*4.5. Polish Perspective*

At the preliminary flood risk assessment stage and based on the Floods Directive Reporting Guidance 2018 to the European Commission, Poland can identify different types of floods e.g., fluvial and pluvial (from rivers or overland runoff), sea water (flooding of land by water from the sea, estuaries, or coastal lakes) and artificial water-bearing infrastructure (flooding of land by water arising from artificial, water-bearing infrastructure, or failure of such infrastructure). Floods resulting from blockages or restrictions may also be identified, a category which would include ice-jam floods. Other mechanisms that fall into this category include blockages of sewerage systems, restrictive channel structures such as bridges or culverts, and natural occurrences, such as landslides. ture, or failure of such infrastructure). Floods resulting from blockages or restrictions may also be identified, a category which would include ice-jam floods. Other mechanisms that fall into this category include blockages of sewerage systems, restrictive channel structures such as bridges or culverts, and natural occurrences, such as landslides.

At the preliminary flood risk assessment stage and based on the Floods Directive Reporting Guidance 2018 to the European Commission, Poland can identify different types of floods e.g., fluvial and pluvial (from rivers or overland runoff), sea water (flooding of land by water from the sea, estuaries, or coastal lakes) and artificial water-bearing infrastructure (flooding of land by water arising from artificial, water-bearing infrastruc-

flood events. "The stretch along the Oder Bruch, formally an inland delta drained for agricultural use, is particularly vulnerable due to its containment through dikes and the sediment accretion of the riverbed to elevations higher than the surrounding land. A catastrophic event of extended flooding throughout the adjacent low-lying area of the Oder Bruch occurred in March 1947, in which ice jams caused backwaters to overtop and breach dikes along the Oder Bruch at two locations, with breach widths of over 100 m. Flooding was extensive, leading to the evacuation of 20,000 people" [14]. The fact that ice jamming has become less frequent along the Oder River in recent decades, plus the advances in flood protection and ice defense measures, warning systems, and corresponding disaster control measures, have led to a lack of perception by the people of the dangers and risks

*Water* **2022**, *14*, x FOR PEER REVIEW 8 of 23

of ice-jam floods along the German Oder riverbanks.

*4.5. Polish Perspective* 

It is still uncertain if flood hazard and risk maps for ice jams will be developed for Poland in the future. Much depends on the results of the next preliminary flood risk assessment and the decisions of government authorities. Currently, the Institute of Meteorology and Water Management, a National Research Institute, is working on various aspects of flood protection, including mathematical modeling of ice jams and determining flood hazards from ice phenomena. Figure 6 shows the results of a preliminary study modelling flood hazard areas from ice jams along the test section of the Oder River. It is still uncertain if flood hazard and risk maps for ice jams will be developed for Poland in the future. Much depends on the results of the next preliminary flood risk assessment and the decisions of government authorities. Currently, the Institute of Meteorology and Water Management, a National Research Institute, is working on various aspects of flood protection, including mathematical modeling of ice jams and determining flood hazards from ice phenomena. Figure 6 shows the results of a preliminary study modelling flood hazard areas from ice jams along the test section of the Oder River.

In the upcoming publishing of the river basin management plan for Poland, locations of past ice jams have been identified and mapped for the Vistula River basin, shown in Figure 7. Concentrations of ice-jam locations are indicative of river stretches with a higher propensity for ice-jam flood hazards.

In regard to the changing climate in Poland, in the headwaters of the Vistula River, in the Carpathian Mountains in southern Poland, mean annual air temperatures at the Beskid Zywiecki station have increased by more than 2 ◦C over the past 40 years. Generally, increases in annual air temperature for stations in the upper Vistula River basin were at the rate of +0.13 ◦C per 10 years to +0.29 ◦C per 10 years (based on the period 1951–2015) [15]. Annual total precipitation has also increased, with an increasing trend of approximately 100 mm over the past 50 years. Lupikasza et al. [15] found trends in annual precipitation at ten stations located in the upper Vistula River basin to vary from −7.2 mm per 10 years up to 16.5 mm per 10 years for the period 1951–2015. The intensity of precipitation has also changed, with the number of days of precipitation totaling more than 5 and 10 mm/day increasing over the same time period. Pinskwar et al. [16] also found that, for the area of the upper Vistula River basin, the number of days with precipitation equal to or above 10 mm as well as 20 mm increased in the period 1991–2015 in comparison to 1961–1990. This has repercussions on the flows along the Vistula River and the degree of substances transported from the catchment area into the receiving waters. The more intense rainfalls

lead to a greater supply of eroded material to the rivers, exacerbated by the increased weathering of rocks and erosion due to rising air temperatures. The additional sediment transported in rivers can lead to increased accretion of the riverbed, particularly at the inlet of reservoirs, as is the case for the Wloclawek Reservoir showcased below in the Section "Ice characterization of a hanging dam". In the upcoming publishing of the river basin management plan for Poland, locations of past ice jams have been identified and mapped for the Vistula River basin, shown in Figure 7. Concentrations of ice-jam locations are indicative of river stretches with a higher propensity for ice-jam flood hazards.

*Water* **2022**, *14*, x FOR PEER REVIEW 9 of 23

**Figure 7.** Possible locations of past ice-jam events along the Vistula River and its tributaries (second update of the river basin management plans). **Figure 7.** Possible locations of past ice-jam events along the Vistula River and its tributaries (second update of the river basin management plans).

In regard to the changing climate in Poland, in the headwaters of the Vistula River, in the Carpathian Mountains in southern Poland, mean annual air temperatures at the Beskid Zywiecki station have increased by more than 2 °C over the past 40 years. Generally, increases in annual air temperature for stations in the upper Vistula River basin were at the rate of +0.13 °C per 10 years to +0.29 °C per 10 years (based on the period 1951–2015) [15]. Annual total precipitation has also increased, with an increasing trend of approximately 100 mm over the past 50 years. Lupikasza et al. [15] found trends in annual precipitation at ten stations located in the upper Vistula River basin to vary from −7.2 mm per 10 years up to 16.5 mm per 10 years for the period 1951–2015. The intensity of precipitation has also changed, with the number of days of precipitation totaling more than 5 and 10 mm/day increasing over the same time period. Pinskwar et al. [16] also found that, for the area of the upper Vistula River basin, the number of days with precipitation equal to or above 10 mm as well as 20 mm increased in the period 1991–2015 in comparison to 1961–1990. This has repercussions on the flows along the Vistula River and the degree of substances transported from the catchment area into the receiving waters. The more intense rainfalls lead to a greater supply of eroded material to the rivers, exacerbated by the increased weathering of rocks and erosion due to rising air temperatures. The additional sediment transported in rivers can lead to increased accretion of the riverbed, particularly Ice phenomena and phenology have also changed in rivers in Poland. Since 1960, ice phenomena generally appear on the Warta River in December or January of each winter, and the trend is that their occurrences have been delayed by approximately three days per decade at the gauge at Pozna ´n. The ice season generally ends in February or March, with the trend in the end dates occurring approximately four days earlier each decade. This leads to a progressive shortening of the ice season, as shown in Figure 8. The figure also reveals a significant increase in the trend in mean air temperatures measured in Pozna´n. An interesting correlation between air temperature and the duration of the ice season can be drawn between the two, as indicated in Figure 9. A suggestion was made at the workshop to model water temperature and apply a correction to the ice phenology due to urban heat islands. One of the workshop participants mentioned that most gauges in Poland, with long-term records of ice phenology and thicknesses, are situated in urban centres, which may result in steeper trends toward shorter ice durations and thinner ice nation-wide compared to potential trends due to climatic conditions alone. The additional heat may stem from effluents such as those from wastewater treatment plants or altered air temperatures. In particular air temperature changes should be tested since a large urban area may be required for a significant effect to occur on river ice. A hydrological modelling system with the capability of simulating river temperatures, e.g., the MESH-RBM

modelling system [17,18], could help determine biases in ice phenology and thicknesses when comparing "actual" water temperatures (due to climate change and urban heating) to "natural" water temperatures (due to climate change alone). Increased transport of dissolved substances, particularly from the application of fertilizers in the surrounding agricultural region, can also lead to a shortening of the ice season. elling system with the capability of simulating river temperatures, e.g., the MESH-RBM modelling system [17,18], could help determine biases in ice phenology and thicknesses when comparing "actual" water temperatures (due to climate change and urban heating) to "natural" water temperatures (due to climate change alone). Increased transport of dissolved substances, particularly from the application of fertilizers in the surrounding agricultural region, can also lead to a shortening of the ice season.

at the inlet of reservoirs, as is the case for the Wloclawek Reservoir showcased below in

Ice phenomena and phenology have also changed in rivers in Poland. Since 1960, ice phenomena generally appear on the Warta River in December or January of each winter, and the trend is that their occurrences have been delayed by approximately three days per decade at the gauge at Poznań. The ice season generally ends in February or March, with the trend in the end dates occurring approximately four days earlier each decade. This leads to a progressive shortening of the ice season, as shown in Figure 8. The figure also reveals a significant increase in the trend in mean air temperatures measured in Poznań. An interesting correlation between air temperature and the duration of the ice season can be drawn between the two, as indicated in Figure 9. A suggestion was made at the workshop to model water temperature and apply a correction to the ice phenology due to urban heat islands. One of the workshop participants mentioned that most gauges in Poland, with long-term records of ice phenology and thicknesses, are situated in urban centres, which may result in steeper trends toward shorter ice durations and thinner ice nation-wide compared to potential trends due to climatic conditions alone. The additional heat may stem from effluents such as those from wastewater treatment plants or altered air temperatures. In particular air temperature changes should be tested since a large urban area may be required for a significant effect to occur on river ice. A hydrological mod-

*Water* **2022**, *14*, x FOR PEER REVIEW 10 of 23

the Section "Ice characterization of a hanging dam".

**Figure 8.** Changes in the duration of ice phenomena on the Warta River in Poznań against the backdrop of the average air temperature in the cool half-year (November to April) in 1961–2020; 1—ice phenomena, 2—average air temperature in the cool half-year, 3—linear trend of ice phenomena in 1964–2020 and 4—linear trend of the average air temperature in the cool half-year in 1961–2020; n no data (source: data from IMWM-NRI and RWMB in Poznań). (data source: IMWM-NRI and RWMB in Poznań). **Figure 8.** Changes in the duration of ice phenomena on the Warta River in Pozna´n against the backdrop of the average air temperature in the cool half-year (November to April) in 1961–2020; 1—ice phenomena, 2—average air temperature in the cool half-year, 3—linear trend of ice phenomena in 1964–2020 and 4—linear trend of the average air temperature in the cool half-year in 1961–2020; n—no data (source: data from IMWM-NRI and RWMB in Pozna´n). (data source: IMWM-NRI and RWMB in Pozna ´n). *Water* **2022**, *14*, x FOR PEER REVIEW 11 of 23

**Figure 9.** Correlation between average air temperature in the cool half-year (November to April) and duration of the ice phenomena on the Warta River in Poznań for the time period 1964–2020 **Figure 9.** Correlation between average air temperature in the cool half-year (November to April) and duration of the ice phenomena on the Warta River in Pozna ´n for the time period 1964–2020.

**5. Ice-Jam Flood Risk Mitigation Measures**  *5.1. Artificial Ice-Cover Breakage*  In order to reduce flood hazards and risks due to ice jamming, ice breakers operate along major waterways to artificially break up ice covers (see Figure 10). The icebreaking operation on the Oder River along the Polish–German border is carried out jointly by the Polish and German waterways administrations. The technical management of the break-On the lower Vistula River, the duration of ice phenomena during the winters in the period 1960–2014 has also decreased [19]. The strongest negative trend was observed in the cross-section of the station situated immediately downstream of the river dam in Wloclawek, approximately −1.5 days/year. Negative trends of −1.64 to −1.97 days/year were also observed at other gauging stations. Negative trends were greater downstream of the Wloclawek Dam than upstream.

age operation of both icebreaker fleets (seven Polish and six German with two reserve icebreakers, one Polish and one German) is exercised by the Polish administration.

**Figure 10.** Icebreakers releasing an ice jam on the Oder River (source: RZGW, Szczecin).

Generally, a permanent ice cover develops first on Dabie Lake in Szczecin (see map in Figure 11), where frazil ice travelling down the Oder River accumulates and juxtapositions upstream along the Oder River's main stem and its tributaries, the Warta and Lusatian Neisse rivers. Icebreaking begins with crushing the permanent ice cover on Dabie Lake and freeing a gutter through the ice cover to make room for ice floes broken upstream along the river. Frontal icebreakers are directed upriver to break the consolidated ice cover along the river, while linear icebreakers crisscross Dabie Lake to prevent the

broken ice from stagnating and refreezing.

#### **5. Ice-Jam Flood Risk Mitigation Measures 5. Ice-Jam Flood Risk Mitigation Measures**

*Water* **2022**, *14*, x FOR PEER REVIEW 11 of 23

#### *5.1. Artificial Ice-Cover Breakage 5.1. Artificial Ice-Cover Breakage*

In order to reduce flood hazards and risks due to ice jamming, ice breakers operate along major waterways to artificially break up ice covers (see Figure 10). The icebreaking operation on the Oder River along the Polish–German border is carried out jointly by the Polish and German waterways administrations. The technical management of the breakage operation of both icebreaker fleets (seven Polish and six German with two reserve icebreakers, one Polish and one German) is exercised by the Polish administration. In order to reduce flood hazards and risks due to ice jamming, ice breakers operate along major waterways to artificially break up ice covers (see Figure 10). The icebreaking operation on the Oder River along the Polish–German border is carried out jointly by the Polish and German waterways administrations. The technical management of the breakage operation of both icebreaker fleets (seven Polish and six German with two reserve icebreakers, one Polish and one German) is exercised by the Polish administration.

**Figure 9.** Correlation between average air temperature in the cool half-year (November to April) and duration of the ice phenomena on the Warta River in Poznań for the time period 1964–2020

**Figure 10.** Icebreakers releasing an ice jam on the Oder River (source: RZGW, Szczecin). **Figure 10.** Icebreakers releasing an ice jam on the Oder River (source: RZGW, Szczecin).

Generally, a permanent ice cover develops first on Dabie Lake in Szczecin (see map in Figure 11), where frazil ice travelling down the Oder River accumulates and juxtapositions upstream along the Oder River's main stem and its tributaries, the Warta and Lusatian Neisse rivers. Icebreaking begins with crushing the permanent ice cover on Dabie Lake and freeing a gutter through the ice cover to make room for ice floes broken upstream along the river. Frontal icebreakers are directed upriver to break the consolidated ice cover along the river, while linear icebreakers crisscross Dabie Lake to prevent the broken ice from stagnating and refreezing. Generally, a permanent ice cover develops first on Dabie Lake in Szczecin (see map in Figure 11), where frazil ice travelling down the Oder River accumulates and juxtapositions upstream along the Oder River's main stem and its tributaries, the Warta and Lusatian Neisse rivers. Icebreaking begins with crushing the permanent ice cover on Dabie Lake and freeing a gutter through the ice cover to make room for ice floes broken upstream along the river. Frontal icebreakers are directed upriver to break the consolidated ice cover along the river, while linear icebreakers crisscross Dabie Lake to prevent the broken ice from stagnating and refreezing. *Water* **2022**, *14*, x FOR PEER REVIEW 12 of 23

**Figure 11.** Main stem of the Oder River and its major tributaries (drawn by the first author). **Figure 11.** Main stem of the Oder River and its major tributaries (drawn by the first author).

Icebreaking on the Oder River continues upstream towards the mouth of the Warta River, where the resources are split, with the larger part of the icebreaker fleet continuing Icebreaking on the Oder River continues upstream towards the mouth of the Warta River, where the resources are split, with the larger part of the icebreaker fleet continuing

breakage along the Oder River towards the Lusatian Neisse River mouth and the remain-

amount of ice floes flowing from the upper sections of the Oder River and its tributaries to the lower reach of the Oder River to create ice jams and, thus, risk inducing a flood

Operational flood warning relies on river gauge monitoring. The federal state of Brandenburg, Germany defines four different flood alert levels, which are specified for representative river gauges used in flood reporting services, see Figure 12. An alert level is proclaimed when water levels exceed a certain alert stage and local authorities must take action. The alert levels require increasing operational flood defense actions with increasing water stages [20]: alert level AI—water level reporting service (German: *Meldebeginn*), AII—control service at flood defense infrastructure such as dikes (German: *Kontrolldienst*), AIII—guard duty (German: *Wachdienst*), and AIV—civil protection (German: *Hochwasserabwehr*). These four alert levels allow quick assessment of the potential severity of a flood across different rivers in Brandenburg. The gauges used for the alert level system require high reliability and redundancy of sensors and communication net-

**Figure 12.** Alert levels for flood warnings (source: Landesamt für Umwelt Brandenburg).

artificially.

works.

*5.2. Flood Warning under Ice Conditions* 

breakage along the Oder River towards the Lusatian Neisse River mouth and the remaining icebreaker force working its way up the Warta River to the Notec River mouth. Caution must be taken not to begin breakage operations too early so as not to create a large amount of ice floes flowing from the upper sections of the Oder River and its tributaries to the lower reach of the Oder River to create ice jams and, thus, risk inducing a flood artificially. tion must be taken not to begin breakage operations too early so as not to create a large amount of ice floes flowing from the upper sections of the Oder River and its tributaries to the lower reach of the Oder River to create ice jams and, thus, risk inducing a flood artificially. *5.2. Flood Warning under Ice Conditions* 

Icebreaking on the Oder River continues upstream towards the mouth of the Warta River, where the resources are split, with the larger part of the icebreaker fleet continuing breakage along the Oder River towards the Lusatian Neisse River mouth and the remaining icebreaker force working its way up the Warta River to the Notec River mouth. Cau-

**Figure 11.** Main stem of the Oder River and its major tributaries (drawn by the first author).

Wroclaw

Poznan

0 100 200 km

**N**

#### *5.2. Flood Warning under Ice Conditions* Operational flood warning relies on river gauge monitoring. The federal state of

*Water* **2022**, *14*, x FOR PEER REVIEW 12 of 23

Dąbie Lake

Baltic Sea

Szczecin

Ratzdorf

Hohensaaten-Finow Kienitz

Operational flood warning relies on river gauge monitoring. The federal state of Brandenburg, Germany defines four different flood alert levels, which are specified for representative river gauges used in flood reporting services, see Figure 12. An alert level is proclaimed when water levels exceed a certain alert stage and local authorities must take action. The alert levels require increasing operational flood defense actions with increasing water stages [20]: alert level AI—water level reporting service (German: *Meldebeginn*), AII—control service at flood defense infrastructure such as dikes (German: *Kontrolldienst*), AIII—guard duty (German: *Wachdienst*), and AIV—civil protection (German: *Hochwasserabwehr*). These four alert levels allow quick assessment of the potential severity of a flood across different rivers in Brandenburg. The gauges used for the alert level system require high reliability and redundancy of sensors and communication networks. Brandenburg, Germany defines four different flood alert levels, which are specified for representative river gauges used in flood reporting services, see Figure 12. An alert level is proclaimed when water levels exceed a certain alert stage and local authorities must take action. The alert levels require increasing operational flood defense actions with increasing water stages [20]: alert level AI—water level reporting service (German: *Meldebeginn*), AII—control service at flood defense infrastructure such as dikes (German: *Kontrolldienst*), AIII—guard duty (German: *Wachdienst*), and AIV—civil protection (German: *Hochwasserabwehr*). These four alert levels allow quick assessment of the potential severity of a flood across different rivers in Brandenburg. The gauges used for the alert level system require high reliability and redundancy of sensors and communication networks.

**Figure 12.** Alert levels for flood warnings (source: Landesamt für Umwelt Brandenburg). **Figure 12.** Alert levels for flood warnings (source: Landesamt für Umwelt Brandenburg).

Flood alert gauges are linked to specific sections of a river which are defined according to the local flood risk and hydraulic conditions, as well as the appropriate administrative units. Figure 13 highlights the river sections used for flood alerts at the lower Oder River in Brandenburg, with panel A showing the sections under normal flood conditions. Since ice-jam flood dynamics are completely different at the lower Oder River, with ice-jams moving upstream, two important adjustments have been implemented. First, river sections upstream of the gauges are used for warning (Figure 13B) and lower alert stages are defined, since ice-jams typically lead to damaged dikes and abrupt rises in water levels are expected. With these adjustments the flood warning system is more representative of ice-flood conditions.

#### *5.3. Ice-Jam Flood Forecasting*

Ice-jam flood forecasting is a key component in any flood management plan to reduce flood hazards and risks. Advances have been made in the development of ice-jam flood forecasting methodologies and systems, particularly for the Athabasca River at Fort McMurray, Alberta [21], the upper reaches of the Saint John River, New Brunswick [22], and the Sanhuhekou bend of the Yellow River in China [23]. These methodologies and systems have also been implemented successfully in ice-jam flood forecasting systems for operational use by the government of Newfoundland and Labrador for the lower (Atlantic) Churchill River [24,25] and by the government of Manitoba for the lower Red River in Manitoba [26]. Requirements for an operational ice-jam flood forecasting system for the Oder River have been laid out in [1] and the need to include such methodologies for ice conditions can be seen in Figure 14, which shows a rapid rise in the backwater levels at Hohensaaten-Finow (see map in Figure 11 for the location) caused by an ice jam downstream of that gauge in February 2021. Forecasts on the rising limb of the event grossly underestimated the water

levels attained by the ice jamming since no river ice processes are integrated in the current hydraulic model used for operational forecasting. The roughness coefficients and the rating curves implemented in the model also require updating to reflect ice-jam backwater effects. An ice-jam hydraulic model has been set up for the Oder River [27] between Ratzdorf and Kienitz (see Figure 11 for locations) and needs to be extended to Dabie Lake to include ice-jam backwater effects at Hohensaaten-Finow. Since ice-jam flood dynamics are completely different at the lower Oder River, with icejams moving upstream, two important adjustments have been implemented. First, river sections upstream of the gauges are used for warning (Figure 13B) and lower alert stages are defined, since ice-jams typically lead to damaged dikes and abrupt rises in water levels are expected. With these adjustments the flood warning system is more representative of ice-flood conditions.

Flood alert gauges are linked to specific sections of a river which are defined according to the local flood risk and hydraulic conditions, as well as the appropriate administrative units. Figure 13 highlights the river sections used for flood alerts at the lower Oder River in Brandenburg, with panel A showing the sections under normal flood conditions.

*Water* **2022**, *14*, x FOR PEER REVIEW 13 of 23

**Figure 13.** Tracking an ice-jam cover progression using the alert level system, which links gauge readings to upstream river reaches (source: Landesamt für Umwelt Brandenburg) with (**A**) for normal flood conditions and (**B**) for ice-jam conditions. **Figure 13.** Tracking an ice-jam cover progression using the alert level system, which links gauge readings to upstream river reaches (source: Landesamt für Umwelt Brandenburg) with (**A**) for normal flood conditions and (**B**) for ice-jam conditions. *Water* **2022**, *14*, x FOR PEER REVIEW 14 of 23

Oder River have been laid out in [1] and the need to include such methodologies for ice conditions can be seen in Figure 14, which shows a rapid rise in the backwater levels at Hohensaaten-Finow (see map in Figure 11 for the location) caused by an ice jam down-**Figure 14.** Measurement and forecast of water levels at the gauge at Hohensaaten-Finow, February 2021 (source: Landesamt für Umwelt Brandenburg). **Figure 14.** Measurement and forecast of water levels at the gauge at Hohensaaten-Finow, February 2021 (source: Landesamt für Umwelt Brandenburg).

#### stream of that gauge in February 2021. Forecasts on the rising limb of the event grossly **6. Technical Advances in River Ice Research 6. Technical Advances in River Ice Research**

#### underestimated the water levels attained by the ice jamming since no river ice processes *6.1. Particle Tracking Velocimetry*

are integrated in the current hydraulic model used for operational forecasting. The roughness coefficients and the rating curves implemented in the model also require updating to reflect ice-jam backwater effects. An ice-jam hydraulic model has been set up for the Oder River [27] between Ratzdorf and Kienitz (see Figure 11 for locations) and needs to be extended to Dabie Lake to include ice-jam backwater effects at Hohensaaten-Finow. *6.1. Particle Tracking Velocimetry*  A novelty presented at the workshop is determining flow velocities of ice using particle tracking velocimetry, which allows the velocities and trajectories of ice floes to be measured remotely. The method tracks the flow velocities of many ice floes using a sequence of images. It includes measuring the camera position and orientation (camera pose), automatic extraction of the water area for feature searching, particle detection and filtering, particle tracking and filtering, and scaling the tracks of (ice floe) velocities [28]. Figure 15 shows such trajectories with velocities of ice floes across the west channel of the A novelty presented at the workshop is determining flow velocities of ice using particle tracking velocimetry, which allows the velocities and trajectories of ice floes to be measured remotely. The method tracks the flow velocities of many ice floes using a sequence of images. It includes measuring the camera position and orientation (camera pose), automatic extraction of the water area for feature searching, particle detection and filtering, particle tracking and filtering, and scaling the tracks of (ice floe) velocities [28]. Figure 15 shows such trajectories with velocities of ice floes across the west channel of the lower Oder River during the February 2021 ice-jam event.

**Figure 15.** Trajectory and flow velocities of ice floes from the ice-jam event of February 2021 in the

In February 2021, a severe ice jam occurred in the upstream end of the Wloclawek Reservoir on the Vistula River near Plock [29]. This area is one of the most ice-jam prone river stretches in Poland. Evidence points to sediment accretion at the reservoir inlet as the reason for the increase in this area's propensity to ice jamming, indicated by a shift in the rating curve over time between the years 2009 and 2020. Efforts to break up the ice jam with ice breakers were hindered by the shallow depth of the reservoir inlet. Dredging works have been cut back in recent years even though the intensity of sedimentation has increased due to greater transport of sediment from the upstream catchment area, with an accretion rate of approximately 5 cm/year downstream of Plock. Despite the shortening of the duration of the ice season, the ice-jam flood risk of this reservoir section remains

west channel of the lower Oder River (source: Technische Hochschule Nürnberg).

*6.2. Ice Characterization of a Hanging Dam*

high.

lower Oder River during the February 2021 ice-jam event.

**Figure 15.** Trajectory and flow velocities of ice floes from the ice-jam event of February 2021 in the west channel of the lower Oder River (source: Technische Hochschule Nürnberg). **Figure 15.** Trajectory and flow velocities of ice floes from the ice-jam event of February 2021 in the west channel of the lower Oder River (source: Technische Hochschule Nürnberg).

**Figure 14.** Measurement and forecast of water levels at the gauge at Hohensaaten-Finow, February

A novelty presented at the workshop is determining flow velocities of ice using particle tracking velocimetry, which allows the velocities and trajectories of ice floes to be measured remotely. The method tracks the flow velocities of many ice floes using a sequence of images. It includes measuring the camera position and orientation (camera pose), automatic extraction of the water area for feature searching, particle detection and filtering, particle tracking and filtering, and scaling the tracks of (ice floe) velocities [28]. Figure 15 shows such trajectories with velocities of ice floes across the west channel of the

#### *6.2. Ice Characterization of a Hanging Dam 6.2. Ice Characterization of a Hanging Dam*

2021 (source: Landesamt für Umwelt Brandenburg).

**6. Technical Advances in River Ice Research** 

lower Oder River during the February 2021 ice-jam event.

*6.1. Particle Tracking Velocimetry* 

In February 2021, a severe ice jam occurred in the upstream end of the Wloclawek Reservoir on the Vistula River near Plock [29]. This area is one of the most ice-jam prone river stretches in Poland. Evidence points to sediment accretion at the reservoir inlet as the reason for the increase in this area's propensity to ice jamming, indicated by a shift in the rating curve over time between the years 2009 and 2020. Efforts to break up the ice jam with ice breakers were hindered by the shallow depth of the reservoir inlet. Dredging works have been cut back in recent years even though the intensity of sedimentation has increased due to greater transport of sediment from the upstream catchment area, with an accretion rate of approximately 5 cm/year downstream of Plock. Despite the shortening of the duration of the ice season, the ice-jam flood risk of this reservoir section remains In February 2021, a severe ice jam occurred in the upstream end of the Wloclawek Reservoir on the Vistula River near Plock [29]. This area is one of the most ice-jam prone river stretches in Poland. Evidence points to sediment accretion at the reservoir inlet as the reason for the increase in this area's propensity to ice jamming, indicated by a shift in the rating curve over time between the years 2009 and 2020. Efforts to break up the ice jam with ice breakers were hindered by the shallow depth of the reservoir inlet. Dredging works have been cut back in recent years even though the intensity of sedimentation has increased due to greater transport of sediment from the upstream catchment area, with an accretion rate of approximately 5 cm/year downstream of Plock. Despite the shortening of the duration of the ice season, the ice-jam flood risk of this reservoir section remains high.

high. To determine the volume of ice in the hanging dam that caused the jamming and the thickness of the hanging dam in relation to the water depth, cross-sections of the ice with depth were surveyed at the upstream end of the hanging dam using a sounding device (weight) to penetrate the hanging dam ice. During these frazil slush penetration tests, changes in the compactness of the ice deposits constituting the hanging dam were also recorded to determine the amount of ice grounded on the reservoir bottom during the ice-jam event. Three grades of compactness were classified:


The cross-section of the depths of the hanging dam is shown in Figure 16, indicating a decrease in the ice compactness with depth. A water layer was still evident between the bottom of the hanging dam and the reservoir bottom (no grounding); however, the hanging dam did fill a substantial percentage of the cross-sectional flow area. Flow velocity on the right side of the cross-section would have been greater where loose slush ice did not deposit, whereas on the left side, flow velocities would have been less, allowing frazil ice to be deposited on the underside of the hanging dam.

#### *6.3. Design of New Ice-Control Structure*

Ice jams are initiated when ice transport conveyance is reduced locally along a river stretch, particularly in meanders or in areas where the river narrows and obstacles are present in the river (e.g., islands, bridge piers, and sand bars) [30]. A continuous high inflow of ice from upstream can also help in the initiation process of ice jams. High volumes of inflowing ice can lead to increases in ice thicknesses constricting the cross-sectional flow area, resulting in the impediment of discharge under the ice jam with an increase in water surface elevations in the section upstream of the jam. One means of reducing the

influx of ice in an ice-jam prone area is to arrest the flow of ice upstream of a potential ice-jam location using an ice-control structure (ICS), shown in Figure 17 [31]. The structure mostly impedes the ice transport further downstream but not the flow of the water, which is allowed to bypass the ice accumulation. A transverse set of piers only partially spans across the channel from one bank; the piers then extend longitudinally upstream parallel to the bank to form a side channel between the longitudinal set of piers and the bank. This side channel provides a passage of water to flow around the accumulation of ice which is held back by the piers. With this design, additional space in an adjacent floodplain to bypass water around the ice accumulation is not required. • loose accumulations—sounder penetrates the accumulation driven by its own weight. The cross-section of the depths of the hanging dam is shown in Figure 16, indicating a decrease in the ice compactness with depth. A water layer was still evident between the bottom of the hanging dam and the reservoir bottom (no grounding); however, the hanging dam did fill a substantial percentage of the cross-sectional flow area. Flow velocity on the right side of the cross-section would have been greater where loose slush ice did not deposit, whereas on the left side, flow velocities would have been less, allowing frazil ice to be deposited on the underside of the hanging dam.

To determine the volume of ice in the hanging dam that caused the jamming and the thickness of the hanging dam in relation to the water depth, cross-sections of the ice with depth were surveyed at the upstream end of the hanging dam using a sounding device (weight) to penetrate the hanging dam ice. During these frazil slush penetration tests, changes in the compactness of the ice deposits constituting the hanging dam were also recorded to determine the amount of ice grounded on the reservoir bottom during the ice-

*Water* **2022**, *14*, x FOR PEER REVIEW 15 of 23

jam event. Three grades of compactness were classified:

• firm accumulations—sounder must be driven into the ice by force. • compact accumulations—sounder remains stationary within the slush.

**Figure 16.** Cross-section of the hanging dam formed at the Wloclawek Reservoir inlet near Plock in February 2021 [29]. **Figure 16.** Cross-section of the hanging dam formed at the Wloclawek Reservoir inlet near Plock in February 2021 [29]. *Water* **2022**, *14*, x FOR PEER REVIEW 16 of 23

water, which is allowed to bypass the ice accumulation. A transverse set of piers only partially spans across the channel from one bank; the piers then extend longitudinally upstream parallel to the bank to form a side channel between the longitudinal set of piers **Figure 17.** Ice-control structure (photo courtesy of US Army Cold Regions Research and Engineering Laboratory, Hanover, New Hampshire; used with permission). **Figure 17.** Ice-control structure (photo courtesy of US Army Cold Regions Research and Engineering Laboratory, Hanover, New Hampshire; used with permission).

#### and the bank. This side channel provides a passage of water to flow around the accumu-*6.4. River Ice Detection Using Optical and Radar Satellite Imagery in Tandem 6.4. River Ice Detection Using Optical and Radar Satellite Imagery in Tandem*

lation of ice which is held back by the piers. With this design, additional space in an adjacent floodplain to bypass water around the ice accumulation is not required. The aim of river ice monitoring using satellite data is to (i) provide spatially constant, frequent information about the presence of ice along a river course, (ii) provide imagery support for water management services, and (iii) detect possible threats and natural disasters provoked by ice jams. The two satellite sensors used in this study were: The aim of river ice monitoring using satellite data is to (i) provide spatially constant, frequent information about the presence of ice along a river course, (ii) provide imagery support for water management services, and (iii) detect possible threats and natural disasters provoked by ice jams. The two satellite sensors used in this study were:


indices calculations to strengthen the classification of the desired coverage classes (water, snow or ice, vegetation, and bare soil) before a composite of the two images was created. There are some limitations with both sensors. For example, misclassifications may occur between smooth black ice covers and open water. Misclassifications are also possible when predefined waterbeds are changed due to changing water levels. For Sentinel-2 imagery, misclassifications may also occur when differentiating between smooth black ice and open water. Image areas can also be misclassified as turbid waters or waters having

Figure 18 provides a combined product of the Sentinel-1 and Sentinel-2 images of the west and east channels of the lower Oder River. Sentinel-1 images were calibrated, speckle

The high-resolution sensors from the Sentinel family proved very useful in detecting ice coverage, including ice jams. The Sentinel-2 optical sensor offers better thematic accuracy and should be treated as a primary source for ice classification, whereas the Sentinel-1 radar sensor offers the possibility of providing observations even under cloudy conditions and can be used as a secondary or auxiliary source for ice classification. Ice jams are well presented in the images from both sensors. Success in these composite images have led to the creation of the high-resolution snow and ice monitoring service, developed be-

an algal bloom. Clouds will also hamper Sentinel-2 image clarity.

tween 2019 and 2021 under the auspices of the Europe Environmental Agency

Figure 18 provides a combined product of the Sentinel-1 and Sentinel-2 images of the west and east channels of the lower Oder River. Sentinel-1 images were calibrated, speckle filtered, terrain corrected, and rescaled to dB, whereas the Sentinel-2 images underwent atmospheric correction, resampling to consistent spatial resolution of 20 m, and spectral indices calculations to strengthen the classification of the desired coverage classes (water, snow or ice, vegetation, and bare soil) before a composite of the two images was created. There are some limitations with both sensors. For example, misclassifications may occur between smooth black ice covers and open water. Misclassifications are also possible when predefined waterbeds are changed due to changing water levels. For Sentinel-2 imagery, misclassifications may also occur when differentiating between smooth black ice and open water. Image areas can also be misclassified as turbid waters or waters having an algal bloom. Clouds will also hamper Sentinel-2 image clarity. *Water* **2022**, *14*, x FOR PEER REVIEW 17 of 23

**Figure 18.** The combined product from Sentinel-1 and Sentinel-2 based classifications. Images credit: https://land.copernicus.eu/pan-european/biophysical-parameters/high-resolution-snow-and-icemonitoring; images acquired on 16 February 2017. **Figure 18.** The combined product from Sentinel-1 and Sentinel-2 based classifications. Images credit: https://land.copernicus.eu/pan-european/biophysical-parameters/high-resolution-snowand-ice-monitoring; images acquired on 16 February 2017.

*6.5. Stochastic Model to Assess Ice-Jam Flood Hazards*  The chaotic nature of ice-jam formation and flooding can be captured using a stochastic modelling approach. In this approach, a deterministic river ice hydraulics model runs many times, with each run having a different set of input values for the parameters and boundary conditions. These values are randomly selected from frequency distributions of each parameter and boundary condition. This results in an ensemble of possible ice-jamming outcomes along a river reach of interest. Such an approach has been newly developed to quantify ice-jam flood hazards and risks [32,33]. Figure 19 conceptualizes The high-resolution sensors from the Sentinel family proved very useful in detecting ice coverage, including ice jams. The Sentinel-2 optical sensor offers better thematic accuracy and should be treated as a primary source for ice classification, whereas the Sentinel-1 radar sensor offers the possibility of providing observations even under cloudy conditions and can be used as a secondary or auxiliary source for ice classification. Ice jams are well presented in the images from both sensors. Success in these composite images have led to the creation of the high-resolution snow and ice monitoring service, developed between 2019 and 2021 under the auspices of the Europe Environmental Agency

the approach, which requires frequency distributions of the boundary conditions (shown at the top of Figure 19) and parameter values (not shown) to be input to the deterministic river ice hydraulic model RIVICE (see [34,35]) for model descriptions). The boundary con-

• upstream water flow Q (Figure 19a) represented by an extreme-value distribution of

• volume of inflowing ice accumulating in the ice jam Vice (Figure 19b), which is a function of Q, with the scatter represented by a confidence band within which ran-

• downstream water level W (Figure 19c), which is a function of the upstream dis-

• Location of the ice-jam lodgment x (Figure 19d), which is represented by a stepped uniform distribution to capture the predisposition of ice jamming in some stretches

the flows at instantaneous water level maxima during ice-jam events,

dom variables are selected,

ditions include:

charge.

over others.

#### *6.5. Stochastic Model to Assess Ice-Jam Flood Hazards*

The chaotic nature of ice-jam formation and flooding can be captured using a stochastic modelling approach. In this approach, a deterministic river ice hydraulics model runs many times, with each run having a different set of input values for the parameters and boundary conditions. These values are randomly selected from frequency distributions of each parameter and boundary condition. This results in an ensemble of possible icejamming outcomes along a river reach of interest. Such an approach has been newly developed to quantify ice-jam flood hazards and risks [32,33]. Figure 19 conceptualizes the approach, which requires frequency distributions of the boundary conditions (shown at the top of Figure 19) and parameter values (not shown) to be input to the deterministic river ice hydraulic model RIVICE (see [34,35]) for model descriptions). The boundary conditions include:


**Figure 19.** Conceptualization of the stochastic modelling framework for ice-jam flood hazard assessment; explanations for each subfigure are provided in the main text (drawn by the first author). **Figure 19.** Conceptualization of the stochastic modelling framework for ice-jam flood hazard assessment; explanations for each subfigure are provided in the main text (drawn by the first author).

Parameters are generally uniform distributions between minimum and maximum values determined through calibration. Parameters are generally uniform distributions between minimum and maximum values determined through calibration.

Using a Monte Carlo approach (Figure 19e), RIVICE runs hundreds of times, with each simulation having a different set of boundary conditions and parameter values chosen randomly from all distributions. One output is an ensemble of backwater profiles (Figure 19f), the results of which can be compiled within a probabilistic context using percentile profiles of exceedance probabilities (Figure 19g). The water level elevations at a gauge can then be compared to the annual exceedance probabilities of the levels recorded at the gauge (Figure 19h). Discrepancies between the simulated and recorded exceedance probabilities can be reduced by adjusting the percentage of the confidence band in the Q vs. Using a Monte Carlo approach (Figure 19e), RIVICE runs hundreds of times, with each simulation having a different set of boundary conditions and parameter values chosen randomly from all distributions. One output is an ensemble of backwater profiles (Figure 19f), the results of which can be compiled within a probabilistic context using percentile profiles of exceedance probabilities (Figure 19g). The water level elevations at a gauge can then be compared to the annual exceedance probabilities of the levels recorded at the gauge (Figure 19h). Discrepancies between the simulated and recorded exceedance probabilities can be reduced by adjusting the percentage of the confidence band in the Q vs.

Vice relationship with the processes, including the Monte Carlo simulations, having to be

breaker as described in the section "Artificial ice-cover breakage" above, can be implemented as an option to mitigate ice-jam flood hazards and risks. This scenario can be simulated within the stochastic modelling framework by removing those stretches in the lodgment location distribution (Figure 20D). It is assumed that ice cannot lodge to form an ice jam in areas that have been artificially broken. Rerunning the Monte Carlo analysis should lead to a change in the elevations of the percentile profiles of the backwater level

At the workshop, research was also presented on the testing of different solutions applied to probability analyses of ice jams and ice covers. Time series of the maximum water stages were determined using: (i) extreme-value approaches using annual maximum water staging [36], and (ii) the peak-over-threshold (POT) method. The main idea of the POT method is to prepare a series of maxima on the basis of all events occurring in the analyzed time period that exceed an assumed threshold value, the so-called cut-off level [37]. The application of the POT method makes it possible to include, within the time series, the fact that ice phenomena may occur several times in one year, and, within the empirical probability, the fact that ice phenomena do not occur every year [38]. The final probability of exceedance for the POT time series was determined using the following tested distributions: log-normal, Gumbel, Pearson III, Gamma, log-Gamma, and Pareto.

repeated.

ensemble.

Vice relationship with the processes, including the Monte Carlo simulations, having to be repeated.

Referring to Figure 20, artificially breaking up the ice cover, for example using an ice breaker as described in the section "Artificial ice-cover breakage" above, can be implemented as an option to mitigate ice-jam flood hazards and risks. This scenario can be simulated within the stochastic modelling framework by removing those stretches in the lodgment location distribution (Figure 20D). It is assumed that ice cannot lodge to form an ice jam in areas that have been artificially broken. Rerunning the Monte Carlo analysis should lead to a change in the elevations of the percentile profiles of the backwater level ensemble. *Water* **2022**, *14*, x FOR PEER REVIEW 19 of 23

**Figure 20.** Adjustment in the boundary condition frequency distribution within the stochastic modelling framework when considering mitigation options, such as artificially breaking the ice cover to hinder ice-jam lodgments; explanations for each subfigure are provided in the main text (drawn by the first author). **Figure 20.** Adjustment in the boundary condition frequency distribution within the stochastic modelling framework when considering mitigation options, such as artificially breaking the ice cover to hinder ice-jam lodgments; explanations for each subfigure are provided in the main text (drawn by the first author).

*6.6. Monitoring with UAVs*  Growing access to unmanned aerial vehicles (UAVs), also known as drones, opens new possibilities to observe ice on rivers and reservoirs. Even low-cost UAVs are now capable of taking nadir photographs with predefined frontal and side overlap. Such images are standard input data for the structure-from-motion (SfM) algorithm, the products of which are dense point clouds and the resulting digital surface models as well as orthophotomaps. Recently, this popular approach has been adopted to determine the spatial extent of snow [39] or ice [40] as well as to reconstruct snow depth [41] or ice thickness [40]. Figure 21 presents aerial imagery of two frozen reservoirs in the Izerskie Mountains (southwest Poland) as well as two fragments of the unfrozen Oder River (west Poland). It At the workshop, research was also presented on the testing of different solutions applied to probability analyses of ice jams and ice covers. Time series of the maximum water stages were determined using: (i) extreme-value approaches using annual maximum water staging [36], and (ii) the peak-over-threshold (POT) method. The main idea of the POT method is to prepare a series of maxima on the basis of all events occurring in the analyzed time period that exceed an assumed threshold value, the so-called cut-off level [37]. The application of the POT method makes it possible to include, within the time series, the fact that ice phenomena may occur several times in one year, and, within the empirical probability, the fact that ice phenomena do not occur every year [38]. The final probability of exceedance for the POT time series was determined using the following tested distributions: log-normal, Gumbel, Pearson III, Gamma, log-Gamma, and Pareto.

#### is apparent from the figure that the visual analysis of imagery leads to the differentiation *6.6. Monitoring with UAVs*

snow-melt floods, or ice jams.

**7. Conclusions** 

between frozen (top row in Figure 21) and unfrozen water (bottom row in Figure 21), enabling the detection of the presence of ice. Additionally, it is simple to discriminate between spatially uneven ice cover on water reservoirs in Rozdroże Izerskie and Polana Izerska and snow-covered banks or bare land. The knowledge about snow depth and extent and ice thickness and extent, acquired just after collecting UAV data within the concept of rapid mapping [42], may be useful, for instance, to assess the risks of avalanches, Growing access to unmanned aerial vehicles (UAVs), also known as drones, opens new possibilities to observe ice on rivers and reservoirs. Even low-cost UAVs are now capable of taking nadir photographs with predefined frontal and side overlap. Such images are standard input data for the structure-from-motion (SfM) algorithm, the products of which are dense point clouds and the resulting digital surface models as well as orthopho-

involved in the field of river ice in their work and research. The venue provided an opportunity to present ideas and exchange knowledge in the field of ice-jam flood hazards and risks and how the subject was approached and applied in each of the EU member's countries. One key takeaway message from the workshop was that ice-jam floods are important components in the flood hazard and risk assessment and should be catalogued in the Flood Management Plans of the EU Floods Directive deliverables, but ice-jam floods do not need to be explicitly expressed within the directive itself. This may be partially due

tomaps. Recently, this popular approach has been adopted to determine the spatial extent of snow [39] or ice [40] as well as to reconstruct snow depth [41] or ice thickness [40]. *Water* **2022**, *14*, x FOR PEER REVIEW 20 of 23

Figure 21 presents aerial imagery of two frozen reservoirs in the Izerskie Mountains (southwest Poland) as well as two fragments of the unfrozen Oder River (west Poland). It is apparent from the figure that the visual analysis of imagery leads to the differentiation between frozen (top row in Figure 21) and unfrozen water (bottom row in Figure 21), enabling the detection of the presence of ice. Additionally, it is simple to discriminate between spatially uneven ice cover on water reservoirs in Rozdroze Izerskie and Polana ˙ Izerska and snow-covered banks or bare land. The knowledge about snow depth and extent and ice thickness and extent, acquired just after collecting UAV data within the concept of rapid mapping [42], may be useful, for instance, to assess the risks of avalanches, snow-melt floods, or ice jams. to the fact that, in rivers of northern and eastern countries, which are members of the European Union, the floodwater levels for a certain annual exceedance probability (or return periods) from ice jamming is generally lower than for those of open-water floods. A more comprehensive examination of other rivers is required for this statement to be thoroughly conclusive. Depending on the region, ice-jam flood hazards and risks have different foci, for example, hydropower operations in Norway and shipping navigation in Poland and Germany.

**Figure 21.** Fragment of single aerial image taken in central projection by a UAV over Rozdroże Izerskie in southwest Poland showing two interconnected frozen reservoirs (**top left**), fragment of the SfM-based orthophotomap of Polana Izerska in southwest Poland centered on a frozen reservoir (**top right**), fragments of the SfM-based orthophotomaps of the Oder River in west Poland in Pomorsko (**bottom left**) and Tarnawa (**bottom right**) showing an unfrozen river channel (source: Department of Geoinformatics and Cartography, University of Wrocław). **Figure 21.** Fragment of single aerial image taken in central projection by a UAV over Rozdroze˙ Izerskie in southwest Poland showing two interconnected frozen reservoirs (**top left**), fragment of the SfM-based orthophotomap of Polana Izerska in southwest Poland centered on a frozen reservoir (**top right**), fragments of the SfM-based orthophotomaps of the Oder River in west Poland in Pomorsko (**bottom left**) and Tarnawa (**bottom right**) showing an unfrozen river channel (source: Department of Geoinformatics and Cartography, University of Wrocław).

#### At the workshop, many new advances were also presented on monitoring and miti-**7. Conclusions**

gating ice-jam flood hazards and risks, including application with particle tracking velocimetry, hanging dam characterization, ice-control structure design, remote sensing of ice covers, and modelling ice-jam flood hazards and hazard reductions. As an outlook, areas that need research furtherance include: • safely measuring flows under covers of loose ice accumulations, • incorporating near-ground (trail cameras), aerial (drones), and space-borne (satellites) remote sensing imagery into integrated monitoring systems for quick response to ice-jam flood hazard developments, and • real-time monitoring of ice-cover elevations as a proxy for ice-thickness measurements. **8. Outlook**  The workshop brought together many scientists and government officials who were involved in the field of river ice in their work and research. The venue provided an opportunity to present ideas and exchange knowledge in the field of ice-jam flood hazards and risks and how the subject was approached and applied in each of the EU member's countries. One key takeaway message from the workshop was that ice-jam floods are important components in the flood hazard and risk assessment and should be catalogued in the Flood Management Plans of the EU Floods Directive deliverables, but ice-jam floods do not need to be explicitly expressed within the directive itself. This may be partially due to the fact that, in rivers of northern and eastern countries, which are members of the European Union, the floodwater levels for a certain annual exceedance probability (or return periods) from ice jamming is generally lower than for those of open-water floods. A more comprehensive examination of other rivers is required for this statement to be

The workshop focused more on the technical aspects of flood risk assessment. A follow-up workshop could include the social aspects of ice-jam flood risk [43], for example

carried out with agent-based modelling (ABM) to incorporate both technical and social

thoroughly conclusive. Depending on the region, ice-jam flood hazards and risks have different foci, for example, hydropower operations in Norway and shipping navigation in Poland and Germany.

At the workshop, many new advances were also presented on monitoring and mitigating ice-jam flood hazards and risks, including application with particle tracking velocimetry, hanging dam characterization, ice-control structure design, remote sensing of ice covers, and modelling ice-jam flood hazards and hazard reductions. As an outlook, areas that need research furtherance include:


#### **8. Outlook**

The workshop focused more on the technical aspects of flood risk assessment. A follow-up workshop could include the social aspects of ice-jam flood risk [43], for example community resilience [44] and socioeconomic vulnerabilities [45]. Research has also been carried out with agent-based modelling (ABM) to incorporate both technical and social aspects in flood risk assessments and management, on the individual [46], household [47] and regional [48] levels, in policy and decision making. An application of ABM specific to ice-jam flood risk assessment and mitigation is currently being explored by Ghoreishi et al. [49,50].

**Author Contributions:** Conceptualization, K.-E.L., K.A., D.C., A.C., D.G., M.H., B.H., N.K., M.K. (Michael Kögel), T.K., M.K.-D., M.K. (Michał Kubicki), Z.W.K., C.L., A.M., W.M., F.M., B.N.-L., T.N., A.P., B.P., I.P., J.R., M.R. (Maik Renner), M.R. (Michael Roers), M.R. (Maksymilian Rybacki), E.S., M.S., G.W., M.W., M.Z. (Mateusz Zagata) and M.Z. (Maciej Zdralewicz); investigation, K.-E.L., K.A., D.C., A.C., D.G., M.H., B.H., N.K., M.K. (Michael Kögel), T.K., M.K.-D., M.K. (Michał Kubicki), Z.W.K., C.L., A.M., W.M., F.M., B.N.-L., T.N., A.P., B.P., I.P., J.R., M.R. (Maik Renner), M.R. (Michael Roers), M.R. (Maksymilian Rybacki), E.S., M.S., G.W., M.W., M.Z. (Mateusz Zagata) and M.Z. (Maciej Zdralewicz); writing—original draft preparation, K.-E.L.; writing—review and editing, K.-E.L., K.A., D.C., A.C., D.G., M.H., B.H., N.K., M.K. (Michael Kögel), T.K., M.K.-D., M.K. (Michał Kubicki), Z.W.K., C.L., A.M., W.M., F.M., B.N.-L., T.N., A.P., B.P., I.P., J.R., M.R. (Maik Renner), M.R. (Michael Roers), M.R. (Maksymilian Rybacki), E.S., M.S., G.W., M.W., M.Z. (Mateusz Zagata) and M.Z. (Maciej Zdralewicz); visualization, K.-E.L., K.A., D.C., A.C., D.G., M.H., M.K. (Michael Kögel), T.K., M.K.-D., M.K. (Michał Kubicki), Z.W.K., W.M., F.M., T.N., B.P., I.P., J.R., M.R. (Maik Renner), M.R. (Michael Roers), M.R. (Maksymilian Rybacki), E.S., G.W., M.W., and M.Z. (Maciej Zdralewicz); supervision, K.-E.L., D.C. and Z.W.K.; project administration, K.-E.L., D.C. and Z.W.K.; funding acquisition, K.-E.L. All authors have read and agreed to the published version of the manuscript.

**Funding:** We would like to thank John Pomeroy, director of the Global Water Futures program at the University of Saskatchewan, for his sponsorship to partly fund this workshop. The drone-related fragments of the manuscript are supported by the National Science Centre of Poland through the research project no. 2020/38/E/ST10/00295—carried out by Tomasz Niedzielski, Matylda Witek, Joanna Remisz, Michał Halicki, and Grzegorz Walusiak—within the Sonata BIS programme.

**Institutional Review Board Statement:** Not applicable.

**Informed Consent Statement:** Not applicable.

**Data Availability Statement:** Not applicable.

**Acknowledgments:** We would like to extend a special thanks to all the participants for attending and contributing to the workshop.

**Conflicts of Interest:** The authors declare no conflict of interest.
