Ice Ice Maybe: Stream Hydrology and Hydraulic Processes During a Mild Winter in a Semi-Alluvial Channel

Abstract

Warm conditions during typically cold winters impact runoff and resulting hydraulic processes in channels where ice-cover would typically dominate. This field study on a short, low-slope reach in Southern Ontario, Canada, examined hydrologic and hydraulic processes with a focus on winter runoff events and subsequent bed shear stress variability. Through winter 2024, six cross-sections over a ~100 m reach were monitored near-weekly to measure hydraulic geometry and velocity profiles. These data characterized channel processes and estimated bed shear stress with law of the wall. In this channel, velocity increased more rapidly than width or depth with rising discharge and influenced bed shear stress distribution. Bed shear stress magnitudes were highest (means ranged ~2–6 N/m²) and most variable over gravel beds compared to the exposed bedrock (means ranged ~0.05–2 N/m²). Through a rain-on-snow (ROS) event in late January, bed shear stress estimates decreased dramatically over the rougher gravel bed, despite minimal changes in water depth and velocity. Pebble counts before, during, and after the event, showed that the proportion of finer-sized particles (i.e., <5 cm) increased while median grain size did not vary. These observations align with findings from both flume and field studies and suggest that milder winters reduce gravel-bed roughness through finer-sized sediment deposition, altering sediment transport dynamics and affecting gravel habitat suitability. Additionally, limited ice-cover leads to lower bed shear stresses and thus finer-sized materials are deposited, further impacting gravel habitat suitability. Results highlight the importance of winter hydrologic variability in shaping channel processes and inform potential stream responses under future climate scenarios.

1. Introduction

Winter generates unique hydrology and hydraulic conditions in fluvial systems that experience winters with temperatures below 0 °C [1,2,3]. Ice-cover along banks and over channels forms during prolonged freezing periods coincident with relatively low discharge, and leads to altered channel geometry and velocity profiles [4,5,6,7]. Ice break-up in the spring can lead to bank failures, intense erosion, and sediment transport, in addition to higher water levels after ice-jam releases (e.g., javes) [1,4]. Recent global climate change is making winters more variable, and in temperate regions warmer [8].

Shifts in winter conditions have impacts beyond winter stream ecology and thermal norms [3]. Milder temperatures and rainfall lead to variable runoff events (e.g., rain-on-snow [ROS] events, rain on frozen ground, snowmelt) and contribute to within winter seasonal variability in flow [9,10,11,12,13]. Rainfall and ROS events during winter, when the ground is often frozen, significantly impact discharge, producing short-lived but intense increases in flow [10,13,14]. Moreover, milder thermal conditions in rivers and streams change ice-cover, leading to various outcomes (e.g., less overall cover, earlier break-up, multiple break-up/formation/break-up cycles) that, like other runoff events, can rapidly create dynamic fluctuations in bed shear stress and sediment transport, and therefore changes to channel morphology [1,2,15,16,17,18].

Fluvial processes in small channels are regulated by flow conditions (e.g., discharge, water level, velocity) and bed morphology (e.g., bedforms, grain size, roughness or resistance) [1,5,7,19,20]. At the reach scale, bed shear stress is driven by water level and velocity, which are in part a function of channel discharge and slope, and by boundary roughness [6,21]. Previous studies indicate that in channels dominated by grain resistance, rougher stream beds (e.g., pebbles, cobbles) tend to produce higher and more variable bed shear stress compared to smoother stream beds such as bedrock and/or finer-grained material [7,21,22]. In natural channels, bed shear stress is expected to exhibit significant variability over grain-scales (i.e., millimetre to centimetre), driven by the heterogeneity in bed roughness. There are few field studies that focus on evaluating velocity and bed shear stress distributions in semi-alluvial channels.

River channels are complex and to help understand the processes and factors that influence these processes, channels are often classified as alluvial or bedrock [23,24,25]. In regions influenced by past glaciation, neither category adequately describes the conditions, and semi-alluvial is the classification used. Semi-alluvial channels are characterized by fine-to-coarse-grained sediment moving through a somewhat cohesive channel (e.g., exposed bedrock, glacial till, over-sized material) [19,23,26,27,28,29]. Despite their prevalence in Southern Ontario [23,28,30], semi-alluvial channels remain underrepresented in hydraulics studies, which typically focus on alluvial or bedrock systems, and often are represented by flumes. Moreover, studies often focus on larger (average discharge > 10 m³/s) systems, as monitoring equipment can be easily utilized. Smaller, low-order channels (average discharge < 10 m³/s) are often overlooked due to challenges in monitoring (e.g., variable water depths, sensitive habitat, flashy runoff response). Sixteen Mile Creek is a prime example of a semi-alluvial system that exhibits considerable variability in bed composition, with sections of gravel, exposed bedrock, and finer-sized sediments (e.g., sands) within a relatively small area (e.g., metres in width and length). These features make Sixteen Mile Creek an ideal site to study interactions between winter flow conditions and bed characteristics in complex (semi-alluvial) channels.

Initially, this study set out to evaluate typical winter conditions in a semi-alluvial channel, however the 2024 field season was not a typical winter, which led to altered study goals and objectives. This study investigates the hydrologic and hydraulic processes in response to varying flow conditions in a low-order, semi-alluvial channel over a mild winter. A 100-metre section along Sixteen Mile Creek in Southern Ontario was used as a representative semi-alluvial reach to collect channel geometry and velocity profiles to derive bed shear stress over gravel beds and exposed bedrock. These data were collected during various discharge conditions in the winter and early spring, including runoff events triggered by ROS. This study addresses two research objectives. Firstly, it examines changes in velocity and bed shear stress magnitude and variability through varying flow conditions. And secondly, it evaluates related changes to bed shear stress over varying bed types (i.e., rough and smoother beds). As argued elsewhere [2,23], semi-alluvial channels are complex and to address research gaps and develop management and restoration strategies suited to these systems, focused field studies in these channel types are necessary.

2. Methods

2.1. Study Reach

Sixteen Mile Creek originates from wetlands and forested swampland near the Niagara Escarpment (Figure 1A) [31]. Flowing southward through forested, agricultural, and urbanized landscapes, the creek empties into Lake Ontario, draining approximately 372 km² across nine sub-watersheds [30]. The hydrological and morphological features in Sixteen Mile Creek result from historical glacial conditions, similar to other watersheds in the region [23,28,30,32]. As the Laurentide Ice Sheet retreated, various landforms comprising glacial till deposited in central upland areas in Southern Ontario, while expansive glaciolacustrine plains formed in coastal areas [30]. Underlying the glacial deposits, and sometimes exposed at the surface, the bedrock throughout the watershed is dominated by shale, limestone, dolostone, and siltstone dated from the upper Ordovician [33]. Flowing from the escarpment base, the creek passes through undulating till, shale, and sand plains and intersects with the Trafalgar moraine (Figure 1A) [34]. The glacial history influences the longitudinal profile, hydrological response, and sediment availability, contributing to the semi-alluvial nature of the study reach (Figure 1A) [28,30].

The watershed around Sixteen Mile Creek is a humid continental climate due to its mid-latitudinal (43°29′40″ N, 79°50′20″ W) location and proximity to the Great Lakes (Figure 1A) [35]. Average winter temperatures range between −4.0 °C and 0.0 °C and summer averages range between 17.0 °C and 23.0 °C [36]. There is no distinct dry season, with snow typically falling in January and February, and rain common through March and April [35]. Freezing temperatures are typical in this region in the winter [9], leading to seasonal ground freezing and ice formation and melt over fluvial systems, influencing stream runoff processes. Seasonal variations in discharge are closely tied to the local climate, especially rainfall and snowmelt events, notably affecting catchment runoff volumes and timing in the region [12]. The complete hydrometric data record between 1959 and 2024 collected at a Water Survey of Canada (WSC) gauging station (02HB005) approximately 4.5 km upstream of the study reach indicates average winter discharge as 1.49 m³/s, and average spring discharge as 1.97 m³/s [37].

The research was conducted along a ~100 m segment (reach) in Sixteen Mile Creek, which was 20 m in width at its widest section, averaged 0.5 m deep, and has an average thalweg slope less than 0.002 (Figure 1B). The segment exhibits morphological diversity in its bed and bank composition typical of semi-alluvial reaches. It features cohesive banks with no well-organized bedforms or geomorphic units, and the bed is characterized by a wide range in particles, from fine-sized grains, such as clay-sized, silt and sand (<0.2 cm) particles, to gravels, including pebbles, granules, cobbles and boulders (>0.2 cm), interspersed with sections of exposed bedrock (Figure 1B). Additionally, channel restoration work in 2021 added vegetated revetments along the left and right banks for stabilization, and five large boulders (at least 10× larger than the average D₈₄ for the channel) placed near the left bank to diversify flow fields. To understand how flow conditions and bed heterogeneity influence velocity and bed shear stress in the channel, six cross-sections were chosen for detailed data collection, with the upstream- and downstream-most transects denoted as cross-sections 1 and 6 (Figure 1B). Cross-sections 2 and 3 are relatively wider, while cross-sections 4, 5, and 6 are relatively narrower (Figure 1B).

2.2. Field Data Collection

Near-weekly site visits during the winter and monthly site visits during the spring were conducted to characterize physical conditions, download sensors, and collect hydrodynamic and morphologic data along each cross-section (

Table 1. Site visit activity summaries at Sixteen Mile Creek for the research period.

Date (Air Temperature *)	Instruments and Methods	Notes
20 December 2023 (0 °C)	ADV, RTK, PC	HOBO loggers installed, reach and XS’s surveyed
9 January 2024 (3 °C)	ADV, RTK	Resistance-depth surveys along XS’s
21/22 January 2024 (−3 °C)	ADV, RTK	Ice-extent survey, Minor ice jamming US of XS1
29 January 2024 (2 °C)	ADV	High flow, following rainfall
6 February 2024 (2 °C)	ADV, RTK	Resistance-depth surveys along XS’s
8 February 2024 (3 °C)	PC + Facies maps	Pebble counts for XS 5 and 6
13 February 2024 (4 °C)	RTK	Survey large boulders and bedrock extent
20 February 2024 (2 °C)	ADV	Thin border ice at XS 2 and 3 (melted)
26 February 2024 (4 °C)	PC, Facies maps	Minor biofilm growth on larger grains (>5 cm) in the downstream portion of the reach
29 February 2024 (−2 °C)	ADV, Facies maps	Minor biofilm growth on larger grains (>5 cm) in the downstream portion of the reach
28 March 2024 (9 °C)	ADV, PC, Facies maps	Minor biofilm growth in the downstream portion of the reach
22 April 2024 (10 °C)	ADV, PC, Facies maps	Minor biofilm growth throughout the reach
14 May 2024 (20 °C)	ADV, PC, Facies maps	Biofilm/Algae throughout reach, Patches of grassy vegetation in low velocity areas

Notes: XS—Cross-Section, PC—Pebble Count, US/DS—Up/Downstream. * Average during site visit (9:00 am to 3:00 pm).

). Three HOBO water level data loggers were used to continuously record pressure and temperature at 15 min intervals (±0.62 kPa). The sensors were deployed on 20 December 2023, in the upstream (cross-section 1, left bank) and middle (cross-section 3, left bank) portion of the reach; additionally, along cross-section 3, another logger was installed above the stream bank to record atmospheric pressure (Figure 1B). This continuous monitoring was used to record water level, water temperature, and air temperature at the site. A trail camera was installed near the downstream HOBO logger, looking upstream over cross-sections 1–3 to collect photos at two-hour intervals (Figure 1B). The sensors and trail camera were downloaded and serviced during site visits to ensure consistent recordings.

A Real-Time Kinematic Global Positioning System (RTK-GPS) was used to conduct topographic surveys along the bed of each monitoring cross-section in a method similar to those outlined in Lotsari et al. (2017) [38] and Smith et al. (2023b) [7]. The RTK-GPS is a precise positioning instrument capable of collecting centimetre-level accurate coordinates (i.e., UTM) and elevation in real-time [39]. The surveys provided a detailed spatial reference for collecting velocity data and analyzing morphological characteristics. Bed and boundary conditions surveyed on 13 February were used to map out large boulders (>25.6 cm) within 1.0 m up- and downstream of the monitored cross-sections (and when they were sitting on exposed bedrock) and the exposed bedrock extent (Figure 1B). Precise boulder surveys beyond this buffer and the exposed bedrock were not conducted. A measuring tape was set up along each cross-section at fixed points along the right and left banks prior to channel measurements to ensure consistency throughout the study period.

Bed morphology was examined along each transect during site visits to observe changes. A Wolman (1954) [40] pebble count was conducted when water level and temperature permitted along cross-sections 5 and 6 during site visits (as noted in Table 1) to estimate substrate size (e.g., median particle size [D₅₀]), shape, and organization) [40]. For this field campaign, the Wolman (1954) [40] pebble count was modified in that particles less than 2 cm were placed in a category of <2 cm. Cross-section 5 and 6 were chosen for pebble counts due to their relatively shallow water depths, which allowed for safe pebble acquisition in the winter. Bed material smoothness (e.g., roundedness), shape (e.g., relative long, intermediate, and short axis ratios) and imbrication were recorded as qualitative observations over gravel bed portions throughout the study reach. Additionally, detailed channel bed field sketches (facies maps) were completed within 1.0 m (upstream and downstream) of all cross-sections during site visits. These sketches included estimated particle sizes and distribution descriptions while also noting various flow features (e.g., low-velocity zones, recirculation zones).

Velocity depth profiles were collected during each site visit throughout the study period (i.e., from December 2023 to May 2024) using a Sontek FlowTracker2 Acoustic Döppler Velocimeter (ADV), SonTek, a Xylem brand, San Diego, CA, USA. The ADV collects three-dimensional velocity data using the Döppler effect by transmitting acoustic beams and measuring the frequency shift (Döppler shift) of reflected sound waves to calculate particle speed within a fixed sampling volume [41]. Velocity depth profiles were delineated by taking data points at different depths above the bed. Profiles were collected at 1 m intervals for cross-sections 1 and 6, and at 2 m intervals for cross-sections 2 to 5. Each velocity profile used at least four points above the bed (e.g., 0.02 m, 0.05 m, 0.1 m, and near-surface), with deeper profiles containing more measurements to represent velocity in the water column. When ice was present, the ADV was positioned as close to the ice edge as possible. Ice-cover was never sufficient to support making measurements through the ice (i.e., Smith et al., 2023b [7]). The average sampling time per point was relatively short (30–45 s) due to the number of points required for this work, recorded at a frequency of 2.0 MHz [41]. The FlowTracker2 ADV used has a 0.001 m depth measurement resolution, 0.0001 m/s velocity resolution, and a minimum depth requirement of 0.02 m [41].

2.3. Data Processing and Data Analysis

Data collected with the ADV were pre-processed using Microsoft Excel in conjunction with pre-set quality control parameters outlined in the instrument’s manual (Supplementary Material Table S1). The signal-to-noise (SNR) ratio compares the reflected acoustic signal strength to detected background noise and is an important quality control metric [41]. Data points with an SNR value greater than ≥4 decibels (a suggested pre-set criteria from Sontek [41] were flagged and not used for analysis. Similarly, velocity standard error, boundary interference, velocity angle and tilt angle properties were also used as quality control parameters (Supplementary Material Table S1) [41].

Bed shear stress estimates were derived using the logarithmic law (log law) approach. This approach considers the velocity profile in a water column and requires mean velocity ($U_y$) data at various heights ($y$) above the bed. The log law approach is a commonly used method to derive bed shear stress in natural settings where bed morphology is variable [5,7,42,43,44,45]. Other methods of deriving bed shear stress were deemed unsuitable (e.g., Reynolds shear stress) for this study as site visit timing and the site extent made the log law method most practical to study local bed shear stress. To improve accuracy when estimating bed shear stress using log law, multiple near-bed velocity measurements (e.g., the bottom 20% of flow depth) were taken [46]. Velocity profiles were considered unsuitable for analysis if they (1) contained fewer than four data points, and (2) contained substantial vertical profile scatter (e.g., R² value < 0.8) [45]. When logarithmic velocity profiles are examined, a linear relationship is expected between the mean velocity ($U_y$) for some point in the water column (y-axis) and height above the bed ($y$) (x-axis) on a semi-log scale [47]. The slope of a line for a linear regression of the measured values of $U_y$ and $\mathit{ln}y$, is defined as b, and the y-axis intercept is a, and can be used to derive bed shear stress using Equations (1) and (2):

$$U_y=a+b\mathit{ln}(y/y_0),$$

(1)

$${\tau }_0=\rho {(b/2.5)}^2.$$

(2)

where $y_0$ (roughness length) is the projected height above the bed where velocity is zero, ${\tau }_0$ is bed shear stress and $\rho$ is the density of water. The slope (b) was also used to determine shear velocity ($U_{*}$) shown in Equation (3):

$$U_{*}=b/2.5.$$

(3)

Using Equation (3) is reasonable while working at the boundary scale (bed) as it is assumed that velocity profile characteristics scale with the bed characteristics given the moderate water depth across the site (e.g., less than 1 m). A linear regression line was fitted to each profile, and the R² value and the equation of the line for each profile ($U_y=a+b\mathit{ln}y$) were generated. R² values falling below 0.8 were not used for analysis as this suggests velocity profiles or outlier data points that do not meet the expected logarithmic shape [45]. Roughness lengths were calculated using Equation (4):

$$y_0=e^{-a/b}.$$

(4)

3. Results

3.1. Sixteen Mile Creek Study Reach Field Conditions

Field visits were conducted under various discharge conditions that were classified into low (≤1.12 m³/s), moderate (1.59–2.09 m³/s), and high (≥2.4 m³/s) (

Table 2. Flow condition classification for velocity profile surveys. High flow days are shown by a square symbol, and the colour indicates the date; moderate flow days are shown with a triangle symbol, the colour indicates the date; and a circle symbol is used.

Flow Condition	Discharge Range (m3/s)	Date (* Calculated Discharge [m3/s])	Symbol
High Flow	≥2.4	29 January 2024 (3.70)
		6 February 2024 (2.40)
Moderate Flow	1.59–2.09	13 February 2024 (2.09)
		22 April 2024 (2.06)
		29 February 2024 (1.59)
Low Flow	≤1.12	14 May 2024 (1.12)
		9 January 2024 (1.06)
		20 February 2024 (1.06)
		28 March 2024 (0.92)

Note: * Discharge calculations from collected ADV profiles at cross-section 6.

). These categories were chosen based on the calculated discharge values at cross-section 6 during each site visit (i.e., between 20 December 2023 and 15 May 2024, Table 1). There were four site visits that were classified as the low flow category, which includes values below the 44th percentile. Three field visits were categorized as moderate flow, which includes values between the 44th and 77th percentile. Two field days were categorized as high flow conditions, which include values exceeding the 77th percentile. ADV measurements collected on 20 December were limited to cross-section 1 due to equipment issues and thus were not included in this classification. Similarly, the partial ice-cover observations on 21 and 22 January were excluded.

Air temperature, water temperature, and water level were continuously monitored between 20 December 2023 and 15 May 2024 using HOBO pressure sensors (Hoskin Scientific, Oakville, ON, Canada) (Figure 2). There is a gap in the water temperature data (Figure 2) due to a logger downloading error. Daily precipitation quantities were obtained from a local weather station within a kilometre of the study reach (Wunderground Station IMILTO103). The air temperature fluctuated throughout the winter with consecutive days below ~−5.0 °C coinciding with the most substantive ice-cover between 15 January and 23 January (Figure 3, Supplementary Material Figure S1). Ice thickness varied (a few centimetres to 10.0+ cm) and was only continuous over cross-section 4 at this time. Often ice was frozen to the banks or in shallow areas of the bed (e.g., cross-sections 2, 3, 4, 5, and 6) (Figure S1). ADV measurements collected on 21 January (Supplementary Material Table S1) were not used in flow classification comparisons due to incomplete ice-cover across the study reach and ice jamming upstream of cross-section 1 (Supplementary Material Figure S1). Partial ice-cover was observed along the banks early in the morning on 14 January and 20 February (Supplementary Material Figure S1A,D); this ice was never strong enough to support through ice observations (e.g., standing on the ice and drilling a hole through it to then place the ADV for measurements). By late March, air temperatures were typically above 0.0 °C. Continuous discharge values are from the WSC gauging station (02HB005) located upstream from the study reach (Figure 2) [37].

The 2024 Southern Ontario winter was the warmest on record since 1948, at approximately 5.4 °C above the average and coincided with a powerful El Niño event [48,49]. Water levels throughout the winter were highly variable due to rainfall, ROS and melting events that occurred (e.g., between 9 January and 28 February) (Figure 2). Data collected on 9 January was during the onset of a 34 mm precipitation (mostly rain and wet snow) event (Figure 2). Due to this event, data collected on this day was at the onset of a steep rising limb (Figure 3, discharge). From 29 January through 20 February, data were collected during the falling limb following a 30.09 mm (combined) rainfall event that fell onto the snow-covered (and likely frozen) ground between 23 and 25 January (Supplementary Material Figure S2). Rainfall began overnight on 23 January and continued through to 25 January, with the rain-inducing snowmelt, as air temperatures remained above freezing (Figure 2). This event resulted in high and moderate flow conditions for the following site visits (29 January, 6 February, and 13 February, Table 2). Water levels and discharge also increased within 24 h following two rainfall events on 3 April (~24.13 mm) and 11 April (~28.96 mm; Figure 2).

During the 2024 season (including December 2023), the average winter (20 December to 19 March) discharge was 1.73 m³/s, and the average spring (20 March to 19 May) discharge was 1.50 m³/s. When considering the complete historical daily discharge record for Sixteen Mile Creek (i.e., 1959–2023, WSC Station [02HB005]), the average winter discharge for this period is 1.49 m³/s, and 1.97 m³/s for the spring (Figure 3) [37]. Compared to the historical averages, discharge during winter 2024 was higher, while discharge during spring 2024 was lower (Figure 3). It is important to note that the ROS event in January contributed to the elevated 2024 winter discharge. Discharge following this event swiftly peaked at 7.56 m³/s, resulting in high flow conditions for site visits on 29 January and 6 February during the falling limb (Figure 3).

Pebble count data collected during site visits between December 2023 and May 2024 were used to monitor changes in the gravel bed at cross-section 5 and 6 (Figure 4). Overall, the median particle size in the gravel sections at cross-sections 5 and 6 decreased over the winter and early spring, reaching its lowest point in April and May (Figure 4). For the entire monitoring season, the D₅₀ for cross-section 5 ranged from 3.3 cm to 7.5 cm, and 3.6 to 7.6 cm for cross-section 6. There were no significant variations in D₈₄ for the same pebble count data collected (Figure 4). Particles within the D₈₄ classification throughout the study reach were dominantly disc-like, sub-angular pieces of local bedrock (Supplementary Material Table S2). Particles within the D₅₀ classification had moderate sphericity and were sub-rounded (Supplementary Material Table S2).

3.2. Velocity Patterns in the 2024 Field Season

To ensure data accuracy and reliability, a thorough quality assurance and quality control process was implemented when analyzing velocity profiles. Each profile collected along cross-sections 1 through 6 was plotted and visually inspected before the R² value was determined from the line of best fit for measured values of $U_y$ (y-axis) and $\mathit{ln}y$ (x-axis) on a semi-log plot and used to assess profile robustness (Figure 5). Cross-sections 1 and 6 have more data points than the others due to more frequent sampling along these transects (i.e., profiles every 1 m across, including observations on 9 January). Profiles with R² values below 0.8 were not used to estimate bed shear stress [45]. Overall, profiles collected in this field campaign exhibited a logarithmic shape, with slower velocities near the bed and increasing velocity toward the water surface. Given the relatively shallow depth at the study site, free-stream velocity zones were not observed. Additionally, despite wide ranging flow conditions (Table 2), there were no patterns associated with flow conditions and profile R² values (Figure 5). At specific cross-sections, especially near the banks or around larger material, there tended to be a cluster of R² values that did not meet the threshold. Cross-section 1 showed several values below the threshold, particularly between 0.0 m to 2.0 m, and 5.5 m to 9.5 m from the right bank (Figure 5A). Cross-sections 2 and 3 followed a similar pattern, with lower R² values typically near the right bank compared to the middle of the transect (Figure 5B,C). R² values for cross-sections 4 and 5 were notably below the threshold along the left bank, where large boulders are sitting on exposed bedrock (Figure 1B and Figure 5D,E). Cross-section 6 had the most data points exceeding the 0.8 R² threshold, with below threshold points recorded during lower flow conditions (e.g., blue circles, Table 2).

Velocity profiles on nine field visits at cross-section 1 were collected at 1.0 m intervals along the transect (Figure 6). Maximum velocity and average velocity below 0 m/s indicate upstream movement (e.g., recirculation zone), which was observed adjacent to the right bank (Figure 6). Both the maximum and average velocities for this cross-section were typically faster in the centre (5.0 m to 6.0 m) (Figure 6A,B). Velocities were overall faster during higher flow conditions (e.g., 29 January, red squares) and slower for lower flow conditions (e.g., blue circles). Calculated bed shear stress values throughout the cross-section varied between <0.1 N/m² and 6.4 N/m², with an outlying value of 20.1 N/m² on 29 January at 5.5 m (Figure 6C). Excluded data points for bed shear stress over the bedrock portion are attributed to abnormal velocity profiles that do not meet the criteria for R² values (Figure 5) [45]. The bed morphology at cross-section 1 consisted of two gravel bed portions (~0.0 m to 5.5 m and ~8.0 m to the left bank) with a large protruding boulder (~0.7 m, b-axis) along the left bank (Figure 6D). Exposed bedrock was located near the centre (~5.5 m to 8.0 m) (Figure 6D). Minor fluctuations in bed elevation were observed over the gravel bed between approximately 0.0 m and 4.0 m, with the lowest point located between approximately 4.3 m and 5.1 m (Figure 6E). The profile gradually increases in elevation toward the left bank (~5.1 m to 9.0 m) before reaching a large boulder on the left bank, resulting in a sharp increase in bed elevation. The gravel bed portions at this cross-section were poorly sorted, containing disc-like, sub-angular cobbles (Supplementary Materials Table S2). Moderate-sphericity, sub-rounded granules and pebbles were observed over the gravel bed near the right bank (between the right bank and ~5.0 m) throughout the monitoring season (Supplementary Materials Table S2).

Velocity profiles at cross-section 2 were collected every 2.0 m along the transect on eight field visits (no measurements on 9 January) (Figure 7). On 29 January, measurements ended at 11.5 m from the right bank due to a battery pack issue in the ADV that was later resolved. Observations on 29 January recorded the fastest velocities throughout the cross-section, while 28 March, the day with the lowest observed discharge, recorded the slowest velocities. The greatest magnitudes for maximum and average velocities were recorded within the thalweg and were slower along the banks (Figure 7A,B). Velocity magnitudes along the left bank (~12.5 m to 16.0 m) were often less than zero due to the large upstream boulder that created a recirculation zone along the left bank (Figure 1B). Calculated bed shear stress values were greatest ~7.5 m to 11.0 m from the right bank, ranging from 8.9 N/m² to 16.8 N/m² and were low (<0.1 N/m²) closer to the left bank (~11.0 m to 16.0 m) (Figure 7C). The bed morphology along cross-section 2 is variable, ranging from fine-grained material (e.g., sand and finer-sized material) covering larger material near the banks, interrupted with gravels (~4.0 m to 11.0 m from the right bank), and exposed bedrock (~2.5 m to 4.0 m from the right bank) through the channel middle (Figure 7D). Small boulders were observed along the right bank (between ~2.5 m and 5.5 m from the right bank; Figure 7D). The topographic profile along this cross-section was lowest over the exposed bedrock (~4.40 m from right bank) and gradually increased in elevation moving toward the left bank. Like cross-section 1, cross-section 2 was poorly sorted with disc-like, sub-angular cobbles, while granules and pebbles were moderately spherical and sub-rounded (Supplementary Materials Table S2).

Velocity profiles at cross-section 3 were taken at 2.0 m intervals across the transect, on eight site visits (excluding 9 January) (Figure 8). On higher flow days (e.g., 29 January and 6 February (squares)), velocity throughout the cross-section was greater than on lower flow days. Maximum velocities (Figure 8A) and average velocities (Figure 8B) were typically highest within the middle section, between 5.0 m and 9.0 m from the right bank. Bed shear stress through this cross-section was relatively low compared to others in the study reach, with the greatest variability between 6.5 m and 10.0 m from the right bank (<0.1 N/m² to 8.2 N/m²) (Figure 8C). The bed morphology at cross-section 3 was characterized by a combination of individual boulders near the right bank, exposed bedrock (~6.0 m to 8.0 m from the right bank), and fine-grained material (~11.2 m from the right bank to the left bank) (Figure 8D). Two distinct gravel bed portions were observed from 0.5 m to 3.5 m from the right bank, and 5.5 m from the right bank to the left bank. The topographic profile shows a relatively classic channel cross-section with the deepest section near the centre (~6–8 m from the right bank) (Figure 8E). Gravels were poorly sorted throughout cross-section 3, dominated by disc-like, sub-angular cobbles and moderately spherical, sub-rounded granules and pebbles (Supplementary Materials Table S2). Boulders were prominent near the right bank portion in this cross-section between the right bank and 3.0 m from the right bank.

Cross-section 4 velocity profiles were collected at 2.0 m intervals across the transect during eight site visits (no measurements on 9 January) (Figure 9). Overall, both the maximum and average velocities were highest during high-flow days (e.g., 29 January, red square). Old bridge abutments upstream of this cross-section (Figure 1B) confine the channel compared to other cross-sections in the study reach and likely account for the higher velocities observed throughout (Figure 9A,B). Limited bed shear stress estimates adjacent to the left bank are attributed to multiple protruding boulders, which made velocity profile sampling challenging. Calculated bed shear stress values did not exceed 10 N/m² through this transect (Figure 9C). Cross-section 4 features a gravel bed between 0.0 m and ~5.5 m from the right bank, with exposed bedrock from ~5.5 m from the right bank to the left bank (Figure 9D). The topographic profile shows the deepest section at the transition from gravel bed to exposed bedrock (~5.7 m from the right bank) (Figure 9E). The gravel bed gradually deepens from the right bank before an abrupt decrease (~0.2 m in depth) over the exposed bedrock, starting at ~5.5 m from the right bank (Figure 9E). The gravel bed portion was poorly sorted throughout the research period, with minor imbrication along the right bank (Supplementary Materials Table S2).

Cross-section 5 velocity profiles were collected on eight site visits (no measurements on 9 January), every 2.0 m along the transect (Figure 10). Discrepancies in maximum and average velocity magnitudes (Figure 10A,B) are attributed to flow interruptions caused by boulders located between 7.5 m and 9.0 m from the right bank, directly upstream of the monitoring transect (Figure 1B). Maximum and average velocities were typically faster in the shallower portion (0.0 m to 7.0 m from the right bank) compared to the deeper section near the left bank (Figure 10A,B). Calculated bed shear stress values between the right bank and ~6.5 m (centre of the channel) ranged from <0.1 N/m² to 18.3 N/m², while bed shear stress variability from the near-centre of the channel to the left bank ranged from <0.1 N/m² to 2.1 N/m² (Figure 10C). Bed morphology at cross-section 5 was characterized by large individual boulders located between 7.0 m and 9.0 m, gravels along the right bank, and exposed bedrock along the left bank (Figure 10D). This transect deepens approximately halfway through the channel (~5.7 m from the right bank), as the bed material transitions from gravel to exposed bedrock (Figure 10D,E). The poorly sorted gravel bed portion contained moderately spherical, sub-rounded D₅₀ particles. In contrast, D₈₄ particles were disc-like and sub-angular (Supplementary Materials Table S2). Cobbles were imbricated throughout the research period at cross-section 5.

Velocity profiles at cross-section 6 were taken at 1.0 m intervals along the transect on nine field days, as described for cross-section 1. Maximum velocity (Figure 11A) contained minor variations moving from right bank to left. Average velocity (Figure 11B) exhibited a slightly different pattern where higher average velocities for a given site visit (Table 2) were found over the deepest points between approximately 7.0 m and 8.0 m from the right bank. Like cross-section 5, calculated bed shear stress values between the right bank and near-centre had higher variability (<0.1 N/m² to 14.1 N/m²) than the left bank side (0.4 N/m² to 4.0 N/m²) of cross-section 6 (Figure 11C). Bed morphology at cross-section 6 showed a notable transition from gravel bed morphology to exposed bedrock approximately 8.5 m from the right bank (Figure 11D). The topographic profile remained relatively flat extending from the right bank (~1.0 m to 6.0 m) before reaching the lowest point. Like cross-section 5, D₅₀ particles were moderately spherical and sub-rounded while the D₈₄ were disc-like and sub-angular (Supplementary Materials Table S2). These gravels were poorly sorted and imbricated throughout the research period (Supplementary Materials Table S2). Additionally, at-a-station hydraulic geometry changes were assessed at cross-section 6 (Supplementary Materials Table S3) and conformed to expected values [50,51].

3.3. Bed Shear Stress and Bed Type

Bed shear stress and roughness lengths varied according to different bed types in each cross-section (Figure 12). Overall, the gravel bed exhibited the highest bed shear stress magnitude and greatest variability compared to the other bed types (Figure 12A). Bed shear stress averages over exposed bedrock were similar at all cross-sections (e.g., <0.3 N/m² to 2.2 N/m²) (Figure 12A). Bed shear stress estimates over exposed bedrock typically had lower variability, except in cross-section 2, where few profiles met the bed shear stress calculation criteria over this bed type (n = 2). There are few bed shear stress estimates over fine-grained material (n = 9) because this bed type was only observed along cross-section 2, where velocity profiles were collected every 2 m (Figure 7). The variability and magnitudes in bed shear stress at cross-sections 1 and 4 are similar between the two bed types (Figure 12A). Cross-sections 1 and 4 are the deepest cross-sections in the study reach with maximum water depths reaching 0.95 m and 0.80 m on 29 January. Roughness lengths exhibited a similar trend to bed shear stress over the different bed types along each cross-section (Figure 12B). On average roughness lengths over the gravel bed were greater than roughness lengths over the exposed bedrock at each cross-section. Cross-section 1 recorded the highest roughness lengths over the gravel bed. Overall, gravel beds showed the highest variability and magnitudes in bed shear stress and roughness lengths throughout the study reach for all site visits. Roughness length estimates are susceptible to variations in individual velocity measurements and fluctuate with repeated trials on the same profile and bed type, making it useful as a reference or validation parameter only and was used as such in this study [44].

Bed shear stress estimates over the gravel bed in general were greater and more variable as discharge increased (panel C on Figure 6, Figure 7, Figure 8, Figure 9, Figure 10 and Figure 11). When these results were evaluated by cross-section, no statistically significant trends were identified (Supplementary Materials Figure S3). Moreover, bed shear stress over exposed bedrock showed little to no changes as discharge increased (Supplementary Materials Figure S3). These findings are included in this work to underline the need to evaluate bed shear stress at a local point on the bed and not just at a cross-section. Moreover, with data in hand it is tempting to run predictions and models to estimate bed shear stress relationships, which is problematic in semi-alluvial channels and/or other systems with complex beds.

Water depth, average velocity, and bed shear stress estimates from two representative stations at cross-section 6 over the gravel bed (i.e., 8.0 m from the right bank) and the exposed bedrock (i.e., 11.0 m from the right bank) were directly compared (Figure 13). These data were investigated following the ROS event (the falling limb of the hydrograph) that began on 23 January (Figure 2 and Figure 13). Data from 9 January was also included to demonstrate lower flow condition prior to the event (Figure 13). Water depth and average velocity remained relatively similar between the two stations for each respective day during the event (Figure 13A,B). However, bed shear stress estimates over the gravel bed demonstrated variability despite small changes to water depth and average velocity (Figure 13C). The highest flow day, 29 January (i.e., discharge was 3.7 m³/s), recorded the second largest bed shear stress estimate at the gravel bed station (i.e., 5.0 N/m², 8.0 m from the right bank at cross-section 6). Bed shear stress at the same station peaked on 6 February (Figure 13C), when discharge was 2.6 m³/s, and was higher than the average and median for this bed type at this cross-section when considering all the observations (Figure 12A). Subsequent visits on the falling limb recorded decreasing bed shear stress estimates over the gravel bed (Figure 13C) and returning to an estimated bed shear stress value near 1.5 N/m² (Figure 12A). Bed shear stress estimates over the exposed bedrock did not decrease at the same rate as the gravel bed and tended to remain relatively low at around 1 N/m² (Figure 13C). Overall, average bed shear stress over the gravel bed decreased throughout the falling limb following the ROS event.

4. Discussion

Mild Winter Flow Regimes Impact Local Bed Hydraulics

Southern Ontario winters are typically characterized by freezing temperatures, with snowfall as the dominant precipitation [11,52]. In low-order channels, winter water levels are lower, flow volumes are reduced, and ice cover can form and persist throughout the cold period [2,11,53,54]. With freezing air temperatures, and near zero water temperatures, riparian and in-stream vegetation growth is reduced (or non-existent) [3,55]. Additionally, when the ground and banks are frozen or near-frozen, sediment availability is also reduced [1,56]. Variations from the norm, cooler than average or warmer than average winters, occur due to natural and anthropogenic climate variability (e.g., El Nino Southern Oscillation, polar vortex, global warming) [8,49]. Warmer regional temperatures during winters in Southern Ontario often result in rainfall rather than snow [52,53,57,58] and can lead to large runoff events (ROS) in the middle of winter [9,11]. Milder winters generate stream conditions that are unique, as temperatures are still cool, some ground may be frozen, but ice-cover may not be persistent [1,3,4]. Additionally, most winter river research is focused on large channels (high order, >30 m width, >1 m depth) and thus the impact of milder winters on smaller (low-order, <20 m width, <1 m depth) channels remains understudied [2,11,54].

Winter 2024 in Southern Ontario was the mildest since 1948 [48]. The average air temperature recorded at the study reach was 0.47 °C, with the longest freezing period taking place between 15 and 23 January (Figure 2). Winter 2024 aligned with a very strong El Nino event [49]. This teleconnection influences the climate in Southern Ontario, resulting in warmer winter conditions [58,59,60]. But with temperatures near zero, the ground was frozen, which meant infiltration into the ground was reduced. These conditions led to several ROS in winter 2024 (Figure 2 and Figure 3). Winter runoff events in January are not unusual, however runoff events with daily discharge greater than 6 m³/s typically happen once every 8–9 years and are much larger than typical winter season discharge daily average (1.49 m³/s) (Figure 3). In late January 2024, a combined ROS and then rainfall event led to a rapid increase in runoff (Figure 2, Supplementary Material Figure S2). The late January ROS and lack of significant snowfalls for the rest of the winter led to a below-average spring runoff in 2024 (Figure 2 and Figure 3).

ROS events result in significant runoff generation that leads to rapid increases in streamflow [10,13]. Temperature and rainfall data as well as images from the trail camera tracked a major ROS event at Sixteen Mile Creek beginning overnight on 23 January through 26 January (Figure 2 and Supplementary Materials Figure S2). A mixture of snow and rain fell throughout the day on 24 January, increasing discharge steadily until a rapid increase on 26 January, which coincided with ~12 mm of rainfall on the partially snow-covered and frozen ground (Supplementary Materials Figure S2). The rapid increase and high flow volume observed on 26 January was significantly greater than other rain-induced events earlier in January (e.g., 10, 13 January) (Figure 2), which is expected as rainfall accelerates the snowmelt process, thus increasing overall runoff [10,13,14]. Rainfall-induced snowmelt events are typically shorter in duration and more intense than those driven by thermal processes alone [10,13,14]. Climate variability forecasts for Southern Ontario suggest that winter variability will increase (e.g., very cold or very mild), thus ROS events, and/or rapid snowmelt events combined with partially frozen ground cover will likely be more common [11,14,61].

The winter 2024 Sixteen Mile Creek field observations broadly align with other field studies of complex channels [1,5,7,19,38]. The fastest velocity at a given transect was along the thalweg, near the water surface during ice free conditions (Figure 6, Figure 7, Figure 8, Figure 9, Figure 10 and Figure 11). At-a-station hydraulic geometry (e.g., width, depth, velocity) changed as discharge increased. Velocity increased faster than the other two parameters (Supplementary Materials Table S3), which is typical for coarse-grained channels with cohesive banks (e.g., [62]). Over the 2024 field season, as flow conditions varied (Table 2), overall average and maximum velocities through a cross-section increased with increasing discharge (Figure 6, Figure 7, Figure 8, Figure 9, Figure 10 and Figure 11). Variation at specific points along the bed resulted in large changes in bed shear stress closely corresponding to bed material roughness (Figure 12).

Bed shear stress magnitudes and distributions in natural channels are difficult to measure as they require high frequency velocity measurements in two or three dimensions [43,45]. Several studies have evaluated the limitations and merits of using law of the wall (log law) to estimate bed shear stress [44,45,46]; the results presented here are an estimate for bed shear stress using log law and should be treated as relative. The relationship between bed shear stress magnitude and the average and maximum velocity along a cross-section within Sixteen Mile Creek was most closely related to changes in relative bed roughness. Bed shear stress estimates were between <0.1 N/m² and 10.0 N/m² for cross-sections 1, 3, and 4 (Figure 6C, Figure 8C and Figure 9C) with the exception of one point in cross-section 1 (i.e., 5.4 m from the right bank on 29 January) (Figure 6C). Conversely, cross-sections 2, 5, and 6 had a larger bed shear stress range (between <0.1 and 19.0 N/m²; Figure 7C, Figure 10C and Figure 11C). However, when looking at individual stations within a cross-section, bed shear stress was distributed unevenly. For example, cross-section 5 had the largest bed shear stress estimates compared to the other cross-sections (Figure 12A). Most notably, large bed shear stress values were only observed over the gravel bed compared to the exposed bedrock (Figure 10C, Figure 11C and Figure 12A). Large bed shear stress magnitudes over hydraulically rougher beds are consistent with other studies evaluating bed shear stress in channels [5,6,7]. Due to the varying bed morphology throughout the study reach, bed shear stress magnitudes were larger and more variable (wider range) over the gravel bed (i.e., hydraulically rougher) in a cross-section (Figure 12A) and not necessarily aligned with the thalweg (i.e., where velocity was fastest). These findings are consistent with a winter 2021 study at Sixteen Mile Creek where a larger bed shear stress range was observed over the rougher beds (<0.1 N/m² to 9.0 N/m²) compared to smoother beds (<0.1 N/m² to 5.0 N/m²) in open flow conditions [7]. Additionally, the winter 2024 findings are similar to various flume studies, where bed shear stress was shown to have a greater magnitude and range over hydraulically rough beds when compared to smoother (e.g., fine-grained, cohesive) beds [21,22,29].

Throughout winter 2024, maximum and average velocity were greater on higher flow days and were associated with increased bed shear stress magnitude and variability over the gravel bed (Figure 6C, Figure 7C, Figure 8C, Figure 9C, Figure 10C, Figure 11C and Figure 12). In previous work at Sixteen Mile Creek [7], ice-cover in the channel changed the depth that maximum velocity occurred within a profile, and thus altered the estimated bed shear stress, which is also observed in flume studies [63,64]. However, in 2024 ice-cover was limited to the bank edges for most of the study reach (Supplementary Materials Figure S2), and thus there were no measured differences in velocity patterns or bed shear stress that is typical in ice-covered conditions. The largest differences in bed shear stress observed at the site occurred through the falling limb of the late January ROS.

Two comparison profiles along cross-section 6, one over gravel (8 m from the right bank) and one over bedrock (11 m from the right bank) (Figure 11D), illustrate how relative bed roughness impacted bed shear stress during the ROS event (Figure 13). Over the gravel bed, despite similar water depth and average velocity on 6 February and 13 February, bed shear stress on 6 February was 7.3 N/m² and less than 1 N/m² on 13 February (Figure 13A–C). Coincidently, there were no large changes in D₅₀ (~7.6 cm decreased to 5.5 cm) or D₈₄ (remained ~12.8 cm) values at cross-section 6 (none at cross-section 5 either) (Figure 4 and Figure 13F). However, the percentage of finer-sized, sub-rounded particles (i.e., sub-centimetre) observed after the event, compared to before increased (Figure 4 and Figure 13F). Several flume experiments have observed that finer-grained infilling is common in gravel beds during the falling limb of an event. Polvi (2021) [29] observed both scour and infilling in their flume experiment and noted a lack of bedform organization. Using a laboratory flume, Khuntia et al. (2018) [18] found that estimated bed shear stress decreased through the falling limb. Sediment along the bed becomes mobilized during the rising limb and is transported downstream until it is then deposited when flow decreases [16,17,29,47]. Mao (2012) [16] argued that the observed decrease in bed roughness is related to the surface layer of finer materials being mobilized with sufficient bed shear stress, but is not sufficient to move larger materials. Although the increase in smaller particles does not significantly change D₅₀ (or D₈₄) values, their presence effectively reduced bed roughness, and thus changed flow physics near the bed (Figure 13) [17,29].

The winter 2024 findings align with previous work that investigated changes to bed shear stress in unsteady flow conditions passing through the rising and falling limbs in a hydrograph [16,17,29]. At Sixteen Mile Creek, following the runoff peak in late January, the bed morphology along the lower part of the Sixteen Mile Creek study reach was altered due to infilling from finer-sized particles (Figure 13). This reduced bed roughness and enhanced particle imbrication. In turn, the roughness reduction resulting from finer-grain infilling in the gravel bed along cross-section 6 explains the disproportionate change in bed shear stress estimates observed through the large runoff event (Figure 13C) and contributes to the overall greater variability in bed shear stress estimates observed throughout the site and field season (Figure 12). These findings have implications for winter hydraulics in typically ice-covered channels.

Milder winters resulting in elevated discharge and reduced bed roughness over gravels will influence geomorphic unit composition, and maintenance [19,20,65]. Removing fine-grained material from gravels is an important process to maintaining gravel beds and aquatic habitat [17,32,47,66]. Ice-cover in small (low-order) channels, works to generate sufficient bed shear stress over winter, such that finer-grained material are removed from gravel beds [1,4,7], especially with low slopes [7]. If milder winter (ice-free) conditions persist, void-spaces within gravel beds may be compromised and thus compromise aquatic habitats [32,67]. Given the projected shift in increased ROS, and thus reduced spring runoff [11,13,14], fine-grained material removal from gravel beds will be further reduced even though their abundance has increased [23,29,56]. This leads to further compromised gravel beds which are important geomorphic features for strong aquatic ecosystem functioning [23,67].

5. Conclusions

This study investigated hydrology and hydraulic processes and morphological characteristics in a semi-alluvial reach within Sixteen Mile Creek during a mild winter. The results demonstrated significant differences in velocity and bed shear stress magnitude and variability throughout the winter, but most notably over rougher stream beds (e.g., gravels). In contrast, exposed bedrock surfaces exhibited a lower bed shear stress range in magnitude and variability. Winter 2024 was a mild winter with limited ice-cover formation and was dominated by runoff generated by a large ROS event in late January. The ground during the event was frozen resulting in a swift and significant runoff event. Looking at the same point over a gravel portion of the bed through that runoff event revealed that bed shear stress did not change in relation to velocity and depth, and that infilling by smaller gravels in that section reduced roughness and thus reduced bed shear stress. Gravel infilling by smaller material is often observed in flume studies and could have consequences for overall sediment transport and habitat suitability in channels that future studies need to take into account.

The findings from this study have broader implications for stream restoration and management in semi-alluvial systems. As observed in Sixteen Mile Creek, a low-order, low slope reach bed morphology is highly sensitive to changes in flow conditions, particularly during a mild winter. Understanding these changes is important in semi-alluvial reaches in this region, where land use change can exacerbate climate variability impacts. As ice-covered stream conditions are reduced (e.g., due to warmer winters), fine-material infilling gravels will occur and potentially degrade habitat suitability. Future stream habitat conservation and restoration efforts certainly need to account for shifting winter conditions. This study stresses the importance of reach-scale work in understanding local bed shear stresses and sediment transport through short-lived events. Field studies are needed to understand local nuance within a site and understand feedback between processes (e.g., increased ROS and reduced ice cover).