Site assessment and evaluation of the structural damages after the flood disaster in the Western Black Sea Basin on August 11, 2021

On August 11, 2021, one of the most destructive flood disasters occurred in the Western Black Sea Basin of Turkey. The flood resulted in the death of 76 individuals, 30,000 people being affected by the disaster. A maximum precipitation depth of 400 mm/day was recorded at one station, indicating a return period exceeding 500 years for the rainfall event. During a two-day site visit immediately following the flooding event, damages to infrastructures, water structures, bridges, retaining walls, roads, and private houses were observed in the Bozkurt and Ayancık regions. Based on the observations, the flood wave propagated through the initial meandering river bed and floodplain, exceeding the channelized river bed capacity. Due to the massive sediment transport and drifting of trees, several bridges have been blocked and overflown where the basements of the structures in these regions were flooded. The enormous flood flow triggered extensive scouring on bridge piers, building foundations, and retaining walls, eventually causing the walls and bridges to collapse. The collapse of structures blocked the waterway and amplified the backwater effect when combined with the sediment transport. The total collapse of the retaining walls in some sections of the stream caused accelerated scouring in the foundations of the nearby buildings. Damages were also observed on the side roads along the river beds. This paper evaluated the driving mechanism of damages caused by flood flow from hydrological, structural, and geotechnical perspectives. Based on these observations and assessments, recommendations on engineering design guidelines for structures close to the floodplain, such as bridges, retaining walls, and side roads, were elaborated. Emphasis was placed on the flood-resistant design of these structures to develop a comprehensive approach for flood risk management.


Introduction
Flood events increase tremendously due to uncontrolled urbanization, global warming, and climate change. As a result, countries suffer not only from the loss of life but also from direct and indirect economic losses. To reduce the adverse effects of the flood, it is of great importance to learn from past events and understand the causes and consequences of the flood. Therefore, investigating and assessing extreme flood events and large-scale structural damages caused by flood flow plays an important role in explaining the physical mechanism.
Flood events deeply affect many countries on a global scale and cause one-third of the total economic losses due to natural disasters (Chikwiramakomo et al. 2021;Loucks and Van Beek 2017;Moel and Aerts 2011;Uddin et al. 2013;Koç et al. 2021; Ozmen 2019; Mogollon et al. 2016;Abdo et al. 2012;Urzica and Grozavu 2021;Chendeș et al. 2015). The most common cause of flooding is heavy rainfall or snowmelt (Paveluc et al. 2021). In addition to natural causes, human intervention in nature plays a vital role in the emergence of floods (Cai et al. 2016). With the increase in population and urbanization, uncontrolled construction interferes with the river beds (Jian et al. 2021). Therefore, the cross-sectional area of the stream may decrease, and a flood occurs as a result of the construction in the floodplain (Dang et al. 2011).
Flood disaster poses a significant risk for the environment and human life (Siddiqui 2011;Cook and Merwade 2009). Millions of people lose their lives, and financial losses are experienced due to floods in various parts of the world. For example, Pakistan, India, and China were severely affected by floods in 2010 and Australia between 2010 and 2011 (Mueller et al. 2019;Kundzewicz 2019;Atta-ur-Rahman and Khan 2013). In 2008, the flood occurred due to the heavy rain and snowmelt that severely affected the US State of Iowa; damaged many residences and workplaces, and 40,000 people had to leave their homes (Sönmez et al., 2013). The maximum damage caused by river flooding in a country in one year was seen in China in 2010 with 51 billion dollars (Kundzewicz 2019). In 2010, almost 2000 people lost their lives due to monsoon floods in Pakistan (Kundzewicz et al. 2014). In May and June 2016, extreme storms in Europe caused heavy rains; the Seine River overflowed in France. These disasters caused an economic loss of approximately US$ 4 billion (https:// en. wikip edia. org/ wiki/ 2016_ Europ ean_ floods). In Serbia, the floods affected 1 million people and resulted in 51 casualties, of which 23 were due to drowning (Dottori et al. 2017). In Bosnia-Herzegovina, over a million people were affected by flooding, almost 90,000 were displaced, and 25 casualties were recorded in 2014 (Dottori et al. 2017).
Many flood events were also observed in Turkey. The Turkey Disaster Database (TABB), which was established in 2009 to collect all documents related to both natural and anthropogenic disasters, reported 1076 flood events that caused 795 deaths and economic losses of US$ 800 million from 1960 to 2014 Koç and Thieken 2018). The Black Sea, Mediterranean, and Western Anatolia regions are the most sensitive places to floods in Turkey (SYGM 2019; Oğuz et al. 2016). The highest frequency of human and economic losses due to flood hazards are seen in the Black Sea region (Koç 2021). During the flood events in Istanbul, Ankara, and Senirkent in 1995, 74 people lost their lives; 46 people were injured; 2000 people lost their homes, and 65 million dollars of economic loss were reported (Ertek 2014;Korkanç and Korkanç, 2006). More than 2 million people were affected by the flood disasters that occurred in the Western Black Sea Region in 1998; more than 30 lives were lost, and 478 houses were completely inundated (Kömüşçü and Çelik 2013;Ceylan et al. 2007;Ergunay 2007;Zeybek 1998). During the Ayamama flood, which took place in Istanbul between 8 and 12 September 2009, 32 people lost their lives, 3816 houses and 1490 workplaces were damaged (Kömüşçü et al. 2011). Fourteen people lost their lives in the flood in Rize on August 26, 2010(AFAD 2018. In addition, 77 and 30 flood incidents in Kastamonu and Sinop Provinces of Turkey from January 01, 1950, to June 01, 2018, were observed (AFAD 2018. A review of national reports on natural disasters shows that floods are the most economically damaging natural disasters after earthquakes (Ozmen 2019). Based on these events, it is crucial and necessary to carry out studies on floods to prevent and reduce the damages and losses in Turkey.
The hydrodynamic forces resulting from flooding cause damage to structures such as bridges, culverts, roads, retaining walls, and buildings. The direct flood damage to the structures is typically observed in two ways, i.e., wall failure or scour under foundations (Chung and Adeyeye 2018). Flood actions that result in scour are a leading cause of bridge and structural failure (Prendergast et al. 2018). For example, washing away the soil from around bridge foundations by hydraulic action causes bridge collapse since it reduces the stiffness and bearing capacity of foundations of bridges located in waterways (Prendergast et al. 2018;Prendergast & Gavin 2014;Arneson et al. 2012). Therefore, there are many examples of scouring failure at bridges and culverts due to flooding (Dyke et al. 2021;Sung and Wang 2013). Moreover, roads and highways are commonly damaged by flooding, resulting in their dysfunctionality or reducing their serviceability (Fathy et al. 2020;Ismail et al. 2019;Lertworawanich 2012;Keller and Ketcheson 2011).
A flood event hit the Western Black Sea Basin on August 11, 2021, resulting in a serious loss of life and property. Based on the extent of the losses, this flood can be considered to be one of the most devastating flood disasters in the country. Following the disaster, a two-day site visit was carried out by the authors to Bozkurt and Ayancık which are among the most affected areas. Observations and site assessments of the Bozkurt and Ayancık regions were reported after the flood disaster on August 11, 2021, to contribute and shed light on the existing studies on flooding. For this purpose, first, the flooded region and the flood event were explained and assessed, along with the photographs taken during the visit. Then, the causes and effects of flood and damages of the flood on infrastructures, bridges, culverts, retaining walls, and roads were discussed. Finally, recommendations were made on the design guidelines of the structures located in or nearby the floodplain, which were subject to potential frequent flooding. The observations and the evaluations were made based on different disciplines of civil engineering; hydraulics, geotechnics, transportation, and structures which gives to this paper a comprehensive perspective.

Study area: Bozkurt & Ayancik regions
Bozkurt and Ayancık regions are within the boundaries of the Black Sea Basin located on the northwest side of Turkey, shown in the red line in Fig. 1a. The Black Sea Basin, which is among the 25 basins in Turkey, is affected by flooding due to high rainfall intensity. The annual average precipitation was reported as 775 mm in Western Black Sea Basin (SYGM 2019).
Bozkurt, the district of Kastamonu city, is located between 42° 0′ 00″-41° 40′ 0″ north latitudes and 33° 45′ 0″-34° 10′ 0″ east longitudes (Fig. 1b). The altitude of the region above sea level varies between 27 and 400 m. The population of the region is about 10,000 to the Turkish Statistical Institute (https:// www. tuik. gov. tr/ indir/ duyuru/ favori_ rapor lar. xlsx). The Ezine Stream flows through the Bozkurt region, with a watershed area of 375 km 2 , and reaches the Black Sea downstream. The stream flows in a south-north direction, and there are high-density residential and commercial areas located on the east and west shores of the Ezine Stream, in Kastamonu.
Ayancık, the district of Sinop city, is located between 42° 0′ 00″-41° 45′ 0″ north latitudes and 34° 20′ 0″-34° 45′ 0″ east longitudes (Fig. 1c). The population of the region was reported to be about 24,000 by the Turkish Statistical Institute (https:// www. tuik. gov. tr/ indir/ duyuru/ favori_ rapor lar. xlsx). Ayancık Stream Watershed has an area of 675 km 2 , and Ayancık Stream reaches the Black Sea downstream. Ayancık Stream flows through the center of the Ayancık region in Sinop, and there is a residential area on the east and west sides of the Ayancık Stream at the downstream part of the watershed. While 28% of the region consists of agricultural areas, the rest of the area consists of forests since the clayey and calcareous character of the soil type provides a suitable environment for the cultivation of forest products (https:// en. wikip edia. org/ wiki/ 2016_ Europ ean_ floods).
According to the General Directorate of Water Management (SYGM) of the Republic of Turkey Ministry of Agriculture and Forestry report (SYGM 2019) a total number of 40 flood events in Kastamonu and a total number of 37 flood events in Sinop are reported between the years 1963 and 2016. In fact, only one major flood event in 1964 in Sinop was observed until 1983. After 1983, the number flood events increased due to the urbanization in the region.

Field observations
There are several factors influencing the magnitude, frequency, and mechanisms of flood generation, i.e., topographical, physical, or man-made. Thus, flood development and its impact differ from one region to another (Wohl 2000). Nevertheless, Bozkurt and Ayancık regions have their particular cases, detailed below, based on the field evidence.

Bozkurt region
The rainfall event in Bozkurt started on the 10th of August, and the destructive flood affected the region occurred almost 24 hours later. Particularly, Kuz and Mamatlar areas, located 20 km away from the city center at elevations between 700 and 1200 m, were significantly affected. Heavy rainfall and the steep topography of the region increased the energy of the flood flow. Due to the forest cover of the region, lots of trees were washed away by the flood, which resulted in a tremendous increase in the damages induced by the flood. Based on the reconnaissance survey findings and individual eyewitness records, the effect of the flood in the Bozkurt city center is detailed below and in Fig. 2a.
• The first indication of the flood was observed close to Grids B1 and C1, located at the intersection point of two sub-branches of the Ezine Stream. The flood carried lots of trees and logs from the forest in the early hours of the event. • A pedestrian bridge named Bridge B5 at Grid B3 was destroyed, and the impact of the flood washed away its deck. While being dragged, the deck of Bridge B5 remained intact until it was stopped by a reinforced concrete Bridge B4 located at Grid B4. The dragged deck of Bridge B5 blocked the flow and reduced the waterway of Bridge B4, and soon backwater effect was observed as given in Fig. 7b. As the flood debris, i.e., timber logs and trees, accumulated in a short period, the depth of the flood behind Bridge B4 increased, and water flowed over the bridge approximately in thirty seconds. Based on the eyewitness videos, the flood wave amplitude was at least 2 m higher than the bridge deck. After the backwater effect was observed, the water flowed over Bridge B4, and the city center located on the floodplain of both sides of the river was completely inundated. Since Bridge B4 was located in the city center, very close to the densely populated area (Grids C4-C5-C6 and D4-D5-D6), the flooding led to many casualties in the residential areas. • The flood damage was more pronounced at Grids C2 to C6 and D2 to D8 located on the eastern side of the river since the ground level was approximately 2 to 4 m lower than the western side. Therefore, the flood extent was almost 200 m on the eastern side, whereas it was about 40 m on the western side. Consequently, the flood effect on the residential buildings on the western side was more limited (Grids B2 to B4), except for the four residential buildings in Grid B5. • The impact of the flood was dramatic both on the river bed and retaining walls along the river. Four residential buildings in Grid B5 (Buildings B2-B3-B4-B5) and almost entire industrial buildings in Grids D7 and D8 suffered from significant or total damage since the retaining structures protecting these shores lost their stability. These failures are discussed in detail in the following sections of the paper.

Ayancık region
Ayancık region, located 50 km east of Bozkurt, was also significantly affected by the heavy rainfall on the 10th and 11th of August. The loss of life and property caused by the flooding in Ayancık was relatively less than the one in Bozkurt. The damages observed in the Ayancık region are presented in Fig. 2b. Locations of scoured areas, collapsed buildings, and bridges can be seen in this figure. Unlike Bozkurt, the damage was observed along the river, and the flood wave did not reach the city center. The main reason for less damage was the river bed's extensive width, which is about 100 m for most sections. The sediment transport over the retaining walls due to the flood was noticeable only at a couple of locations along the river-side. However, the most dramatic and extensive damages were observed in the bridges in this region. In this context, emphasis is given to the failure of the bridges in Ayancık in this paper.

Hydrologic and hydraulic evaluations
The causes and effects of flooding were presented based on the recorded data and observations done after post-disaster as follows: The total precipitation recorded during the storm event at 4 rain gauge stations, i.e., Abana, Mamatlar, Kuzköy, and Devrakani, was used to find the average rainfall for the Bozkurt disaster area with Thiessen Polygon Method. The precipitation data were obtained from the Turkish State Meteorological Service (TSMS) Republic of Turkey Ministry of Environment, Urbanization and Climate Change. The weighted rainfall depth was calculated as 296.8 mm/day (Table 1). This value was by far greater than the maximum total precipitation recorded between the years 1930 and 2021. The second-largest value was 104.7 mm/day measured in 1953 in Kastamonu (https:// www. mgm. gov. tr/ verid egerl endir me/ il-ve-ilcel er-istat istik. aspx?m= KASTA MONU).
The total precipitation recorded during the storm event at 3 rain gauge stations, i.e., Akören, Ayancık, and Çangal, was used to find the average rainfall for the Ayancık disaster area with Thiessen Polygon Method. The weighted rainfall depth was calculated as 223.5 mm/day (Table 1). This value was greater than the maximum total precipitation recorded between 1930 and 2021. The second-largest value was 203.2 mm/day measured in 1948 in Sinop (https:// www. mgm. gov. tr/ verid egerl endir me/ il-ve-ilcel er-istat istik. aspx?m= SINOP).  To support and further strengthen the above argument, the hydrological models of both Ezine ( Fig. 3a) and Ayancık ( Fig. 3b) Stream Watersheds were developed by using the Hydrologic Engineering Center Hydrologic Modeling System (HEC-HMS). Firstly, the Watershed Modeling System (WMS) program was used to obtain watershed area, slope, boundary, DEM, land use, and soil group type. In addition, Coordination of Information on the Environment (CORINE) 2018 land cover data were obtained from the European Union Copernicus Land Monitoring Service (https:// land. coper nicus. eu/ pan-europ ean/ corineland-cover). Soil map was obtained from the Food and Agriculture Organization Harmonized World Soil Database (FAO HWSD) (http:// www. fao. org/ soils-portal/ soil-survey/ soil-maps-and-datab ases/ harmo nized-world-soil-datab ase-v12/ en). Soil Conservation Service-Curve Number (SCS-CN) method was selected as a rainfall-runoff model using land cover and obtained soil maps. The weighted CN value for the watershed was calculated through the WMS program by taking into account the land use, hydrologic soil groups and the areas covered by these values in the watershed.
In the next step of the modeling, the data obtained from WMS are exported to the HEC-HMS. Soil Conservation Service (SCS) Curve Number (CN) model was selected in the hydrological model. The CN was calculated as 65 for both watersheds and the percent of the impervious area was chosen as 5%. Models were simulated under the August 9-12, 2021 rainfall events (Fig. 3c, d). The all precipitation data was obtained from the Turkish State Meteorological Service (TSMS). The precipitation data recorded during the storm event at 4 rain gauge stations, i.e., Abana, Mamatlar, Kuzköy, and Devrakani, for the Bozkurt disaster area and at 3 rain gauge stations, i.e., Akören, Ayancık, and Çangal for the Ayancık disaster area was defined into the HEC-HMS model with Thiessen Polygon Method as shown in Fig. 3a, b. And hydrographs at the outlets of the watersheds for the Ezine Stream in Fig. 3e and the Ayancık Stream in Fig. 3f have been obtained. Based on the HEC-HMS models outputs, the maximum flow rates in the watershed were calculated as 1026 m 3 /s for the Ezine Stream and 1603 m 3 /s for the Ayancık Stream ( Table 2). The flow calculated by HEC-HMS represents the total flow rate generated over the whole basin whereas the estimated flow rate was observed at a particular time, at a particular point and within a defined cross section in the video. Thus, observed flow rate may not be the peak flow rate. The calculated peak flow rate for the storm event using HEC-HMS model, the observed peak flow rate using videos, and the 500 year peak flow rate (Q 500 ) calculated by SYGM (2019) for the Ezine and Ayancık Stream watersheds are given in Table 2.
• Based on the rainfall records and flow observations for the flood event, one can argue that the flood event that occurred on August 11, 2021, falls into the category of at least a 500 year storm or even a storm with a greater return period than a 500 year. The flood  (Fig. 4). • The residential areas exposed to flooding in the Bozkurt region are located in the original stream bed (Fig. 5a) and floodplain, where the channelized river ( Fig. 5b) bed capacity was exceeded. The original streambed in Fig. 5a had a meandering shape when the topography of the area was examined. In areas where retaining walls lose their stability, the stream bed tends to return to its original position with severe signs of scour. This finding is supported by the scouring locations compatible with the original stream bed as given in the 1968 map in Fig. 5b with solid red zones. • The heaviest damages to structures were observed where the original or channelized stream bed has meanders. The reason was attributed to the existence of big momentum fluxes on meanders which causes big reaction forces exerted on the fluid by the structural components or soils lying under the structure (Schuurman et al. 2016;Posner 2011;Chang 1988;Davies and Tinker 1984;Ferguson 1983). The reaction force effects on meander are shown in the schematic view (Fig. 6a) and photograph (Fig. 6b). • The water passages through water structures such as bridges and culverts on the stream bed were blocked with the deposit of trees and sediment, resulting in capacity losses in the cross-sections (Fig. 7a). Clogging of the bridges causes the backwater effect, and the water accumulated behind the bridges overflows from the stream bed towards the residential areas located on the floodplain of the stream. Particularly, these blockages may be the major reason of flooding in Ayancık Region as the channel capacity is only slightly exceeded there. The backwater effect occurred in the Ezine Stream at Bridge B4 (Fig. 7b). The location of Bridge B4 (Grid B4) is shown in Fig. 2a. • The big magnitude of flood flow along with high flood velocity and high flood depth has caused extreme scouring on bridge piers, the foundation of buildings, and retaining walls on the sides of the channelized river beds. In addition, the deposit of large pieces of wood and logs carried from the forest by the flood amplified the damage to the piers and the decks. As a result, total collapse of the walls and bridges was observed (Fig. 8a, b). • A combination of the said effects of flood, i.e., big momentum fluxes on meanders, extreme scouring, and deposition, also result in partial damage on the side roads and highways next to the river beds (Fig. 8c, d).

Observed damage to structures
The reconnaissance study carried out by our group a few days after the flood disaster covered an extensive area between Bozkurt and Ayancık regions where the regional centers and connection roads were included. Based on the findings of this survey, the damaged structures due to floods can be categorized as follows; • Retaining walls (RW) • Residential and commercial buildings, • Transportation network: roads, bridges, and culverts.
The level of damage observed in the buildings can be classified as ranging from low to very high, mainly based on their proximity to the river and the land's topography. The following sections of this paper present the observed damage and the possible causing mechanisms.

Observed damage to retaining structures
One of the most devastating effects of the flood event was on the retaining walls. Total or partial failures and collapses occurred in the retaining walls, which eventually caused amplification in the damage levels for the nearby roads and structures. It can be argued Visual observations revealed that the walls along the riverside were either reinforced concrete T type or trapezoidal masonry walls in the flood region. Severe damage was observed in the regional center and along the connection roads and mountainous roadsides for both wall types. Along the flooded river, the failure modes of reinforced concrete retaining walls can be listed as moving out of the alignment, toppling, and separation at the joint planes. Structural failures were also observed. Masonry walls also had similar types of failures in addition to the collapse of the masonry body. It has been observed that the partial damage or total collapse on the retaining walls; • sometimes occurred with a similar mechanism along the entire riverside section • but in many cases, the damage was concentrated at a particular section of the wall and • even different damage conditions could be observed on the walls on either side of the stream.
Typical damages observed in retaining walls due to the flood effect are discussed using the findings of a river section located at Grid C7 as given in Fig. 9. The authors defined three locations as L1, L2, and L3 and further evaluations are made through photographs taken during the field survey at these locations.
RW-A and RW-B were gravity-type retaining walls, RW-C was a T-type reinforced concrete wall. One of the most important findings was that the damage levels observed at RW-B and RW-C were more significant than RW-A. This finding may be attributed to the bend scour mechanism in the meander regions where high shear stress zones occurred. This phenomenon was also expressed in Sect. 3.2 of this paper as the presence of large reaction forces at these points due to big momentum fluxes. Observations made Fig. 9 Key plan for the damaged retaining walls at Grid C7 1 3 at L1 and L2 locations give evidence through Figs. 10 and 11 for why and how these retaining walls may have collapsed. Walls that toppled out of the alignment, weak joint connections, structural damage in the walls, and scour below the foundations are visible in Fig. 10. Although the foundation depths of the walls were not known for certain, the walls had shallow (almost surficial) foundations, and the reinforced concrete retaining walls did not have any shear keys, which could have increased lateral stability. • The photograph taken from the L2 location is given in Fig. 11 and shows no leftover of RW-B since the entire wall totally collapsed and was washed away with the flood sediment. Therefore, only the concrete basement of the gravity-type retaining wall could be observed. As seen in Fig. 11, the bedding concrete of the RW-B gravity walls was still visible without any sign of scouring. • The photograph taken at the L3 location is Fig. 12, which shows a collapsed retaining wall section where it is probable that the driving forces behind the retaining wall increased significantly due to the strong water flow. The water came from the back of the retaining wall and increased the driving forces, which caused the retaining wall to collapse. This effect was accompanied by the erosion of most of the backfill material.
Although heavy damage was observed on the reinforced concrete and masonry walls, there was almost no structural damage on the mass concrete walls with artificial concrete blocks to protect the wall footings (Fig. 13). These walls were located through Grids B2 to B4 in Fig. 2a. Retaining sections of these walls were supported with a concrete block along the toe for scour protection, which was probably one of the reasons to help these walls stay stable. These retaining walls were located along the west side of the river upstream of Bridge B4 and remained stable, while those walls downstream of Bridge B4 were significantly damaged. This was an important finding as it was consistent with the progress of the flood given in Sect. 3.1.1.
Based on the above observations, the possible mechanisms for the failure of the retaining walls can be listed as follows; • Due to the high flow rate and velocity of the flood, the alluvial soils on which the retaining wall foundations were located experienced extensive scour. This caused two Fig. 12 The backfill of the gravity type RW-B was heavily eroded at L3 critical problems for the retaining wall stability; loss of passive pressure at the foundation level and loss of bearing capacity at the foundation base. • Due to high flood instantaneous velocity (up to 7 m/s in some locations), large hydrodynamic forces combined with impact forces due to the carried tree logs caused significant lateral forces on the retaining walls. • One of the possible reasons is that the increased depth of the water increased the uplift pressures; thus, retaining structures lost their vertical stability and probably lost contact with the foundation soil, resulting in a complete loss of stability. • The observations revealed significant soil erosion at the subsoil under building foundations and the backfills behind the retaining walls. Since the flood water washed away fine particles from the voids between the large grains, the soil skeleton became looser, resulting in lower strength and stiffness values. • Logs and rocks carried by strong river flow directly hit the retaining walls resulting in structural damage. • Due to the increased water levels, effective stresses changed behind the walls and below the foundations. For soils, effective stress is the main concept in geotechnical engineering which governs the compressibility and shear strength of soils. However, for the cases observed in this paper, the effect of this concept on the stability of the structures seems negligible compared to the effects mentioned above.
Although all these mechanisms have separate significant effects on the retaining walls, the dominant ones on the observed damages or collapses cannot be identified because of the complexity of the flood event.

Damage to residential structures
A significant number of residential buildings was affected by the flood. Following the flood, governmental authorities conducted field surveys, and the damage distribution of buildings was listed in Table 3. The surveys were carried out for Kastamonu, Sinop, and Bartın cities and according to observed damage levels, the structures were categorized from slightly damaged to collapsed. Based on the number of buildings affected by the flood, the  Table 3). Four of the collapsed buildings were multi-storey structures used as residences. The remaining buildings were single-storey and were used as warehouses or workplaces.
Some typical photographs of building damages are presented in Figs. 14, 15, 16, 17. Building damages can be classified into two different groups; 1. Damage to infill walls due to large hydrodynamic and impact forces: Typical examples are shown in Fig. 14. McBean et al. (1988) expressed that a water velocity of 3 m/s acting over a 1 m depth can produce a force sufficient to exceed the design capacity of a Fig. 14 Infill walls collapsing due to hydrodynamic forces typical residential wall. As given in the previous sections of this paper, the flood height increased to approximately 3-4 m above the ground level, and the flood instantaneous velocity was around 7 m/s at Grid B3 in the Bozkurt region. This velocity was calculated by using eyewitness video records through a very simplified method where the displacements of the dragged objects were tracked in time. Flood height and velocity of these magnitudes are expected to cause high lateral forces on the residential building walls. It is anticipated that the impact forces due to the materials carried by the flood amplified the lateral forces significantly. 2. Damage due to scour: Typical examples are shown in Figs. 15,16,and 17. This type of damage ranged from, a. local scour of the foundations, which caused none, to partial structural damage below the foundations (as exemplified in Fig. 15) to b. total scour of the foundation materials, which led to partial or total collapse in the building (as seen in Figs. 16 and 17).
One of the most striking effects of the flood on the buildings occurred in the Bozkurt region in Grid B5 of Fig. 2a. Three residential buildings entitled B3, B4, and B5 in Fig. 17a suffered partial or total structural collapse causing several fatalities after a significant amount of scouring occurred below their foundations. All buildings in Fig. 17 were located on shallow foundations on loose alluvial deposits, which consisted of loose clays, silts, sands, and gravels. The series of events for buildings in Fig. 17 are explained below and are schematized in Fig. 18.
• In Fig. 17, the collapsed buildings were located at varying lateral distances to the retaining walls. The three collapsed buildings given in Fig. 17a were approximately 12 to 25 m far from the walls, but the building that collapsed in Fig. 17b was almost 35 m far from the retaining walls. In both cases, the buildings collapsed due to the loss of stability of the retaining walls. Once the retaining walls collapsed, the build- ings founded on shallow foundations suffered partial or total collapse due to the scour of their foundation soils. Buildings as far as 35 m away from the initial main stream were affected due to the scour of the foundation subsoil mainly due to the collapse of the retaining walls. It should be recalled that their location corresponds to the former riverbed which makes these structures more vulnerable. • In Grid B5 of Fig. 2a, total collapse of the retaining walls occurred for a long portion of the riverside, which made the foundation subsoil of the buildings vulnerable to scour since the loose alluvial subsoil has a high potential for this effect. Some local people argued that a discontinuity was present in the retaining walls in this area due to some logistic and shipping purposes. This discontinuity probably accelerated the overflow of the flood and the loss of stability of the retaining walls. Eventually, the foundation subsoils experienced different scouring levels dominating the observed structural damage levels. • Figure 18 presents the interrelation of the stability of the retaining walls and the stability of the building on shallow foundations in Fig. 17. Figure 18a represents a typical building on a shallow foundation in locations where the retaining walls retained their stability. In these cases, the buildings did not suffer from any scour on the foundation soil and therefore the structural stability was provided. Figure 18b represents the buildings B1, B4, and B5 of Fig. 17. Although it was not possible to observe the exact foundation depth of these three buildings, the foundation depth can be accepted to be about 2 m since it is a typical implementation in the area. These buildings experienced a significant amount of scour in the foundation subsoil resulting in the partial collapse of the building. Building B3 of Fig. 17a is schematized in Fig. 18c. For this building, the foundation depth was about 2 m, and since a great portion of the subsoil was lost, this resulted in the total collapse of the building. Building B4 in Fig. 17a is schematized in Fig. 18d. B2 was located at a depth of 5 m. This depth saved the foundation subsoil from being scoured and this building remained undamaged.

Effects of flood on transportation network
The disaster, which was defined by experts as the most devastating flood in the history of Turkey, affected three provinces and 12.000 km 2 in total in the Western Black Sea Region. The extraordinary rainfall in Kastamonu, Bartın, and Sinop provinces caused various damage to the transportation infrastructure. These damages resulted in both increased loss of life and property, difficulty in post-disaster first aid, and search and rescue efforts. Efforts were made to open the roads to traffic and to keep the main arteries open by establishing portable bridges to enable rapid transportation in the disaster area and also first aid. Figure 19 shows the Western Black Sea Region of Turkey, and the road network of the flooded provinces of Kastamonu, Sinop, and Bartın located in this region, which were devastated by the flood (URL 6). The flood was observed to be devastating and caused various levels of damage in an area of 240 km in length, especially in areas close to the coast. Damages in the transportation infrastructure were reviewed by two separate classes; road and hydraulic structures.

Road damages
The total area of the three provinces where precipitation occurred is 21,300 km 2 , and the areas where floods were devastating in these provinces are approximately 12,000 km 2 . In this area, there is an average of 96 m of road per km 2 . All of the roads are asphalt paved roads (URL 7). Kastamonu Province has a road network of 553 km, and a total of 59.5 km of damage was detected at different spots. 54 km of damage occurred in 564 km of road network in Sinop Province, and 41 km of road damage occurred in 111 km of road network in Bartin Province. In total, 154.5 km of roads were destroyed due to the flood and had to be rebuilt. Although there was no structural damage in some areas, roads and bridges were flooded due to rising water levels and disrupted the traffic. The roads could be made functional only after the floodwaters receded. As a result, 967 km of roads were damaged at various levels, partially or completely closed to traffic, and affected by flooding.
Damages on the roads can be reviewed under three headings; 1. Major collapses caused by floodwaters hitting the road platform 2. Partial collapse and asphalt damage due to deposits on the road platform 3. Flooding and collapses on the road platform due to insufficient capacity of the culvert.
Two pictures taken from two different locations are given in Fig. 20. As seen in Fig. 20a, it has been observed that the damages occurred mostly in the sections where the hydrodynamic forces act perpendicular to the road platform, depending on the flow direction.
Apart from this kind of damage, the insufficient excess capacity of the culvert under the road platforms also caused road damage. Especially with the overflow of the culverts, the body of the road was carved from the bottom, and there were collapses on the roads. In Fig. 20b, the damaged road platform due to the culvert with insufficient capacity is shown.
In the second type of damage, a part of the road platform collapsed due to the flood and scouring along the streamline. Examples of such damage, parallel to the streamline and the road direction are given in Fig. 21 where ruptures in the asphalt layer are observed. Roads were partially closed to traffic in such places, or driving safety was reduced.

Bridge damages
There are many bridges that have been damaged with different lengths and widths due to the streams flowing through the neighborhoods and even different blocks. Most of the bridges over the main arteries were built between 1955 and 1965 by Turkish Directorate of Highways for construction and maintenance. These bridges constructed between 1955 and 1965 were designed according to a 100 year flood, taking into account the H20-S16 load class (AASHTO 2002). Bridges on roads with less Annual Average Daily Traffic (AADT) were designed according to the H15-S12 load class. These bridges were built with simple support beams or Gerber beams. In general, they have shallow foundations without piles, with reasonable embedment depths. Between 1965 and 1975, bridges were constructed with reinforced concrete slabs. The foundations of some of these bridges were piled depending on the local soil conditions. Their load class was H20-S16.
Bridges connecting inner roads that are under the responsibility of local administrations were constructed with simple techniques, and load classes were generally H15-S12 (AASHTO 2002). Therefore, the damage levels were more severe on these bridges. These bridges were typically reinforced concrete slab bridges built between 1965 and 1973, consisting of 12.95 m spans. These bridges were 26 m, 39 m, 52 m, and 65 m long depending on the number of spans. They often had shallow foundations. Due to the increasing flood disasters caused by global warming and other reasons, consecutive changes have been made in the legislation. Formerly, the bridges over the main arteries were designed considering the 100 year flood depth according to the bridge design criteria set by the General Directorate of Highways. These criteria were revised 5 years ago due to the frequent flood events observed within the region and 500 year flood depth has been taken into consideration since then.
However, almost all of the bridges damaged in flood in the Western Black Sea Region were bridges designed and built about 40 years ago according to the previous legislation. Therefore, the hydraulic sections of the bridges were insufficient and severe damages were observed after the flooding on August 11, 2021. Damages on bridges can be summarized under two main categories.
1. Completely damaged bridges which were washed away by flood 2. Partially damaged bridges which can be opened to traffic in a short period of time with emergency repairs The main arteries, towns and neighborhoods in this region are connected by 115 bridges. The total length of these bridges is approximately 4600 m. There are 27 bridges on the main arteries that are under the responsibility of the State Highways Administration, and their total length is 1100 m. Six of these bridges were completely damaged, and twelve were partially damaged. Of the destroyed bridges, three are in Bartın, two are in Kastamonu, and one is in Sinop (KGM, 2021a).
An example of a collapsed bridge, is the Şevki Şentürk Bridge (Fig. 2b-Bridge A1), which failed due to the scouring of the shallow foundations, is seen in Fig. 22a. This bridge was built in 1965 to connect neighborhoods and is under the responsibility of local governments. As a result of the collapse of the piers, the bridge lost its stability, and the middle spans crumbled. Another destroyed bridge was the 68 m long Çatalzeytin Bridge on the İnebolu-Abana-Çatalzeytin road (Fig. 22b). The location of Çatalzeytin can be seen on Fig. 19. Çatalzeytin Bridge was a Gerber girder bridge built in 1963 with spans of 21.30 m + 24.75 m + 21.45 m (KGM 2021b). The cross-section of the destroyed bridge is shown in Fig. 23a. Since the foundations of the bridge were superficial, the original soil, in which the bridge abutments were founded, has been carved because of the flood, and the bridge completely collapsed and was washed away. A new bridge with pile foundations was designed based on the 500 year flood flow and For partially damaged bridges that could be reopened to traffic with rapid intervention, it was observed that most of the damage in these bridges occurred due to scouring at the approach embankment. Figure 24a shows an example of this kind of bridge damage whose approach embankment was damaged due to flooding. Another important factor in the bridge damage was flood debris. The damage to the piers increased as large trees and logs, transported by the flood, hit the bridge piers laterally. In addition, these trees and logs blocked the waterway of the bridges, resulting in additional horizontal pressure on the bridges, as seen in Fig. 24b.
Striking damage to the bridge piers was observed on the 90 m long Ayancık Bridge (Fig. 2b-Bridge A2 at Grid C6), which consists of four middle piers. The superstructure of the bridge was Gerber beams. The oval geometry of the bridge piers was chosen to reduce the scour effect; and the foundations of the bridge were not on piles. During the field investigations, severe damages were observed on Pier 1 and Pier 2 (Fig. 25). Pier 1 has indications of severe scouring, whereas Pier 2 has structural damage, most likely due to the interaction with the superstructure. The plan view in Fig. 25 shows that the stream bed has a curved geometry with an estimated radius of 350-380 m. The eyewitness records clearly showed that a massive amount of debris was carried by the flood very close to the shoreline as the river geometry curves toward the bridge location. As the dragged trees and logs hit the piers with the flooding, an additional impact and turbulence were induced on Piers 1 and 2. The flow velocity was higher outside the bend with a corresponding higher erosion and scour power. The observed damage on these piers may be attributed to the curved geometry of the river, which caused additional centrifugal impact forces on Piers 1 and 2. Piers 3 and 4, which were less affected by the impact of the transported materials, on the other hand, remained stable after the flood. The widening of the river bed on the left bank just downstream of the bridge may also have produced local turbulence, and related active scour. The damage observed in this bridge reveals that the geometry of the stream at the bridge location and the magnitude of the flood can make some of the bridge piers more vulnerable to flood damage than other neighboring piers.

Hydrologic and hydraulic perspective
The following recommendations are made from the hydrological perspective to reduce the flood damages based on the lessons learned from the flood event: a. The observations show that the classical approach of flood damage evaluation, which was done only with respect to flood depth and flood extent, was not sufficient for flood management. The structures may suffer from the forces due to the large momentum fluxes and high velocities, which will result in huge scouring and partial or total collapse of the structures within the area. Therefore, the engineering design guidelines of the structures, including bridges, culverts, retaining walls, and side roads and highways located close to the floodplain, should be elaborated by taking into account the flood conditions. b. Settlements in the original stream bed and floodplain, downstream of the watershed, and in curvatures of meanders should be avoided. c. The base elevations of the structures built on the riversides should be at least above the 100 year flood water level or possibly the 500 year flood water level. d. The design of the water structures, especially the ones close to the residential areas, should be done by taking into account the 500 year flood flow. In addition, the hydraulic dimensions of all existing major bridges located in regions where frequent heavy rainfall is observed should be reviewed, and in case of necessity, cross-section enlargements should be recommended and implemented. e. Flood management plans, which involve flood hazard and flood risk maps for 500 year floods for each basin of Turkey, were prepared by the General Directorate of Water Management and revised every 6 years. In light of these plans, it is of utmost importance to plan and implement the measures which should be taken before, during, and after floods, especially for regions with high flood risk. f. The effect of sediment transport due to flood or other solid materials which are washed off with flood flow should be simulated by using a sediment transport model incorporating into the flood modeling. g. "Best Management Practices" such as detention ponds and sediment traps to reduce the effect of flood flow in the watershed and water structures to control the sediment transport due to the flood should be planned and implemented. h. It is important to establish an early flood warning system in regions under extreme flood risk.

Structural and geotechnical perspective
a. The damage patterns observed in this disaster revealed that the most damage-prone areas in case of flood include high-velocity areas and meander bend zones, where higher shear stresses occur. Therefore, special precautions should be taken for these sections, such as retaining walls. b. Retaining wall stability during floods becomes significantly essential for decreasing the losses due to flood damage. This requirement can be achieved by maintaining lateral resistance during floods. Deeper shallow foundations combined with solid scour meas-ures or deep foundations below the retaining walls should be considered, especially in urban areas. Retaining walls should be constructed continuously to ensure no weak zones or discontinuities. If such discontinuities exist, especially in bending zones, these zones may govern the flood damage. c. The foundation depth is of utmost importance for buildings close to the river bed. In case the retaining walls fail during the flood event, a significant amount of foundation subsoil may be lost due to accelerated scour, resulting in structural collapses. Since these areas are generally loose alluvial soils, the vulnerability to scour is already very high. d. The countermeasures for the foundations of the retaining walls against scour were not adequate for this flood. Foundation protection against scouring for these structures should be designed conservatively for both the depth and the width of the scour protection elements. Present retaining walls probably lacked appropriate scour protection countermeasures for the flood. In any case, scour measures in urban areas should use materials resistant to erosion in case ordinary soil measures do not work.

Roads and transportation structures perspective
a. The major collapse on the roads was observed with a direct impact of the flood to the road platform accompanied by the asphalt damage due to settlements on the road platform. At some sections, the insufficient capacity of the culverts also triggered the damages. b. The bridges were destroyed due to the loss of foundation stability induced by scouring in the stream bed and the rotations and displacements of the piers. In addition, the bridge piers, which became unstable because of the scouring of the stream bed, were either with piers or shallow foundations. c. The performance of the bridges in Ayancık should be evaluated based on the fact that the bridges were about 40-50 years old and were designed according to the 100 year flood, and some strengthening options may be recommended. d. Based on the field observations, it can be expressed that bridges with piled foundations were either undamaged or partially damaged and did not collapse. Another finding was that bridges with both pile foundations and continuous superstructures performed better against the destructive effect of flood. Considering that these bridges were built 40-50 years ago and were designed according to the 100 year flood, one can argue that bridges with pile foundations performed well. e. From the design perspective, it can be concluded that even though the use of shallow foundations can be adequate from a geotechnical point of view, in regions prone to flood risk, the use of pile foundations should be recommended unless it is proven that shallow foundations can protect the bridge piers from flood induced loads and effects.

Conclusions
On August 11, 2021, one of Turkey's most destructive flood disasters occurred in the Bozkurt and Ayancık regions in the Western Black Sea Basin. To contribute and shed light on the existing studies on flooding, this paper presents the observations and site assessments of the Bozkurt and Ayancık regions after the flood disaster. For this purpose, first, the flooded region and the flood event were explained along with the photographs taken during the two-day site visit. The causes, effects of flood, and damages of flood on infrastructures, bridges, retaining walls, and roads were discussed within an integrated framework involving hydraulics, geotechnics, structures, and transportation perspectives.
The following evaluations and recommendations were made for the design guidelines of the structures located in or nearby the floodplain. With the high risk for the structures in curvatures of meanders or in places where the original stream bed was located before stream rectification, the requirement that the 500 year flood flow should be considered in the design guidelines of critical infrastructures such as bridges, highways, and retaining walls are emphasized. The floodplain areas should not be allowed for settlement, or the design and implementation of flood-resistant structures should be guaranteed by the authorities. The retaining wall stability in urban areas should be provided at all costs because in case they fail, the neighboring buildings with inadequate foundation depth may suffer different levels of damage. The importance of piled foundations and continuous superstructures for better bridge performance against the destructive effect of the flood is clearly observed. Based on the findings of this paper, pile foundations in bridge piers should be considered for areas with high flood risk unless proven otherwise. Although the typical flood hazard microzonation studies consider the depth and the extent of the flood as the only damaging criteria, the damages observed in this flood showed that this should be accepted as the primary damage-generating phenomenon. The floods inherently involve significant secondary damage-making phenomena, such as high energy and momentum flux on neighboring structures and meanders, and these secondary effects should be incorporated into flood hazard assessment studies so that a comprehensive approach can be developed.