Relationship Between Runoff and Sediment Transfer in a Slope–Gully Cascade System During Extreme Hydrological Events in the Lublin Upland, East Poland

Abstract

Erosion monitoring was carried out between 2003 and 2022 using a hydrological station with a Thomson overflow, a water gauge, and a limnigraph installed at the outlet of the Kolonia Celejów gully system. The study area is located in the north-western part of the Lublin Upland in the Nałęczów Plateau mesoregion (SE Poland). The total amount and intensity of precipitation were measured using an automatic station and water runoff and suspended sediment yield (SST) were also continuously measured. High variability in water runoff was observed during this period (max. of about 76,000 m³ and mean > 26,000 m³), and as a result of numerous heavy rains, a significant increase in SST (max. of about 95 Mg to about 1200 Mg and mean of 24 Mg to about 215 Mg) was noted in the second half of the measurement period. Most of the material removed at that time came from the cutting of the gully bottom and from the redeposition of material transported from the catchment used for agricultural purposes. In order to determine the volume of material delivered to the slope–gully cascade system in November 2012, a second station was installed at the gully head, which only operated until June 2013. However, the measurements covered all snowmelts and summer runoffs, as well as the June downpours. At the same time, these measurements represent the first unique attempt to quantify the delivery of material from the slope subcatchment to the gully system. The year 2013 was also important in terms of water runoff from the loess gully catchment area (about 40,000 m³) and was a record year (SST > 1197 Mg) for the total amount of suspended material runoff (7.6% and 33.5% of the 20-year total, respectively). During the cool half of the year, 16,490 m³ of water (i.e., 42% of the annual total) flowed out of the gully catchment area, and during the warm half of the year, 23,742 m³ of water (59% of the annual total) flowed out. In contrast, 24,076.7 m³ of water flowed out of the slope subcatchment area during the year, with slightly more flowing out in the cool half of the year (12,395.9 m³ or 51.5% of the annual total). In the slope and gully subcatchment areas, the suspended sediment discharge clearly dominated in the warm half of the year (98% and 97%). The record-breaking SST amount in June was over 1100 Mg of suspended sediment, which accounted for 93% of the annual SST from the gully catchment area and over 94% in the case of the slope subcatchment area. The relationships in the slope–gully cascade system in 2013 were considered representative of the entire measurement series, which were used to determine the degree of connectivity between the slope and gully subsystems. During summer downpours, the delivery of slope material from agricultural fields accounted for approx. 15% of the material removed from the catchment area, which confirms the predominance of transverse transport in the slope catchment area and longitudinal transport in the gully. The opposite situation occurs during thaws, with as much as 90% of the material removed coming from the slope catchment area. At that time, longitudinal transport dominates on the slope and transverse transport dominates in the gully.

1. Introduction

A dense network of loess gullies plays an important role in shaping surface runoff and provides significant amounts of terrigenous material [1,2,3,4,5,6,7]. When a gully drains directly into a river valley, some of the material reaches the riverbed and feeds the fluvial system [8,9,10,11,12,13]. Gullies most often concentrate and accelerate periodic water runoff from slope catchments, draining source catchments and increasing the risk of flash floods [4,13,14,15]. In heavily dissected loess areas, gullies are also responsible for small-scale flooding, muddy floods, and reservoir sedimentation [13,16,17,18,19]. According to [4], gully erosion is the main process that delivers slope sediments, accounting for between 10 and 95% of the total mass of sediments deposited in catchment areas, even when gullies occupy less than 5% of the catchment area.

In contrast, the role of gully drainage area size on the initiation of gully formation and on the intensity of gully erosion processes has been unanimously emphasised, especially in modelling studies [4,16,20,21]. Ref. [5] analysed whether a critical drainage area is a necessary factor for producing sufficient runoff to concentrate and initiate gullying. The long periodic surveys conducted on China’s Loess Plateau showed that reducing the catchment area effectively slowed gully development [15]. Research on gully head-cut retreat also showed that there is a significantly correlation between the gully erosion rate index and the runoff-contributing area of the gully [22]. A study carried out on the Nałęczów Plateau [23] showed that the shape of a gully catchment also determines the concentration and volume of water runoff, thus determining the dynamics of gully erosion processes.

On the other hand, the importance and influence of extreme events on the gully erosion intensity and the development of the loess gully network is well established [4,5]. An exceptionally high intensity of sediment transport (2289.9 Mg km⁻²) in the loess gully system during the summer rainstorm of 1969 was documented in the Lublin Upland in [18]. Similar ratios with extreme sediment runoff of 2–10 × 10³ Mg km⁻² for the few downpours of the 20th century in this loess region were also compiled in [24]. According to [19], the specific gully erosion rates under heavy rainfall can reach as high as 10,000 Mg km⁻² or more, comparable to annual gully erosion rates in the Loess Upland in China [15,25]. It has also been documented that most gullies formed during historical periods in Europe, with extreme rainfalls associated with the expansion of farmland and deforestation [5,26]; for example, this occurred in Germany [27], Poland [1], Slovakia [12,28], and Romania [29].

In addition, the area of the gully catchment uncut by gullies is also an important source of material for the gully system [30]. However, the most commonly identified feeding areas in a gully catchment are the gully bottom and gully head. The main sources of slope material and pathways in the gully system are activated during intensive water runoff, which effectively dissects the gully bottom and causes an increase in gully volume and gully head expansion [12,17,18,19]. Nevertheless, it is important to empirically quantify the amount of material delivered directly from the slope subcatchment area to the gully system. The connection between slopes and gully systems during extreme events and the share of soil erosion products carried out of the gully catchment area also require attention. However, there is a lack of quantitative assessments of this material contribution to the erosion balance as the usually old and forested gully systems are generally inactive and only have episodic runoff, which makes monitoring studies difficult.

Studies of slope runoff in the loess areas of the Lublin Upland have shown that, compared to dry loess valleys used for agriculture, runoff from old forested gullies is rare and small in volume [23,24,30,31,32,33]. In addition, measurements of runoff from three gully catchment areas in the Nałęczów Plateau have shown that local conditions can influence runoff, causing changes that differ by an order of magnitude, even in neighbouring catchment areas [23]. A good site for this type of monitoring study is the small and homogeneous gully catchment in Kolonia Clejów (NW part of the Lublin Upland). This loess gully catchment is nearly square shaped and has a periodic runoff, which is fed in wet years by a small spring located in the main gully head. Therefore, the main objective of the monitoring undertaken in this study was to determine the connectivity of the slope and gully systems during extreme events and the volume of slope material delivered to the gully system during intense snowmelt and summer downpours. The determination of the detailed conditions and mechanism of sediment transport in the loess gully catchment area can contribute to the development of effective techniques to reduce gullying in sensitive loess areas with dense gully networks.

2. Study Area

The study area is located in the north-western part of the Lublin Upland in the Nałęczów Plateau mesoregion (Figure 1). The area is covered by Vistulian loess with a thickness of up to 30 m [34]. The main feature of the first-order relief of this area is a fairly flat and even plateau, bounded by accumulation edges. The second-order reliefs are created by extensive periglacial erosion and denudation valleys with flat and accumulative bottoms. Trough-shaped valleys and slope troughs fragment the slopes and sides of the river valleys, while numerous closed depressions without drainage occur on the plateaus and in the heads of the dry valleys [35,36,37]. As a result of intensive and long-term agricultural use, a dense network of gullies (average of 5 gullies km km⁻²) has formed, cutting through the plateau and the bottoms of the dry valleys [38]. The micro-relief is enriched with contemporary anthropogenic landforms, e.g., edges and terraces of cultivated land, sunken lanes, road gullies, road troughs, and road embankments [30].

The Kolonia Celejów gully catchment is located at the edge of an area with a high gully density of about 5 gullies km⁻², with some areas exceeding 10 km² (Figure 1). A small, homogeneous and nearly square-shaped catchment can represent loess areas with a well-developed surface runoff network consisting of dry valleys and gullies. These favourable conditions were exploited by covering the catchment since the hydrological year 2003, with continuous monitoring of the water and suspended sediment runoff.

2.1. Hydrometeorological Conditions

The climate of the western part of the Polish Uplands is characterised by a moderate transitional climate with characteristic large fluctuations and contrasts in temperature and humidity throughout the seasons and throughout the year. The average temperature in January is around −4 °C and around 18 °C in July, with an annual average of around 7.5 °C. The total annual precipitation varies, ranging from 550 to 700 mm. Precipitation is most abundant in summer, averaging 70–100 mm in July. The maximum snow cover thickness is 20–50 cm, and the water equivalent in snow at the end of winter is 40–60 mm but can reach 100 mm [39]. This dissected loess area is characterised by a frequent occurrence of intense summer downpours, most often from May to July, and sometimes until September, which average 30 mm in 60–120 min [39]. On the other side, there are often long periods without rainfall, which cause droughts and low water levels and also pose a threat [40,41]. Most often, rivers with a complex ground–meltwater–rainwater supply regime experience one main meltwater flood (March–April) and sometimes a second rain flood, usually in July.

The rather variable climatic conditions of this region, especially the occurrence of wet periods with summer downpours, also result in large variations in the intensity of soil and gully erosion. In the slope systems in the study area, due to the intensive agricultural use and high relief, the intensity of soil erosion is very high, especially during heavy rainfall and spring thaws [24,30,31,32,33,41,42,43,44]. Similarly, in the case of the gully systems, the highest intensity of water runoff and sediment transport processes are only observed during extreme events. These events sometimes cluster in series, which have been used to separate the phases of intensive gully erosion (e.g., 1997–1999 and 2010–2018), which are punctuated by periods of low activity or stabilisation of the gully systems [30,31,41,45]

2.2. Landscape Features and Land Use

A branched and forested gully system located in Kolonia Celejów, with a total length of 7.5 km and a catchment area of 1.24 km², has been monitored since 2003 [19,36] and was selected for detailed research. It consists of two main erosional forms: Celejów Dół and Grabczyna, which are approximately 1 km long and connected via a 200 m outlet section, with side forms that are 100–300 m long (Figure 1). The slopes of the gully and its edges are stabilised by oak–hornbeam forests but numerous piping–erosion holes have developed at the forest edge and in side branches, creating characteristic badland reliefs [38]. The bottom of the gully is wide and accumulates periodic drainage. The largest watercourse, with a flow rate of up to 1 dm³·s⁻¹, flows out at the Celejów Dół gully head. Heavy rainfall in 1997 caused erosion dissection (secondary gully) that cut through the colluvium and bedrock in the lower section of this main gully. Over the course of 16 years, its volume reached 1000 m³, with a length of 160 m, a width of 2.5–4.5 m, and a depth of 1.5–3 m

The relative heights in the gully catchment area do not exceed 214 m above sea level, and the elevation differences are 48 m (

Table 1. Morphological characteristics of the Kolonia Celejów gully catchment area.

Drainage Basin	Area [km2]	Height Min [m a.s.l.]	Height Max [m a.s.l]	Height Range [m]	Height Mean [m a.s.l.]	Flow Length [m]	Runoff System
Grabczyna gully	0.51	167.0	214.0	47.0	190.5	1200	Episodical
Slope subcatchment	0.20	177.5	206.5	29.0	192.0	650	Ephemeral
Celejów Dół gully	0.43	166.0	202.6	31.4	184.3	1409	Perennial
Kolonia Celejów catchment	1.24	166.0	214.0	48.0	190.0	2135	Perennial

). The slopes have a complex, convex–concave profile and vary in length from 50 m to 500 m. The slope angle is similarly varied (from 2° to >45°), and the average slope angle does not exceed 8°. The depth of the main gullies varies between 5 and 15 m. The main branch gullies have flat and accumulative bottoms, with a width ranging from 5 m to 10 m.

On the thicker loess deposits, fertile grey soils have developed, forming a characteristic catena of eroded soils [30,42]. Currently, in the Kolonia Celejów gully catchment area, agricultural land covers approx. 65% of the area, while forests and wasteland cover 30%. The catchment area is dominated by arable land (65%), and the crop structure consists of cereals, with a significant share of berry plantations and root crops (potatoes and sugar beet) and, recently, ornamental shrubs (thuja and spruce). The land belongs to individual farmers, with small farms (7–8 ha on average) usually divided into several plots. Since the 1970s, the fields near the gully edges have been left fallow and are now covered by a secondary succession of oak–hornbeam forest (Tilio-Carpinetum).

The choice of a representative study site is limited because the control of the gullies by the forest weakens their hydrogeomorphological activity. In an area with a high gully density, which is characteristic of loess areas with significant height differences, many gullies do not meet the criteria of a controlled catchment. Typically, gully catchments also have no permanent drainage and are almost entirely dissected by the gully system. The investigated gully system has periodic water runoff and has shown exceptional geomorphological activity since the late 1990s. This is largely due to the intensive agricultural activity in the slope subcatchment, with agricultural land occupying 30% of the entire catchment.

3. Materials and Methods

3.1. Hydrometeorological Measurements

Detailed studies of the sediment transport in a cascade system were conducted in the catchment area of the Kolonia Celejów gully, which consists of three subcatchment areas with diverse hydromorphological conditions (Figure 2). The total amount and intensity of precipitation were measured using a TPG-023 automatic rain gauge manufactured by A-STER in the subcatchment area of the Grabczyna gully. Water levels were recorded at the outlet of the catchment area at a triangular Thompson weir with a water gauge patch (Figure 3). The recording has been carried out continuously since 2003 using an electronic limnigraph (THALIMEDES by OTT). In order to determine the amount of sediment transported in the slope–gully cascade system, a similar station was installed in November 2012 at the gully head on an abandoned farm (Figure 3A). Because the new owner reactivated the pond that had once existed there, this station only operated until June 2013. The water flow was calculated using Thompson’s formula for small streams, which was determined empirically for a triangular weir with an opening angle of 90°:

Q = 1.4⋯h^5/2,

where Q is the flow [m³ s⁻¹] and h is the water depth [m].

3.2. Sediment Transport Measurements and Regression Models

Water samples for suspended sediment concentration (SSC) analysis were collected by an observer using a bottle bathymeter (0.5 dm³) depending on the size and quality of the runoff. During small, stable periodic runoff, samples were collected once a week, and during short, intense floods, samples were collected up to 4–6 times per hour. The frequency and timing of water sampling depended on the course of the flood wave. Based on experimental measurements conducted in this catchment area, the selected sampling frequencies ensured that the entire flood peak was adequately represented. The water samples were analysed in the MCSU laboratory using the Brański gravimetric method with filter cartridges [46].

Regression models based on the correlation between water flow and suspended sediment concentration were used to estimate the total suspended sediment runoff (SST). The generated regression equations were used to calculate the amount of suspended sediments for each instantaneous flow rate during periods when no samples were taken (during storms and at night).

During the heavy rainfall on 23 June 2013, the water runoff at the upper overflow in the main gully head was blocked; therefore, the water and suspended sediment runoff from the slope subcatchment area were reconstructed. The high linear correlation of the analysed water runoff parameters between the upper and lower overflows during the 11 June downpour allowed for the use of a linear regression method to replace any missing data.

Unfortunately, the short operation period of the hydrological station at the main gully head until June 2013 did not allow us to obtain a longer measurement series, which would have allow for a full analysis of the connectivity of the slope and gully subsystem and the slope material supply on a time scale longer than 1 hydrological year. Future research should focus on studying the delivery of material from the slope to the gully system under different varying runoff conditions, e.g., during a series of downpours or continuous rainfall, and for other types of land use or newly introduced crops.

4. Results

4.1. Hydrometeorological Conditions in 2013

4.1.1. Precipitation Distribution and Flow Regime

In the 2013 hydrological year, the total precipitation in the analysed catchment area reached 558.9 mm, which was 14.1 mm below the long-term average [35]. Precipitation in the warm half of the year amounted to 352 mm, accounting for 63% of the annual total. Precipitation in the cool half of the year reached 206.9 mm, accounting for 27% of the annual total. June was particularly heavy in terms of rainfall (140.1 mm), which accounted for 39.8% of the total for the warm half of the year (

Table 2. Hydrological conditions of water and suspended sediment runoff in the Kolonia Celejów gully catchment in the 2013 hydrological year.

Month	Rainfall [mm]	Upper Weir			Lower Weir
		Mean Flow [m3 s−1]	Water Runoff [m3]	Suspended Sediment [Mg]	Mean Flow [m3 s−1]	Water Runoff [m3]	Suspended Sediment * [Mg]
XI	26.4	0	0	0	0.4	1030	0.2
XII	25.8	0	0	0	0.2	628	0.1
I	54.5	0.9	1403.3	0.6	0.7	1987	0.6
II	35.6	1.4	3468.2	0.6	1.4	3479	0.6
III	21.8	1.3	3360.6	4.3	1.3	3572	4.5
IV	42.8	1.6	4163.8	30.3	2.2	5794	40.0
V	91.9	1.0	2642.4	13.3	1.3	3554	17.0
VI	140.1	5.6	5824.4	1104.5	6.3	16,455	1116.9
VII	46.1	1.4	3214.1	15.4	1.4	3649	17.9
VIII	17.5	0	0	0	0	84	0
IX	49.1	0	0	0	0	0	0
X	7.3	0	0	0	0	0	0
Sum/mean	558.9	1.1	24,076.7	1169.0	1.4	40,232	1197.8

Mg: megagram (1 Mg = 1000 kg); * calculated data.

).

The precipitation distribution was reflected in the periodic flow regime of the drainage stream in the Kolonia Celejów gully catchment area. In general, low flow dominated throughout the year and only exceeded 10 dm⁻³ s⁻¹ during a few thaw and precipitation episodes (Figure 4). Maximum flows occurred after the June downpours, and from August onwards, the outflow disappeared, similar to the beginning of the hydrological year.

The flows in the upper overflow, which closes the slope subcatchment area, were well correlated with the flows in the lower cross-section at the mouth of the gully catchment area. During winter and spring thaws, the maximum flows in both overflows were similar, which indicates a delay in runoff in the gully catchment area due to the high retention capacity of the accumulative gully bottom. However, during summer downpours, flows at the catchment closure increased by over 200%, mainly as a result of an increase in the drained area.

4.1.2. Variability in Water Runoff and Suspended Solids

In the period 2003–2017, there was very high variability in water runoff (max. of about 76,000 m³ and mean > 26,000 m³). The suspended sediment yield also showed high variability. In the second half of the measurement period (after 2010), a significant increase in the sediment yield was noted. For comparison, in the 2003 hydrological year, the SST index was about 95 Mg, and in 2013, it increased to about 1200 Mg. In the first period of 2003–2009, the mean annual SST index was only 24 Mg and in the second period (2010–2023), it increased to about 318 Mg (10 times higher), mainly as a result of numerous heavy rainfall events. The water runoff from the Kolonia Celejów gully catchment area in the 2013 hydrological year was more than twice the long-term average for 2003–2022 and half as high as the maximum values recorded in 2017. During the cool half of the year, 16,490 m³ of water (i.e., 42% of the annual total) flowed out of the entire gully catchment area, and during the warm half of the year, 23,742 m³ (59%) flowed out. In contrast, 24,076.7 m³ of water flowed out of the slope catchment area during the year, with slightly more flowing out in the cool half of the year (12,395.9 m³ or 51.5% of the annual total). In the slope catchment area, the maximum water runoff was recorded in June, with relatively high monthly runoffs also recorded between February and April (Figure 5A). The monthly water runoff from the gully catchment area was similar to that from the slope catchment area, except for the record-breaking values in June, which were over 300% higher (Figure 5B).

The total annual suspended sediment yield from the gully catchment area and the slope subcatchment area was similar (1169.8 Mg and 1197.8 Mg, respectively), which are record highs (over 500% higher than the long-term average) for the entire measurement period of 2003–2022. In both catchment areas, the suspended sediment discharge in the warm half of the year clearly dominated, accounting for 98% and 97% of the yearly total. June delivered a record-breaking yield of over 1100 Mg of suspended sediment, which accounted for 93% of the annual suspended sediment discharge from the gully catchment area and over 94% in the case of the slope catchment area. In the remaining months, the suspended sediment yield was low in both catchment areas and did not exceed 40 Mg (Figure 5). The amount of suspended sediment transported was therefore determined based on the June downpours, i.e., two events on 11 and 23 June. The course of both events therefore requires a detailed description. For 2013, the amount of snowmelt was also significant, which was also used to determine the connectivity of the slope and gully subsystems.

4.2. Water and Sediment Runoff in a Slope–Gully Cascade During Extreme Events

4.2.1. The Course of the Spring Thaw

Snow cover during the winter of 2012/2013 formed several times, and its continuity was interrupted by thaws (Figure 6). Continuous snow cover appeared at the beginning of March despite unfrozen ground, was replenished several times by snowfall, and survived until the thaw at the beginning of April. The peak of the runoff occurred on 11 April, in the form of rain and snowmelt runoff. In total, during the thaw, 14,832 m³ of water and 45.7 Mg of suspended sediment were discharged from the gully catchment area, which accounted for 36.9% of the annual water discharge and 3.8% of the annual suspended sediment yield. On 12 April, a record water discharge of over 1000 m³ was recorded, while on 11 April, the suspended sediment yield peaked (Figure 7A,B). These two days therefore determined the amount of water discharge and suspended sediment yield during the entire thaw period. During the final thaw, and especially on 11 April, strong links between the slope and gully subsystems were also observed. The supply of material from the slope subcatchment area determined the total suspended sediment yield from the Kolonia Celejów gully catchment area.

4.2.2. Water and Suspended Sediment Yield in a Cascade System on 11 April

On 11 April 2013, the meltwater flows in the described catchment area reached their peak (Figure 8A,B). In the overflow closing the gully catchment area, the flood phase was clearly flattened, which confirmed the retention of runoff water by the wide and accumulative bottom of the main gully. The suspended sediment concentrations in both overflows followed the flood wave, with their peak showing a complex pattern and preceding the flood peak. A rapid increase and decrease in the suspended sediment concentration is characteristic of episodic phenomena, which can be linked to the availability of transported materials, especially in slope subsystems. Increased suspension concentrations occurred in the peak phase in both overflows. However, its maximum values were recorded synchronously at 12:20 p.m. during the peak flow. A secondary concentration of suspended sediments was also recorded after 1:10 p.m. in both overflows, which was associated with the delivery of soil material from the upper part of the catchment area.

On 11 April, there was a significant outflow of meltwater accompanied by increased suspended sediment transport. In general, the course of the water flow and suspended sediment yield was similar in both overflows, with higher values of water flow and suspended sediment yield recorded at the lower overflow (Figure 9A,B). This shows the important role of the slope catchment area in supplying water and soil material to the gully system during thaws. The maximum water discharge at the lower overflow occurred with a 50 min delay, and its volume was over 4 m³ higher. This confirmed the retention function of the snow cover lying at the bottom of the gully. Interestingly, during the peak, over 600 Mg more material was discharged from the slope catchment area. This material was redeposited at the bottom of the main gully.

4.3. June Downpours

Table 3. Comparison of the runoff patterns and effects during the June downpours of 11 and 23 June 2013.

Parameter	11 June		23 June
	Upper overflow	Lower overflow	Upper overflow	Lower overflow
Rainfall start and end	5.00–11.10		7.20–8.00
Rainfall duration	4.10 min		0.40 min
Total rainfall sum	38.3 mm		19.2 mm
Maximal rainfall intensity	1.4 mm min−1		0.9 mm min−1
Mean rainfall intensity	0.15 mm min−1		0.48 mm min−1
Water runoff time	5.20–11.00	5.20–11.10	7.30–13.30	7.40–13.40
Water runoff duration	4.40 h	4.40 h	6.00 h	6.00 h
First flow peak/time [h]	0.12 m3 s−1/5.40	0.34 m3 s−1/6.50	0.34 m3 s−1/8.00	0.86 m3 s−1/8.10
Maximal flow/time [h]	0.34 m3 s−1/6.50	0.12 m3 s−1/5.50	0.34 m3 s−1/8.00	0.86 m3 s−1/8.10
Suspended sediment concentration range	7.9–58.7 g dm−3	7.0–168.6 g dm−3	4.3–37.1 g dm−3	5.1–179.5 g dm−3
Mean suspended sediment concentration	14.8 g dm−3	44.4 g dm−3	57.6 g dm−3	48.2 g dm−3
Total water runoff	1442.1 m3	5143.9 m3	2092.6 m3	5634.8 m3
Total suspended sediment yield	58.7 Mg	424.0 Mg	73.5 Mg	481.66 Mg

shows the water sediment runoff during the June downpours. Both downpours had similar characteristics and their effects were similar in both the slope and gully subcatchments.

4.3.1. The Downpour of 11 June 2013

Within 20 min after the rainfall on 11 June, concentrated surface runoff formed in the catchment area, which generated large erosion effects (Figure 10A,B).

A rapid increase in water flow was recorded during both overflows (Figure 11A,B), with the first peak occurring at the upper overflow (Q = 0.12 m³ s⁻¹). The second, significantly higher maximum (Q = 0.34 m³ s⁻¹) occurred at 6.50 a.m., which was linked to an increase in precipitation intensity in the catchment area (Appendix A.1). The hydrograph in Figure 11 documents a very rapid increase in flows during the peak phase. The recession phase consisted of three stages: the first stage had a very rapid decrease in flow, the second stage involved the effect of overlapping runoff waves from side valleys, and the third stage showed a gradual decay of the runoff. The concentration of suspended sediments followed the flow and did not exceed 60 g dm⁻³ in the upper overflow.

At the lower overflow, the water flow and suspended solid concentration followed a similar pattern to those at the upper overflow but their amplitudes were several times higher (Figure 11A,B). During the downpour on 11 June, a high water discharge intensity was recorded, especially at the lower overflow. Two discharge maxima occurred at both overflows, separated by approximately 150 min (Figure 12A,B). The maximum water discharge at the lower overflow was over 200% higher than at the upper overflow. In addition, in the slope catchment area, the water discharge amplitudes were significantly lower and its course was more gentle. The suspended sediment yield was also bifurcated, with synchronised maxima in both overflows occurring during the peak flow. However, at the lower overflow, the suspended sediment yield was several times higher, which indicates a significant supply of transported material from the cutting of the main gully bottom [19].

4.3.2. The Downpour of 23 June 2013

During the rainfall, a concentrated surface runoff quickly formed in the catchment area, which utilised the drainage network created during the downpour on 11 June (Figure 13A,B). At the upper overflow, a rapid increase in water flow was recorded just 10 min after the start of the downpour, with the maximum occurring after 40 min (Figure 14A). At the lower overflow, the delay in these flow phases was 10 min, and the maximum flow was more than twice as high (Figure 14B).

At both overflows, the peak phase was very short, followed by a rapid drop in flow after several minutes and a very long recession phase. The water flow and suspended sediment concentration showed almost identical courses. The water flow and SSC at the upper overflow were calculated using applied statistical correlation.

Table 4. Regression models used to calculate the water states and suspension sediment concentrations at the upper overflow on 23 June 2013.

	Rising Phase		Recession Phase
Parameter	Regression Model	Determination coefficient R2	Regression Model	Determination coefficient R2
Water flow depth	y = 1.134x1.11	0.71	y = −0.04x2 + 4.32x − 33.33	0.77
Suspended sediment concentration	y = 0.022x1.94	0.95	y = 1.31x0.64	0.74

x—measured variables at lower overflow; y—calculated variables at upper overflow.

shows the regression models used to reconstruct the missing data for the upper overflow. A good fit of the models to the empirical data was obtained (R² > 0.7), which allowed us to perform a detailed analysis of the course of water runoff at the upper overflow during this rainstorm (Appendix A.2).

At the upper overflow, the maximum concentration of suspended sediment occurred after a 10 min delay. This can be linked to the lack (exhaustion) of available material in the slope subcatchment area, which was carried away during the previous downpour. In contrast, at the lower overflow, the maximum concentration of suspended sediment occurred simultaneously with the peak flow, which indicates that there was a supply of slope material (long-distance transport). On the other hand, a more than fourfold increase in suspended sediment concentration occurred as a result of increased erosion of the ‘underloaded’ stream, which could have intensively cut into the bottom of the main gully.

The course of the water flow and suspended sediment yield on 23 June 2013 in the Kolonia Celejów gully catchment area differed from those of the previous June downpour (Figure 15A,B). The maximum discharge intensity was also more than twice as high here. However, the suspended sediment yield in both overflows was synchronised. In the upper overflow, the maximum suspended sediment yield was recorded 10 min after the peak water discharge. This delay can be explained by the depletion of the available capacity to transport slope material, which had been removed during the previous rainfall.

The subsequent increase in the suspended sediment yield resulted from the development of slope runoff and the increased supply of material from soil erosion on the watershed. At the lower overflow, the maximum suspended sediment yield coincided with the peak in water discharge, which confirmed long-range sediment transport from the slope catchment area. On the other hand, the more than twofold increase in suspended sediment yield at the lower overflow confirmed the increased share of material originating from the cutting of the gully bottom.

During extreme events in the Kolonia Celejów catchment, high rates of gully erosion were recorded. The specific erosion rate reached 31.4 Mg km⁻² in the gully catchment area during the final phase of the snowmelt (

Table 5. Effects of snow melt and June downpours in the Kolonia Celejów gully catchment in the 2013 hydrological year.

Time	Total Rainfall (mm)	Maximal Flow Depth (m)	Max Flow (m s−1)	Total Water Runoff (m3)	Runoff Coefficient (%)	Sediment Yield [Mg]	Specific Erosion Rate (Mg km−2)
Snowmelts
Slope subcatchment
31 March–14 April	-	0.227	0.034	3077.6	-	31.4	157.0
11 April	-	0.227	0.034	770.7	-	13.2	66.0
Gully subcatchment
31 March–14 April	-	0.243	0.005	4756.0	-	38.9	31.4
11 April	-	0.243	0.005	953.7	-	14.0	11.3
Downpour on 11 June
Slope subcatchment
24 h values	38.4	0.565	0.34	1425.0	18.6	66.8	334.0
5.00–12.00	36.9	0.565	0.34	1422.0	19.3	66.5	332.5
Gully subcatchment
24 h values	38.4	0.812	0.7	5253.8	11.0	425.8	343.4
5.00–12.00	36.9	0.812	0.7	5143.9	11.2	424.0	341.9
Downpour on 23 June
Slope subcatchment
24 h values	19.2	0.570	0.34	2094.0	54.5	73.7	368.5
6.00–16.00	19.0	0.570	0.34	2092.6	55.1	73.5	367.5
Gully subcatchment
24 h values	19.2	0.824	0.9	5641.1	23.9	482.9	389.4
6.00–16.00	19.0	0.824	0.9	5634.8	22.1	481.5	388.3

Mg: megagram.

). During the June torrential downpours, the rate was 10 times higher in both subcatchments.

5. Discussion

5.1. Intensity of Gully Erosion

During the measurement period (2003–2023), there were no extraordinary events with an exceptionally high gully erosion intensity comparable to the events described in the Lublin Upland in 1969 [18] or 1981 [19], which had extreme sediment runoffs of 2–10 × 10³ Mg km⁻² in the Kolonia Celejów gully catchment area. However, the 2013 hydrological year should be considered “above average” as 7.6% of the 20-year total water flow and a record 33.5% of suspended sediments were washed away [30]. The calculated snowmelt erosion indices were relatively high, but two times lower than the highest values recorded in 2003 [30] despite relatively the heavy rainfall (39.3 mm) between 31 March and 13 April. The condition of the soil played an important role: in 2003, the soil was deeply frozen and icy, while in 2013, it did not freeze due to a warm winter and insulation by a permanent snow cover in March. At the same time, high erosion rates were recorded in the subcatchment area (157 Mg km⁻²), which are comparable to the results obtained in the slope catchment areas of the Lublin Upland during the rapid thaws of 1996, but significantly lower than the extreme thaw indices of 1964 or 1956 [24]. The exceptionally high daily melt erosion rates on 11 April accounted for approximately 30% of total melt erosion in the gully catchment area and as much as 50% in the slope subcatchment area. These rates are comparable to the annual ephemeral gully erosion rates [4]. During the June downpours, the specific erosion rates were similar, ranging from 332 to 389.4 Mg km⁻² in both subcatchment areas. Meanwhile, these indicators were more than twice as high in the slope subcatchment area and more than ten times higher in the gully subcatchment area compared to during snowmelt (Table 3). This indicates the dominant role of heavy rainfall in the development of loess gullies in the last 20 years and the declining role of snowmelt in Central and Eastern Europe [30,32,33,34,40,41,42,43,44,45,47]. The obtained erosion intensity indices are three times higher than those reported for the loess catchment area in [44] and fall within the class of extreme events with high erosion intensities that was compiled for the Lublin Upland in [24]. The high effectiveness of the analysed downpours indicates that their values can be compared with the annual ephemeral gully erosion indices compiled in [4] as well as the average annual gully erosion indices for the Chinese Loess Plateau [11]. These values confirmed the impact of individual hydrometeorological events on gully erosion intensity [30,31,32,33,34,41,42,43,44,45] and indicate the directions and dynamics of loess gully development [1,4,5,6,7,17,18,19,22,23,24,25,26,27,28,29,30,31,47,48].

5.2. Connectivity of Slope and Gully Subsystems

Between 2003 and 2020, 41 effective geomorphological events were recorded, during which, a connection between the slope and gully subsystems was observed [30,31,41]. Only during intense snowmelt and heavy rainfall was the delivery of water and sediment material from the slope to the gully system observed and it was only possible to estimate the magnitude of this supply in the cascade system in the 2013 hydrological year. During the 2013 thaw, longitudinal transport dominated in the slope catchment area, mainly in the form of concentrated runoff in the axis of the basin valley (Figure 16A). On poorly dissected slopes, the runoff concentrated at the edge of the gully or its side branches, but it mainly fed areas without drainage: potholes and piping holes [30,45]. Often, in the gully edge zone or at the field–forest boundary, the surface runoff was blocked and material was deposited (Figure 16B). Deposition also occurred at the bottom of the main gully below the head of the Celejów Dół gully and at the mouth of the Grabczyna gully.

The above observations confirmed the dominance of transverse transport in the gully system during snowmelt: from the slopes of the gully or its edges cut by piping potholes to the bottom of the gully [30,45]. During the thaw in 2023, the water supply from the slope subcatchment area accounted for 64.7% of the total water runoff from the gully catchment area. At the peak of the thaw on 11 April, this share was even higher at 80.8%. In the case of suspended sediment delivery, the proportion from slopes was even higher (80.7%) for the entire thaw period and as much as 94.3% at its peak. During the June downpours, the water supply from the slope subcatchment area to the gully catchment area amounted to over 21% and 37.1%, respectively. On the other hand, slope material, mainly originating from farmland, accounted for approx. 15% of the material removed from the catchment area. The decrease in the share of the slope subcatchment area in the supply of water and sediment material indicates the predominance of transverse transport in the slope catchment area. Soil erosion material is also often redeposited in the middle part of slope: on crop boundary, balks, dirt roads, flat areas, and along the gully edges [30,31,42]. During heavy rainfall, concentrated runoff and longitudinal transport of material dominate in gullies [30,31,41,42,45].

5.3. Sources of Sediment

During thaws, the main source of sediment runoff from the gully catchment area was arable land and fields with thuja and currant seedlings (so-called ‘sztobry’ in Polish). A secondary source was the cutting of the main gully bottom (Figure 17) and the piping potholes cutting the gully edges. During the June downpours, there was significant deposition of soil erosion material above the main gully edge, which limited its supply to the gully. The secondary gully provided the most sediment. During the downpour on 11 June, the runoff concentrated in the gully was enriched with sediment material from the bottom of the cut that was produced by multigelation of its walls during the thaw (Figure 17A). These processes also increased the suspended sediment concentration. As a result, the next runoff was underloaded, with greater erosive power, which caused the incision to deepen and caused intense head-back erosion processes (Figure 17B). In addition, these processes contributed to the cutting of sandy formations, which provided the material transported in the gully. No major supply of material from piping potholes was observed [30,45].

The delivery and timing of material delivery to the gully system varied, mainly as a result of the frequent redeposition of material in the cascade slope–gully system. Most often, the maximum suspended sediment yield preceded the peak flow, which is related to the availability of material to transport, similar to river channels [8,49,50]. In the analysed cascade, both flow parameters were observed simultaneously, which indicates a complex mechanism of material delivery and long-range transport of slope material [9,49,50]. On the other hand, a more than fourfold increase in suspended sediment transport at the lower overflow occurred as a result of increased erosion of the ‘underloaded’ stream. There was a lack (exhaustion) of available material in the slope subcatchment area due to it being carried away during the previous downpour, which could have intensively cut into the bottom of the main gully [30].

6. Conclusions

In the period of 2003–2022 in the loess gully catchment, high variability in water runoff was observed (with a max. of about 76,000 m³). In the second half of this period, a large increase in SST (to about 1200 Mg) was noted as a result of numerous downpours. In the extremely effective year of 2013, the water runoff was about 40,000 m³ and the SST was >1197 Mg, which account for 7.6% and 33.5% of the 20-year totals, respectively. During the cool half of the year, the gully discharged 16,490 m3 of water (42% of the annual total) and had an SST of about 60 Mg. During the warm half of the year, 23,742 m³ (59%) of water was discharged and the SST was about 1137 Mg (95%). June delivered a record-breaking amount of over 1100 Mg of suspended sediment (92% of the annual total SST).

The study carried out during the 2013 hydrological year showed the important role of the slope subcatchment (about 25% of the catchment area) in the formation of water runoff and as a source of sediment for the gully system. During thaws, longitudinal transport dominated on the slope and as much as 90% of the SST came from the slope subcatchment area. On the other hand, branched and forested gully systems play a significant retention role by capturing and delaying snowmelt runoff. In gully systems during snowmelt, transverse transport with piping dominates. During summer downpours, the delivery of slope material from agricultural fields accounted for approx. 15% of the material removed from the gully catchment area, which confirmed the predominance of transverse transport in the slope catchment area and longitudinal transport in the gully. During summer downpours, material from soil erosion was repeatedly redeposited on local denudation bases and, as a result, only a limited amount reached the gully. In the gully during downpours, longitudinal transport dominated and the main source of SST was the gully bottom, which was mainly cut by linear erosion processes (deep and back erosion).

A comparison of the activity of the two main arms of the gully showed that the complete cutting of the catchment area limited the supply of water to the gully system. Cutting only the lower part of the catchment area left a significant feeding area in the form of a slope catchment area, which resulted in an increase in water runoff from the gully catchment area. In addition, this effect was enhanced in the catchment as a result of the establishment of thuyas plantations above the head of the main gorge in 2013. The research carried out in the gully catchment simultaneously highlighted the role of the agricultural activity on the slope subcatchment in shaping the retention capacity of the catchment and surface water runoff. The introduction of ornamental crops, e.g., thuya, especially in the early years, did not provide sufficient soil protection during snowmelt and summer downpours. In 2013, high concentrations of runoff were observed in thuya plantations, and there was accelerated soil erosion and an increase in gully erosion rates. It appears that in addition to the implementation of extensive and sustainable farming in sensitive areas, it is advisable to introduce turf to the areas between crop rows, especially in the early years of establishing a plantation. It is also necessary to continue research into the impact of ornamental plants, whose acreage is increasing significantly, with a focus on their protective function and ability to shape surface runoff in gully catchments.