Wind erosion in the central Ebro Basin under changing land use management. Field experiments with a portable wind tunnel

Journal of Arid Environments 73 (2009) 996–1004

Contents lists available at ScienceDirect

Journal of Arid Environments journal homepage: www.elsevier.com/locate/jaridenv

Wind erosion in the central Ebro Basin under changing land use management. Field experiments with a portable wind tunnel W. Fister*, J.B. Ries Department of Physical Geography, Trier University, Behringstr., 54286 Trier, Germany

a r t i c l e i n f o

a b s t r a c t

Article history: Received 25 March 2007 Received in revised form 1 May 2009 Accepted 19 May 2009 Available online 21 June 2009

The agricultural landscape in the semi-arid central Ebro Basin is changing from dry farming towards land abandonment. This study aims to describe quantitatively the influence of this land use change onto wind erosion susceptibility in this region. Additionally, the effects of tillage operations on wind erosion rates were evaluated. A portable wind tunnel was used to assess the relative sediment loss rates at three test sites near Zaragoza. Three different land use systems varying in crust disturbance level were investigated – (1) fallow land with undisturbed physical soil crusts, (2) simulated sheep trampling and (3) conventional tillage (dry farming). The results show that simulations on undisturbed crusted soils produce little soil loss. Consequently, wind erosion can be considered as negligible on these surfaces. Simulated sheep trampling during wind tunnel test runs produce 10 times higher sediment losses than simulations on undisturbed crusted soils. Highest sediment losses (50 times) were observed from rolled surfaces. Because of the ongoing extensification process, the distributions of physical soil crusts will most probably further increase. According to the results, this would lead to a reduction of wind erosion susceptibility in the central Ebro Basin depending on intensity and time of sheep pasturing and tillage. Ó 2009 Elsevier Ltd. All rights reserved.

Keywords: Animal trampling Extensification Physical soil crusts Tillage simulation Wind tunnel

1. Introduction Since the early 1990s, an increasing area of the central Ebro Basin has been abandoned from agricultural use. For instance, during the period from 1990 to 2003 the arable land of the province of Arago´n has been reduced by 146.000 ha (w9%) to a level of about 1.5 Mio. ha (Eurostat, 2006). Major reason for this change from arable to fallow or abandoned land is the low profitability in this semi-arid region. Although since 1992 large set-aside programs, which are subsidised by the European Union, play an important role as well. Following this land use change the relative importance between wind and water erosion is changing and farmers have to adapt their agricultural operations. For example, if arable land is abandoned, physical soil crusts develop more frequently. During rainstorms, crusted fields may cause higher surface runoff and therefore higher sediment losses due to water erosion than prior abandonment (Rice et al., 1996; Ries and Hirt, 2008). Usual mitigation measures by farmers would be tillage operations to reduce crusting, but this would increase the intensity of wind erosion. In their ‘‘Soil erosion map of Western Europe’’, De Ploey et al. (1989) consider the central Ebro Basin as susceptible to wind erosion. * Corresponding author. Tel.: þ49 651 201 2998; fax: þ49 651 201 3976. E-mail address: w.fi[email protected] (W. Fister). 0140-1963/$ – see front matter Ó 2009 Elsevier Ltd. All rights reserved. doi:10.1016/j.jaridenv.2009.05.006

However, as Lo´pez et al. (1998) emphasise, no actual data exist about wind erosion in this region. In contrast to wind erosion the amount of water erosion in the central Ebro Basin has been examined quite extensively (Lasanta et al., 1994, 2000; Navas, 1990; Ries, 2002; Ries and Langer, 2002; Seeger, 2007). With their research, Lo´pez et al. (1998) started to fill this gap in knowledge. The subsequent WELSONS-project (Wind Erosion and Loss of Soil Nutrients in semi-arid Spain) focused on the effects of conventional tillage and reduced tillage onto wind erosion (Gomes et al., 2003a) and the improvement of an existing dust emission model (Gomes et al., 2003b). To extend the information about wind erosion in the Ebro Basin, a portable wind tunnel was used in 46 test runs to assess the relative sediment loss rates by wind erosion at three test sites near the city of Zaragoza. During two field campaigns in September 2005 and April 2006, three different land use systems varying in crust disturbance level were investigated – (1) fallow land with undisturbed physical soil crusts, (2) simulated sheep trampling and (3) conventional tillage. In contrast to the WELSONS-project, this study aimed on solving following research questions: (1) Does the present land use change, with focus on soil crusts and sheep trampling, have a quantitative impact on the wind erosion susceptibility of silty soils in the central Ebro Basin?

W. Fister, J.B. Ries / Journal of Arid Environments 73 (2009) 996–1004

(2) How much does the sediment loss by wind erosion increase on tilled surfaces compared to undisturbed crusted surfaces?

a

997

Wind Rose Airport Zaragoza in % N 20

NNW

NNE

2. Methods and materials NW

2.1. Geographical location and characteristics of the study area The research area is located close to Marı´a de Huerva inside the Val de las Lenas, an eastern tributary of the Huerva River, which lies about 15 km southwest of Zaragoza (Province of Arago´n, NE – Spain). Primary reason why this area was chosen is its good comparability to large areas of the Ebro Basin. Secondly, the availability of rainfall simulation data from the same fields by Ries and Langer (2002) was very valuable and finally, the close vicinity to some of the test sites of Lo´pez et al. (1998) and Gomes et al. (2003a). The average elevation above sea level is 410 m. A regional overview and a detailed localisation of the test area are given in Appendix 1 (electronic version only). There are three reasons why this area is prone to wind erosion: the semi-arid climate, poor silty and loamy soils, and the dry farming system. This system is commonly used, but because of long periods with fallow stages, it is inappropriate for protection against wind erosion (Lo´pez et al. 1998). The annual average temperature in the study area is 14.6  C with very high summer temperatures up to 40  C and low winter temperatures. Annual average precipitation in this cold steppe is 270 mm (Appendix 2, electronic version only), with a very high inter annual variability and torrential rainfall pattern. Due to high evaporation losses and low rainfall during summer time, Cuadrat Prats (2004) calculated a mean annual water deficit of almost 1800 mm for the region of Zaragoza. Another important climatic factor in Arago´n is wind. The main wind direction is WNW, which is determined by the orientation of the Ebro Basin and the surrounding mountain ranges, the Pyrenees, Iberian Range, and Cantabrian Mountains (Gomes et al., 2003a). This prevailing wind, which is locally called Cierzo, brings very cold air in winter and cool air in summer. Fig. 1a shows a wind rose of Zaragoza Airport weather station for the period from 1986 to 2007 and Fig. 1b presents the corresponding mean wind velocities for each wind direction. In both figures, the predominance of the wind directions NW and WNW can be observed very clearly. Own observations from September 2005 show that gust speeds can reach maxima above 25 m s1 at 2 m above ground during Cierzo periods. Confirming these short time observations, Biel and Garcı´a de Pedraza (1962) mention that during the period from 1943 to 1960 wind velocities above 30 m s1 in 10 m above ground were very common. Calculations for the period from 1986 to 2007 for Zaragoza Airport weather station show that on average 214 days per year the wind velocity in 10 m above ground exceeds 10 m s1. Puicercu´s et al. (1994) give an average annual wind speed of almost 5 m s1 at a height of 10 m. This overview shows clearly the potentially high erosivity of the local wind system. The soils in the study area are developed on Holocene alluvial fills, which are up to 20 m thick and mainly consist of alternating layers of silty loam, sandy silt, and loamy clay. These layers of fine material are discontinuously interrupted by stone and gravel layers with a sandy matrix (Ries and Langer, 2002). Because of the high gypsum content, the soils are developed as Leptic Haplogypsid (USDA classification). The very high silt contents above 50%, together with a scarce vegetation cover, and rapid wetting/drying cycles, are the main reasons for frequent soil sealing and crusting on fallow and abandoned fields in this region. Basically, physical soil crusts form due to breakdown of surface aggregates into finer particles, which then are transported and deposited somewhere else (Bresson and Valentin, 1994). Due to their genesis, four main

15

18.5

NE

10 8.2

WNW 17.8

W

ENE

5 11.9

0

9.9

E

8.3 WSW

ESE

SW

SE SSW

SSE S

b Mean Wind Velocity in km*h-1 N 25

NNW

NW

NNE

20 18.5

24.3

NE

15 WNW 20.7

10

ENE

8.1

5 W

6.1

16.3

0

E

10.4

11.1 6.9

12.9 10.2

WSW

ESE

SW

SE SSW

SSE S

Fig. 1. a) Wind rose of Zaragoza Airport weather station and b) mean wind velocity for each sector. Values below 6% and 6 km h1 are left out for visibility reasons. Frequency of calms is 8.2%. (own calculation with data from Instituto Nacional de Meteogologia, period 1986–2007).

crust types can be differentiated, the so-called ‘‘structural crusts’’, ‘‘erosion crusts’’, ‘‘sedimentary or depositional crusts’’, and ‘‘cryptogamic crusts’’ (Bresson and Valentin, 1994). Cryptogamic crusts need a longer period of growth for establishment and are therefore not existent on arable and negligible on young fallow land at our test sites. Ries and Hirt (2008) distinguish five different structural and depositional crust types that occur on arable and abandoned land in this region: young and old ridge crust, young and old furrow crust, and alluvial fan crust. Crusts on ridges are dominantly structural developed and crusts in furrows and alluvial fans are

998

W. Fister, J.B. Ries / Journal of Arid Environments 73 (2009) 996–1004

sedimentary formed. Whereas the first type is formed through processes resulting from raindrop impact and slaking as well as rapid wetting and drying of the soil surface (Chen et al., 1980), the second one forms through deposition of the transported sediment (Valentin and Bresson, 1992). Ries and Hirt, 2008 show that young furrow crusts of thicknesses up to 1 cm can form in less than three months. The direct neighbourhood of ridges, furrows, and small alluvial fans cause a high spatial variability in crust type on the field or even on a single test plot. Because of their binding forces and therefore limited amount of erodible material, soil crusts are generally regarded as protection against wind erosion, until they are somehow disturbed or abraded by saltating grains (Gillette, 1978a; McKenna Neuman et al., 2005; Rice and McEwan, 2001). Main reasons for disturbance are tillage operations, animal grazing and trampling, or vehicular traffic. 2.2. Characteristics of wind erosion test sites The wind erosion test sites were in different fallow stages of the ˜o y conventional cereal-fallow rotation land use system (cultivo an vez) which is commonly used in this semi-arid region to store more water in the soil for the following growing season (see chapter 2.4). For characterisation of each test plot, vegetation cover was classified visually in 5% intervals ranging from 0 to 100% and soil moisture contents were measured gravimetrically (drying temperature 30  C) for depths of 0–2 cm and 2–10 cm. Soil surface roughness was measured with the chain method and a roughness index (R) was calculated after Saleh (1994). Only the total roughness was registered, no differentiation between oriented and random roughness was made. The first test site on young fallow land (site 1) was completely covered with well-developed physical soil crusts that showed large contraction cracks. The ridges and furrows from the last tillage operation of the previous year were hardly visible, but many larger clods, which were embedded in the crust, were present. The roughness was therefore quite low with measured roughness indices (R) between 3.0 and 9.6. A definite categorization of the crust type is very difficult without micro-morphological investigations, but because of not existing ridges and furrows and their thickness of up to 1 cm, they seem to be mainly of sedimentary nature. This site was scarcely overgrown (0–20%) with exclusively annual plants like Hordeum murinum. There was heavy rainfall just one week before the fieldwork started in autumn 2005. Because of this, the soil moisture content at this time was in the depth of 0–2 cm, slightly higher (average 3.0 mass %) than in spring 2006, after a longer period with only little rainfall (0.4 mass %). Below the depth of 2 cm, the water contents in both years showed comparable values slightly above 10 mass %. The second test site was situated in a neighbouring tributary valley. It was recently ploughed and harrowed, but had experienced at least one heavy rainfall that caused a light superficial fixation of the fine silt particles. Because of their initial stage, they were characterised as young ridge and furrow crusts, following the categorisation of Ries and Hirt (2008). Apart from some crop residues and seeds from the last harvesting, no standing vegetation was present (coverage < 5%). Roughness indices ranged from 11.1 to 13.64 and the surface moisture content (0–2 cm) was below 1 mass %. The third test site was located further up the valley. It was divided in a freshly harrowed section (site 3a) and a completely sealed and crusted part (site 3b) with standing remnants from the last grain-harvest (coverage 10–30%), a plain surface (R ¼ 2.0), and a low moisture content (0.4 mass %). The surface crust showed far less contraction cracks and broke less intensively during sheep trampling simulation. In some areas, it seemed more like a ductile deformation with pulverisation of the surface than a crust break-up. The freshly harrowed section (site 3

a) had even less vegetation coverage than the second test site and a very loose aggregation with much fine material. In longitudinal direction of the ridges and furrows, roughness indices were below 5. In orthogonal direction, they reached values of 16.7. The soil moisture contents were as low as on the other part of the field. The genuinely harrowed wind tunnel tests were carried out on this field and the second test site.

2.3. The portable wind tunnel The priority during development and construction of the wind tunnel was to get highest mobility with best possible approximation to the natural wind conditions. There is a simple reason why mobility was preferred to a fully developed turbulent boundary layer that is in equilibrium to the soil surface. To increase the quantitative amount of data about wind erosion, it is necessary to choose test sites not only according to their accessibility, but also because of their relevance to wind erosion. The necessary length to develop a thick boundary layer would considerably decrease its transportability. Other more or less portable wind tunnels, which have been successfully used in field researches, were reported from Zingg (1951), Gillette (1978b), Leys and Raupach (1991), Leys et al. (2002), and Maurer et al. (2006). The portable wind tunnel is a push type one. A 5.5 hp fan with 163 cm3 and an axial propeller with a length of 0.62 m generates the airflow. Its two vanes can be variably adjusted to different angles to change both the wind power and wind velocity. The fan itself is mounted on an aluminium frame with wheels and can therefore easily be transported in the field. Each test run was conducted with maximum engine power, which resulted in an average wind velocity of 8 m s1, measured in 15 cm above ground. The wind profile is pre-formed by the position of the wind machine in combination with the honeycomb, a wire mesh, and a tripping fence. A 2 m long transition section, which is made of strong PVC plastic sheet (thickness ¼ 1 mm), leads the turbulent rotating airflow to a honeycomb. The honeycomb is 15 cm long and has the same cross-section as the working section. Its structure is made of 289 PVC tubes with a diameter of 4 cm each. Lacking the space for a longer fetch distance, the lowest tube row is blocked with gaffertape to simulate a 4 cm high tripping fence to reduce the wind velocity near ground. Upstream of the honeycomb a double layer of wire mesh with open spacing of 0.5 cm and a blend are attached. The blend is made of plywood and used to deflect the airflow from the upper 20 cm downwards to reduce wind velocities on the tunnel roof where pressure nozzles of a rainfall simulator can be installed. Laboratory calibration tests with a pietot tube on different surfaces (polystyrene plates, roofing cardboard, corrugated sheets with different heights and spacing) showed that due to the short length of the tunnel, the pre-shaped profile adjusts very little to the present surface. The thickness of the thin existing boundary layer varies from 15 to 20 cm. This induces that suspension can only be modelled properly for particles transported below this height. Further velocity measurements and tests with induced smoke indicate that the airflow within the lower 30 cm of the tunnel is sufficiently homogenous across the tunnel (deviation from mean 0.4–0.55 m s1) and that the propeller induced rotating vortex is removed by the honeycomb. Calculations from the profile measurements in combination with the measured mean wind velocity during test runs result in a reference wind velocity of 10.6 m s1 in 10 m above ground. This velocity does not match the strongest gust speeds, which can reach values up to 30 m s1, but it resembles very common medium to strong winds. With a velocity above 5.3 m s1 in 2 m above ground, it can be designated as erosive wind (Skidmore, 1965). The test duration of 10 min was chosen

W. Fister, J.B. Ries / Journal of Arid Environments 73 (2009) 996–1004

because during this period, all transportable soil particles are detached and no visible transport is evident. The working section is 3 m long and rectangular in shape (0.7 m  0.7 m). Its sidewalls are made of three segments of 1 m long aluminium and Perspex sheets, which can be folded up completely for transport. The sheets are stabilised and connected by three aluminium frames. Four steel bars, each 1.5 m long and 0.3 cm thick, are attached to the sidewalls for connection with the ground. To avoid gaps between tunnel sidewall and test surface, the bars are driven 5 cm into the ground. In this way, the floor of the tunnel is open and creates a 2.1 m2 test area that can easily be accessed by opening the complete sidewall of Perspex sheets. With these dimensions, the cross-section of the tunnel is quite similar to the ones reported by Leys and Raupach (1991), width 1.2 m and height 0.9 m, and Maurer et al. (2006), width 0.6 m and height 0.7 m. The length of their working sections ranges from 5 to 10 m and is significantly longer than the one used in this study (3 m). Because of the additional length, they are reported to have a fully developed turbulent boundary layer of varying thickness up to 0.5 m. However, together with this better accuracy in modelling natural wind conditions comes their major disadvantage, the semiportability. Although they have been used in fieldwork, their utilisation needs a lot of work force and sometimes even heavy machinery like trucks and trailers. With partially the same length, but much smaller cross-sections, the reported wind tunnels by Gillette (1978b), width 0.15 m and height 0.15 m, and Leys et al. (2002), width 0.1 m and height 0.5 m, switch to the other extreme. They are highly portable, but simulate natural conditions in a very restricted way. Despite the aerodynamic limitations, both studies show that it is possible to characterise wind erosion on different surfaces in relation to each other. The wind tunnel discussed here is, because of its size, the good portability and its thin, but existing, boundary layer, situated in between these two contrary methodological approaches. Therefore, the main use of the wind tunnel is to gain relative information about the wind erosion susceptibility of different soils and soil treatments. Because of its methodological limitations, an extrapolation of the data from plot to larger scales should be avoided. The main reason for choosing the height of 0.7 m was the planned installation of a rainfall simulator for simultaneous measurements of wind and water erosion rates. This height is on one hand, the absolute minimum for representative rainfall simulations with pressure nozzles and on the other hand the maximum height that the fan was able to work with. The detached sediment from the test area is caught by the sediment catching area, which is made of commercial tarpaulin. It has a base of 3 m  5 m and vertical boundaries of 1 m (side) and 1.5 m (back) height. For stabilisation of the vertical boundaries, wooden poles were fixed to the plane and connected by strings. A view of the built-on tunnel is shown in Fig. 2. Although the wind velocity on the catching area is considerably reduced by 6 m s1, it is evident that an undefined amount of silt and clay particles might not be caught. On the other hand, it is also possible that dust from the surrounding air deposits on the sediment catching area. Generally, the unknown output should exceed the input rates by far, so that an underestimation of the real wind erosion rates can be assumed. On soils with higher sand contents, the importance of this error diminishes almost completely. The main advantage of this sediment catching area compared to commonly used sediment catchers with tested efficiency values, like the Modified Wilson and Cook Sampler (Wilson and Cooke, 1980), Big Spring Number Eight (Fryrear, 1986) or Guelph-Trent Wedge Trap (Nickling and McKenna Neuman, 1997) is the total amount of caught material. Considering the low amount of detachable material on a crusted 2.1 m2 surface, the methodological error of this experimental investigation

999

decreases with increasing amount of caught material. Moreover, further supplementary chemical and physical analyses are possible if there is enough collected material. 2.4. Soil surface preparation The traditional dry farming system in the semi-arid Ebro Basin ˜ o y vez) is characterised by a two-year rotation of cereal (cultivo an and fallow stages. During the fallow stage, farmers plough their fields several times during the rainy season (October–August) to increase the infiltration of rainwater. Fields are harrowed and rolled directly after the rainfall in order to destroy capillaries and seal the soil surface against evaporation loss. As a result, there is more water stored in the soil to support the next growing season of barley or wheat in the second or third year (September–June). Throughout the period between harvesting and the next rainy season (July– September), the fields are used for extensive sheep grazing (GarciaRuiz et al., 1996; Wagner, 2001, 246f.). To represent this system, three different soil surface treatments were selected for investigation in this study: (1) Undisturbed fallow land with developed physical soil crust (2) Crust disturbance by different intensities of simulated sheep trampling on fallow land (3) Tillage methods (ploughing, harrowing, and rolling). These different surface conditions (shown in Fig. 3) were either genuinely existent or simulated by an operator. The decision about the simulation methods was based on their applicability, visible representativeness, and roughness measurements with the chain method. As test sites for fallow and abandoned land only areas, which were completely covered with undisturbed physical soil crusts were selected. Due to the high silt content and the heavy rain, this surface is predominant on fallow surfaces in this area and was therefore of major interest. Although a least possible crust disturbance was aspired, complete prevention of crust breaking while driving the steel bars into the ground was not possible. Intensity of crust disturbance was directly linked to crust thickness. It might therefore be possible that an unknown proportion of the deflated material, which was blown out during test runs on crusted surfaces, resulted from this initial disturbance. Like Leys and Eldridge (1998), we decided to simulate sheep trampling instead of using naturally disturbed areas, because of the unknown intensity and variability. The simulation with living animals inside the tunnel is undoubtedly not possible and an artificial method was therefore necessary. Unfortunately, no ultimate simulation method has been developed yet. Major problems are the simulation of genuine impulse forces and the exact course of motion of the animal. A hoof impact simulator, developed by Walker et al. (2005), is for example capable of simulating impulse forces by animal trampling with good accuracy and precision, but the material uplift by the moving hoof cannot be simulated. Moreover, its application within the wind tunnel would take a lot of time and create a big obstacle to the airflow (length 1.1 m, width 0.3 m, height 0.25 m). With a diameter of 0.4 m and a width of 0.5 m, the sheep’s foot roller by Leys and Eldridge (1998) forms also a big obstacle if applied during the test runs. The weight of 35 kg or more makes it very difficult to use inside the tunnel. In contrast to the hoof impact simulator, this device can simulate, at least partially, the material uplift with its artificial hooves (rectangularshaped steel bars). Without a better applicable alternative, sheep trampling was simulated in this study by hitting the ground with the edge of a 1 kg hammer. The head of the hammer was 4 cm by 10 cm and the triangular impact area approximately 4.5 cm2 in size. Main aim of this treatment was the destruction of the soil crust and

1000

W. Fister, J.B. Ries / Journal of Arid Environments 73 (2009) 996–1004

Fig. 2. Wind tunnel on test site 3 near Zaragoza (NE-Spain).

not the effect of trampling on soil compaction. However, to increase reproducibility, the hammer was brought down as closely as possible from 30 to 40 cm height without any additional force impulse by the operator. A rough estimation of the applied pressure, ranging from 0.4 to 0.6 kg cm2, shows similar dimensions like

genuine sheep trampling, ranging from 0.5 to 0.7 kg cm2, for a sheep of 30–40 kg. The absence of the material uplift into the air stream by the motion of a sheep hoof is partially compensated by the uplift due to the impact kinetic energy. Visually both destroyed crust surfaces, simulated and genuine, resembled each other well.

Fig. 3. Examples of the different investigated soil surface treatments.

W. Fister, J.B. Ries / Journal of Arid Environments 73 (2009) 996–1004

Two disturbance levels, 50 prints m2 and 100 prints m2, were used. Additionally, both intensities were applied in separate test runs prior as well as during the actual wind tunnel tests. The tillage operations ploughing, harrowing and rolling were also difficult to simulate, because the size of the tunnel is too small for any real size or miniature agricultural machinery. Ploughing was therefore simulated with a spade by driving the shovel pan as deep as possible into the ground and turning the loosened soil upside down. Most surfaces were so dense and sealed that the depth was never more than 10–15 cm, which is clearly less than the common 20–30 cm of a real plough. Additionally, average roughness indexes (R) of 14.52 in direction of the wind tunnel compared to 16.74 across the wind tunnel quantitatively show that it was not possible to simulate proper ridges and furrows with this method. Since no roughness measurements on genuinely ploughed surfaces were made, it can only be assumed that the soil roughness on simulated ploughed surfaces was less than on genuinely ploughed fields. On closer examination, the simulated ploughed surfaces resembled more chiselled fields than the conventional mouldboard ploughed fields described by Lo´pez et al. (1998) for this region. Although this simulation method is very basic, its main purposes, the destruction of crusts, the turning of soil, and the change of soil structure were successfully reached. Test runs on harrowed surfaces include both, genuinely harrowed fields and simulated harrowing. Comparing both surface types (Fig. 3d and e) visually, it stands out that by using a rake to reduce aggregate sizes and simulate harrowing it is again not possible to simulate corresponding ridges and furrows. Moreover, it seems as if the broken up aggregate clods accumulate on the surface, protecting the fine dust particles underneath. By comparing the average total roughness index on genuinely harrowed surfaces (R w 17) with that on simulated harrowed surfaces (R w 7); the difference in roughness due to missing ridges and furrows becomes evident. Simulation of rolling was achieved by hitting the ground with the flat side of a 1 kg hammerhead with the size of 10  4 cm. Again, the pressure is kept as constant as possible, but this time an additional impulse by the operator is needed. At first glance, the procedure seems to be very different from genuine rolling with a flat-roller, but the effect on the soil surface is quite similar – clod destruction, soil compaction, and creation of a smooth surface. The very low average roughness index of R ¼ 2.75 shows clearly the smooth and aggregate free surface. Regarding the soil compaction, a rough estimation of the simulated pressure, ranging from 0.08 to 0.13 kg cm2, shows similar dimensions like genuine rolling with a common flat-roller (weight 0.5–0.8 t, width 3–4 m), ranging from 0.06 to 0.17 kg cm2. 3. Results The sediment losses of in total 46 wind tunnel test runs with seven distinct soil surface treatments at all test sites in Marı´a de Huerva (NE-Spain) are shown in Fig. 4. It is important to note that the organic material was removed before weighing to better analyse the influence of the applied surface treatments. For crusted soil surfaces (column 1) we obtained very low sediment loss rates through wind erosion (median 0.45 g m2 10 min1), ranging from 0.15 to 2.37 g m2 10 min1. Even when the soil crusts were clearly destroyed by sheep trampling, the increase of sediment output in calm wind conditions with amounts around 0.93 g m2 10 min1 cannot be interpreted as significant (column 2). Remarkable, compared to this low increase is the multiplication by more than 10 times due to simulated sheep trampling during a wind event (column 3). The amount of wind erosion on simulated ploughed (column 4) and simulated harrowed fields (column 5) was much

1001

lower than that caused by sheep trampling during wind events. On fields which were harrowed by farmers a few days prior to the tests, sediment loss showed a high variability, ranging from 0.31 to 6.55 g m2 10 min1 with a median of 2.71 g m2 10 min1 (column 6). By far the highest sediment output and the highest internal variability was found on rolled fields with median values of 21.38 g m2 10 min1 (column 7) and maximum values above 50 g m2 10 min1. The absolute results from all simulated sheep trampling tests were separated into two main categories (Fig. 5), a first group with simulations of sheep trampling prior the actual wind tunnel tests and a second group with simulations of sheep trampling during the test runs. For each group two different trampling intensities were tested, 50 hoof prints per m2 and 100 hoof prints per m2. The presented data show a very high variability within the absolute values of the different groups. Generally lower, but also characterised by a considerable variability are the test runs prior wind simulations. Especially the values for the test runs number 4, 9, and 12 with their extremely low amounts seem not to fit well into their groups. Remarkably, a double increase in disturbance intensity (50 / 100 prints per m2) during wind simulation does not lead to an increase in sediment loss. 4. Discussion Field measurements on natural surfaces are generally characterised by a high variability, because of heterogeneity in soil properties, vegetation cover, soil moisture content, and soil treatment. Additionally, an internal methodological variability exists that has to be taken into account during interpretation of measured data, especially when only 4–11 repetitions per group have been made. Therefore, only variations of different magnitudes were interpreted in this study. At first glance, Figs. 4 and 5 show a high variability of data for each treatment. After closer examination, the outstanding differences between the groups of soil treatments are of such dimensions that reproducibility can be presumed. The treatments soil crust, trampling before wind, ploughing and simulated harrowing can be aggregated in one group of more or less low sediment loss. Because of their internal variability, a further differentiation is not possible. Sheep trampling during wind and genuine harrowing show the same range of values, but excluding individual outliers, they can be further distinguished. Most values for harrowing and trampling during wind are grouped around their medians, 2.7 and 5.3 g m2 in 10 min respectively. Although the test runs for rolled surfaces show the highest variability, it seems to be clear that they compose their own group with the highest sediment loss rates. Measurements by Rehberg (1999) and Ries et al. (2000) with a previous version of the wind tunnel on the same test sites confirm these results. As Fig. 4 clearly shows, the sediment output by wind erosion on undisturbed, crusted soil is almost negligible. This result, which is due to soil sealing and fixation of fine soil particles, corresponds with our expectations and the results from previous studies by Zobeck (1991) and Hupy (2004) who measured sediment flux rates on soils with different clay contents. Especially compared to the erosion control target of 5 g m2 s1 for a 65 km h1 wind measured in 10 m height, which is proposed by Leys and Eldridge (1998), this observation seems obvious. A relatively clear difference between the medians of the different sheep trampling simulations can be seen in Fig. 4. Although the soil crusts were clearly destroyed in both simulation processes, the soil loss in the scenario combining sheep trampling with wind is more than five times higher than in the scenario without simultaneous wind simulations. The destruction of the crust is not the only factor for the increase in soil detachment. The fact that the impact kinetic energy of the hammerhead dislodges and lifts large loosened crust

1002

W. Fister, J.B. Ries / Journal of Arid Environments 73 (2009) 996–1004 N=5

N=4

60

N=7

N=4

N = 11

N=7

N=8

60

sediment loss [g*m-2*10min]

single test runs median

50

50

40

40

30

30 21.38

20

20

10

10 5.27 0.45

0

2.71

0.93

0.58

0.51

0

soil crust trampling trampling ploughed harrowed harrowed (gen.) before wind during wind (sim.) (sim.) (gen.) (sim.) (sim.) #2 #1 #3 #4 #5 #6

rolled (sim.) #7

soil treatments Fig. 4. Results of 46 test runs in seven surface classes assessed with the portable wind tunnel in Marı´a de Huerva; (sim.) ¼ manually simulated surfaces by an operator; (gen.) ¼ genuine surfaces either naturally developed like soil crusts or mechanically created by farmers.

fractions up into the moving air stream is more important. This process is quite similar to the entrainment of soil material through the kinetic impact energy of saltating particles, a process already described by Bagnold (1941). Findings of aggregates, larger than 2 cm in size, on the sediment catching area give evidence that the momentum of the hammer exceeds the one by saltating particles by far. In comparison, Lancaster and Nickling, 1994 give an upper threshold for saltating particles of 1000 mm, which is clearly lower than the found aggregate sizes on the sediment catching area. Obviously, neither airflow nor saltating grains would have been able to erode particles of this size without the dislodgement and uplift by the hammer impact. The absolute values of the sheep trampling test runs in Fig. 5 show a high variability. The discrepancy of the test runs number 4, 9, and 12 in their groups are for the most part caused by a difference in soil crust stability and development stage at this test site 3b. In contrast to the well-developed thick and multiple layered crusts on young fallow land (test site 1), which showed many contraction cracks, the crust on test site 3b was very thin with a ductile consistency and

almost no contraction cracks. Therefore the crusts on test site 3b did not break when sheep trampling was simulated, causing much less sediment loss in all three test runs. Excluding these three test runs from the calculation of the median, because of their different crust type, sediment losses due to sheep trampling increase for both simulations (before 1.28 g, during test run 5.85 g). This strengthens the conclusion that sheep trampling increases wind erosion rates considerably. With the exception of these three explainable outliers, the absolute values in Fig. 5 confirm the conclusion from Fig. 4 that especially the time of sheep trampling is important for determining the soil loss rate. Surprisingly, the intensity of sheep trampling simulation does not seem to have a significant impact on soil loss rates. A comparison with other results from wind erosion research with sheep trampling is quite difficult due to their scarcity and the lack of comparability between methods. Leys and Eldridge (1998) investigated the influence of sheep trampling on loamy and sandy soils covered with cryptogamic crusts. They observed that an increase in disturbance intensity by sheep trampling from moderate (228 hoof prints per m2) to severe (456 hoof prints per m2) leads to

sediment loss [g*m-2*10min-1]

10

8

7.44 6.91 5.68

6

5.27 3.95

4 2.40

2 0.93

1.18 0.71

0.60 0.02

0 #1

#2

#3

#4

50 prints/m²

before wind simulation

0.12

#5 100 prints/m²

#6

#7

#8

#9

50 prints/m²

#10

#11

#12

100 prints/m²

during wind simulation

Fig. 5. Absolute sediment loss of all test runs with sheep trampling on physically crusted soils in Marı´a de Huerva.

W. Fister, J.B. Ries / Journal of Arid Environments 73 (2009) 996–1004

1003

an increase in sediment flux by almost 10 times. In contrast to these results, our much lower but also doubling intensity increase from 50 hoof prints per m2 to 100 hoof prints per m2 surprisingly did not cause a significant increase in soil loss. These contrasting results make clear that better simulation methods and more measurements are necessary to improve the knowledge about the importance of sheep grazing on soil erosion rates. Even though soil roughness measurements with the chain method show a decrease in mean surface roughness from ploughed (R w 15) to simulated harrowed surfaces (R w 7), the sediment losses do not significantly differ from each other. A probable explanation for this phenomenon is the protection of fine particles by the micro-topography, which was created by fragment accumulation of the former soil crusts. This so-called aggregate roughness cannot be adequately measured with the chain method. In some of the test runs, the surface looked as if it was covered completely with non-erodible aggregates, sizes bigger than 0.84 cm Chepil (1950). Corresponding sediment losses were very low. Logie (1982) demonstrated in wind tunnel experiments, that apart from the size of aggregates, their spacing plays an important role on threshold velocity. Depending on aggregate size, an inversion point of cover density exists where the influence changes from protection to activation. Another reason for the uncharacteristically low sediment losses on ploughed and harrowed surfaces is the test procedure itself. Through tillage operations, soil from deeper horizons with higher moisture content was brought to the surface. After surface preparation there was no time left in between the test runs, to give the soil time to evaporate the surplus water. The additional adhesive forces of the water can bond more particles and therefore the threshold friction velocity increases (Fecan et al., 1998). Apart from the general problems to simulate the selected surfaces with the chosen simulation methods, this is probably the major drawback for the chosen tillage simulation techniques. The fact that the expected increase in soil loss only occurred on fields that were harrowed by a farmer (increase by 5 times) underlines the problems of the tillage simulations. By far the highest sediment loss was observed on rolled surfaces which reach quantities similar to the amount of particle loss through water erosion that were measured by Ries and Langer (2002) and Ries et al. (2000) on crusted soils at the same test sites. In comparison between the tillage operations, the sediment loss increases from ploughed to harrowed and rolled surfaces. In absence of bonding agents, seals, or soil crusts, the most probable explanation for this fact is soil roughness. Soil moisture content can be excluded, because no specific pattern that might explain the differences could be found in the data. The protective influence of increasing soil roughness against wind erosion is mainly based on a reduction of wind velocity near surface and an increase of critical threshold shear velocity (Nickling, 1994). Because of the problem with simulation of ridges and furrows, the influence of ridge roughness is very important. Du¨wel et al. (1996) and Skidmore (1994) show that sediment transport rates change with respect to orientation and size of ridges. Conform to our results, they show a transport increase from rolled surfaces to aggregate roughness, which corresponds with harrowing, to ridge roughness that corresponds with genuine ploughing.

useful research instrument. An additional problem for interpretation of the results is due to the simulation methods of the surface treatments. For example missing ridges and furrows, too small aggregate sizes, too high soil moisture contents or inadequate simulation of course of movements cause a clear reduction in their representativeness to genuine soil treatments. The land use change from arable to fallow or abandoned land will eventually cause an increase in areas covered with physical soil crusts. The results presented here suggest that this will lead to a significant reduction in wind erosion susceptibility on silty soils in the central Ebro Basin. The determining factor of wind erosion rates will thereafter be the intensity and the time of sheep grazing, although it was shown that better simulation methods for grazing need to be invented and further investigations are necessary to support this conclusion. Perhaps, intensive pasturing ought to be limited to periods with low wind velocities to minimise soil loss by wind erosion. Additionally, the results show that the conventionally used dry farming system is inadequate to protect the soil from wind erosion, although it is perfectly adapted to the low amount of rainfall and employed to reduce the threat by water erosion. Furthermore, the results prove that sediment loss by wind erosion from simulated rolled surfaces compared to sediment loss from undisturbed crusted surfaces is more than 40 times higher. In a reduced manner, this observation is also true for genuinely harrowed surfaces. For simulated ploughing and harrowing, no significant increase could be found. Problems in treatment simulation, e.g. aggregate size, soil moisture content and so on, and the internal variability prevent well-founded conclusions. Therefore, more wind tunnel test runs need to be made and more accurate and easy to handle simulation methods, need to be invented and used in future investigations. It can be expected that by combining this wind erosion information and data about the intensity of water erosion from the same study area (Ries and Langer, 2002); it is possible to further quantify the relative importance of these main soil degradation processes in the Central Ebro Basin.

5. Conclusion

Supplementary data associated with this article can be found in the online version, at doi:10.1016/j.jaridenv.2009.05.006.

Regarding the method employed in this study, it can be concluded that with careful interpretation of internal data variability, useful findings can be drawn from the test results. Although such a small wind tunnel is not able to simulate all aerodynamic parameters, which are necessary to represent exactly natural wind conditions. The mobility of the tunnel together with its ability to produce replicable data from different soil surfaces makes it an

Acknowledgements This investigation was funded by the Deutsche Forschungsgemeinschaft within in the project RI 835/3. The first author would like to thank the State of Rheinland-Pfalz for their financial support through a postgraduate scholarship. Additionally the authors are also grateful to Nadine Lux, Marie Roche, Stefan Wingler and all other helpful students for their assistance during fieldwork and laboratory tests. Moreover, we are grateful for constructive discussions with Dr. Tilmann Sauer, Dr. Manuel Seeger, and Dr. Reinhard-Gu¨nther Schmidt (Trier University) as well as Prof. Dra. Marı´a Teresa Echeverrı´a Arnedo (Zaragoza University) for sharing her knowledge about the local environmental circumstances. We would also like to thank the reviewer Dr. Andreas Baas for his critical comments, which helped us further improving the paper. Appendix. Supplementary data

References Bagnold, R.A., 1941. The Physics of Blown Sand and Desert Dunes. second ed., London. Biel, A., Garcı´a de Pedraza, L., 1962. El clima en Zaragoza y ensayo climatologico para el valle del Ebro, Publ. Serie A (Memorias) 36, Servicio Meteorolo´gico Nacional. Ministerio del Aire, Madrid.

1004

W. Fister, J.B. Ries / Journal of Arid Environments 73 (2009) 996–1004

Bresson, L.-M., Valentin, C., 1994. Soil surface crust formation: contribution of micro morphology. In: Ringrose, A.J., Humphreys, G.S. (Eds.), Soil Micro Morphology: Studies in Management and Genesis. Proc. IX Int. Working Meeting on Soil Micro Morphology, Townsville/Australia, pp. 737–762. CHE (Confederacio´n Hidrogra´fica del Ebro), 2006. Banco de Datos Precipitatio´n Relleno Mensual URL: http://oph.chebro.es (accessed 08.08.06). Chen, Y., Tarchitzky, J., Brouwer, J., Morin, J., Banin, A., 1980. Scanning electron microscope observations on soil crusts and their formation. Soil Science 130, 49–55. Chepil, W.S., 1950. Properties of soil which influence wind erosion: II: dry aggregate structure as an index of erodibility. Soil Science 69, 403–414. ˜ a, J.L., Longares, L.A., Cuadrat Prats, J.M., 2004. El clima de Arago´n. In: Pen Sa´nchez, M. (Eds.), Geografı´a Fı´sica de Arago´n. Aspectos generales y tema´ticos. Universidad de Zaragoza e Institucio´n Fernando el Cato´lico, Zaragoza, pp. 15–26. De Ploey, J., Auzet, A.V., Bork, H.R., Misopolinos, N., Rodolfi, G., Sala, M., Silleos, N.G., 1989. Soil Erosion Map of Western Europe. Catena, Cremlingen-Destedt, Germany. Du¨wel, O., Scha¨fer, W., Kuntze, H., 1996. The effect of soil surface roughness on soil transport by wind. In: Buerkert, B., Allison, B.E., von Oppen, M. (Eds.), Proceedings of the International Symposium ‘Wind erosion in West Africa: the problem and its control’, 5–7 December 1994, Hohenheim, Germany, pp. 357–364. Eurostat, 2006. Statistisches Amt der Europa¨ischen Gemeinschaften URL: http:// epp.eurostat.ec.europa.eu (accessed 01.09.06). Fecan, B., Marticorena, B., Bergametti, G., 1998. Parametrization of the increase of the aeolian erosion threshold wind friction velocity due to soil moisture for arid and semi-arid areas. Annales Geophysicae 17, 149–157. Fryrear, D.W., 1986. A field dust sampler. Journal of Soil and Water Conservation 41, 117–120. Garcia-Ruiz, J.M., Lasanta, T., Ruiz-Flano, P., Ortigosa, L., White, S., Gonza´lez, C., Martı´, C., 1996. Land-use changes and sustainable development in mountain areas: a case study in the Spanish Pyrenees. Landscape Ecology 11.5, 267–277. Gillette, D.A., 1978a. A wind tunnel simulation of the erosion of soil. Effect of soil texture, sandblasting, wind speed and soil consolidation on dust production. Atmospheric Environment 12, 1735–1743. Gillette, D.A., 1978b. Tests with a portable wind tunnel for determining wind erosion threshold velocities. Atmospheric Environment 12, 2309–2313. Gomes, L., Arru´e, J.L., Lo´pez, M.V., Sterk, G., Richard, D., Gracia, R., Sabre, M., Gaudichet, A., Frangi, J.P., 2003a. Wind erosion in a semiarid agricultural area of Spain: the Welsons project. Catena 52, 235–256. Gomes, L., Rajot, J.L., Alfaro, S.C., Gaudichet, A., 2003b. Validation of a dust production model from measurements performed in semi-arid agricultural areas of Spain and Niger. Catena 52, 257–271. Hupy, J.P., 2004. Influence of vegetation cover and crust type on wind-blown sediment in a semi-arid climate. Journal of Arid Environments 58, 167–179. Lancaster, N., Nickling, W.G., 1994. Aeolian sediment transport. In: Abrahams, A.D., Parsons, A.J. (Eds.), Geomorphology of Desert Environments. Chapman & Hall, London, pp. 447–473. Lasanta, T., Garcı´a-Ruiz, J.M., Pe´rez-Rotome´, M.C., Sancho-Marce´n, C., 2000. Runoff and sediment yield in a semi-arid environment. The effect of land management after farmland abandonment. Catena 38, 265–278. Lasanta, T., Pe´rez-Rontome´, M.C., Garcı´a-Ruiz, J.M., 1994. Efectos hidromorfolo´gicos de differentes alternativas de retirada de terras en ambientes semia´ridos de la depressio´n del Ebro. In: Garcı´a-Ruiz, J.M., Lasanta, T. (Eds.), Efectos gem˜ ola de Geomorfologı´a, orfolo´gicos del abandono de tierras. Sociedad Espan Zaragoza, pp. 69–82. Leys, J.F., Strong, C., McTainish, G.H., Heidenreich, S., Pitts, O., French, P., 2002. Relative dust emission estimated from a mini-wind tunnel. In: Lee, J.A., Zobeck, T.M. (Eds.), Proceedings of ICAR5/GCTE-SEN Joint Conference. 22–25 July 2002, Lubbock, Texas, USA, pp. 117–121. Leys, J.F., Eldridge, D.J.,1998. Influence of cryptogamic crust disturbance to wind erosion on sand and loam rangeland soils. Earth Processes and Landforms 23, 963–974. Leys, J.F., Raupach, M.R., 1991. Soil flux measurements using a portable wind erosion tunnel. Australian Journal of Soil Research 29, 533–552.

Logie, M., 1982. Influence of roughness elements and soil moisture of sand to wind erosion. Catena 1 (Suppl. l), 161–173. Lo´pez, M.V., Sabre, M., Gracia, R., Arru´e, J.L., Gomes, L., 1998. Tillage effects on soil surface conditions and dust emission by wind erosion in semiarid Arago´n (NE Spain). Soil and Tillage Research 45, 91–105. Maurer, T., Herrmann, L., Gaiser, T., Mounkaila, M., Stahr, K., 2006. A mobile wind tunnel for wind erosion field measurements. Journal of Arid Environments 66, 257–271. McKenna Neuman, C., Maxwell, C., Rutledge, C., 2005. Spatial and temporal analysis of crust deterioration under particle impact. Journal of Arid Environments 60, 321–342. Navas, A., 1990. The effect of selected physiographic factors on dissolved gypsum transport by simulated runoff on gypsiferous soils. Catena 17, 409–416. Nickling, W.G., 1994. Aeolian sediment transport and deposition. In: Pye, K. (Ed.), Sediment Transport and Depositional Processes, pp. 293–350. Oxford. Nickling, W.G., McKenna Neuman, C., 1997. Wind tunnel evaluation of a wedgeshaped aeolian sediment trap. Geomorphology 18, 333–345. Puicercu´s, J.A., Valero, A., Navarro, J., Terre´n, R., Zubiaur, R., Martı´n, F., Iniesta, G., 1994. Mapa eo´lico de Arago´n. Serie: Datos Energe´ticos de Arago´n. CIRCEGobierno de Arago´n, Zaragoza. Rehberg, C., 1999. Experimentelle Untersuchungen zur Winderosion mit einem mobilen Windkanal. In: Ries, J.B. (Ed.), Bodenwasserhaushalt und aktuelle Geomorphodynamik auf Brachfla¨chen in Arago´n/Spanien (APT-Berichte 10), pp. 135–144. Rice, M.A., McEwan, I.K., 2001. Crust strength: a wind tunnel study of the effect of impact by saltating particles on cohesive soil surfaces. Earth Surface Processes and Landforms 26, 721–733. Rice, M.A., Willetts, B.B., McEwan, I.K., 1996. Wind erosion of crusted soil sediments. Earth Surface Processes and Landforms 21, 279–293. Ries, J.B., Langer, M., Rehberg, C., 2000. Experimental investigations on water and wind erosion on abandoned fields and arable land in the central Ebro Basin, Arago´n /Spain. Zeitschrift fu¨r Geomorphologie N.F. (Suppl. 121), 91–108. Ries, J.B., 2002. Geomorphodynamics on fallow land and abandoned fields in the Ebro Basin and the Pyrenees – monitoring of processes and development. Zeitschrift fu¨r Geomorphologie N.F. (Suppl. 127), 21–45. Ries, J.B., Langer, M., 2002. Runoff generation of abandoned fields in the Central Ebro Basin. Results from rainfall simulation experiments. In: Garcı´a-Ruiz, J.M., Jones, J.A.A., Arnae´z, J. (Eds.), Environmental Change and Water Sustainability. Instituto Pirenaico de Ecologı´a, Zaragoza, pp. 65–81. Ries, J.B., Hirt, U., 2008. Permanence of soil surface crusts on abandoned farmland in the Central Ebro Basin/Spain. Catena 72, 282–296. Saleh, A., 1994. Measuring and predicting ridge-orientation effect on soil surface roughness. Soil Science Society of America Journal 58, 1228–1230. Seeger, M., 2007. Uncertainty of factors determining runoff and erosion processes as quantified by rainfall simulations. Catena 71 (1), 56–67. Skidmore, E.L., 1965. Assessing wind erosion forces: directions and relative magnitudes. Soil Science Society of America Proceedings 29, 587–590. Skidmore, E.L., 1994. Wind erosion. In: Lal, R. (Ed.), Soil Erosion Research Methods, pp. 265–294. Delray Beach. Valentin, C., Bresson, L.-M., 1992. Morphology, genesis and classification of surface crusts in loamy and sandy soils. Geoderma 55, 225–245. Wagner, H.-G., 2001. Mittelmeerraum. Wissenschaftliche Buchgesellschaft, Darmstadt. Walker, M.J., Kutsch, R., Miller, W.W., Cirelli, A., Donaldson, S., 2005. A consistent hoof impact simulator. Soil Science Society of America Journal 69, 257–259. Wilson, S.J., Cooke, R.U., 1980. Wind erosion. In: Kirkby, M.J., Morgan, R.P.C. (Eds.), Soil Erosion, pp. 217–251. Chichester. Zingg, A.W., 1951. A portable wind tunnel and dust collector developed to evaluate the erodibility of field surfaces. Agronomy Journal 43, 189–191. Zobeck, T.M., 1991. Abrasion of crusted soils: influence of abrader flux and soil properties. Soil Science Society of America Journal 55, 1091–1097.