Influence of higher rain intensities on phosphorus movements in the upper half meter of macroporous clay soil

Abstract

In climate change scenarios, the frequency of high-intensity rain events in Sweden is assumed to increase. In a plot experiment at Ultuna, Uppsala, the influence of rain intensities on phosphorus (P) transport in the uppermost 0.5 m of a clay soil was studied at 16 locations. A rain simulator, 0.5 × 0.5 m and mounted 1 m above the soil surface, was used to simulate 85–500 min rain sequences causing small (4–9 mm h⁻¹) and large (22–28 mm h⁻¹ and one extreme at 37 mm h⁻¹) steady water fluxes (intensity) in the underlying soil profile. Water percolated to a zero-tension collector tray at 0.5 m depth where drain water and its sediment load was sampled at discrete time intervals. The total P (TP) mass flux ranged, at low intensity, between 12–92 μg m⁻² min⁻¹ (average 28.1 μg m⁻² min⁻¹) and, at high intensity, between 83–375 μg m⁻² min⁻¹ (average 168.5 μg m⁻² min⁻¹) and 648 μg m⁻² min⁻¹ at the extreme intensity. The soluble reactive (inorganic) P (SRP) mass flux ranged, at low intensity, between 1–65 μg m⁻² min⁻¹ (average 10.0 μg m⁻² min⁻¹) and, at high intensity, between 6–205 μg m⁻² min⁻¹ (average 47.9 μg m⁻² min⁻¹) and 495 μg m⁻² min⁻¹ at the extreme intensity. Thus, in the intensity range 4–28 mm h⁻¹, TP and SRP increased, on average, by approximately 12% (μg m⁻² min⁻¹) per unit increase in intensity (mm h⁻¹). The results of this study demonstrate increased sediment and P loss/mobility for clay soil under increased precipitation intensity predicted under climate change.

Keywords:

• climate change
• in-field lysimeter
• preferential flow
• rain intensity
• rain simulator
• phosphorus leaching

Introduction

In climate change scenarios, the frequency of high-intensity rain events in Sweden is assumed to increase. Results, based on the climate change scenario data of the Swedish Meteorological and Hydrological Institute, reveal that maximum precipitation intensity (average of yearly maxima during a 30 year period for precipitation intensity below 30 min) is predicted to increase by 5–10% in northernmost region, 20–30% in southwest and 10–20% in the rest of Sweden (Svenskt Vatten 2011). Furthermore, Hernebring et al. (2012) reported that the intensity of heavy rainfall in summer is predicted to generally increase by 10–15%, based on a study where climate signals (changes in climate scenario data between a control period and the future) were investigated for five selected Swedish cities based on a total of 11 different climate scenarios. The spread between the different scenarios is, however, very large (from unchanged intensity up to an increase of more than 40%). In terms of the temperature change, and water vapor storage capacity (and thus indirectly the intensity of extreme rainfalls), one can expect to find a 7% increase in extreme rain intensity per degree of warming. Moreover, there is an indication according to literature data, that the percentage increase for the most extreme events will be doubled over a threshold value for the average daily temperature (around 12°C), also applies to Swedish conditions (Hernebring et al. 2012).

The scenarios of increasing frequency of larger rainfall intensities raise the concern that phosphorus (P) movement over the soil surface, by surface erosion (Mainstone et al. 2008), as well as leaching through soil macropores (Jarvis 2007), may increase in the future, and thus, in turn, result in increased P loads to surface waters and groundwater. Despite a large adsorption potential of P in many soils, P may be transported quickly through the soil via subsurface flow through macropores (e.g. Skaggs et al. 1994; Djodjic & Bergström 2005) and into drain tiles (e.g. Baker 1964; Bottcher et al. 1985). While subsurface drain P concentrations are typically lower than surface runoff P concentrations, annual P loads from subsurface drainage may exceed loads from surface pathways (e.g. Algoazany et al. 2007; Lamba 2010). Phosphorus export via artificial subsurface drainage systems has been shown to occur rapidly and episodically, a large portion of the yearly losses can occur in connection with a few runoff events (Ulén 1995), and is in many cases originating, after mobilization, from sources in the topsoil (Grant et al. 1996; Øygarden et al. 1996). Phosphorus transport and losses include both dissolved and particulate forms of P, making estimations of P loads and management particularly difficult (Sharpley et al. 1994; Sims et al. 1998).

The concerns above motivate the need for further studies on P leaching and P mobility in the soil profile as related to predicted increasing rainfall intensities. The results of such studies can be utilized for calibration and improvement of models for scenario analyses of the impacts of climate change on P mobility. In addition, results can be helpful for determining strategies for adaption, mitigation, or modification in drainage system design and management as well as crop, tillage, and fertilizer management.

The objective of the present study is to estimate the influence of high rainfall intensity on P mobility at steady flow conditions in the uppermost half meter of a clay soil. The experimental setup consisted of an in situ rain simulator which controlled the input rate of water, and a zero-tension collection tray at half a meters depth, through which the percolating water outflow rate as well as P load was measured.

Site and methods

The experiments were carried out during the period late May to June in eight experimental plots (two locations/plot), within a field recently planted with spring barley (Hordeum vulgare L.), at Ultuna in Uppsala (59^°48′35″N, 17^°39′00″E). The soil was silty clay to clay (Typic Eutrochrepts; Soil Survey Staff 1975), which is relatively representative for soils of high potential for agriculture production in Sweden, and is characterized by structural dynamics and occurrence of macropores. About 20 kg P had been annually applied during the actual and preceding seasons. A summary of ranges of climate properties (during the course of the experiments) and soil properties (in the different plots) are given in Table 1.

Table 1. Duration of experiment and ranges of climate parameters (climate station) and soil properties (in the plots studied).

Dates	23 May–21 June
Cumulative rain	40 mm
Average daily temperature	10.6–22.6°C
Maximum daily temperature	14.4–29.4°C
Clay content	39–59% (0.1–0.15 m), 39–63% (0.3–0.35 m)
Antecedent P-AL	2.6–5.1 mg 100 g^−1 (0.1–0.15 m), 0.9–4.1 mg 100 g^−1 (0.3–0.35 m)
Antecedent water content	25–35% (0.1–0.2 m), 30–40% (0.2–0.5 m)

A description of the development of the method used in the present study is given in Joel et al. (2012). A tent covering the whole measurement area in each of the 16 locations (5 × 4 m) was installed one day before each measurement occasion and taken away after the last measurement had taken place. During each measurement occasion, a rain simulator, 0.5 × 0.5 m with 481 drippers (capillary tubes with 0.8 mm inner diameter and 2.3 mm external diameter), was placed 1 m above the soil surface and vertically above a drainage collection tray placed at 0.5 m depth below the soil surface (accuracy ±0.05 m due to irregularities on the soil surface). In order to install the collection tray a trench was dug to 1 m depth. In the wall of the trench, a cavity was constructed by aid of a motor driven earth auger, with its roof at the depth 0.5 m and being 0.7 m wide, 0.4 m high, and about 0.9 m into the soil profile wall. The roof of the cavity was prepared with a spade and a spatula so that the natural pore system was not disturbed. The collection tray – constructed in stainless steel with dimensions of 0.57 × 0.57 m, 0.01 m high edges, v-shaped bottom plate and an outflow device on the one side – was filled with spherical plastic beads (4 mm diameter) to ensure good soil contact and free passage of sediments. It was placed in contact with the roof of the cavity by the aid of a scissor jack, with its front side at least 0.2 m from the border of the trench wall. The collection tray, in this way, composed the base of a zero-tension (or gravity) lysimeter setup, i.e. water moved into it by gravity and no suction was applied to the overlying soil. A zero-tension setup was chosen since sediment and P transport in the present study are generated at relatively high water flow rates and therefore the majority of the transport was assumed to occur via macropore flow (Steenhuis et al. 1995; Schmidt & Lin 2008).

On the soil surface, vertically above the cavity, a 0.525 × 0.525 m metal frame with 0.05 m high rims, composing a surface runoff collector, was placed at least 0.2 m from the border of the trench wall and inserted into the soil to delimit the area of infiltration. In order to prevent lateral outflow from the lysimeter area, a narrow trench (10–15 mm wide) was cut with a chain saw, around the surface runoff collector down to 0.4 m depth: the trench was filled with bentonite and subsequently watered so that the bentonite could expand and seal the narrow trench wall.

Simulated rain was applied following two sequence orders in the trench of each plot. First, at the excavated location in one end of each trench, a lower intensity, opting for percolation rates of 5–10 mm h⁻¹, was applied until a steady drain outflow rate from the collector tray was reached. This rate was immediately followed by an increase in application rate to a higher intensity, opting for 20–30 mm h⁻¹, that was as for the lower intensity run until a new steady outflow rate was reached. This sequence order will here be called L-H (standing for low to high intensity). In the second sequence order, here called H-L, which was carried at the excavation in the other end of the trench, the above procedure was repeated but starting with the higher rainfall intensity and proceeding to the lower, both run to steady outflow rate conditions. Thus, totally 16 rain simulation runs were made, two in each of the eight plots of which one was run from low to high (L-H) intensity and one from high to low (H-L) intensity. The L and H intensities were chosen to represent relatively high and very high intensities, respectively.

During runs the amount of water applied by the rainfall simulator was measured continuously as the difference between readings on a balance on which the water supply container was placed. The actual input intensities of the rain simulator to create the lower intensities (L) above varied in most cases between 10 and 11 mm h⁻¹ (occasionally down to 7 and up to 13 mm h⁻¹), whereas the input to create the higher intensities (H) mostly varied between 33 and 35 mm h⁻¹ (occasionally down to 24 and up to 47 mm h⁻¹). The outflow from the collector tray was measured volumetrically at time intervals of 5, 10, or 20 min depending on flow rate. Leachate from the collection tray, including suspended particles, was sampled at strategic time intervals during each simulation in bottles and stored at 4^°C or frozen until analyzed. Thus, samples were taken during unsteady increasing and decreasing water outflow intervals as well as during steady water outflow intervals. In the present study, however, only samples that represent the intervals with steady outflow rates have been considered.

Water samples were analyzed in the laboratory with regard to total P (TP) and soluble reactive P (SRP) in accordance with Swedish Standard (EN 1189). Separation of particulate P from soluble P was done by filtering the samples through a 0.45 μm filter.

In the present study, thus, only the parts of the hydrographs representing steady water outflow, and the P samples representing these parts, were considered. Given the seasonal timing and the procedure of the experiment, this approximately corresponds to a situation of heavy rains at the early development stage of the crop which wet up the soil so that steady flow is attained over a cross-sectional area in the soil. The soil matrix is thus wetted and, given the high input intensity, the flow in macropores is assumed to be saturated or near-saturated.

Results and discussion

No surface runoff originating from the water from the rain simulator was observed during any of the measurement occasions. Macropores were visible throughout the experiment period in the form of earthworm channels and desiccation cracks, in some parts of the horizons evenly spread within small distances, in other more compact parts more irregular and with larger space in between them.

Drain outflow and P mass flux rates for the drip intensity sequences from one of eight plots is shown in Figure 1 to illustrate the procedure. It is seen in the sequence L-H (Figure 1a), that outflow from the collector tray starts 230 min after initiation of simulated rainfall from the rain simulator and that the outflow rate after this breakthrough increases toward steady value (when there is no change in outflow rate with time) at about 6 mm h⁻¹. The breakthrough of TP occurs simultaneously as the water breakthrough, whereas SRP shows detectable breakthrough in the sample taken 100 min after water breakthrough. TP increases toward a value at about 20 μg m⁻² min⁻¹. Soluble reactive P ends at 1 μg m⁻² min⁻¹. After changing from low to high intensity, outflow rate increases, first rapidly and then slower, toward steady value at about 23 mm h⁻¹, and TP and SRP increases to values around 170 μg m⁻² min⁻¹ and 13 μg m⁻² min⁻¹, respectively.

Figure 1. Drain outflow rates and P mass rates for the drip intensity sequences – example from one of eight plots: (a) low to high intensity, total P (TP) and soluble reactive P (SRP); (b) high to low intensity, TP and SRP. The input rain simulator intensities were in (a) during time 0–440 min = 10.3 mm h⁻¹, during time 440–540 min = 33.7 mm h⁻¹; and in (b) during time 0–248 min = 33.5 mm h⁻¹, during time 248–338 min = 10.1 mm h⁻¹.

In the sequence H-L (Figure 1b), outflow from the collector tray starts earlier than in L-H, i.e. 53 min after the onset of the rain simulator, and then increases toward steady value at about 22 mm h⁻¹, and after the decrease in input intensity decreases toward 6 mm h⁻¹. Note that the H and L intensities, i.e. by comparing the results in Figure 1b with those in Figure 1a, have similar steady outflow rates, respectively, irrespective of sequence order. The breakthrough of TP occurs simultaneously as the water breakthrough (Figure 1b) and it increases toward 116 μg m⁻² min⁻¹, whereas SRP is detectable in the sample taken 35 min after water breakthrough and increases toward 20 μg m⁻² min⁻¹. At the subsequent L, TP and SRP decrease toward 17 μg m⁻² min⁻¹ and 3 μg m⁻² min⁻¹, respectively. As for the outflow rate, both TP and SRP mass fluxes at H and L, respectively, are fairly similar irrespective of sequence order.

In the following, only water outflow rates and TP and SRP fluxes at steady or assumed steady water outflow conditions are considered, i.e. one value for each of these parameters per L and H in each sequential run as identified in Figure 1 above. The selected value is the average of the data points representing steady conditions, e.g. for outflow rates in Figure 1a: three points for L and four for H, and in Figure 1b: six for H and four for L. When analyzing the results from all plots it was observed that the rain simulations generated two steady water outflow rate (intensity) ranges: (1) low intensity equaling 4–9 mm h⁻¹, which corresponds to 1 hr duration rainfall intensity with less than 0.5-year-recurrence interval (Dahlström 2010) and (2) high intensity equaling 22–28 mm h⁻¹, corresponding to 1 hr duration and 10-year-recurrence interval. Furthermore, one extreme H outflow rate at 37 mm h⁻¹ (1 hr duration and 20-year-recurrence interval) was observed.

The larger steady water flow intensities generate larger mass fluxes of TP as well as of SRP (Figure 2). Exponential least square fits (Figure 2a, 2b) explain the increases better than linear fits (Figure 2c, 2d). Treating sequence 1 and sequence 2 as one common sample, i.e. fit least square lines to all data in Figure 2a and 2b, respectively, give the relationships TP = 11.86e^{0.104 SFI} (R² = 0.83) and SRP = 1.67e^{0.121 SFI} (R² = 0.56), where SFI = steady flow intensity. These similarities to the relationships in Figure 2a and 2b underscore the conclusion above that outflow rates (SFI), TP and SRP mass fluxes at both L and H, respectively, are fairly similar irrespective of sequence order. The least square fits in Figure 2 should be regarded as approximate trend lines since they are based on two groups of data, and data in the range 10–20 mm h⁻¹ are missing. However, these relationships are interesting for discussing possible influences of rain intensity changes on P movements. Complementary studies can be carried out to verify the whole ranges of intensities.

Figure 2. Relation between P mass flow and steady flow intensity for water sequence 1 (S1) and sequence 2 (S2) for all 16 experimental locations: (a) total P (TP), exponential; (b) soluble reactive P (SRP), exponential; (c) TP, linear; (d) SRP, linear.

The results shown in Figure 2 raise the question, what are the implications of increasing rainfall intensities on P mobility in soil? If we, for example, look at the increment between the intensities in the range 4–28 mm h⁻¹, an average increase by approximately 12% (µg m⁻² min⁻¹) per unit increase in intensity (mm h⁻¹) can be observed for TP (11%) and SRP (13%) (Figure 2). As mentioned in the introduction, the maximum precipitation intensity is predicted to increase by 10–20% in western Sweden (Svenskt Vatten 2011). An average 15% increase in intensity would then imply an increase in TP of about 2.5 μg m⁻² min⁻¹ and in SRP of about 0.4 μg m⁻² min⁻¹ at lower intensities, and 76 and 20 μg m⁻² min⁻¹ for TP and SRP, respectively, at higher intensities (Table 2). To give a hint about the relative importance these differences may have in practice, these figures are converted to kg ha⁻¹ h⁻¹ (Table 2) and compared to a very general figure of total P leaching in Sweden, i.e. 0.3 kg ha⁻¹ year⁻¹ (SCB 2011). At the lower intensities, thus, these differences correspond to 0.5% of the yearly TP leaching (Table 3). One hour duration rainfall, with recurrence of 2–3 times per year (Dahlström 2010), would generate 1.0–1.5% of the yearly leaching. Correspondingly, at the higher intensities, the differences in Table 2 correspond to 15.0% of the yearly TP leaching. The recurrence at these intensities are 1 per 10 years for a 1 hr duration rain, 2 per year for a 15 min rain, and several times per year for a 5 min duration rain. As an example, thus, the higher intensity would generate altogether 7.6% of the yearly leaching during two 15 min rainfalls.

Table 2. Increases in leaching of total P (TP) and soluble reactive P (SRP) at 0.5 m depth if intensity of heavy rainfall is predicted to generally increase by 10–20% (average 15%) at average low intensity (=6.5 mm h⁻¹) and average high intensity (=25 mm h⁻¹) using exponential relationships between P and steady flow intensity (x) on the combined data-sets of sequence 1 and sequence 2 (TP = 11.86e^0.104x, SRP = 1.67e^0.121x).

	Leaching (at 0.5 m depth) (μg m^−2 min^−1)	Leaching after 15% intensity increase (μg m^−2 min^−1)	Difference (μg m^−2 min^−1)	Difference (kg ha^−1 h^−1)
TP low	23.3	25.8	2.5	0.0015
TP high	159.7	235.8	76.1	0.0457
SRP low	3.7	4.1	0.4	0.0002
SRP high	34.4	54.1	19.7	0.0118

Table 3. Increase in total P (TP) leaching at 0.5 m depth as expressed in percentage of total P leaching in Sweden (=about 0.3 kg ha⁻¹ year⁻¹ (SCB 2011)) and related to duration and recurrence of corresponding rain events, i.e. 6.5 and 25 mm h⁻¹ for low and high intensity, respectively.

	1 hr duration	15 min duration	5 min duration
TP low	0.5%
	(2–3 times per year)
TP high	15.2%	3.8%	1.3%
	(1 time per 10 year)	(2 times per year)	(several times per year)

As mentioned in the method section, the rain simulator input intensities in most cases were larger than the steady water outflow rates. Part of the differences may be due to that some inflow of water from macropores into soil matrix may have still occurred even though, at steady outflow conditions, the soil matrix was assumed to be wetted to the degree that further inflow rates into matrix were negligible compared to rates in the macropore system. Part of the differences may be due to that some leakage of water have occurred through weaknesses in the bentonite barrier, which had the function of delimiting the flow area and reduce lateral flows, especially in the lower parts between the collector tray upper boundary and bentonite barrier lower boundary. Despite these obstacles, we have assumed that the steady outflow rates from the collector tray, which equal steady volume flow per tray area and time (flux) into the tray, were approximately equal to the volume flow per area and time (flux) also in the overlying half meter of soil. The intensities, based on measured outflow rates, as presented in our study, are supposed to reflect rain intensities for the situation outlined in the end of the method section: i.e. of heavy rains at the early development stage of the crop which wet up the soil in a field to the level that steady flow equal to the rain intensity is attained over the cross-sectional flow area, with most part of the flow taking place in the macropores.

It should be recognized that our study only considers point measurements and P flows in the uppermost half meter of the soil. The conditions and properties in the subsoil and down to the drains (at about 1 m depth) or groundwater may, however, strongly affect P flows to the environment. In some soils, like clay soils with well developed macroporosity, P flow may be large even in the subsoil, whereas in other soils, the P may become sorbed, for example in sandy subsoils with iron and aluminum oxides (Andersson et al. 2013). In the soil studied in the present article, macropores are present also below half a meters depth and hence P mobility may be high even in the subsoil. Nevertheless, the results in the present study demonstrate the degree of sensibility for increasing downward transport of P in the upper soil profile that increasing rain intensities may encounter.

Disclosure statement

No potential conflict of interest was reported by the authors.

Additional information

Funding

The authors wish to thank the Swedish Farmers' Foundation for Agricultural Research for their financial support of this project [grant number H0970242].