et al. 2011).Alluvial plains are usually the regions most impacted by nitrate pollution, particularly nitrate groundwater contamination (Cinnirella et al. 2005, Mastrocicco et al. 2011a).The most widely used method for determining water and agrochemical losses from agricultural land is use of process-based mathematical models that describe the water movement, transport and transformations of dissolved species through the soil profile, such as GLEAMS (Knisel and Davis 2000) and HYDRUS (Šimunek et al. 2008).However, simpler models such as classification indices (vulnerability or risk indices) use fewer and more accessible data such as climatic data, topography and general soil physical properties (Gogu and Dassargue 2000).Some popular indices for describing the intrinsic vulnerability of the groundwater to pollution are the DRASTIC index (Aller et al. 1985), SINTACS index (Civita and De Maio 1997), RISKE index (Petelte-Giraude et al. 2000), COP (Zwahlen 2003) and MERLIN (Aveline et al., 2010).The basic problem of these indices is the use of weights and ratings based on subjective criteria; they can be used only in regions where an aquifer system exists and they do not take or partly take into account the different climatic conditions that tend to change nitrogen cycle processes.These facts lead to limitations in their use, leading also to results that are not comparable between regions with different climatic conditions.Due to these limitations, a new approach to describing water and nitrogen losses from agricultural land was followed for the development of LOS indices (Aschonitis et al. 2012); it considers climate data, nitrogen cycle processes and gives special attention to the topsoil (unsaturated zone), which constitutes the door to groundwater pollution.LOS indices have the advantage that they are already calibrated via regression analysis using the results of the GLEAMS model (Knisel and Davis 2000) for the reference crop under various combinations of climate, soil and topographic conditions.
The aim of this study is to present the first application of the LOS indices methodology for the assessment of the intrinsic vulnerability of agricultural land to water and nitrogen losses, in a manner to compare results among different regions.To test the performance of this approach, two heavily impacted agricultural regions with different climate, soil types, land uses and groundwater attributes were selected: the deltaic lowland area of Ferrara Province (FP) in Northern Italy and the upland area of Sarigkiol basin (SB) in Greece.

MATERIALS AND METHODS
The LOS indices (Aschonitis et al. 2012) were developed to assess the intrinsic vulnerability of agricultural land to water (LOSW) and nitrogen (LOSN) losses through percolation and runoff based on the simulation results of the GLEAMS V3.0 model (Knisel and Davis 2000).Simulations were performed using reference field-crop characteristics, which concern the top 30 cm of a uniform reference square area of 1 ha with a homogeneous soil profile, cropped with perennial grass (with an extensive, uniform surface of dense actively growing coverage).The simulation scenarios consisted of various combinations of saturated hydraulic conductivity K s , soil organic matter content OM% (excluding organic soils with carbon:nitrogen ratios >30 and with high immobilization rates), surface slope S%, different climatic conditions (temperature T, precipitation PCP, potential evapotranspiration PE) and irrigation IR (non-irrigated and irrigated cases).Irrigation was adjusted to keep the soil moisture between 20% and 100% of the plant available water content in the soil profile.Fertilization type and rate was constant in all simulations.The parameters of K s , S%, PCP, PE and IR from the simulation scenarios were used as independent variables in multiple regression analysis to derive the following indices based on the GLEAMS results: the LOSW-P which describes the annual water losses through percolation (mm year -1 ), the LOSW-R which describes the annual water losses through runoff (mm year -1 ) and the LOSW-PR as the sum of water losses obtained from the previous two indices (mm year -1 ).Apart from the water losses, the LOSW indices can provide an indication of the fate of conservative pollutants.For the case of nitrogen, the same procedure was followed by adding another two independent variables, OM% and T, in order to derive the following indices: the LOSN-PN which describes the annual nitrogen losses through percolation (kg N ha -1 year -1 ), the LOSN-RN which describes the annual nitrogen losses through runoff (kg N ha -1 year -1 ) and the LOSN-PRN as the sum of nitrogen losses obtained from the previous two indices (kg N ha -1 year -1 ).The estimated water and nitrogen losses from the aforementioned indices were used to develop additional indices that provide the mean relative concentration of nitrogen in the percolated water CPW (mg N L -1 ) and the respective relative concentration of nitrogen in the runoff water CRW (mg N L -1 ).The depth of the groundwater table and the K s values were used to estimate the relative transit time TT (days) of percolated water to reach groundwater.The formulas defining the aforementioned indices, together with an explanation of their derivation, can be found in Aschonitis et al. (2012) while additional LOS indices describing the intrinsic rates of nitrogen transformations are provided in Aschonitis et al. (2013).
LOS indices were applied in two case studies: the FP (Italy), covering an area of 2623.4 km 2 , and the SB (Greece), covering an area of 469.2 km 2 (Fig. 1).The minimum, maximum and mean altitude for the FP are -11.0,38.0 and 0.5 m above sea level (a.s.l.), and for the SB are 640.0,1796.1 and 952.4 m a.s.l., respectively.Slope models for each region were obtained from national digital elevation models (DEMs).
The FP is part of the Upper Adriatic climatic sector and can be divided into two zones: a "coastal zone" stretching from the sea to 30 km inland and a "peri-coastal zone" located on the west side of the province.The whole area has a temperate climate with sub-continental characteristics, cold winters and warm summers.The SB is characterized by a semi-arid, Mediterranean climate with the most rainfall in autumn and spring, while summers are usually dry.Precipitation PCP and temperature T maps were developed from surface interpolation of data obtained at local meteorological stations.Potential evapotranspiration, PE, was determined using the Thornthwaite and Mather method (Black 2007) in both case studies.Irrigation, IR, was estimated as the sum of monthly deficits of PCP to cover PE (Aschonitis et al. 2013).
Land use for the study areas was identified using CORINE Land Cover 2000, while crop distribution was obtained from the reports of the regional administrative units.For FP, surface integration of hydraulic conductivity, K s , was carried out using values that were determined for the 12 soil texture classes according to the USDA soil taxonomy based on the Green-Ampt equation approach (Liu et al. 2008).Surface integration of organic matter was based on a wide number of field measurements that cover the entire region (SGSS, http://www.regione.emilia-romagna.it/wcm/geologia/index.htm).For SB, surface integration of K s was carried out using the average value of two pedotransfer functions (PTFs): the parametric sand dependent PTF of Ferrer-Julia et al. (2004) and the clay dependent PTF of Dane and Puckett (1994); organic matter was determined using Dexter's PTF (2004).
The FP surface waters consist of both natural and artificial bodies.The latter, used for drainage purposes, have a total length of more than 3500 km.In the FP, the unconfined aquifer texture is sand and silty-sand, with higher spatial heterogeneity due to the presence of palaeochannels and alluvial fans interconnected with alluvial fan and marsh lagoon environments (Amorosi et al. 2003).The unconfined aquifer thickness usually ranges between 2 and 7 m, but in some cases may exceed 45 m (Mastrocicco et al. 2011b).The water table depth ranges between 0.5 and 5.5 m below ground level (b.g.l.).Significant surface water bodies do not exist in the SB except for the Soulou torrent (or Eordaios), which was artificially excavated for use as the main drainage ditch to support secondary ditches.The alluvial deposits in the centre of the SB host an unconfined aquifer superimposed on successive confined or semi-confined aquifers, covering an area of approx.60 km 2 .The depth of the water table in the alluvial aquifer ranges from 7 to 75 m b.g.l., i.e. 585-730 m a.s.l.For both regions, maps of groundwater elevation were developed using surface interpolation of piezometric data.
The thematic base maps for both the FP and the SB were developed using ArcGIS 9.3 ESRI software and their mean pixel values are given in Fig. 1.

Case study of Ferrara Province (FP) -Italy
The application of LOSW indices in FP (Fig. 2(a)-(c)) showed that the water losses through percolation (LOSW-P) (Fig. 2(a)) are greater in the coastal region, where the soil texture is dominated by the sand of the dunes having high K s values.The water losses through surface runoff (LOSW-R) (Fig. 2(b)), are higher in the western and south-central regions of FP due to a combination of higher S, lower K s values and higher PCP.The resulting total water losses (LOSW-PR) (Fig. 2(c

Case study of Sarigkiol Basin (SB) -Greece
The application of LOSW indices in SB (Fig. 3(a)-(c)) showed that water losses through percolation (LOSW-P) (Fig. 3(a)) are lower in the centre of the basin and increase towards the fringes due to the transition from fine textured soils with low K s to coarser soils with higher K s , especially in the northeastern region.The water losses through surface runoff (LOSW-R) are higher in the eastern region due to higher slopes while medium values are observed in the centre of the region due to low K s values (Fig. 3(b)).The total water losses (LOSW-PR) show that the most vulnerable zone of SB is the northeastern region (Fig. 3(c)).Nitrogen losses through percolation (LOSN-PN) show a distribution analogous to LOSW-P (Fig. 3(d)).Water losses through surface runoff (LOSN-RN) (Fig. 3(e)) indicate higher values in the eastern region due to higher slopes.The total nitrogen losses (LOSN-PRN) (Fig. 3(f)) indicate that the most vulnerable zone is the northeastern region and the areas close to the boundaries of the aquifer.The highest nitrogen concentrations in percolated water (CPW) (Fig. 3(g)) are observed in the regions close to the edge of the aquifer, while the highest nitrogen concentrations in the runoff water (CRW) (Fig. 3(h)) occur in the northeastern region of SB.The relative transit time map (Fig. 3(i)) is restricted to only the region defined by the aquifer boundaries and values gradually increase from the lowland central-west part to the eastern upland region.
while LOSW-R values were approximately 18% higher in FP than in SB, leading to higher total water losses (LOSW-PR), a difference which can be mainly ascribed to the higher annual precipitation and the lower potential evapotranspiration in FP.The mean values of LOSN-PN, LOSN-RN, LOSN-PRN, CPW and CRW were similar in both regions.Taking into account the TT values, FP can be considered more vulnerable than SB due to the higher water table and the shorter transit time necessary to reach the groundwater for both water and nitrogen species.

Fig. 1
Fig. 1 Location of Ferrara province in Italy (FP) and of Sarigkiol basin (SB) in Greece, and the mean pixel values for each base map.

Table 1
Minimum, maximum and mean pixel values for each LOS index.