Facies analysis, depositional activity, and internal structure of sieve deposits on an active alluvial fan

Sieve lobes typically appear in gravel‐rich and matrix‐poor alluvial fans. Despite being extensively studied, the sieve‐lobe facies has been defined largely based on qualitative field observations without quantitative sedimentological analyses. Additionally, depositional activity of sieve lobes has not been monitored over extended periods (monthly to annually) and not directly associated with specific precipitation triggers. Furthermore, the internal geometry of sieve‐lobe built alluvial fans has not yet been imaged by subsurface methods. We performed a multi‐method analysis of sieve lobes in the Julian Alps (NW Slovenia) on an alpine alluvial fan composed of carbonate gravels. We performed a detailed textural and structural sedimentological analysis of 11 recent sieve lobes differing in size and age. A three‐year aerial survey of the alluvial fan surface with a small unmanned aircraft and photogrammetric modelling was used to detect active sieve‐lobe evolution. Detected sieve‐lobe formation events and volumetric surface changes were paired with triggering precipitation events. Ground‐penetrating radar (GPR) profiling depicted the geometry of the sieve‐lobe built alluvial fan. The sieve‐lobe facies consists of over 80% poorly sorted, open‐framework gravels and less than 2% mud. Lobes exhibit downward coarsening and increase in clast mean size. These textural and structural characteristics are present in all sieve lobes regardless of their age and size. Sieve lobes form with a sub‐annual frequency, usually following 24 h rainfall events exceeding 50 mm. Over 1000 m3 of sediment was deposited during these events. The GPR profiles confirm that the studied alluvial fan is formed predominantly by stacked sieve lobes. Quantitative sedimentary analysis of sieve lobes, monitoring of their recent evolution, and depiction of their subsurface geometry—demonstrated in this study—reinforce the challenged concept that sieve lobes are one of the main building blocks of alluvial fans. This work also demonstrates that, under specific conditions, sieving may become the dominant alluvial fan‐forming process.

of the catchment area by a suite of different processes, ranging from dilute flows (stream and sheet flows) to dense ones (gravity flows).
A particular sedimentary facies characteristic of alluvial fans is the sieve-lobe facies (Hooke, 1967). Sieve-lobe deposits are an end product of the sieve deposition process and occur primarily on alluvial fans, however, they can also occur in other depositional environments, such as proglacial outwash fans and perennial streams (Milana, 2010). The requirement for sieve deposition is a significant amount of coarse-grained sediment devoid of fine-grained material transported as a bedload on an inclined surface. Such sediment is transported with discharges moderate enough to allow infiltration into permeable and unsaturated ground (Hooke, 1967;Milana, 2010). The resulting sieve-lobe deposit consists of matrix-poor, clast-supported, moderately sorted open-framework particles ranging from sands to very coarse-grained gravels (Bull, 1977;Hugenholtz, 2011;Morgan & Craddock, 2017). Sieve-lobe growth is a gradual process in which a single lobe develops from multiple stacked sublobes, deposited on top of each other during one depositional event (sensu Milana, 2010), which typically occurs during short and intense precipitation events (Milana, 2010;Morgan & Craddock, 2017). Sieve-lobe formation was also documented in permanent streams with stable discharge, where the sediment bed became more permeable (Milana, 2010). A number of intertwined sieve lobes form a sieve-lobe complex and lobes may represent major or even sole building blocks of a fan sequence that has a sigmoidal shaped surface from apex to toe (Hugenholtz, 2011;Milana, 2010;Morgan & Craddock, 2017;Nemec & Postma, 1995).
Sieve-lobe deposits and the sieve deposition process were first studied in a laboratory experiment (Hooke, 1967), which later served as an analogue for interpreting naturally occurring deposits on fans and cones. However, the sieve-lobe paradigm was subsequently opposed by several studies (Blair & McPherson, 1992, 1994, 2009, which proposed instead that open-framework sieve-lobe deposits are the end product of winnowing of primary matrix-rich debris flows by water. Despite the documentation of Holocene sieve deposits on Crete (Nemec & Postma, 1993, in a study later challenged by Blair and McPherson (1995), the sieve depositional process remained a disputed topic. Only recently has Hooke's original sieve-lobe paradigm been confirmed by documenting the deposition of multiple recent sieve lobes via the sieve deposition process (Milana, 2010), which corroborated the ideas of Hooke's (1967) original laboratory experiment. Modern studies document coarse-grained (pebble to cobble) (Colombo & Rivero, 2017;G omez-Villar & García-Ruiz, 2000;Milana, 2010;Morgan & Craddock, 2017) and sandgrained (Hugenholtz, 2011) lobes deposited via sieve deposition processes after intense precipitation events. In addition, morphological features on Mars have also been interpreted as deposited by sieve depositional processes (Brož et al., 2019).
Although the sieve-lobe depositional process paradigm has been widely accepted recently and sieve lobes have been documented in nature, quantitative sediment analysis, monitoring of their evolution over a longer period, and analysis of the precipitation conditions that form this sedimentary feature have not yet been extensively published in the scientific literature. First, the sieve-lobe facies has been described only qualitatively, without detailed quantitative analyses of the sedimentary structure and texture of individual sieve lobes. In a stratigraphic sequence of alluvial fans, a sieve-lobe deposit can easily be mistaken for another open-framework gravel facies formed by other processes (Lunt & Bridge, 2007;Zhang et al., 2021). Second, the depositional conditions (i.e. the triggering conditions) for sieve-lobe deposition-such as rainfall intensity, rainfall quantity, and amount of transported sediment-have rarely been studied. These depositional conditions could be investigated by detailed monitoring of surface changes over a longer period (monthly to annually). Milana (2010) attributed sieve deposition predominantly to storm events during the rainy season, without defining the rainfall quantities. Hugenholtz (2011) described the formation of sand-grained sieve lobes during rapid snowmelt. Morgan and Craddock (2017) linked the deposition of recent sieve lobes to recorded triggering precipitation events of high intensity and low frequency. None of the studies observed active sieve-lobe deposition over an extended (i.e. annual) period and directly associated the quantity of transported sediment to the intensity of a particular precipitation event. In the case of episodically activated sieve lobes, the minimum amount of precipitation required to trigger the sieve-lobe transport and deposition is unknown. The lack of monitoring of sieve-lobe formation over a longer period raises the question of whether the sieve-lobe formation is an infrequent deposition event driven by extreme and infrequent precipitation, or whether it occurs more frequently under regular meteorological conditions. Third, it is proposed that alluvial fans can be predominantly or even entirely built of multiple stacked sieve lobes (Milana, 2010). However, the subsurface geometry of such a fan has not yet been documented.
In this study, we investigated the active deposition of sieve lobes on a gravel-rich alpine alluvial fan by performing a detailed sedimentary, topographic, and geophysical analysis. Our objective was to answer, at least in part, the open questions listed in the previous sections by: (i) providing a detailed quantitative facies analysis based on the sedimentary structure and texture of recent sieve lobes differing in size and age; (ii) monitoring the active sieve-lobe depositional dynamics; (iii) linking their depositional activity to triggering precipitation events; and (iv) depicting the subsurface geometry of the sievelobe built alluvial fan.

| Study site
The study was performed on the Suhi vrh alluvial fan in the Planica Valley in the Julian Alps in NW Slovenia (Figures 1a and b). The valley slopes are predominantly composed of Upper Triassic carbonates (Gale et al., 2015) and the valley floor is covered by various Quaternary sedimentary bodies, of which the Holocene alluvial fans are the most numerous (Novak et al., 2018). The alluvial fans have ephemeral streams in which the sediment (mainly gravel) is very actively deposited by water flows and sporadic debris floods (Novak et al., 2018(Novak et al., , 2020. The studied Suhi vrh alluvial fan is located on the eastern slope of the valley below Suhi vrh Mountain (Figures 1 and 2a (ARSO, 2006(ARSO, , 2009(ARSO, , 2021a(ARSO, , 2021bFigure 1c). In the study area, there are an average of six to eight precipitation events per year where more than 50 mm of precipitation occurs within a 24 h period and the snow cover persists from late November up to early May. The combination of active gravel-rich alluvial fans lacking fine-grained sediment and detailed meteorological records makes this study site an ideal location for studying active sieve-lobe deposition.

| Sampling strategy and sedimentary analysis
We catalogued the sedimentary structures and general characteristics (size and morphology) of the sieve lobes in the field. Sieve lobes differ in their relative age and stage of development (i.e. fully developed lobes and sublobes). During fieldwork, several sublobes were found that have not formed into fully developed sieve lobes and therefore represent an initial development stage of incomplete sieve lobes (sensu Milana, 2010). Eleven individual sieve lobes and sublobes evenly distributed on the active surface of the alluvial fan were distinguished and sampled (Figures 2a, b, c). The relative age of each lobe was determined by the intensity of grey lichen coating of clast (Figures 2d,3a and b), with older lobes having a more intense and darker coating. All catalogued sublobes lacked lichen coating, and no sublobes with coating were found ( Figure 3c). We assume that all sublobes are relatively young and were freshly deposited just prior to fieldwork. Using these criteria, we sampled three types of sieve lobes.
We sampled four larger and relatively old fully developed sieve lobes (designated SO 1, SO 2, SO 3, and SO 4), four larger and relatively young fully developed sieve lobes (designated SY 5, SY 6, and SY 7), and three relatively young sublobes (designated SL 9, SL 10, and SL 11). For each sampled sieve lobe, the samples were collected from proximal and distal parts of the lobe and, where possible, from the middle part ( Figure 2d). Only lobes that showed no evidence of reworking (i.e. partial erosion or coverage) by subsequent sieve deposition processes were sampled. Depending on the lobe size, each sample contained between 7 and 30 kg of dry weight; a total of 618 kg of sediment was analysed. Granulometric analysis was performed using a Haver and Boecker EML 200 sieve shaker, following previous studies (Dufresne et al., 2016). Samples were oven dried for 48 h at a temperature of 40 C and dry sieved using standard sieve pans with diameters ranging from 32 mm to 63 μm. Clasts larger than 64 mm were measured manually, and each size group was dry weighted. Granulometric analysis of the dry-weighted sediment was performed using Gradistat software (Blott & Pye, 2001) and following grain size and texture classification of coarse sediment particles (Blair & McPherson, 1999;Blott & Pye, 2012). Particles finer than 63 μm were quartered to obtain 1 g of representative sediment and then measured using the Fritch Analysette 22-28 laser granulometer with dynamic image analysis. Each sample was measured three times and an average was calculated from the three measurements. The calculated average of 1 g was extrapolated for the remaining amount of particles below 63 μm and extrapolated to the total amount of sediment. Grain shape, roundness, and fabric were determined visually according to Illenberger (1991). A slope map was created to determine the surface inclination of the area where sieve-lobe deposition occurs. We used a digital elevation model (DEM) derived from airborne laser scanning (ALS) with a resolution of 0.5 m, and classified the inclination in five classes (0-19 , 20-29 , 30-44 , 45-54 , and >55 ). Data from ALS was obtained from the publicly available ALS dataset of Slovenia (ARSO, 2021c). The inclination map was created with the QGIS program (QGIS, 2021a) using the QGIS Raster Terrain Analysis plugin (QGIS, 2022).  (Figure 2b). GCP coordinates were obtained with a rapid static GNSS survey, which was periodically repeated to control the long-term stability of the GCP network. Surveys using the DJI Phantom 4 RTK UAV also used a post-processed kinematic method to directly georeference flight paths and imagery, but no significant difference in survey precision was observed compared to other UAVs used in the study. UAV surveys were repeated several times per year following seasonal changes and major rainfall events. In this study, we analyse data from 10 surveys covering the period from April 2019 to Surface changes were linked to recorded precipitation events from the Rateče Meteorological Station (ARSO, 2021d) by selecting the most intense rainfall events that could potentially trigger sediment transport. Following the studies of Guzzetti et al. (2007), ARSO (2009ARSO ( , 2021b, and Novak et al. (2020), the threshold for a potentially triggering precipitation event was set at 50 mm of rainfall in 24 h. Intense 48 h rainfall events were also considered as potential triggers. Additional meteorological factors such as snowfall and snow cover were also considered as factors that could reduce or increase sediment transport.

| Ground-penetrating radar
The ground-penetrating radar (GPR) technique was applied to understand the subsurface geometry of Suhi vrh fan. This geophysical method has been used successfully for alpine alluvial fans (Franke et al., 2015;Mills & Speech, 1997), as well as for coarse-grained sedimentary bodies (talus slopes) built of carbonate gravels (Sass & Krautblatter, 2007).
The Mala ProEx GPR common-offset survey was used with a 50 MHz unshielded rough terrain antenna (RTA). According to research by Sass and Krautblatter (2007), the 50 MHz antenna offers the best compromise between penetration depth and resolution of subsurface structures of sedimentary bodies composed of carbonate gravels. The collected raw GPR data was processed using RadExplorer software by DECO Geophysical (2005). The processing workflow included editing, time-zero correction, removal of DC, and topography correction. The DEM from the 10/6/2021 UAV survey (closest in time to the GPR measurements) was used as an elevation reference.
Three GPR profiles were created in the active area of the alluvial fan ( Figure 2b). The profiles were labelled as GPR-1, GPR-2, and GPR-3. Longitudinal profile GPR-1 is 187 m long and extends from the fan's toe up to the fan's apex orientated parallel to the direction of sediment transport. GPR-2 and GPR-3 are 84 and 91 m long, respectively, and orientated perpendicular to GPR-1 and thus the active area of the Suhi vrh fan (Figure 2). The GPR profiles cover only the active area of the fan as the inactive part is too densely vegetated and difficult to access. The radargrams were interpreted according to nomenclature of Sangree and Widmier (1979  3.1.2 | Sediment texture: Grain size, internal grain distribution, grain shape, roundness, and fabric The majority of the samples belong to the textural group of gravel. Only four samples (the proximal sample of SO 3, the proximal and the middle samples of SL 9, and the proximal sample of SL 11) classify as sandy gravel (Figures 5-7). In all samples, the percentage of fines was very low, with less than 2.5% of mud (<63 μm) and samples contained between 0.005 and 0.295% of clay (<2 μm). The distribution curves are unimodal in all samples, except the proximal sample of SO 1. In the field, the lobes appear to have well-sorted sediment; however, the gravel fraction ranges from granules to coarse cobbles (Figure 8   Appendix). In six lobes it was also possible to extract samples from the middle part of the lobe (Figures 4-6). In general, the samples from the middle parts of the lobes have a larger gravel content than the proximal parts and a lower gravel content than the distal parts. For most sieve lobes, there is an increase of one size class from the proximal to the distal part of individual lobes (Appendix), following the gravel size classification of Blair and McPherson (1999).
The general trend of increasing downward coarsening of the sediment differs between younger, older, and sublobes (Table 1)

| Surface changes and detected triggering events
Our aerial surveys during the period from 27/4/2019 to 17/11/2021 yielded 10 DEMs, from which we calculated 9 DoDs that reveal various surface changes on the Suhi vrh alluvial fan (Figures 9-11). The surveying periods, volumetric changes (erosion and deposition of sediment), dates of the triggering rainfall events, rainfall amount, detected

| DISCUSSION
The present study provides a detailed and quantitative facies analysis of the sieve deposits observed in the natural environment of the Suhi vrh alluvial fan. Their depositional activity is directly related to the amount and intensity of triggering precipitation events, while subsurface results show that they also represent major building blocks of the studied gravel-rich alluvial fan. To place the results in a broader context of gravel-rich alluvial fans formed by sieve lobes, we compare the results of this study to previous sieve-lobe studies (Hooke, 1967;Milana, 2010;Nemec & Postma, 1993 in the following sections.

| Sieve-lobe facies
The quantitative results of the sedimentary facies analysis in this study corroborates previous qualitative analyses conducted in the field and in the laboratory (Hooke, 1967;Milana, 2010;Nemec & Postma, 1993 (Lunt & Bridge, 2007;Zhang et al., 2021). However, the difference in mean clast size is only in the range of one clast size class, which could be difficult to detect during mapping in outcrop scale (especially for smaller grain sizes that are difficult to distinguish in the field). Therefore, granulometric analysis is required to determine whether a layer is a sieve lobe. A comparable proximal-to-distal grain size variation was observed in grain flow deposits where the front of the lobe has coarser clasts than the back and the roundness of the clasts may be similar to sieve lobes (Van Steijn, 2011;Van Steijn et al., 1995). However, compared to sieve lobes, grain flows occur on considerably steeper surfaces (>33 ) of colluvial deposits (Bertran et al., 1997). Such grain flows should not be confused with the fluidized grain flows observed by Milana (2010), which form at the sieve-lobe front and represent sublobes formed during sieve-lobe growth.
The downward gradation of the individual lobes is almost uniformly pronounced in all the sieve lobes we have examined, T A B L E 1 Average amounts of gravel, sand, mud, and clay particles in old, young, and sublobes  discrepancy in percentages can be attributed to the sampling strategy. The contacts between the sieve lobes are poorly defined, and it is possible that some of the samples were also partially extracted from the lower deposited lobe.
Older sieve lobes tend to have a higher percentage of smallgrained particles (sand and mud) compared to younger ones. This could be attributed to physical weathering of clasts in older deposits, which produces a larger quantity of sand particles. The Conzen dolomite is highly fractured, and physical weathering of clasts following deposition could result in production of finer particles. In addition, the greater amount of smaller particles in older sieve lobes could derive from secondary sedimentation processes following sieve deposition, such as aeolian dust or pedogenesis (Hooke, 1993;Milana, 2010).
Measurements of the higher proportion of small-grained particles in older sieve lobes do not confirm the mechanism of winnowing of primary matrix-rich debris flows by waterflow proposed by Blair and McPherson (1992). The results of our study clearly confirm an increase in fine-grained material with age in the sieve lobes. Despite bedload transport, which should orient the a-axis of the clasts transverse to the transport direction, the clasts did not appear orientated.
However, the clasts are highly spherical (Figure 8), making orientation challenging to measure.
In some lobes, we detected crude stratification based on sharp vertical changes in mean grain sizes (Figure 3d). Similar observations were made by Nemec and Postma (1993) during studies in Crete. We suggest that this phenomenon is probably due to the fact that a single sieve lobe is gradually built up from several sublobes stacked on top of each other, which are hierarchically one level lower than a complete sieve lobe. Milana (2010)  and wet deposits, whereas on dry deposits, morphology was smoothed and difficult to discern. We documented only relatively freshly deposited sublobes and no old ones. This suggests that the sampled sublobes were the initial building blocks of undeveloped sieve lobes that did not continue to form due to a variety of reasons, such as lowering discharge or lack of sediment.

| Detection of surface changes and creation of sieve lobes
Three years of surface monitoring allowed us to detect and quantify surface changes and sieve-lobe formation. The changes are directly related to a specific triggering rainfall event with known quantities.   (Figure 1c; ARSO, 2009ARSO, , 2021c. This strongly suggests that sieve-lobe formation may occur frequently and several times per year. During substantial events there was an increase in sediment volume by several hundreds of cubic metres, which was transported as bedload and deposited predominantly as sieve lobes. Such changes occurred annually during rainfall events, with more than 60 mm of rainfall in 24 h (Table 2). A major amount of positive versus negative changes in sediment volume indicate that surface changes are caused not only by redeposition of pre-existing sediment, but also by sediment influx from the catchment.
Moderate changes were caused by either 24 or 48 h events with more than 50 mm of precipitation ( Table 2). The magnitude of precipitation events that caused moderate changes was similar to that of major changes in some cases. However, these precipitation events generally transitioned from rain to snowfall, and therefore failed to cause substantial sediment transport. An exception is the triggering event of 4/11/2021, which did not cause substantial deposition, despite the large amount of precipitation (67.5 mm). We assume that rainfall did not occur in a short period of a few hours but extended throughout the entire day. This facilitated concurrent water infiltration so that little surface runoff occurred. During moderate changes, sediment volume increased only up to a few tens of cubic metres, indicating that redeposition of pre-existing sediment was the predominant mechanism, with only little or no sediment influx from the catchment.
Minor changes happened during less intense triggering rainfall events which did not exceed 50 mm of rainfall in 24 h. The only exception was the 3/5/2021 event, which had a precipitation amount comparable to a substantial event. However, this event later transitioned to snowfall, which could not cause significant sediment movement.
In our study site, sieve lobes were deposited either at the channel mouth or inside the distributary channel, which corroborates previous observations of sieve-lobe generation (Hooke, 1967;Milana, 2010;Nemec & Postma, 1995 (Figures 9a and c, 11a), where the prevailing depositional mechanism was bedload accumulation in the form of sieve-lobe complexes due to total decay of stream shear stress, a process already documented by Milana During two substantial events (Figures 9a and c), the middle and lower parts of the distributary channel were filled with sieve lobes, which lead to avulsion. Alternatively, avulsion might also occur due to channel plugging with sediment of previous moderate and minor events (cf. de Haas et al., 2018). During such events the sediment was deposited inside the distributary channel, predominantly in the form of sieve lobes (Figures 9b, 10b and c, 11c). These lobes, confined in a very narrow channel, could cause a blockage, and avulse the subsequent depositional events. The documented sieve lobes did not damage vegetation, indicating that the transport energy of sieve lobes is low, and the sediment is transported as bedload. The same findings were described by Milana (2010). A low transport energy is also indicated by low surface inclination (20 or less), and the stacked and intertwined lobes exhibiting no erosional contact.
With minor and moderate surface changes, the sieve lobes grew on the fan's surface in the middle or even distal parts with no direct connection to the fan's catchment area. Similar phenomena were observed on the alluvial fans of Crete (Nemec & Postma, 1993). Nemec and Postma (1993) Milana (2010) suggested that alluvial fans may be predominantly or entirely built of sieve-lobe deposits, which is corroborated by the research of Argentine fans where the fan surfaces are almost entirely covered by sieve lobes. The study of Hooke (1967) shows that fans can be simultaneously bult either by sieve lobes or other sedimentary deposits, whereas in the findings of Nemec and Postma (1993) sieve lobes account for only a minor percentage of alluvial fan composition. These studies were based on surface analyses of alluvial fans with little or no subsurface information. Our results from the GPR profiles (Figures 12 and 13) confirm previous estimates that entire fans may be predominantly or entirely built of sieve lobes. The maximum depth reached by the GPR signal was up to 15 m, below which the signal was most likely attenuated due to underground water, which is usually present in highly porous sediments. Therefore, the bottom surface of the fan and the deeper sediment geometry were not reached. Consequently, the total thickness of the Suhi vrh alluvial fan is unknown. However, the radargrams show reflection patterns that correspond to the surface topography of exposed recent sieve lobes and are interpreted as such. Reflections in profile GPR 1 clearly show lobate cross-bedding, interpreted as inclined stacked sedimentary layers dipping parallel to the recent surface. Milana (2010) described that the slope of the surface of an alluvial fan built by sieve-lobe deposition exhibits a sigmoidal shape that derives from rapid extraction of water from the transported sediment. The recent surface of the Suhi vrh alluvial fan has a very pronounced sigmoidal shape, which is also visible in the subsurface data. The proximal part exhibits less pronounced sigmoidal shapes due to the dominance of sediment transport over deposition. Radargrams 2 and 3 are interpreted as an undulating F I G U R E 1 3 Radargrams 2 and 3 exhibiting stratified hummocky and discontinuous reflections of up to 10 m long (examples marked with red lines). [Color figure can be viewed at wileyonlinelibrary.com] morphology of intertwined sieve lobes. Such a morphology is present on the recent surface ( Figure 4) and the reflectors have the same shape below the surface.

| Subsurface geometry of a sieve-lobe built fan
The shape and orientation of the reflectors resemble the parallel and perpendicular cross-sections of the individual sieve lobes that occur on the surface of the Suhi vrh alluvial fan. The GPR data indicate that the upper 10 to 15 m of the studied alluvial fan consist predominantly of stacked sieve lobes, supporting Milana's idea of alluvial fans built entirely of sieve lobes (Milana, 2010).

| CONCLUSION
This study provides a quantification of sieve-lobe sedimentary facies, depositional activity, and triggering precipitation conditions The depositional process occurs as bedload transport, with coarsegrained sediment deposited predominantly as sieve-lobe deposits.
Sieve lobes can be deposited inside or outside the channel and occur at slope angles lower than 30 . The most severe triggering rainfall events that resulted in substantial surface changes and deposition of several sieve lobes had rainfall amounts greater than 60 mm within a 24 h period. Such precipitation events are very common for the study site and therefore sieve-lobe deposition occurs regularly with a sub-annual frequency under regular meteorological conditions. The most substantial event resulted in the deposition of more than 1000 m 3 of sediment. The internal architecture of the studied alluvial fan derived from the GPR data resembles the surface deposits and confirms previous research that some coarse-grained alluvial fans can be almost entirely built by sieve lobes.