Flood-induced ground effects and flood-water dynamics for hydro-geomorphic hazard assessment: the 21–22 October 2019 extreme flood along the lower Orba River (Alessandria, NW Italy)

ABSTRACT

The knowledge of flood-induced ground effects and flood-water dynamics is a crucial issue for hydro-geomorphic hazards assessment and mapping, and thus for river management and land use planning. This paper and the related 1:15,000 map illustrate the lower Orba River (NW Italy) and its adjacent floodplain geomorphic response to the 21–22 October 2019 extreme flood. This was estimated to be a 500-years flood and caused severe damage to cultivated fields, structures, and infrastructures. The research is based on extensive post-flood field surveys, ante- and post-flood GNSS surveys, and aerial photographs interpretation. Intense in-channel sediment mobilization, bank retreats, and channelization structures collapses were surveyed. Furthermore, alluvial gullies, overbank deposits, crevasse splays, and surficial-erosion evidences were mapped over the floodplain, along with the flooded area (17.65 km²) and the flood-water features. A specific legend developed for flood-related and anthropogenic elements mapping in a typical lowland agricultural landscape with regulated rivers is proposed.

KEYWORDS:

• Floods
• geomorphic effectiveness
• geomorphological mapping
• alluvial gully
• Orba River
• north-western Italy

1. Introduction

River-related floods occur frequently in large areas of the world (Bathrellos et al., 2018), and are the most prevalent natural hazard and among the most damaging events in Europe (EEA, 2010, 2016; Gaume et al., 2009; Kundzewicz et al., 2013, 2018; Llasat et al., 2010). Long lists of past inundations and relative damage testify the relevant flood-proneness of Italy (Brandolini et al., 2020; Faccini et al., 2015, 2016; Luino et al., 2002; Mandarino et al., 2019a, 2020a; Pavese et al., 1992; Roccati et al., 2020; Salvati et al., 2012; Zani, 2000), where there were 581 deaths and 171,764 evacuated and homeless due to floods in the 50-years period 1969–2018 (CNR-Polaris, 2019), and about thirty floods and a hundred victims from 2010 to 2019 (Paliaga et al., 2020).

Floods are natural and common phenomena associated with river dynamics, which become ‘natural disasters’ when impact on humans and their activities (Benito & Hudson, 2010; Talbot et al., 2018). Man’s use of catchments and anthropogenic interventions along fluvial stems have impacted worldwide upon the severity and consequences of floods (Casale et al., 1998; Gregory, 2006), often increasing river-related risks against a generalized riverine-area occupation and riverbed channelization (Cencetti et al., 2017; Luino, 1999; Luino et al., 2012; Mandarino et al., 2019b; Serrano-Notivoli et al., 2017; Wyzga, 1996). Floods can cause loss of lives and severe damage to structures, infrastructures, and human activities, and may result in catastrophic social, economic, and environmental consequences (Merz et al., 2010, EEA, 2010). The knowledge of floods in terms of flood-induced ground effects (FIGEs), i.e. riverbed and floodplain geomorphic response and anthropogenic-elements damage, and flood-water dynamics (FWDs), such as inundated area, flood-flow direction, and flood-water levels, is a crucial issue for hydro-geomorphic hazards assessment and mapping, and thus for the implementation of measures for river-related hazards and risks reduction (Hooke, 2015).

Numerous researches focusing on flooded area mapping, generally based on models or remotely-sensed data (De Musso et al., 2018; Guerriero et al., 2018, 2020; Romanescu et al., 2017; Segura-Beltrán et al., 2016), and on riverbed geomorphic response analysis and mapping (Fuller, 2008; Heritage et al., 2004; Hooke, 2016; Morche et al., 2007; Rinaldi et al., 2016; Thompson & Croke, 2013; Yousefi et al., 2018), were conducted worldwide to investigate high-magnitude floods dynamics. Furthermore, several studies analyzed ground effects triggered by intense rainfalls and floods specifically in mountain and hilly areas (Bartelletti et al., 2017; Borrelli et al., 2015; Cevasco et al., 2012, 2015, 2017; Pepe et al., 2019a, 2019b; Rago et al., 2017; Santo et al., 2017; Tessitore et al., 2011). In contrast, relatively little research was carried out focusing on detailed field-based flood-induced geomorphic effects mapping in lowland areas (Magilligan et al., 2015).

In this frame, this paper and the related map illustrate the lower Orba River (LOR) (NW Italy) and its adjacent floodplain geomorphic response to the 21–22 October 2019 extreme flood. This was estimated to be a 500-years flood and originated from very intense rainfall that hit large portions of the Orba River catchment. This research is based on extensive post-flood geomorphological field surveys, GNSS surveys and aerial photographs interpretation. Specific legend entries related to floods and anthropogenic elements are proposed, and this case study represents a significant example of what can be mapped after a high-magnitude flood in lowland mainly-agricultural landscapes of European/Mediterranean catchments presenting regulated rivers. In this paper the FIGEs and the FWDs are overall referred to as flood-related elements (FREs).

2. Study area

The Orba River origins West of Genova (NW Italy), at the closest part of the Ligurian-Po divide to the Ligurian Coast (i.e. only 5 km of distance), and flows Northward from the Ligurian Alps to the Bormida River, south of Alessandria (Figure 1(a)).

Figure 1. (a) Location of the study reach. See the main map for data information. (b) Representative stretches of the Orba River floodplain reach planform features. The gray dotted lines represent artificial levees. In (b1) the artificial-levee-delimited flood corridor (LFC), that is the floodplain comprised between the bank and the artificial levee, is indicated. The left LFC (1), the active channel (2), and the right LFC (3) are shown along the AA’ segment.

The main fluvial stem is about 73 km long and its catchment spreads over 797 km². The catchment elevation ranges from 1287 to 88 m a.s.l., at the Mt. Beigua and the outlet, respectively. The all major tributaries are from the right slope of the catchment and are, in order of their occurrence, the Stura, Piota, Albedosa, and Lemme streams. The Orba River has a mean daily discharge of about 15 m³/s, at the Casalcermelli gauging station (Figure 1(a)) (Regione Piemonte, 2007).

In the southern part of the catchment the main outcropping lithotypes are serpentinites, metagabbros, metabasites, metasediments and mantle peridotites belonging to the Voltri Massif and the Sestri-Voltaggio Zone (Allasinaz et al., 1971; Capponi et al., 2009a; Ferraris et al., 2012; Molli et al., 2010). The middle part presents conglomerates, sandstones and marls of the Eocene-upper Miocene sedimentary succession of the episutural Tertiary Piermont Basin (Capponi et al., 2009b; Festa et al., 2015). Moreover, a narrow belt of Pliocene silty and clayey marls is situated between reliefs and the Quaternary terraced floodplain (Boni & Casnedi, 1970; Piana et al., 2017). This latter is located in the northern sector of the catchment and is part of the wider Alessandria-Tortona floodplain (Cortemiglia, 1998). The Orba River floodplain is mainly composed of coarse sediments (Arpa Irace et al., 2009; Piemonte, 2019a), and presents a wide series of fluvial terraces partly covered by aeolian deposits (Biancotti & Cortemiglia, 1981) and delimited by scarps up to a few tens of meters high.

Concerning land use, most of the upper part of the Orba River catchment is covered by forest, while the hilly and floodplain sectors present mainly agricultural areas.

The climate is generally characterized by hot and dry summers and cold and wet winters, with rainfalls mainly concentrating in autumn (Mandarino, 2018). The catchment mean annual rainfall is about 950 mm (ADBPo, 2001) and ranges approximately from 600 mm to over 1700mm, at the Bormida River confluence and close to the main Ligurian-Po divide, respectively (Mandarino, 2018).

Over the upper part of the catchment, the closeness to the sea and the relief topographic effect result in high annual rainfall values and frequent intense rainfall events associated with an atmospheric circulation prevailing over the Ligurian Sea that conveys moist air masses from the sea to the reliefs around Genova (Acquaotta et al., 2018; Sacchini et al., 2012, 2016). This makes the Orba River catchment prone to severe floods.

During the twentieth century several Orba River high-magnitude floods were documented (Mandarino, 2018 and references therein). The most severe events occurred in 1935, caused by a secondary-dam collapse (Bonaria, 2013), and 1977 (Govi, 1978; Tropeano, 1989). Consecutive high-magnitude floods even associated with very intense rainfalls recently occurred (Table 1), resulting in severe geomorphic effects, large flooding, and relevant damage (ARPA Piemonte, 2002, 2011, 2014a, 2014b, 2014c; Luino & Turconi, 2017).

Table 1. Maximum water level and related discharge measured and estimated, respectively, at the Casalcermelli gauging station for the most recent high-magnitude floods (see Figure 1 for location).

Flood peak	Water level (m)	Discharge (m^3/s)
26/11/2002	4.36	–
05/11/2011	6.8	2200
26/12/2013	4.73	1300
13/10/2014	7.07	–
15/11/2014	6.45	2000

This paper focuses on the LOR, that is the Orba River floodplain reach, from Pratalborato to the Bormida River confluence and its adjacent floodplain (Figure 1(a)). The LOR is 23.8 km long, unconfined, and presents a quite sinuous, single-thread channel deeply entrenched into the floodplain (Figure 1(b)). The riverbed is on average 80–125 m width, generally shows riffles and pools, and its predominant sediments consist of cobbles, pebbles, fine gravels and sands. Banks are overall high, steep, non-cohesive or composite (Rinaldi et al., 2015), and largely stabilized through anthropogenic structures. Nonetheless, a widespread and growing bank instability has been documented over the last 20 years particularly in the lower part of the study reach (Mandarino, 2018). The floodplain adjacent to the active channel can be generally classified as recent terrace and only narrow strips of modern floodplain can be noticed (Rinaldi et al., 2015). Artificial levees border the riverbed along most of the study reach. Over the last century, and particularly after the 1950s, the LOR experienced severe in-channel alterations mainly consisting of sediment extraction and channelization (Mandarino, 2018). Cultivated fields occupy most of the floodplain adjacent to the riverbed and they generally spread up to the bank edge (BE).

3. The 21–22 October 2019 hydro-meteorological event

From 18 to 24 October 2019 intense and prolonged rainfalls affected large parts of the north-western Italy. The Orba River catchment was mainly hit from 18 to 21 October (ARPA Liguria, 2019; ARPA Piemonte, 2019b), when significant rainfall episodes repetitively affected both the uplands and the lowland areas. The Orba floodplain was largely flooded by the minor hydrographic network in the night and early morning of 21 October. However, the most critical phase of the event occurred in the afternoon and evening of the same day, and it was associated with a thunderstorm cell formed on the Ligurian Sea and then extended northward particularly on the southern part of Piemonte, where it remained stationary for some 12 h (ARPA Piemonte, 2019b). This dynamic resulted in exceptional rainfalls in terms of both cumulative values and intensity (Table 2, Figure 2(a)).

Figure 2. (a) 21 October rainfall values measured by the Rossiglione and Gavi rain gauges (time is referred to CEST). They are the rain gauges that registered the maximum daily rainfall during the event (Table 2). (b) Hydrometric level recorded by the Casalcermelli gauging station (see Figure 1 for location; time is referred to CEST).

Table 2. Rainfall values measured within the Orba River catchment from 18 to 23 October 2019.

	Cumulative daily rainfall (mm)						Cumulative rainfall (mm)		Max mm/h from 18 to 23
Rain gauge station	18	19	20	21	22	23	18–23	14 of 21–2 of 22 (12 h)	Value	Hour, day
1) Piampaludo (O*)	37.4	126	39.4	144.6	82.2	4.4	434	144.4	40.2	1–2, 22
2) Urbe (O**)	89.2	108.8	25.6	181.2	172.4	4.6	581.8	217.2	72.4	1–2, 22
3) Bric Berton (O*)	10.4	33.8	47.4	45.8	3	4.6	145	15.6	23.2	18–19, 20
4) Ovada (O*)	8.4	26.8	50.8	150.2	6	2.2	244.4	90.6	35.4	20–21, 21
5) Turchino (S**)	103.6	85.6	5.8	51.8	43	0.6	290.4	38	29.8	2–3, 22
6) Prai (S**)	42	48	54	248.8	237.4	2.2	632.4	370.2	100.6	1–2, 22
7) Campo Ligure (S**)	46.8	58.6	51	333	217.4	1	707.8	450.6	97.8	0–1, 22
8) Rossiglione (S*)	21.2	33.6	66.8	336.8	95.6	2.4	556.4	295.2	78	17–18, 21
9) Capanne di Marcarolo (P*)	54.6	56.2	26.8	110.2	52	1.2	301	107.6	23.4	15–16, 21
10) Lavagnina (P*)	21	30.4	45.8	327.4	94.8	1.4	520.8	369	61.4	0–1, 22
11) Fraconalto (L*)	49.8	79.8	33.8	185.4	35.4	0.6	384.8	137	53.2	14–15, 21
12) Bric Castellaro (L*)	13.6	15.8	32.2	341.6	36	0	439.2	329.2	50	16–17, 21
13) Gavi (L*)	10.6	19.4	55.8	452.8	33.2	0.4	572.2	428.4	70.2	18–19, 21
14) Basaluzzo (L*)	7.6	10	28.6	155.4	3.4	0.4	205.4	61	29.2	4–5, 21

The letter in brackets indicates the rain gauge station location in terms of sub-catchment. (O) Orba – upstream of the Stura Stream confluence; (S) Stura; (P) Piota; (L) Lemme. The average and range referred to the four maximum daily rainfall values, registered on 21 October in each sub-catchment, were 325.4 and 271.6 mm, respectively; these values changed in 372.3 and 125.4 mm, respectively, without considering the Orba sub-catchment. Raw data from the Regional Agency for Environmental Protection of Piemonte (*) and Liguria (**). Time is referred to CEST. See Figure 1 for rain gauge station location.

As a response, the Orba River experienced a very high-magnitude flood. At the Casalcermelli gauging station its water level rose up to 7.5 m during the evening of 21 October, underlining a lag time of the order of a few hours (Figure 2(b)). This value was estimated to correspond to a peak discharge of 2700–2800 m³/s, that is approximately a 500-year flood according to the Flood Risk Management Plan (ARPA Piemonte, 2019b). However, it could be underestimated due to the flood-water overflowing the artificial-levee-delimited flood corridor (LFC) (Figure 1), upstream of the measurement cross-section. The event magnitude is higher than the most recent floods anyway (Table 1). Considering the rain gauge measurements, a substantial discharge contribute derived from the Orba River main tributaries (Table 2). The most affected catchment areas were the Lemme, Albedosa, Piota, and Stura valleys.

At the catchment scale, this extreme hydrological event resulted in widespread slope-instability and river-related geomorphic processes, along with lowland areas flooding (Mandarino et al., 2020b). All of this caused one casualty and severe damage to agricultural activities, structures, and infrastructures.

4. Materials and methods

An extensive field survey campaign was undertaken along the LOR immediately after the 21–22 October 2019 flood. In this phase the FREs and the local geomorphic features were mapped in a GIS environment by using Qfield (Qfield Development Team, 2019) installed on a tablet along with its GNSS device. The survey focused particularly on the identification of flood-induced erosional and depositional landforms, damage to anthropogenic structures, flooded areas, maximum water levels, and main flood-water direction. By using QGIS (QGIS Development Team, 2019), field data were combined with a 15-cm-resolution orthophotograph taken a few days after the flood (ADBDPo, 2019), and a detailed mapping of FREs was performed.

The ante-flood active-channel polygon (Cencetti et al., 2017; Mandarino et al., 2019a; Winterbottom, 2000), was digitized using the 2019 Google Earth image visualized at 1:1,000–2,500 scale as a base. This data-source was also used to assess the ante-flood geomorphological features of the study area and to map its land-use and land-cover (LULC), in order to distinguish between newly-formed and reactivated landforms and to investigate the LULC classes that were mainly affected by the flood, respectively. Specific LULC categories were defined to describe the local setting. The general geological and geomorphological features of the LOR were mapped considering field surveys, previous researches, technical reports, and official river management plans (e.g. Mandarino, 2018 and references therein; AIPO, 2011; Piana et al., 2017).

The all aforementioned elements were manually-digitized with a few decades of centimeters to a few meters spatial accuracy. The study area is constituted of the floodplain area that was flooded as consequence of the Orba River overflowing, plus a 150 m buffer.

As a result, the 1:15,000 scale Main Map showing the FREs, the geomorphological features and the LULC of the LOR was realized. The FREs and the fluvial and anthropogenic elements were classified and mapped according to (i) literature (Arnaud-Fassetta et al., 2009; Carey et al., 2015; Ramasco & Rossanigo, 1988; Shellberg & Brooks, 2012), (ii) the classification scheme developed in the frame of the Reform Project (Rinaldi et al., 2015), and (iii) the Italian Institute for Environmental Protection and Research guidelines (Campobasso et al., 2018). However, a number of legend entries were specially developed to represent the FREs and the local geomorphological features. The all mapped elements converged in a large database organized in four main distinct categories: (i) FREs, differentiated into ‘Landforms and deposits’, ‘Effects on anthropogenic elements’, that are overall the FIGEs, and FWDs; (ii) ‘Fluvial landforms and deposits’; (iii) ‘Anthropogenic landforms and deposits’; (iv) LULC. These were furtherly classified considering whether or not they existed before the flood or according to their activity. The identified FREs were quantitatively analyzed through the computation of basic statistics in a GIS environment. The active channel corresponds to the ‘Fluvial deposits of the active channel’ and ‘Water course’ legend entries.

The LOR presents four GNSS monitoring sites where the BE and major features have been periodically mapped over the last years through a GNSS antenna in RTK mode. This allowed the ante- (survey on 13 and 15 September 2019) and post-flood (survey on 31 October and 2 November 2019) BE and alluvial-gully headcut position to be compared to assess the impact of the investigated event. Planform displacements lower than 0.5 m were mapped as no migration.

5. Flood-induced ground effects and flood-water dynamics

A total area of approximately 17.65 km² was flooded along the LOR because of the Orba River overflowing. Furtherly, large floodplain areas were simultaneously flooded by tributaries, irrigation channels, ditches, and human-modified small creeks draining the fluvial terraces. This aspect, together with the absence of representative flood markers, made the flooded-area limit pinpointing uncertain at some points. Substantially, the entire floodplain located between the main terrace scarps was flooded from Pratalborato to the Lemme Stream confluence. Downstream, on the left bank, the terrace scarp and the levee contained the flood water up to Portanova and from Portanova to the Barco area, respectively. In contrast, on the right bank, the flood was not completely contained by the almost-continuous levee system and, as a result, large areas were flooded. The LFC was almost completely flooded, and maximum water levels up to approximately 2–3 m above the main ground level were documented. Figure 3 illustrates the percentage area of LULC classes affected by the Orba River flood. The flood spread over 95% of the 200-year flood flow area (the so-called ‘A belt’ according to the river management regulation), 34.4% of the 200-year flooded area (‘B belt’), and 4% of the >200-year flooded area (‘C belt’). Furthermore, 70.9% of the overall flooded area corresponds to the A belt, 20.3% to the B, and 8.8% to the C.

Figure 3. Percentage area of distinct LULC classes affected by the Orba River flood. Wc: water course, Ql: quarry lake, Nat: natural and semi-natural area, Agr: agricultural area, Cuf: continuous urban fabric, Duf: discontinuous urban fabric, Ind: industrial area, Qs: quarried sediment processing site, Q: quarry, R: road infrastructure

The flood triggered erosional and depositional processes that affected both the riverbed and the floodplain. Within the active channel a generalized intense sediment mobilization was noticed, vegetated bars became largely bare, but no channel pattern changes occurred (Figure 4(a)). The active channel area increased by 6.9%, from 223 ha to 238.5 ha. This newly-formed active channel area belonged to agricultural and natural/semi-natural areas by 1.3% and 98.7%, respectively. A generalized slight channel widening involved the modern floodplain strips, generally included within the river corridor defined by ancient (approximately dated back to the twentieth century, before the 1980s) and recent (after the 1980s) bank protections, in terms of both BE retreat and overbank deposition (Figure 4(b)). Furthermore, intense bank erosion processes locally affected the recent terrace downstream of the downstream-most abstraction weir, mostly fostering processes that were already ongoing before the flood under discussion (Mandarino, 2018) (Figure 4(c)).

Figure 4. The Orba River active channel after the 21–22 October 2019 flood. (a) Bar affected by intense sediment mobilization and vegetation cover reduction. (b) Bank retreat process involving the modern floodplain. (c) Bank retreat process involving the recent terrace that was already ongoing before the flood. The white line represents a real height of 1.8 m. (d) Collapsed bank protection; most of these structures were already partly or totally collapsed before the flood.

At the present day, 21.7% of banks is affected by retreating process. Moreover, bank protection structures cover 45.5% of banks, of which 12.3% are partly or totally collapsed (Figure 4(d)). These percentages changes into 62.3% and 10.9%, respectively, considering the total channelization, that is the overall structures within the river corridor. Most of bank instability processes were already ongoing before the flood.

The rarely-possible comparison of GNSS data related to the before- and after-flood BE location (Hooke, 2016) revealed that the four monitored sites experienced different dynamics. The first and the third sites show the major retreats with a mean value of 7.5 and 5.4 m, and an eroded area of 2369 and 1341 m², respectively. In the second the BE was substantially stable, whereas in the fourth it resulted stable and retreating in the upstream and downstream parts, respectively, with a total eroded area of 921 m².

The flood reactivated two flood channels shaped by previous floods (Table 1) in close proximity of the active channel. The former, at the downstream-most abstraction weir, is progressively becoming part of the riverbed, whereas the latter, downstream of Retorto, increased the amount of sediment that has progressively deposited inside the ancient quarry it flowed into.

Overbank deposition was notably documented in the interior part of a number of river bends, in particular at Passalacqua, Grava, downstream of Retorto and of Casalcermelli. Moreover, floodplain areas covered by sediment coming from the floodplain itself were registered. This sediment often derived from the damage or collapse of road embankments of both main and dirty roads, as the cases of Capriata, close to the bridge on both banks, Panatiani, Garzaia and the Bormida confluence (Figure 5(a)). On the whole, 3.8% of flooded areas was covered by coarse sediment from sands to boulders up to approximately 0.5–0.7 m thick. Silty sediment was deposited mainly in depressed areas or upstream of some road embankments.

Figure 5. Effects of erosion and deposition processes over the floodplain. (a) Dusty road removed by the flood (white dotted line) and sediment lobes formed downstream. (b) Evidences (worked-soil removal and elongated scour hole formation) of intense superficial erosion that affected agricultural areas within the LFC. (c) Damage to the 179 Provincial Road embankment. (d) Damage to the road between Capriata and Bruno/Ospedale and elongated scour hole downstream of the road itself. These linear infrastructures (c, d) were overtopped and eroded. Downstream intense erosion and deposition processes occurred. It is noteworthy the draining-pipes-induced erosion downstream of the embankment (d). The white arrows (c, d) indicate the same point on the post-flood orthophotograph (PFO) (ADBDPO, 2019) and the field photograph (Pc, Pd); the yellow dot represents the location of photographs (Pc) and (Pd). The blue arrows indicate the main flood-water direction.

In contrast, 3.3% of inundated floodplain was affected by intense surficial erosion that caused the partial or total soil loss, up to the formation of localized 1–2 m depth elongated scour holes (Figure 5(b)). This process, as documented after previous floods (ARPA Piemonte, 2011, 2014a, 2014b, 2014c; Mandarino, 2018), was particularly intense downstream of Bruno, upstream of the Bormida confluence, and downstream of the two main road crossing the flooded areas perpendicularly to the flood-water flow, that is between Capriata and Ospedale, and along the PR179. In these last cases, the road embankment overtopping together with the flood-water passage into drainage pipes resulted in severe surficial erosion downstream of the infrastructure and in its damage and partial collapse (Figure 5(c,d)).

Sediment deposition occurred in both agricultural and natural/semi-natural areas. On the contrary, surficial erosion was noticed only in agricultural areas, especially in recently-sown fields. These last were affected by rill erosion in two sites, where a certain slope characterizes the ground level.

A number of alluvial gullies (AGs) (Brooks et al., 2009) reactivated or originated along steep scarps and eroded into the adjacent floodplain, where a concentrated runoff from floodplain entered the active channel (Figure 6(a)) or depressed areas due to quarrying activity (Figure 6(b)), following the Carey et al. (2015) development scheme (Figure 6(c)). The AGs mapped as punctual elements generally presented roundish depletion zone delimited by a jagged, semi-circular or bracket-like headcut (Figure 6(a–c)). Linear and continuous scarp front AGs (Shellberg & Brooks, 2012) were also mapped (Figure 6(a,d)). Some AGs developed or reactivated at the GNSS bank monitoring sites. In the site 1 a new AG formed and in the site 3 the existing one substantially remained stable. In the site 2 increases of 49.3% (from 630 to 941 m²) and 19.3% (from 305 to 364 m²) of the AG area were measured for the AGs included and not-included within the LFC, respectively.

Figure 6. AGs formed or reactivated over the flooded area. (a) Example of AGs reactivated during the 21–22 October 2019 flood downstream of Garzaia, located outside (Pa1) and within (Pa2) the LFC. The white arrows a1 and a2 indicate the same point on the PFO (ADBDPO, 2019) and the field photograph (Pa1) and (Pa2), taken at the yellow dot. (b) AGs developed along steep anthropogenic scarps shaped by quarrying activity at Orbetta-Pitocca (b1) and Bruno (b2). (c) AG formation scheme. Newly-formed AG upstream of Garzaia (c1). The schematic not-to-scale cross section AA’ (c2), dashed line in (c1), highlights the AG (c1) formation scheme: the concentrated flood-water from floodplain entering the active channel (blue arrow) triggered a scarp regressive erosion process (dashed arrow) that resulted in the AG formation and enlargement by the AG headcut progressive retreat (series of dotted lines). Vertical arrows in (c1) and (c2) indicate the same scarps, respectively. (d) Continuous scarp front AG developed along a quarry scarp downstream of Pennaceto. In (d1) the eroded scarp is between two ancient quarries; the eroded scarp in (d2) is upstream of the (d1) site. The black dashed pattern indicates the anthropogenic depression (b, d); the blue arrows indicate the main flood-water direction.

At some places the flood overflowed the LFC because of levee overtopping, newly-formed or reactivated breaches, burrows, and unmanaged pipes. These dynamics resulted in serious damage to the levee system (Figure 7(a,b)). Most of breaches present a scour hole a few meters deep (Figure 7(b)). Furthermore, levee breaches shaped large crevasse splays principally involving agricultural areas. Something like the levee failure consequences was registered upstream of Garzaia where the flood-water reactivated a scour hole resulting from the dusty-road embankment overtopping by the most recent extreme floods, triggering severe erosion and deposition processes downstream.

Figure 7. Damage to the levee system at Garzaia. (a) Structural damage to the levee embankment. (b) Levee breach and scour hole; the dotted line represents the levee cross section and the blue arrow indicate the main flood-water direction. The vertical white line represents a real height of 2.2 m.

Considerable volumes of floated large woody debris were trapped by riparian vegetation, single obstacles such as trellises, and woody areas located within the LFC.

Noteworthy is the critical issue of waste fill areas close to the active channel. The flood involved these areas and severely fostered the already-ongoing bank retreat at the Bormida confluence where a thick layer of buried wastes that composes the bank itself has been progressively eroding.

According to available information (ARPA Mandarino, 2018; Piemonte, 2019b), the Orba River high-magnitude floods occurred over the last decade were characterized by similar dynamics and resulted in similar, but less intense, geomorphic effects and damage, with respect to the 21–22 October 2019 flood.

In order to map both the geomorphic effects related to the flood and the geomorphic features of the study area, a specific legend derived from the combination of consolidated and original legend entries was developed. The original legend entries include anthropogenic elements and FREs, and are not site-specific (Figure 8). The former are substantially river-related man-made structures and lowland anthropogenic elements. Symbols were realized within a systematic framework, that is with common geometries and styles referring to distinct groups of elements, to make easier their comprehension. This aspect allowed for the definition of some entries and related symbols beyond the needs of this research, that anyway could be used or furtherly developed in future researches.

Figure 8. The newly-developed legend entries and related symbols aiming to map in detail the FREs after a high-magnitude flood and a number of anthropogenic elements, in a typical lowland agricultural landscape with regulated rivers.

6. Final remarks

The synergic use of extensive field surveys, GNSS surveys and aerial photographs interpretation allowed for a detailed and quantitative characterization and mapping of FIGEs and FWDs referred to the 21–22 October 2019 extreme flood that affected the LOR.

As a result, a 1:15,000 scale flood-related geomorphological map was realized in a GIS environment.

In general, an intense in-channel sediment mobilization, a widespread reactivation of bank protection scouring, and a locally-relevant channel widening were documented. However, in contrast with other cases (Cencetti et al., 2017; Clerici et al., 2015; Mandarino et al., 2020c; Nardi & Rinaldi, 2015; Pellegrini et al., 2008), the active channel did not experience severe planform changes despite the very high magnitude of the event. This aspect is most probably related to the progressive incision and channelization occurred along the LOR over the twentieth century (Mandarino, 2018), that restricted the natural dynamics of the river.

Large lowland areas were flooded, also outside the LFC, and erosional and depositional processes widely shaped newly-formed and existing landforms and deposits. These dynamics resulted in severe damages to cultivated fields, transport infrastructures, and buildings. The FIGEs and the flood markers indicated that the flood water overall flowed rather quickly over the lowland areas reaching notable water level, in particular within the LFC. Furthermore, the FIGEs spatial distribution and the main flood-flow directions over the floodplain highlighted the anthropogenic linear structures (roads and levees) interference on flood propagation and thus on triggering erosional and depositional processes, as already documented both here, after previous floods, and elsewhere (Bellardone et al., 1998; Horacio et al., 2019; Mandarino et al., 2020a). The flooding of areas outside the LFC caused not only by unrepaired breaches, water-loss due to unmanaged pipes or burrows, and levee failures, whose causes could not be identified, but also by levee overtopping, should arise serious issues concerning the LFC suitability. Furthermore, this event represents a further case demonstrating the abandoned channels to be flood-water flow favorite locations over the floodplain, with all the consequences this implies in terms of hydro-geomorphic hazards, and the essential need not to urbanize the flood-prone areas, considering that some construction projects along the LOR were recently-proposed (Boggian, 2019; Carbone, 2011).

This work highlighted the importance of using combined approaches for geomorphological mapping and the need of extensive post-flood field surveys to depict in detail the flood scenario and to validate remote-sensing data.

The legend entries specially developed in this research contribute to increase the wide spectrum of available geomorphological mapping resources. The anthropogenic elements could enhance the most commonly used legends (Campobasso et al., 2018), while the FREs could represent a new reference for detailed flood mapping.

The realized map represents a fundamental tool for hydro-geomorphic hazards assessment and zonation. Moreover, the outcomes constitute a fundamental basis (i) for sustainable and effective land use planning and river management measures definition, aiming to mitigate the river-related risks and restore the fluvial environment in the frame of the European Water Framework Directive and the European Flood Directive (European Commission, 2000, 2007), and (ii) for further research activities on geomorphic effectiveness of high-magnitude floods implementation.

Software

The all data processing and the entire map sheet design was performed by using the free and open source software QGIS.

Acknowledgements

The present study is an unexpected expansion of a wider research, within which the long-term GNSS monitoring sites are included, on the Orba River floodplain reach geomorphological evolution over the last two centuries, recently conducted by Andrea Mandarino (A.M.). The map was realized by A.M.; photographs by A.M. and Piero Mandarino (P.M., Aree Protette del Po Vercellese-Alessandrino). The authors thank all people who provided historical data and flood-related information, in particular Daniele Cermelli, Mauro Nizzo, and Claudio Monferino. Furthermore, they thank Gianfranca Bellardone (Regione Piemonte), Luca Lanteri (ARPA Piemonte), and Claudia Giampani (ARPA Piemonte) for sharing information and materials, and the Po River Basin District Authority for providing the post-flood orthophotograph. They are sincerely grateful to Luigi Moisio (Provincia di Alessandria), Gian Franco Giacobbe (Provincia di Alessandria), and, overall, P.M., for their support during field surveys. Finally, the authors wish to thank the reviewers Heike Apps, Diego Di Martire, and Dorina Camelia Ilies, and the Editors Chiara Cappadonia and Alessandro Chelli, for their observations and positive comments.

Disclosure statement

No potential conflict of interest was reported by the author(s).