<?xml version="1.0" encoding="UTF-8"?>
<!DOCTYPE article PUBLIC "-//NLM//DTD Journal Publishing with OASIS Tables v3.0 20080202//EN" "journalpub-oasis3.dtd">
<article xmlns:xlink="http://www.w3.org/1999/xlink" xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:oasis="http://docs.oasis-open.org/ns/oasis-exchange/table" xml:lang="en" dtd-version="3.0" article-type="research-article"><?xmltex \makeatother\@nolinetrue\makeatletter?>
  <front>
    <journal-meta><journal-id journal-id-type="publisher">TC</journal-id><journal-title-group>
    <journal-title>The Cryosphere</journal-title>
    <abbrev-journal-title abbrev-type="publisher">TC</abbrev-journal-title><abbrev-journal-title abbrev-type="nlm-ta">The Cryosphere</abbrev-journal-title>
  </journal-title-group><issn pub-type="epub">1994-0424</issn><publisher>
    <publisher-name>Copernicus Publications</publisher-name>
    <publisher-loc>Göttingen, Germany</publisher-loc>
  </publisher></journal-meta>
    <article-meta>
      <article-id pub-id-type="doi">10.5194/tc-15-4031-2021</article-id><title-group><article-title>Ice roughness estimation via remotely piloted
aircraft and photogrammetry</article-title><alt-title>Ice roughness estimation via RPA</alt-title>
      </title-group><?xmltex \runningtitle{Ice roughness estimation via RPA}?><?xmltex \runningauthor{J.~Ehrman et al.}?>
      <contrib-group>
        <contrib contrib-type="author" corresp="no">
          <name><surname>Ehrman</surname><given-names>James</given-names></name>
          
        </contrib>
        <contrib contrib-type="author" corresp="yes">
          <name><surname>Clark</surname><given-names>Shawn</given-names></name>
          <email>shawn.clark@umanitoba.ca</email>
        </contrib>
        <contrib contrib-type="author" corresp="no">
          <name><surname>Wall</surname><given-names>Alexander</given-names></name>
          
        </contrib>
        <aff id="aff1"><institution>Department of Civil Engineering,
15 Gillson St., University of Manitoba, Winnipeg,
MB R3T 5V6, Canada</institution>
        </aff>
      </contrib-group>
      <author-notes><corresp id="corr1">Shawn Clark (shawn.clark@umanitoba.ca)</corresp></author-notes><pub-date><day>24</day><month>August</month><year>2021</year></pub-date>
      
      <volume>15</volume>
      <issue>8</issue>
      <fpage>4031</fpage><lpage>4046</lpage>
      <history>
        <date date-type="received"><day>7</day><month>January</month><year>2021</year></date>
           <date date-type="rev-request"><day>29</day><month>March</month><year>2021</year></date>
           <date date-type="rev-recd"><day>16</day><month>July</month><year>2021</year></date>
           <date date-type="accepted"><day>20</day><month>July</month><year>2021</year></date>
      </history>
      <permissions>
        <copyright-statement>Copyright: © 2021 </copyright-statement>
        <copyright-year>2021</copyright-year>
      <license license-type="open-access"><license-p>This work is licensed under the Creative Commons Attribution 4.0 International License. To view a copy of this licence, visit <ext-link ext-link-type="uri" xlink:href="https://creativecommons.org/licenses/by/4.0/">https://creativecommons.org/licenses/by/4.0/</ext-link></license-p></license></permissions><self-uri xlink:href="https://tc.copernicus.org/articles/.html">This article is available from https://tc.copernicus.org/articles/.html</self-uri><self-uri xlink:href="https://tc.copernicus.org/articles/.pdf">The full text article is available as a PDF file from https://tc.copernicus.org/articles/.pdf</self-uri>
      <abstract><title>Abstract</title>
    <p id="d1e96">The monitoring of fluvial ice covers can be time intensive, dangerous, and costly if detailed data are required. Ice covers on a river surface cause resistance to water flow, which increases upstream water levels. Ice with a higher degree of roughness causes increased flow resistance and therefore even higher upstream water levels. Aerial images collected via remotely piloted aircraft (RPA) were processed with structure from motion photogrammetry to create a digital elevation model (DEM) and then produce quantitative measurements of surface ice roughness. Images and surface ice roughness values were collected over 2 years on the Dauphin River in Manitoba, Canada. It was hypothesized that surface ice roughness would be indicative of subsurface ice roughness. This hypothesis was tested by comparing RPA-measured surface ice roughness values to those predicted by the Nezhikhovskiy equation, wherein subsurface ice roughness is proportional to ice thickness. Various statistical metrics were used to represent the roughness height of the DEMs. Strong trends were identified in the comparison of RPA-measured ice surface roughness to subsurface ice roughness values predicted by the Nezhikhovskiy equation, as well as with comparisons to ice thickness. The standard deviation and interquartile range of roughness heights were determined to be the most representative statistical metrics and several properties of the DEMs of fluvial ice covers were calculated and observed. No DEMs were found to be normally distributed. This first attempt at using RPA-derived measurements of surface ice roughness to estimate river ice flow resistance is shown to have considerable potential and will hopefully be verified and improved upon by subsequent measurements on a wide variety of rivers and ice covers.</p>
  </abstract>
    </article-meta>
  </front>
<body>
      

<sec id="Ch1.S1" sec-type="intro">
  <label>1</label><title>Introduction</title>
      <p id="d1e108">The consequences of ice on the flow regimes of rivers in cold climates can be dramatic, sometimes leading to loss of life and damage to infrastructure. In-stream infrastructure such as bridge piers, hydraulic control structures, and hydro-electric generating stations are subject to immense forces due to river ice, which is a critical factor in the design of such structures. Understanding fluvial ice roughness is a critical step in better understanding the evolution and hydraulic impacts of fluvial ice covers. Currently, fluvial ice roughness is either estimated through empirical means, such as the <xref ref-type="bibr" rid="bib1.bibx23" id="text.1"/> equation, or through complex and expensive methods, such as hydraulic modelling. Direct measurements can also be made <xref ref-type="bibr" rid="bib1.bibx5 bib1.bibx8" id="paren.2"/>, or roughness can be inferred from a measured velocity profile <xref ref-type="bibr" rid="bib1.bibx16" id="paren.3"/>. However, these direct measurement methods require personnel to conduct work on ice covers, which are frequently unsafe, thus limiting the types of ice cover that can be studied.</p>
      <?pagebreak page4032?><p id="d1e120">The surface roughness of sea ice and land ice (typically glaciers) has been more extensively researched <xref ref-type="bibr" rid="bib1.bibx13 bib1.bibx9 bib1.bibx35" id="paren.4"/> than that of fluvial ice covers. This discrepancy is due in part to the scale of the these ice sheets, which allows for high-altitude remote sensing from manned aircraft using lidar and imagery and satellites using synthetic aperture radar <xref ref-type="bibr" rid="bib1.bibx9" id="paren.5"/>. The size and thickness of these ice formations also makes in situ measurements generally more feasible from a safety perspective. The goal of obtaining roughness data for glaciers and sea ice surfaces often relates to the determination of aerodynamic roughness length, an important parameter in the estimation of heat fluxes <xref ref-type="bibr" rid="bib1.bibx13" id="paren.6"/>, although <xref ref-type="bibr" rid="bib1.bibx9" id="text.7"/> evaluated sea ice roughness for the use of transportation planning.</p>
      <p id="d1e135">An obvious solution to improve the safety of fluvial ice cover studies is through the use of aerial vehicles. Helicopters, small fixed-wing aircraft, and satellites have long been used for the study of Earth surface phenomena. All are prohibitively expensive to be solely dedicated to the study of fluvial ice covers, and none can produce images of sufficient resolution for surface roughness studies. Recently, remotely piloted aircraft (RPA) have become much more accessible, inexpensive, and reliable. Coupled with high-resolution image-stabilized digital cameras, they offer the opportunity to document and study otherwise inaccessible areas at a fraction of the cost of any other method. Structure-from-motion photogrammetry has been used extensively to process RPA-acquired digital photos (RPA photogrammetry) <xref ref-type="bibr" rid="bib1.bibx7" id="paren.8"/>. The evaluation of surface roughness has also been studied using RPA photogrammetry on land surfaces <xref ref-type="bibr" rid="bib1.bibx22" id="paren.9"/> and non-fluvial ice surfaces <xref ref-type="bibr" rid="bib1.bibx9 bib1.bibx6" id="paren.10"/>.</p>
      <p id="d1e147">Although qualitative assessments of river ice roughness have been made based on visual observation, and quantitative estimates have been made through hydraulic modelling efforts, to date there has been no reliable means of quantitatively assessing river ice roughness over a large area. This study uses RPA photogrammetry to address this need and provides the first detailed description of the non-uniform three-dimensional roughness of a river ice surface. This research hypothesizes that the surface ice roughness of a newly frozen fluvial ice cover is indicative of the subsurface ice roughness of the same cover. The basis of this theory stems from field observations of ice mechanics on the Dauphin River, as observed by <xref ref-type="bibr" rid="bib1.bibx31" id="text.11"/>. Smooth, thermally grown ice was observed to have a smooth texture both at the top and bottom of the ice cover. Ice pans that flowed downstream and met an obstruction were observed to stack in a fashion that presumably had similar surface and subsurface roughness. When the external forces acting on the ice cover overcame its internal strength, the ice would consolidate, becoming thicker and noticeably rougher on the top surface. Even though direct measurements of the underside of the ice cover were not possible, increases in water level upstream indicated that an increase in flow resistance from a rougher bottom of ice was likely. Subsurface ice roughness investigations have been conducted on mature ice covers <xref ref-type="bibr" rid="bib1.bibx4 bib1.bibx5 bib1.bibx8" id="paren.12"/>. It is likely that subsurface ice roughness measurements taken well after freeze-up will under-predict peak ice roughness due to smoothening of the subsurface ice over time by flowing water. While the surface and subsurface of an ice cover are subject to very different external forces, the hypothesis that the roughness of the ice cover at the surface is proportional to the subsurface roughness focuses on newly formed ice covers. Given past experience with this river system, it is expected that the surface and subsurface ice roughness values will not appreciably change within 1 week of ice cover formation.</p>
      <p id="d1e157">The objectives of this study are as follows:
<list list-type="bullet"><list-item>
      <p id="d1e162">evaluate the capabilities of a consumer-grade RPA coupled with a professional photogrammetry software package for the measurement of surface roughness of fluvial ice covers,</p></list-item><list-item>
      <p id="d1e166">present quantitative metrics of surface ice roughness measurements for a range of river ice roughness conditions,</p></list-item><list-item>
      <p id="d1e170">test the hypothesis that the surface ice roughness of a newly frozen fluvial ice cover is indicative of the subsurface ice roughness of the same cover.</p></list-item></list></p>
      <p id="d1e173">To the authors' knowledge, no investigations of surface or subsurface ice roughness have been conducted on newly frozen fluvial ice covers. This work constitutes the first attempt at using RPA photogrammetry for the purposes of ice roughness measurements. Since the scope of this work was limited to a single study site, any conclusions drawn from this work would benefit from further evaluation at other study sites.</p>
</sec>
<sec id="Ch1.S2">
  <label>2</label><title>Background</title>
<sec id="Ch1.S2.SS1">
  <label>2.1</label><title>Field site description</title>
      <p id="d1e191">The Dauphin River is located approximately 250 km north of the city of Winnipeg, in Manitoba, Canada, as shown in Fig. <xref ref-type="fig" rid="Ch1.F1"/>. The indicated sites are part of a larger network of monitoring sites numbered sequentially from upstream (Lake St. Martin) to downstream (Lake Winnipeg). The prefix of each site (DRLL) stands for Dauphin River Levelogger, indicating a site which has equipment for continuous water level monitoring. A Water Survey of Canada (WSC) gauge station (05LM006), located <inline-formula><mml:math id="M1" display="inline"><mml:mo>≈</mml:mo></mml:math></inline-formula> 100 m downstream of site DRLL03, logs water surface elevation and flow at 5 min intervals and reports daily values. The data are periodically adjusted for ice effects during the winter.</p>

      <fig id="Ch1.F1" specific-use="star"><?xmltex \currentcnt{1}?><?xmltex \def\figurename{Figure}?><label>Figure 1</label><caption><p id="d1e204">Key map of the study location.</p></caption>
          <?xmltex \igopts{width=341.433071pt}?><graphic xlink:href="https://tc.copernicus.org/articles/15/4031/2021/tc-15-4031-2021-f01.png"/>

        </fig>

      <p id="d1e213">The Dauphin River is 52 km long and has steep, shallow banks that range between 110 and 160 m wide. The surficial geology of the area is composed of till with erratics, boulders, cobbles, and gravels observed throughout the channel. The most upstream 40 km of channel (Upper Dauphin River) has a mild slope (0.029 %) and is meandering. The bed composition of the Upper Dauphin River was observed to be primarily silt. The most downstream 12 km of channel (Lower Dauphin River) transitions into a well-defined riffle-pool system with a relatively high slope (0.16 %). Riffle sections in the Lower Dauphin River were observed to have a gravel bed with some boulders and erratics. Pool sections were observed to be silt bottomed. During winter ice formation, dramatic<?pagebreak page4033?> ice consolidation events, jams, and flooding have been reported by <xref ref-type="bibr" rid="bib1.bibx31" id="text.13"/> on the Lower Dauphin River. Lake Winnipeg water levels can have a significant effect on the most downstream 2 km of this reach, which is typically where the largest toe of the ice jam would form.</p>
</sec>
<sec id="Ch1.S2.SS2">
  <label>2.2</label><title>Flow resistance</title>
      <p id="d1e227">Surface roughness is an important parameter in the prediction of fluid flow along solid boundaries. This roughness creates drag along fluid boundary layers, generating the logarithmic fluid velocity distribution observed in open channel hydraulics. Rougher surfaces have been shown to exhibit greater flow resistance; however, it is not straightforward to quantify non-uniform, three-dimensional roughness elements. <xref ref-type="bibr" rid="bib1.bibx26" id="text.14"/> helped develop the concept of roughness height through equivalent sand grain roughness representing the height of sand particles fixed to the inside of pipes. More recently, an extensive discussion of methods used to represent the roughness of a heterogeneous three-dimensional surface layer from a surface profile was provided in <xref ref-type="bibr" rid="bib1.bibx15" id="text.15"/>. Many of these methods involve statistical analysis of the entire sample or some subset of the sample (i.e. peaks, valleys, etc.). <xref ref-type="bibr" rid="bib1.bibx18" id="text.16"/> used the difference between peaks and a locally derived average bed surface for the investigation of gravel bed roughness. <xref ref-type="bibr" rid="bib1.bibx25" id="text.17"/> recorded surface data derived from natural gravel point bars and found that the second-order moment of the frequency distribution yielded a suitable estimate of roughness height when compared to in situ field measurements. <xref ref-type="bibr" rid="bib1.bibx1" id="text.18"/> also investigated higher-order statistics but confirmed the use of sample standard deviation (SD) as an appropriate representation of gravel bed roughness height. For non-normal data, the interquartile range (IQR) is a more suitable representation of the spread of the data.</p>
      <p id="d1e245">For uniform flow conditions, Manning's equation is used to relate the discharge in an open-water or ice-covered channel to the water level. Flow resistance in this equation is introduced using Manning's <inline-formula><mml:math id="M2" display="inline"><mml:mi>n</mml:mi></mml:math></inline-formula>, which can be estimated from the roughness height of the channel boundary (in addition to many other modes of flow resistance that are outside the scope of the current study). Equation (<xref ref-type="disp-formula" rid="Ch1.E1"/>) shows a widely used quantitative method of estimating Manning's <inline-formula><mml:math id="M3" display="inline"><mml:mi>n</mml:mi></mml:math></inline-formula> from roughness height (<inline-formula><mml:math id="M4" display="inline"><mml:mi>D</mml:mi></mml:math></inline-formula> [m]) measurements proposed by <xref ref-type="bibr" rid="bib1.bibx28" id="text.19"/>. Strickler suggested a value of <inline-formula><mml:math id="M5" display="inline"><mml:mrow><mml:msub><mml:mi>c</mml:mi><mml:mi>n</mml:mi></mml:msub><mml:mo>=</mml:mo><mml:mn mathvariant="normal">0.047</mml:mn></mml:mrow></mml:math></inline-formula> for general use, but this can be adapted for specific applications <xref ref-type="bibr" rid="bib1.bibx29" id="paren.20"/>.
            <disp-formula id="Ch1.E1" content-type="numbered"><label>1</label><mml:math id="M6" display="block"><mml:mrow><mml:mi>n</mml:mi><mml:mo>≈</mml:mo><mml:msub><mml:mi>c</mml:mi><mml:mi>n</mml:mi></mml:msub><mml:msup><mml:mi>D</mml:mi><mml:mrow><mml:mn mathvariant="normal">1</mml:mn><mml:mo>/</mml:mo><mml:mn mathvariant="normal">6</mml:mn></mml:mrow></mml:msup></mml:mrow></mml:math></disp-formula></p>
      <?pagebreak page4034?><p id="d1e317">Fluvial ice formation has a significant impact on the roughness characteristics of northern rivers. Ice covers increase hydraulic resistance in fluvial systems by replacing the relatively friction-free air–water boundary with a rougher ice–water boundary. This expands the wetted perimeter of the channel and, if the ice cover completely bridges the channel, may or may not pressurize flow. The added source of roughness and constriction of flow results in upstream staging <xref ref-type="bibr" rid="bib1.bibx4" id="paren.21"/>. As with estimates of channel boundary roughness, ice roughness can also be judged qualitatively based on general observations with some success. The <xref ref-type="bibr" rid="bib1.bibx23" id="text.22"/> equation is a widely used empirical formula that provides a quantitative estimate of ice roughness in the form of Manning's <inline-formula><mml:math id="M7" display="inline"><mml:mi>n</mml:mi></mml:math></inline-formula>, as illustrated in Eq. (<xref ref-type="disp-formula" rid="Ch1.E2"/>). In this equation, <inline-formula><mml:math id="M8" display="inline"><mml:mrow><mml:msup><mml:mi>n</mml:mi><mml:mi>i</mml:mi></mml:msup></mml:mrow></mml:math></inline-formula> is the Manning's <inline-formula><mml:math id="M9" display="inline"><mml:mi>n</mml:mi></mml:math></inline-formula> of the underside of the ice cover and <inline-formula><mml:math id="M10" display="inline"><mml:mrow><mml:msub><mml:mi>t</mml:mi><mml:mi>i</mml:mi></mml:msub></mml:mrow></mml:math></inline-formula> [m] is the ice thickness in metres. In both cases, the subscript <inline-formula><mml:math id="M11" display="inline"><mml:mi>i</mml:mi></mml:math></inline-formula> refers to parameters related to ice.
            <disp-formula id="Ch1.E2" content-type="numbered"><label>2</label><mml:math id="M12" display="block"><mml:mrow><mml:msub><mml:mi>n</mml:mi><mml:mi>i</mml:mi></mml:msub><mml:mo>≈</mml:mo><mml:mn mathvariant="normal">0.0252</mml:mn><mml:mi>ln⁡</mml:mi><mml:mo>(</mml:mo><mml:msub><mml:mi>t</mml:mi><mml:mi>i</mml:mi></mml:msub><mml:mo>)</mml:mo><mml:mo>+</mml:mo><mml:mn mathvariant="normal">0.0706</mml:mn></mml:mrow></mml:math></disp-formula></p>
      <p id="d1e401">This relationship is based on measurements conducted on rivers in Russia several decades ago, and it has served well as an estimation tool for engineering applications. Using more complex data, Eq. (<xref ref-type="disp-formula" rid="Ch1.E1"/>) was adapted by <xref ref-type="bibr" rid="bib1.bibx4" id="text.23"/> for use in the estimation of the roughness of newly formed ice jams, resulting in Eq. (<xref ref-type="disp-formula" rid="Ch1.E3"/>).
            <disp-formula id="Ch1.E3" content-type="numbered"><label>3</label><mml:math id="M13" display="block"><mml:mrow><mml:msub><mml:mi>n</mml:mi><mml:mi>i</mml:mi></mml:msub><mml:mo>≈</mml:mo><mml:mn mathvariant="normal">0.095</mml:mn><mml:msup><mml:mi>D</mml:mi><mml:mrow><mml:mn mathvariant="normal">1</mml:mn><mml:mo>/</mml:mo><mml:mn mathvariant="normal">2</mml:mn></mml:mrow></mml:msup><mml:msup><mml:mi>R</mml:mi><mml:mrow><mml:mn mathvariant="normal">1</mml:mn><mml:mo>/</mml:mo><mml:mn mathvariant="normal">3</mml:mn></mml:mrow></mml:msup></mml:mrow></mml:math></disp-formula></p>
      <p id="d1e446">The value given for <inline-formula><mml:math id="M14" display="inline"><mml:mrow><mml:msub><mml:mi>c</mml:mi><mml:mi>n</mml:mi></mml:msub><mml:mo>=</mml:mo><mml:mn mathvariant="normal">0.095</mml:mn></mml:mrow></mml:math></inline-formula> has been determined to be representative for ice jams. Additionally, the inclusion of the hydraulic radius <inline-formula><mml:math id="M15" display="inline"><mml:mi>R</mml:mi></mml:math></inline-formula> [m] accounts for the fact that the roughness elements of ice jams are often of such magnitude as to increase relative roughness to the point where it has a significant impact on the hydraulic radius. This relationship is only valid for newly formed ice jams; immediately after formation, the ice is subject to shear forces from the water flowing underneath, which slowly smoothens the subsurface of the ice cover <xref ref-type="bibr" rid="bib1.bibx3" id="paren.24"/>.</p>
</sec>
<sec id="Ch1.S2.SS3">
  <label>2.3</label><title>Remotely piloted aircraft photogrammetry</title>
      <p id="d1e482">RPAs equipped with high-resolution digital cameras have been used extensively in the collection of near-surface photographic and topographic data <xref ref-type="bibr" rid="bib1.bibx7 bib1.bibx30" id="paren.25"/>. They are smaller and more cost-effective than conventional aircraft, allowing for much more versatile data collection. Compared to manual surveying methods they can collect a greater volume of data in less time and greatly reduce risk to personnel. For the purposes of topographic data collection, the current common practice is to further process images collected with an RPA using structure from motion (SfM) photogrammetry <xref ref-type="bibr" rid="bib1.bibx14" id="paren.26"/>. SfM photogrammetry is a technique that infers a three-dimensional structure from a series of overlapping, offset two-dimensional images <xref ref-type="bibr" rid="bib1.bibx34" id="paren.27"/>. <xref ref-type="bibr" rid="bib1.bibx24" id="text.28"/> used this method to monitor the progression of the Super-Sauze landslide, a task too dangerous to monitor manually. <xref ref-type="bibr" rid="bib1.bibx10" id="text.29"/> employed RPA photogrammetry to document the layout of ancient ruins in Peru, which if done manually would have risked the integrity of the site. <xref ref-type="bibr" rid="bib1.bibx19" id="text.30"/> found use for RPAs in the monitoring of riverbank erosion. RPAs were even used by <xref ref-type="bibr" rid="bib1.bibx2" id="text.31"/> in the mapping of river ice in Norway.</p><?xmltex \hack{\newpage}?>
</sec>
</sec>
<sec id="Ch1.S3">
  <label>3</label><title>Methodology</title>
      <p id="d1e517">Five field sites were selected in this study, their relative location along the bed profile of the Lower Dauphin River is illustrated in Fig. <xref ref-type="fig" rid="Ch1.F2"/>. Data were gathered during the winter months of 2017–2019. A relatively smooth, unconsolidated ice cover has been observed to form at DRLL03b in all previous study years, due to its low bed slope (0.029 %). Sites DRLL05 and DRLL06 exhibited substantial ice dynamics, as they are within the higher gradient (0.16 %) portion of the Lower Dauphin River <xref ref-type="bibr" rid="bib1.bibx32" id="paren.32"/>. Sites DRLL08 and DRLL08a had much milder water surface slopes, due to the backwater effect from Lake Winnipeg. The toe of an ice jam has formed in previous years near sites DRLL08a and DRLL08. These sites were selected in an effort to compare the efficacy of the RPA photogrammetry method on different ice conditions and to determine if the methods can distinguish roughness differences between sites.</p>

      <?xmltex \floatpos{t}?><fig id="Ch1.F2" specific-use="star"><?xmltex \currentcnt{2}?><?xmltex \def\figurename{Figure}?><label>Figure 2</label><caption><p id="d1e527">Channel bed profile of the Lower Dauphin River, with selected study locations. Flow is 86 m<inline-formula><mml:math id="M16" display="inline"><mml:msup><mml:mi/><mml:mn mathvariant="normal">3</mml:mn></mml:msup></mml:math></inline-formula>/s.</p></caption>
        <?xmltex \igopts{width=341.433071pt}?><graphic xlink:href="https://tc.copernicus.org/articles/15/4031/2021/tc-15-4031-2021-f02.png"/>

      </fig>

<sec id="Ch1.S3.SS1">
  <label>3.1</label><title>Field methods</title>
<sec id="Ch1.S3.SS1.SSS1">
  <label>3.1.1</label><title>Photogrammetry</title>
      <p id="d1e559">Remote monitoring of weather conditions informed the selection of field visit dates, with most data collected less than 1 week after the onset of ice formation on the study sites. A total of ten 1 m<inline-formula><mml:math id="M17" display="inline"><mml:msup><mml:mi/><mml:mn mathvariant="normal">2</mml:mn></mml:msup></mml:math></inline-formula> high-visibility medium-density fibreboard targets were distributed on the grounded ice near the left bank of each site and on snow near Provincial Road (PR) 513. A typical layout of targets is shown in Fig. <xref ref-type="fig" rid="Ch1.F3"/>, which illustrates how the targets are placed exclusively on the left bank of the river. The targets were grouped in this way since the right bank was inaccessible. Ideally, the targets would have been evenly distributed across the entire study area <xref ref-type="bibr" rid="bib1.bibx2 bib1.bibx17" id="paren.33"/>. In Sect. <xref ref-type="sec" rid="Ch1.S3.SS1.SSS3"/>, the effects of target distribution on digital elevation model (DEM) accuracy are tested.</p>

      <?xmltex \floatpos{t}?><fig id="Ch1.F3" specific-use="star"><?xmltex \currentcnt{3}?><?xmltex \def\figurename{Figure}?><label>Figure 3</label><caption><p id="d1e580">Typical target distribution.</p></caption>
            <?xmltex \igopts{width=341.433071pt}?><graphic xlink:href="https://tc.copernicus.org/articles/15/4031/2021/tc-15-4031-2021-f03.jpg"/>

          </fig>

      <p id="d1e589">After targets were placed, their centres were surveyed using a Leica Viva GS14<sup>®</sup> survey-grade real-time kinematic (RTK) Global Navigation Satellite System (GNSS) base-and-rover system, which is typically observed to have an in-field reported horizontal error of <inline-formula><mml:math id="M18" display="inline"><mml:mrow><mml:mo>≈</mml:mo><mml:mn mathvariant="normal">2</mml:mn></mml:mrow></mml:math></inline-formula> cm and a vertical error of <inline-formula><mml:math id="M19" display="inline"><mml:mrow><mml:mo>≈</mml:mo><mml:mn mathvariant="normal">3</mml:mn></mml:mrow></mml:math></inline-formula> cm. The Canadian Geodetic Vertical Datum of 2013 (CGVD2013) geoid was used in the recording of all surveyed elevations. Localization was assessed using a Manitoba Infrastructure (MI) benchmark located near DRLL03 and verified using the Natural Resources of Canada Canadian Spatial Reference System Precise Point Positioning (CSRS-PPP) service. Further benchmarks were established using the CSRS-PPP and “leap-frogging” to further benchmarks. In the 2019–2020 season, some RPA flights were completed without targets to allow for more flights to be completed during the field visit. A comparison between the representative metrics calculated from a DEM with and without geo-rectification is presented in Sect. <xref ref-type="sec" rid="Ch1.S4.SS1"/>.</p>
      <?pagebreak page4035?><p id="d1e618">Once all field personnel had finished active tasks, the RPA was launched, and field staff remained stationary for the duration of the RPA flight. A DJI Phantom 4 Professional<sup>®</sup> RPA was flown at an approximate altitude of 30 m, with overlapping photos taken every 10 m, at a 0<inline-formula><mml:math id="M20" display="inline"><mml:msup><mml:mi/><mml:mo>∘</mml:mo></mml:msup></mml:math></inline-formula> or 20<inline-formula><mml:math id="M21" display="inline"><mml:msup><mml:mi/><mml:mo>∘</mml:mo></mml:msup></mml:math></inline-formula> camera tilt. The on-board 20 megapixel camera had an 84<inline-formula><mml:math id="M22" display="inline"><mml:msup><mml:mi/><mml:mo>∘</mml:mo></mml:msup></mml:math></inline-formula> field of view with a 1 in. CMOS sensor. The RPA flight transected the river and included PR 513 and forest on the left and right bank, respectively. The RPA was flown only if wind speeds measured by a handheld digital anemometer were less than 36 km/h. Since light conditions could drastically impact the quality of images taken, the RPA was flown only during daytime and during clear or lightly overcast conditions. Typical capture dimensions of an RPA flight were 90 m in the stream-wise ordinate and 230 m across the river. During the 2019–2020 field season, the RPA mission planning application Pix4Dcapture<sup>®</sup> was used to plan and automate RPA flights over study areas. This greatly reduced the required flight time and produced similar (if not better) photo coverage.</p>
</sec>
<sec id="Ch1.S3.SS1.SSS2">
  <label>3.1.2</label><title>Hydraulic parameters</title>
      <p id="d1e662">Water pressure was recorded every 8 min at the study sites using Solinst Levelogger<sup>®</sup> Edge 3001 M5 pressure transducers and accompanying nearby Solinst Barologger<sup>®</sup> Edge 3001. The listed accuracy of these devices is <inline-formula><mml:math id="M23" display="inline"><mml:mrow><mml:mo>±</mml:mo><mml:mn mathvariant="normal">0.003</mml:mn></mml:mrow></mml:math></inline-formula> m and <inline-formula><mml:math id="M24" display="inline"><mml:mo>±</mml:mo></mml:math></inline-formula>0.05 kPa, respectively. These instruments were installed before the ice season began (typically October), removed for download and maintenance after the end of the ice season (typically April–May), and were then subsequently reinstalled for summer observations. During installation, the water surface was surveyed for use in post-processing to determine the absolute water surface elevation of the observations in metres above sea level (m a.s.l.). Additionally,<?pagebreak page4036?> the observed barometric pressure, converted to its equivalent depth of water, was subtracted from the pressure observations. The observed water level and previously measured channel bathymetry were used to estimate hydraulic radius at each of the study sites using a one-dimensional at-a-station hydraulic model based on Manning's equation.</p>
      <p id="d1e688">During the 2019–2020 field season, holes were drilled in the ice cover at safe locations to determine ice thickness; however, not all locations allowed such convenient means of measurement. Ice thickness measurements of a rough, consolidating ice cover are impossible to conduct, and thus indirect measurement procedures were necessary, depending on the local conditions. In some cases, when it was clear that the ice cover was comprised of one or two layers of ice pans of known dimension, ice thicknesses were estimated through visual observation and  photographs taken by stationary trail cameras. For highly consolidated thick ice covers, mid-winter RTK surveys were conducted to measure top of ice along the Lower Dauphin River. Lateral ice transects were conducted at locations where ice became grounded. The previously surveyed ground elevations were subtracted from these top-of-grounded-ice measurements to provide estimates of ice thickness. Late in the winter (typically February) after a stable ice cover had formed, a top-of-ice elevation survey was undertaken along the Lower Dauphin River using the base-and-rover system. Truncated transects of ice thickness were also surveyed at locations where ground elevation had previously been surveyed. A transect was performed at the DRLL06 and DRLL05 sites but not at DRLL08 due to safety concerns. During the 2019–2020 field season, holes were drilled in the ice cover at safe locations following established safe work procedures to determine ice thickness.</p>
</sec>
<sec id="Ch1.S3.SS1.SSS3">
  <label>3.1.3</label><title>Field accuracy tests</title>
      <p id="d1e699">There was a need to quantify the impact of the required ground control grouping on the left bank of the study area. The field methods described in Sect. <xref ref-type="sec" rid="Ch1.S3.SS1.SSS1"/> were repeated at River's Edge Nursery in La Barriere, Manitoba. A fully dry land study area of equivalent size to typical study areas flown at the Dauphin River was delineated, and 15 targets were distributed. The targets were conceptually grouped into three areas: typical, middle, and end. The “typical” group represented a target distribution that was generally produced during fieldwork at the Dauphin River sites. The “middle” and “end” target groups were supplemental and would be added or subtracted from the photogrammetry analysis to test their respective impacts on DEM accuracy. The distribution of targets in the study area is represented in Fig. <xref ref-type="fig" rid="Ch1.F4"/>. Finally, after the RPA flight was conducted, 10 independent and unmarked locations were captured by RTK-GNSS survey as a check for accuracy in subsequent data analysis.</p>

      <?xmltex \floatpos{t}?><fig id="Ch1.F4"><?xmltex \currentcnt{4}?><?xmltex \def\figurename{Figure}?><label>Figure 4</label><caption><p id="d1e708">Accuracy test experimental setup: <bold>(a)</bold> typical, <bold>(b)</bold> middle, <bold>(c)</bold> end.</p></caption>
            <?xmltex \igopts{width=241.848425pt}?><graphic xlink:href="https://tc.copernicus.org/articles/15/4031/2021/tc-15-4031-2021-f04.png"/>

          </fig>

<?xmltex \hack{\newpage}?>
</sec>
</sec>
<sec id="Ch1.S3.SS2">
  <label>3.2</label><title>Laboratory methods</title>
<sec id="Ch1.S3.SS2.SSS1">
  <label>3.2.1</label><title>Photogrammetry</title>
      <p id="d1e745">The photogrammetry processing software selected for this study was PhotoScan Professional<sup>®</sup> from AgiSoft LLC. <xref ref-type="bibr" rid="bib1.bibx17" id="text.34"/> compared their custom research-grade photogrammetry algorithms to results obtained from  Pix4UAV Desktop<sup>®</sup> and PhotoScan Professional<sup>®</sup>. Their findings suggested that these commercial packages performed similarly to their software, with PhotoScan Professional<sup>®</sup> performing somewhat better than Pix4UAV Desktop<sup>®</sup>. PhotoScan Professional<sup>®</sup> is also considered to be a relatively fully featured and complex <xref ref-type="bibr" rid="bib1.bibx11" id="paren.35"/> tool compared to other options.</p>
      <p id="d1e773">Images were imported into PhotoScan Professional<sup>®</sup>, and aligned to create a sparse point cloud of tie points. Where targets were used, they were identified in all images containing them, and their coordinates were imported to provide geo-rectification of the resultant point cloud. A dense point cloud was then generated, followed by a DEM. An example point cloud consisting of <inline-formula><mml:math id="M25" display="inline"><mml:mrow><mml:mo>≈</mml:mo><mml:mn mathvariant="normal">15</mml:mn></mml:mrow></mml:math></inline-formula> million points and the corresponding DEM are shown in Figs. <xref ref-type="fig" rid="Ch1.F5"/> and <xref ref-type="fig" rid="Ch1.F6"/>, respectively.</p>

      <?xmltex \floatpos{t}?><fig id="Ch1.F5"><?xmltex \currentcnt{5}?><?xmltex \def\figurename{Figure}?><label>Figure 5</label><caption><p id="d1e795">Example point cloud: DRLL06, 21 November 2018.</p></caption>
            <?xmltex \igopts{width=241.848425pt}?><graphic xlink:href="https://tc.copernicus.org/articles/15/4031/2021/tc-15-4031-2021-f05.png"/>

          </fig>

      <?xmltex \floatpos{t}?><fig id="Ch1.F6"><?xmltex \currentcnt{6}?><?xmltex \def\figurename{Figure}?><label>Figure 6</label><caption><p id="d1e807">Example DEM: DRLL06, 21 November 2018.</p></caption>
            <?xmltex \igopts{width=236.157874pt}?><graphic xlink:href="https://tc.copernicus.org/articles/15/4031/2021/tc-15-4031-2021-f06.png"/>

          </fig>

</sec>
<sec id="Ch1.S3.SS2.SSS2">
  <label>3.2.2</label><title>Accuracy testing</title>
      <p id="d1e824">The impact of placing all control points on an extreme end of the study area was tested through a detailed trial on an open field. Groups of targets were used as input to the photogrammetry software, and the resultant DEMs were compared to the 10 independent survey points. The DEM generated using<?pagebreak page4037?> all available targets was assumed to be the most correct representation of the land surface, against which all other target groupings were compared. A maximum acceptable vertical difference of 0.03 m between the DEMs and the independent survey points was adopted. This value was chosen based on the typical error observed in the data gathered by the RTK GNSS base-and-rover system. This system was the limiting factor for accuracy in this study since it was the tool which informs the absolute spatial position of all field equipment. The following target scenarios were tested: “all points”, utilizing all ground control targets; “typical points”, using all the targets identified in the typical subset; “three points”, using a subset of three targets from the typical subset; and “two points”, using a subset of two targets from the typical subset. In the two- and three-point tests, the most spatially distributed targets within the typical subset were selected. An additional test was required to determine if systemic errors were introduced in DEMs generated without the use of geo-rectification targets. This scenario is referred to as the “no points” case. Data collected at site DRLL06 on 13 November 2019 were prepared with and without the inclusion of control point data. A maximum acceptable error of 5 % was adopted to evaluate the results of this test.</p>
</sec>
<sec id="Ch1.S3.SS2.SSS3">
  <label>3.2.3</label><title>Roughness characterization</title>
      <p id="d1e835">To avoid unwanted influences in the surface slope and texture, a 50 m<inline-formula><mml:math id="M26" display="inline"><mml:msup><mml:mi/><mml:mn mathvariant="normal">2</mml:mn></mml:msup></mml:math></inline-formula> subsample from the centre of the river was taken, which excluded all overbank objects and sections of the ice cover that were near to the bank. Additionally, a three-dimensional plane of best fit linear model (LM) was found for each sample and then subtracted from the surface data. The goal of this was to normalize each dataset, setting the average surface elevation to 0 and removing the river slope from the sample data. <xref ref-type="bibr" rid="bib1.bibx15" id="text.36"/> noted that the average surface elevation is the most commonly used and most sensible reference standard from which to assess roughness height. By shifting the elevation data down to a base elevation of 0 and removing unwanted patterns, each data point was transformed from an elevation to a roughness height.</p>
      <p id="d1e850">A two-dimensional fast Fourier transform (FFT) was then applied to each subsample, with the goal of filtering the input data and removing other surface trends beyond those addressed with the plane of best fit. The combination of the LM and FFT adjustment and filtering processes will be referred to as LMFFT. Through an analysis of dominant frequencies it was found that the lowest frequencies (<inline-formula><mml:math id="M27" display="inline"><mml:mrow><mml:mo>&lt;</mml:mo><mml:mn mathvariant="normal">1</mml:mn></mml:mrow></mml:math></inline-formula> m<inline-formula><mml:math id="M28" display="inline"><mml:msup><mml:mi/><mml:mrow><mml:mo>-</mml:mo><mml:mn mathvariant="normal">1</mml:mn></mml:mrow></mml:msup></mml:math></inline-formula>) had the largest amplitudes, while the highest-frequency signals (<inline-formula><mml:math id="M29" display="inline"><mml:mrow><mml:mo>&gt;</mml:mo><mml:mn mathvariant="normal">5</mml:mn></mml:mrow></mml:math></inline-formula> m<inline-formula><mml:math id="M30" display="inline"><mml:msup><mml:mi/><mml:mrow><mml:mo>-</mml:mo><mml:mn mathvariant="normal">1</mml:mn></mml:mrow></mml:msup></mml:math></inline-formula>) had the lowest amplitudes. A bandpass filter constructed with a low-pass value of 0.08 m was generally found to produce the best results. The high-pass component of the filter was adjusted through extensive iterative visual analysis of the image. High-pass cutoff values that were too aggressive caused obvious edge distortion, while values that were too conservative caused insufficient trend removal. The chosen high-pass wavelength cut-off values for all images ranged from 70 to 70.5 m. Figure <xref ref-type="fig" rid="Ch1.F7"/>a and b show the DEM before and after the application of this filter.</p>

      <?xmltex \floatpos{t}?><fig id="Ch1.F7" specific-use="star"><?xmltex \currentcnt{7}?><?xmltex \def\figurename{Figure}?><label>Figure 7</label><caption><p id="d1e901">Summary of impacts of filtering through Fourier analysis for DRLL06, 13 November 2019: <bold>(a)</bold> original ice surface DEM and <bold>(b)</bold> LMFFT-processed ice surface DEM.</p></caption>
            <?xmltex \igopts{width=341.433071pt}?><graphic xlink:href="https://tc.copernicus.org/articles/15/4031/2021/tc-15-4031-2021-f07.png"/>

          </fig>

      <p id="d1e917">A representative value of roughness height was required from each dataset for further analysis and comparison. Based on a review of relevant literature in the fields of photogrammetry, fluvial geomorphology, and roughness characterization, various statistical methods for roughness height characterization were considered, and several were chosen for further consideration in this study (Table <xref ref-type="table" rid="Ch1.T1"/>). The full two-dimensional processed roughness height data, hereafter referred to as LMFFT-full, was then further analysed to create a second dataset that was comprised of only peak values of roughness height. This dataset, called LMFFT-peaks, was developed by extracting peak roughness heights using a three-dimensional implementation of the “peakpick” algorithm <xref ref-type="bibr" rid="bib1.bibx33" id="paren.37"/> available in the R programming language.</p>

<?xmltex \floatpos{t}?><table-wrap id="Ch1.T1" specific-use="star"><?xmltex \currentcnt{1}?><label>Table 1</label><caption><p id="d1e928">Selected statistical metrics used to quantify RPA-measured ice surface roughness.</p></caption><oasis:table frame="topbot"><oasis:tgroup cols="3">
     <oasis:colspec colnum="1" colname="col1" align="left"/>
     <oasis:colspec colnum="2" colname="col2" align="left"/>
     <oasis:colspec colnum="3" colname="col3" align="left"/>
     <oasis:thead>
       <oasis:row rowsep="1">
         <oasis:entry colname="col1">Method</oasis:entry>
         <oasis:entry colname="col2">Data group</oasis:entry>
         <oasis:entry colname="col3">Reference</oasis:entry>
       </oasis:row>
     </oasis:thead>
     <oasis:tbody>
       <oasis:row>
         <oasis:entry colname="col1">IQR</oasis:entry>
         <oasis:entry colname="col2">LMFFT-full</oasis:entry>
         <oasis:entry colname="col3">
                      <xref ref-type="bibr" rid="bib1.bibx1" id="text.38"/>
                    </oasis:entry>
       </oasis:row>
       <oasis:row>
         <oasis:entry colname="col1">SD</oasis:entry>
         <oasis:entry colname="col2">LMFFT-full</oasis:entry>
         <oasis:entry colname="col3">
                      <xref ref-type="bibr" rid="bib1.bibx1" id="text.39"/>
                    </oasis:entry>
       </oasis:row>
       <oasis:row>
         <oasis:entry colname="col1">Minimum peak value</oasis:entry>
         <oasis:entry colname="col2">LMFFT-peaks</oasis:entry>
         <oasis:entry colname="col3">
                      <xref ref-type="bibr" rid="bib1.bibx15" id="text.40"/>
                    </oasis:entry>
       </oasis:row>
       <oasis:row>
         <oasis:entry colname="col1">Maximum peak value</oasis:entry>
         <oasis:entry colname="col2">LMFFT-peaks</oasis:entry>
         <oasis:entry colname="col3">
                      <xref ref-type="bibr" rid="bib1.bibx15" id="text.41"/>
                    </oasis:entry>
       </oasis:row>
       <oasis:row>
         <oasis:entry colname="col1">Average of peaks</oasis:entry>
         <oasis:entry colname="col2">LMFFT-peaks</oasis:entry>
         <oasis:entry colname="col3"><xref ref-type="bibr" rid="bib1.bibx15" id="text.42"/>, <xref ref-type="bibr" rid="bib1.bibx18" id="text.43"/></oasis:entry>
       </oasis:row>
       <oasis:row>
         <oasis:entry colname="col1">84th percentile of peaks</oasis:entry>
         <oasis:entry colname="col2">LMFFT-peaks</oasis:entry>
         <oasis:entry colname="col3">
                      <xref ref-type="bibr" rid="bib1.bibx4" id="text.44"/>
                    </oasis:entry>
       </oasis:row>
     </oasis:tbody>
   </oasis:tgroup></oasis:table></table-wrap>

      <p id="d1e1052">Once the various metrics of roughness height at each site were determined using the LMFTT-full and LMFTT-peaks datasets, the corresponding hydraulic radius for each site was determined using a simple one-dimensional hydraulic model based on Manning's equation. The model used the observed water level, estimated ice thickness, surveyed channel cross section, and an assumed specific gravity of ice of 0.916. Equation (<xref ref-type="disp-formula" rid="Ch1.E3"/>) was then used to calculate the RPA-measured ice surface Manning's <inline-formula><mml:math id="M31" display="inline"><mml:mi>n</mml:mi></mml:math></inline-formula> for each site.</p>
      <p id="d1e1064">In addition to a numerical characterization of roughness, a qualitative classification of roughness type was undertaken using <inline-formula><mml:math id="M32" display="inline"><mml:mi>k</mml:mi></mml:math></inline-formula>-means clustering, which is the most commonly used clustering approach <xref ref-type="bibr" rid="bib1.bibx20" id="paren.45"/>. The input data were IQR and kurtosis of the LMFFT-full and LMFFT-peaks data and the median of the LMFFT-peaks data. Mean and median of the LMFFT-full data were excluded since they were set to 0 through the LMFFT process. Interestingly, the kurtosis was found to be a more useful metric for this analysis than the mean, standard deviation, or skewness. Since they was used in the computation of kurtosis, SD and skewness were<?pagebreak page4038?> removed from both datasets. The data were then mean-centred, and the optimal number of clusters was determined using the average silhouette method. The Euclidean distance formula was used in determination of the <inline-formula><mml:math id="M33" display="inline"><mml:mi>k</mml:mi></mml:math></inline-formula>-means clustering.</p>
</sec>
<sec id="Ch1.S3.SS2.SSS4">
  <label>3.2.4</label><title>Roughness comparison</title>
      <p id="d1e1092">The RPA-measured ice surface Manning's <inline-formula><mml:math id="M34" display="inline"><mml:mi>n</mml:mi></mml:math></inline-formula> values from each site and using each of the roughness metrics were compared to Manning's <inline-formula><mml:math id="M35" display="inline"><mml:mi>n</mml:mi></mml:math></inline-formula> values computed using the commonly used Nezhikhovskiy equation (Eq. <xref ref-type="disp-formula" rid="Ch1.E2"/>), as it has been found to make reasonable predictions of flow resistance of the underside of an ice cover. A similar comparison between RPA-measured ice surface Manning's <inline-formula><mml:math id="M36" display="inline"><mml:mi>n</mml:mi></mml:math></inline-formula> and the site ice thickness was also made. This was undertaken since ice thickness is the sole input parameter of the Nezhikhovskiy equation and could perhaps provide another perspective on the data. A linear regression was computed for the data in each plot. The quality of regressions were evaluated using the correlation coefficient (<inline-formula><mml:math id="M37" display="inline"><mml:mrow><mml:msup><mml:mi>R</mml:mi><mml:mn mathvariant="normal">2</mml:mn></mml:msup></mml:mrow></mml:math></inline-formula>), <inline-formula><mml:math id="M38" display="inline"><mml:mi>F</mml:mi></mml:math></inline-formula> statistic (<inline-formula><mml:math id="M39" display="inline"><mml:mi>F</mml:mi></mml:math></inline-formula>), <inline-formula><mml:math id="M40" display="inline"><mml:mi>p</mml:mi></mml:math></inline-formula> value (<inline-formula><mml:math id="M41" display="inline"><mml:mi>p</mml:mi></mml:math></inline-formula>), and root-mean-square error (RMSE) values. The <inline-formula><mml:math id="M42" display="inline"><mml:mi>p</mml:mi></mml:math></inline-formula> value corresponds to the probability that the data show a significant trend through random chance and not an actual relationship. The criteria for significance in the <inline-formula><mml:math id="M43" display="inline"><mml:mi>p</mml:mi></mml:math></inline-formula> value was set to <inline-formula><mml:math id="M44" display="inline"><mml:mrow><mml:mi mathvariant="italic">α</mml:mi><mml:mo>≤</mml:mo><mml:mn mathvariant="normal">0.05</mml:mn></mml:mrow></mml:math></inline-formula>. The <inline-formula><mml:math id="M45" display="inline"><mml:mi>F</mml:mi></mml:math></inline-formula> value is a component of the <inline-formula><mml:math id="M46" display="inline"><mml:mi>p</mml:mi></mml:math></inline-formula> value and is another method of testing the significance of a linear regression when compared to its critical value <inline-formula><mml:math id="M47" display="inline"><mml:mrow><mml:msub><mml:mi>F</mml:mi><mml:mi mathvariant="normal">crit</mml:mi></mml:msub></mml:mrow></mml:math></inline-formula>, which is calculated in Sect. <xref ref-type="sec" rid="Ch1.S5.SS2"/>. The validity of using linear regression modelling with these data was evaluated by testing the normality of residuals using the Shapiro–Wilk test for normality (<inline-formula><mml:math id="M48" display="inline"><mml:mrow><mml:mi>p</mml:mi><mml:mo>&gt;</mml:mo><mml:mn mathvariant="normal">0.05</mml:mn></mml:mrow></mml:math></inline-formula>). This test was selected due to its wide usage in data analysis and its superior power compared to many other widely used normality tests <xref ref-type="bibr" rid="bib1.bibx27" id="paren.46"/>. Q–Q plots of residuals were used to confirm the test statistics.</p>
</sec>
</sec>
</sec>
<sec id="Ch1.S4">
  <label>4</label><title>Results</title>
<sec id="Ch1.S4.SS1">
  <label>4.1</label><title>RPA performance</title>
      <p id="d1e1245">The RPA chosen for this study performed well during all field visits in various weather and cloud conditions. In extreme cold (<inline-formula><mml:math id="M49" display="inline"><mml:mrow><mml:mo>-</mml:mo><mml:mn mathvariant="normal">20</mml:mn></mml:mrow></mml:math></inline-formula> <inline-formula><mml:math id="M50" display="inline"><mml:msup><mml:mi/><mml:mo>∘</mml:mo></mml:msup></mml:math></inline-formula>C and below) the RPA performed all functions well, although the battery life was reduced by approximately 50 %. It was found that the RPA had to be powered on in a warm area, such as the heated cab of the field vehicle. Once powered on, it could then be placed outside for normal operations.</p>
      <p id="d1e1267">RPA photogrammetry performed very well across a variety of scenarios in the land-based field accuracy tests. The worst observed DEM accuracy was found in the two points scenario, with an average vertical difference of 6.30 m calculated between the DEM and the 10 independently surveyed test points. The three points, typical points, and all points<?pagebreak page4039?> scenarios all had the same average vertical difference of 0.03 m. These three scenarios were all within the acceptable vertical difference criteria of 0.03 m identified in Sect. <xref ref-type="sec" rid="Ch1.S3.SS2.SSS2"/>. The maximum error of the DEM was determined to be the sum of the expected vertical accuracy of the RTK-GNSS unit, 0.03 m, and the observed vertical difference between the DEMs and surveyed locations, 0.03 m, for a total expected error of 0.06 m. The results of the no points test case, which sought to determine if a lack of geo-rectification targets introduces systemic errors into the DEM data, are presented in Table <xref ref-type="table" rid="Ch1.T2"/>.</p>

<?xmltex \floatpos{t}?><table-wrap id="Ch1.T2" specific-use="star"><?xmltex \currentcnt{2}?><label>Table 2</label><caption><p id="d1e1277">Percent and absolute difference between various statistical metrics computed from geo-rectified and non-geo-rectified (no ground control points) DEMs produced from data collected at DRLL06, 13 November 2019.</p></caption><oasis:table frame="topbot"><oasis:tgroup cols="7">
     <oasis:colspec colnum="1" colname="col1" align="left"/>
     <oasis:colspec colnum="2" colname="col2" align="right"/>
     <oasis:colspec colnum="3" colname="col3" align="right"/>
     <oasis:colspec colnum="4" colname="col4" align="right"/>
     <oasis:colspec colnum="5" colname="col5" align="right"/>
     <oasis:colspec colnum="6" colname="col6" align="right"/>
     <oasis:colspec colnum="7" colname="col7" align="right"/>
     <oasis:thead>
       <oasis:row rowsep="1">
         <oasis:entry colname="col1"/>
         <oasis:entry colname="col2">Mean</oasis:entry>
         <oasis:entry colname="col3">Median</oasis:entry>
         <oasis:entry colname="col4">IQR</oasis:entry>
         <oasis:entry colname="col5">SD</oasis:entry>
         <oasis:entry colname="col6">Skew.</oasis:entry>
         <oasis:entry colname="col7">Kurt.</oasis:entry>
       </oasis:row>
     </oasis:thead>
     <oasis:tbody>
       <oasis:row>
         <oasis:entry colname="col1">Absolute</oasis:entry>
         <oasis:entry colname="col2"><inline-formula><mml:math id="M51" display="inline"><mml:mo>&lt;</mml:mo></mml:math></inline-formula>0.01 m</oasis:entry>
         <oasis:entry colname="col3"><inline-formula><mml:math id="M52" display="inline"><mml:mo>&lt;</mml:mo></mml:math></inline-formula>0.01 m</oasis:entry>
         <oasis:entry colname="col4"><inline-formula><mml:math id="M53" display="inline"><mml:mo>&lt;</mml:mo></mml:math></inline-formula>0.01 m</oasis:entry>
         <oasis:entry colname="col5"><inline-formula><mml:math id="M54" display="inline"><mml:mo>&lt;</mml:mo></mml:math></inline-formula>0.01 m</oasis:entry>
         <oasis:entry colname="col6">0.02</oasis:entry>
         <oasis:entry colname="col7">0.08</oasis:entry>
       </oasis:row>
       <oasis:row>
         <oasis:entry colname="col1">Percent</oasis:entry>
         <oasis:entry colname="col2"><inline-formula><mml:math id="M55" display="inline"><mml:mo>&lt;</mml:mo></mml:math></inline-formula>1</oasis:entry>
         <oasis:entry colname="col3">6</oasis:entry>
         <oasis:entry colname="col4">2</oasis:entry>
         <oasis:entry colname="col5">2</oasis:entry>
         <oasis:entry colname="col6">5</oasis:entry>
         <oasis:entry colname="col7">2</oasis:entry>
       </oasis:row>
     </oasis:tbody>
   </oasis:tgroup></oasis:table></table-wrap>

</sec>
<sec id="Ch1.S4.SS2">
  <label>4.2</label><title>Dauphin River results</title>
      <p id="d1e1418">During the 2018–2019 and 2019–2020 field seasons, the Dauphin River experienced much lower flows than those observed in the previous few years. The mean seasonal flow between November and March for each season was 74 and 90 m<inline-formula><mml:math id="M56" display="inline"><mml:msup><mml:mi/><mml:mn mathvariant="normal">3</mml:mn></mml:msup></mml:math></inline-formula>/s for 2018–2019 and 2019–2020, respectively, compared to 178 and 195 m<inline-formula><mml:math id="M57" display="inline"><mml:msup><mml:mi/><mml:mn mathvariant="normal">3</mml:mn></mml:msup></mml:math></inline-formula>/s in 2016–2017 and 2017–2018, respectively, although in prior years lower flows were noted. This resulted in notably thinner ice covers, more thermal ice growth, and less extensive ice jamming than was reported by <xref ref-type="bibr" rid="bib1.bibx31" id="text.47"/>. The average values of observed ice thickness reduced from 2.9 m at site DRLL05 and DRLL06 in 2017–2018 to 1.8  and 0.8 m in 2018–2019 and 2019–2020, respectively.</p>
<sec id="Ch1.S4.SS2.SSS1">
  <label>4.2.1</label><title>Statistical properties of ice roughness height distributions</title>
      <p id="d1e1449">Several different forms of ice roughness were observed in DEMs produced using RPA photogrammetry. Figure <xref ref-type="fig" rid="Ch1.F8"/> illustrates three different ice roughness forms observed at a single site, and their appearance in cross section along the indicated transects. The “rough” form of ice roughness was classified visually as any type of ice formed by the accretion and consolidation of frazil pans or broken pieces of border ice. The “smooth” form of ice roughness was classified as ice that appeared to have formed under quiescent conditions from a combination of transported and thermally grown ice that did not consolidate. Ice that exhibited pressure ridges on otherwise smooth ice was termed “ridged”.</p>

      <?xmltex \floatpos{t}?><fig id="Ch1.F8"><?xmltex \currentcnt{8}?><?xmltex \def\figurename{Figure}?><label>Figure 8</label><caption><p id="d1e1456">Examples of three types of roughness observed in ice surface roughness samples along the indicated transects (dotted lines) at DRLL08a, 13 November 2019.</p></caption>
            <?xmltex \igopts{width=236.157874pt}?><graphic xlink:href="https://tc.copernicus.org/articles/15/4031/2021/tc-15-4031-2021-f08.png"/>

          </fig>

      <p id="d1e1465">Two samples were observed to contain ridged ice, both of which occurred at site DRLL08a on 23 November 2017 and 23 November 2019. Ridged ice presented a unique situation for the evaluation of ice roughness based on surface ice characteristics. In sea ice, it is generally the case that the height of an observed pressure ridge above the ice cover (sail) is less than the depth to which the ridge extends below the ice cover (keel) <xref ref-type="bibr" rid="bib1.bibx21" id="paren.48"/>. In the fluvial setting this relationship is unknown, potentially complicating the hypothesis of this work. Therefore, samples exhibiting pressure ridge formations were discarded from subsequent comparisons between calculated ice roughness and <inline-formula><mml:math id="M58" display="inline"><mml:mi>k</mml:mi></mml:math></inline-formula>-means clustering. However, these data were retained for the general discussion of surface ice roughness characteristics found in Sect. <xref ref-type="sec" rid="Ch1.S5.SS3"/>.</p>
      <p id="d1e1481">Cluster analysis was conducted using <inline-formula><mml:math id="M59" display="inline"><mml:mi>k</mml:mi></mml:math></inline-formula>-means clustering. The optimal number of clusters was found to be two when using average silhouette analysis. Table <xref ref-type="table" rid="Ch1.T3"/> illustrates the grouping of clusters in the data. The within cluster sum of squares was found to be 5.44 and 11.96 for clusters 1 and 2 respectively.</p>

<?xmltex \floatpos{t}?><table-wrap id="Ch1.T3"><?xmltex \currentcnt{3}?><label>Table 3</label><caption><p id="d1e1496">Categorization of ice roughness via cluster analysis.</p></caption><oasis:table frame="topbot"><oasis:tgroup cols="3">
     <oasis:colspec colnum="1" colname="col1" align="left"/>
     <oasis:colspec colnum="2" colname="col2" align="center"/>
     <oasis:colspec colnum="3" colname="col3" align="center"/>
     <oasis:thead>
       <oasis:row rowsep="1">
         <oasis:entry colname="col1">Site</oasis:entry>
         <oasis:entry colname="col2">Date</oasis:entry>
         <oasis:entry colname="col3">Clus.</oasis:entry>
       </oasis:row>
     </oasis:thead>
     <oasis:tbody>
       <oasis:row>
         <oasis:entry colname="col1">DRLL06</oasis:entry>
         <oasis:entry colname="col2">11/21/2018</oasis:entry>
         <oasis:entry colname="col3">1</oasis:entry>
       </oasis:row>
       <oasis:row>
         <oasis:entry colname="col1">DRLL06</oasis:entry>
         <oasis:entry colname="col2">2/20/2019</oasis:entry>
         <oasis:entry colname="col3">1</oasis:entry>
       </oasis:row>
       <oasis:row>
         <oasis:entry colname="col1">DRLL06</oasis:entry>
         <oasis:entry colname="col2">11/12/2019</oasis:entry>
         <oasis:entry colname="col3">1</oasis:entry>
       </oasis:row>
       <oasis:row>
         <oasis:entry colname="col1">DRLL06</oasis:entry>
         <oasis:entry colname="col2">11/13/2019</oasis:entry>
         <oasis:entry colname="col3">1</oasis:entry>
       </oasis:row>
       <oasis:row>
         <oasis:entry colname="col1">DRLL06</oasis:entry>
         <oasis:entry colname="col2">11/26/2019</oasis:entry>
         <oasis:entry colname="col3">1</oasis:entry>
       </oasis:row>
       <oasis:row>
         <oasis:entry colname="col1">DRLL03b</oasis:entry>
         <oasis:entry colname="col2">11/14/2019</oasis:entry>
         <oasis:entry colname="col3">2</oasis:entry>
       </oasis:row>
       <oasis:row>
         <oasis:entry colname="col1">DRLL05</oasis:entry>
         <oasis:entry colname="col2">11/12/2019</oasis:entry>
         <oasis:entry colname="col3">2</oasis:entry>
       </oasis:row>
       <oasis:row>
         <oasis:entry colname="col1">DRLL05</oasis:entry>
         <oasis:entry colname="col2">11/13/2019</oasis:entry>
         <oasis:entry colname="col3">2</oasis:entry>
       </oasis:row>
       <oasis:row>
         <oasis:entry colname="col1">DRLL08</oasis:entry>
         <oasis:entry colname="col2">11/13/2019</oasis:entry>
         <oasis:entry colname="col3">2</oasis:entry>
       </oasis:row>
       <oasis:row>
         <oasis:entry colname="col1">DRLL08a</oasis:entry>
         <oasis:entry colname="col2">11/26/2019</oasis:entry>
         <oasis:entry colname="col3">2</oasis:entry>
       </oasis:row>
     </oasis:tbody>
   </oasis:tgroup></oasis:table></table-wrap>

</sec>
</sec>
</sec>
<sec id="Ch1.S5">
  <label>5</label><title>Discussion</title>
<sec id="Ch1.S5.SS1">
  <label>5.1</label><title>Accuracy of the RPA photogrammetry method for fluvial ice analysis</title>
      <p id="d1e1667">The results of the dry-land RPA photogrammetry accuracy test showed that if three or more targets were used, vertical differences of no more than 0.03 m were observed<?pagebreak page4040?> between resultant DEMs and independently surveyed points. This amount of vertical difference was deemed acceptable when compared to the maximum acceptable vertical difference of 0.03 m established in Sect. <xref ref-type="sec" rid="Ch1.S3.SS2.SSS2"/>. No excessive tilt was observed in DEMs as a result of grouping targets on one side of the study area. AgiSoft PhotoScan Professional<sup>®</sup> appeared to be able to find an adequate number of tie points between images principally comprised of snow.</p>
      <p id="d1e1675">As described in <xref ref-type="bibr" rid="bib1.bibx12" id="text.49"/>, systemic errors causing deficiencies in DEM accuracy differ due to local-scale errors causing deficiencies in DEM precision. Systemic errors include those incurred by improper sampling technique and by limitations in the analysis. These errors were largely assumed to have been handled to the maximum extent possible in this research by the automated processes in AgiSoft PhotoScan Professional<sup>®</sup>. <xref ref-type="bibr" rid="bib1.bibx11" id="text.50"/> found that AgiSoft PhotoScan Professional<sup>®</sup> also performed well in reproducing the texture of complex natural surfaces. Direct comparisons could not be made in this research between the naturally occurring ice surfaces and the RPA photogrammetry reproductions. However, the magnitude of such results as the maximum ice peak value matched visual observations and field notes. The accuracy test performed at the La Barrier field site confirmed that with appropriate ground control points this method could accurately reproduce snow-covered land surfaces. It also showed that the method could precisely reproduce features of the same order of magnitude as the chunks of ice expected to be measured on the Dauphin River.</p>
      <p id="d1e1690">Comparison of geo-rectified and non-geo-rectified DEMs yielded deviations for most metrics well within the 5 % threshold. The percent difference of the median was greater than 5 %; however, its absolute deviation was <inline-formula><mml:math id="M60" display="inline"><mml:mrow><mml:mo>&lt;</mml:mo><mml:mn mathvariant="normal">0.01</mml:mn></mml:mrow></mml:math></inline-formula> m, which is much less than the maximum vertical error (0.06 m). These results confirm that geo-rectification was not strictly necessary for the analysis of surface ice roughness, since the analysis is principally concerned with surface variation rather than absolute location and orientation. However, geo-rectification would be necessary for most other applications of RPA photogrammetry of ice surfaces, such as those proposed in Sect. <xref ref-type="sec" rid="Ch1.S5.SS4"/>.</p>
</sec>
<sec id="Ch1.S5.SS2">
  <label>5.2</label><title>Comparison of ice roughness estimates</title>
      <p id="d1e1713">It was unclear which statistical metric would produce the most representative value of roughness height for the LMFFT-full and LMFFT-peak data. Since this has not been attempted in the field of fluvial ice research, a total of six potential options for the calculation of roughness height were selected, and once converted to Manning's <inline-formula><mml:math id="M61" display="inline"><mml:mi>n</mml:mi></mml:math></inline-formula> using Eq. (<xref ref-type="disp-formula" rid="Ch1.E3"/>), they were compared to the Nezhikhovskiy-predicted subsurface ice Manning's <inline-formula><mml:math id="M62" display="inline"><mml:mi>n</mml:mi></mml:math></inline-formula> (Fig. <xref ref-type="fig" rid="Ch1.F9"/>) and ice thickness (Fig. <xref ref-type="fig" rid="Ch1.F10"/>). A linear regression was attempted in each plot, and the corresponding statistical parameters are presented in Table <xref ref-type="table" rid="Ch1.T4"/>. The reported <inline-formula><mml:math id="M63" display="inline"><mml:mi>F</mml:mi></mml:math></inline-formula> statistic (<inline-formula><mml:math id="M64" display="inline"><mml:mrow><mml:msub><mml:mi>F</mml:mi><mml:mrow><mml:mn mathvariant="normal">1</mml:mn><mml:mo>,</mml:mo><mml:mn mathvariant="normal">4</mml:mn></mml:mrow></mml:msub></mml:mrow></mml:math></inline-formula>) has 1 and 4 degrees of freedom for the numerator and denominator, respectively. Using the significance value of <inline-formula><mml:math id="M65" display="inline"><mml:mrow><mml:mi mathvariant="italic">α</mml:mi><mml:mo>≤</mml:mo><mml:mn mathvariant="normal">0.05</mml:mn></mml:mrow></mml:math></inline-formula> established in Sect. <xref ref-type="sec" rid="Ch1.S3.SS2.SSS4"/>, a value of <inline-formula><mml:math id="M66" display="inline"><mml:mrow><mml:msub><mml:mi>F</mml:mi><mml:mi mathvariant="normal">crit</mml:mi></mml:msub></mml:mrow></mml:math></inline-formula> of 0.0045 was found. Values of <inline-formula><mml:math id="M67" display="inline"><mml:mrow><mml:msub><mml:mi>F</mml:mi><mml:mrow><mml:mn mathvariant="normal">1</mml:mn><mml:mo>,</mml:mo><mml:mn mathvariant="normal">4</mml:mn></mml:mrow></mml:msub></mml:mrow></mml:math></inline-formula> greater than this value indicate a significant relationship.</p>
      <p id="d1e1804">Figure <xref ref-type="fig" rid="Ch1.F9"/> highlights the fact that the Nezhikhovskiy-predicted subsurface ice Manning's <inline-formula><mml:math id="M68" display="inline"><mml:mi>n</mml:mi></mml:math></inline-formula> was generally greater than the RPA-measured values and plotted over a fairly narrow range. Most RPA-derived roughness values did not fall on the <inline-formula><mml:math id="M69" display="inline"><mml:mrow><mml:mn mathvariant="normal">1</mml:mn><mml:mo>:</mml:mo><mml:mn mathvariant="normal">1</mml:mn></mml:mrow></mml:math></inline-formula> line, indicating that the RPA-measured ice surface Manning's <inline-formula><mml:math id="M70" display="inline"><mml:mi>n</mml:mi></mml:math></inline-formula> was not identical to the values predicted using a common empirical relationship. It should be noted, however, that ice cover with relatively high RPA-measured surface ice Manning's <inline-formula><mml:math id="M71" display="inline"><mml:mi>n</mml:mi></mml:math></inline-formula> had high Nezhikhovskiy-predicted subsurface ice Manning's <inline-formula><mml:math id="M72" display="inline"><mml:mi>n</mml:mi></mml:math></inline-formula> and vice versa.</p>

<?xmltex \floatpos{t}?><table-wrap id="Ch1.T4" specific-use="star"><?xmltex \currentcnt{4}?><label>Table 4</label><caption><p id="d1e1853">Performance statistics of applied linear models.</p></caption><oasis:table frame="topbot"><oasis:tgroup cols="6">
     <oasis:colspec colnum="1" colname="col1" align="left"/>
     <oasis:colspec colnum="2" colname="col2" align="right"/>
     <oasis:colspec colnum="3" colname="col3" align="right"/>
     <oasis:colspec colnum="4" colname="col4" align="right"/>
     <oasis:colspec colnum="5" colname="col5" align="right"/>
     <oasis:colspec colnum="6" colname="col6" align="right"/>
     <oasis:thead>
       <oasis:row>
         <oasis:entry namest="col1" nameend="col6">RPA-measured ice surface Manning's <inline-formula><mml:math id="M73" display="inline"><mml:mi>n</mml:mi></mml:math></inline-formula> vs. Nezhikhovskiy-predicted </oasis:entry>
       </oasis:row>
       <oasis:row rowsep="1">
         <oasis:entry namest="col1" nameend="col6">subsurface ice Manning's <inline-formula><mml:math id="M74" display="inline"><mml:mi>n</mml:mi></mml:math></inline-formula></oasis:entry>
       </oasis:row>
     </oasis:thead>
     <oasis:tbody>
       <oasis:row rowsep="1">
         <oasis:entry colname="col1">Metric</oasis:entry>
         <oasis:entry colname="col2">Slope</oasis:entry>
         <oasis:entry colname="col3"><inline-formula><mml:math id="M75" display="inline"><mml:mrow><mml:msup><mml:mi>R</mml:mi><mml:mn mathvariant="normal">2</mml:mn></mml:msup></mml:mrow></mml:math></inline-formula></oasis:entry>
         <oasis:entry colname="col4"><inline-formula><mml:math id="M76" display="inline"><mml:mrow><mml:msub><mml:mi>F</mml:mi><mml:mrow><mml:mn mathvariant="normal">1</mml:mn><mml:mo>,</mml:mo><mml:mn mathvariant="normal">4</mml:mn></mml:mrow></mml:msub></mml:mrow></mml:math></inline-formula></oasis:entry>
         <oasis:entry colname="col5"><inline-formula><mml:math id="M77" display="inline"><mml:mi>p</mml:mi></mml:math></inline-formula></oasis:entry>
         <oasis:entry colname="col6">RMSE</oasis:entry>
       </oasis:row>
       <oasis:row>
         <oasis:entry colname="col1">SD</oasis:entry>
         <oasis:entry colname="col2">0.493</oasis:entry>
         <oasis:entry colname="col3">0.913</oasis:entry>
         <oasis:entry colname="col4">42.136</oasis:entry>
         <oasis:entry colname="col5">0.003</oasis:entry>
         <oasis:entry colname="col6">0.001</oasis:entry>
       </oasis:row>
       <oasis:row>
         <oasis:entry colname="col1">IQR</oasis:entry>
         <oasis:entry colname="col2">0.675</oasis:entry>
         <oasis:entry colname="col3">0.913</oasis:entry>
         <oasis:entry colname="col4">42.077</oasis:entry>
         <oasis:entry colname="col5">0.003</oasis:entry>
         <oasis:entry colname="col6">0.002</oasis:entry>
       </oasis:row>
       <oasis:row>
         <oasis:entry colname="col1">Min. peak</oasis:entry>
         <oasis:entry colname="col2">0.913</oasis:entry>
         <oasis:entry colname="col3">0.840</oasis:entry>
         <oasis:entry colname="col4">21.074</oasis:entry>
         <oasis:entry colname="col5">0.010</oasis:entry>
         <oasis:entry colname="col6">0.004</oasis:entry>
       </oasis:row>
       <oasis:row>
         <oasis:entry colname="col1">Max. peak</oasis:entry>
         <oasis:entry colname="col2">1.360</oasis:entry>
         <oasis:entry colname="col3">0.888</oasis:entry>
         <oasis:entry colname="col4">31.760</oasis:entry>
         <oasis:entry colname="col5">0.005</oasis:entry>
         <oasis:entry colname="col6">0.004</oasis:entry>
       </oasis:row>
       <oasis:row>
         <oasis:entry colname="col1">Mean peak</oasis:entry>
         <oasis:entry colname="col2">1.146</oasis:entry>
         <oasis:entry colname="col3">0.819</oasis:entry>
         <oasis:entry colname="col4">18.139</oasis:entry>
         <oasis:entry colname="col5">0.013</oasis:entry>
         <oasis:entry colname="col6">0.005</oasis:entry>
       </oasis:row>
       <oasis:row rowsep="1">
         <oasis:entry colname="col1">84th percentile of peaks</oasis:entry>
         <oasis:entry colname="col2">1.331</oasis:entry>
         <oasis:entry colname="col3">0.822</oasis:entry>
         <oasis:entry colname="col4">18.492</oasis:entry>
         <oasis:entry colname="col5">0.013</oasis:entry>
         <oasis:entry colname="col6">0.006</oasis:entry>
       </oasis:row>
       <oasis:row rowsep="1">
         <oasis:entry namest="col1" nameend="col6">RPA-measured ice surface Manning's <inline-formula><mml:math id="M78" display="inline"><mml:mi>n</mml:mi></mml:math></inline-formula> vs. ice thickness </oasis:entry>
       </oasis:row>
       <oasis:row rowsep="1">
         <oasis:entry colname="col1">Metric</oasis:entry>
         <oasis:entry colname="col2">Slope</oasis:entry>
         <oasis:entry colname="col3"><inline-formula><mml:math id="M79" display="inline"><mml:mrow><mml:msup><mml:mi>R</mml:mi><mml:mn mathvariant="normal">2</mml:mn></mml:msup></mml:mrow></mml:math></inline-formula></oasis:entry>
         <oasis:entry colname="col4"><inline-formula><mml:math id="M80" display="inline"><mml:mrow><mml:msub><mml:mi>F</mml:mi><mml:mrow><mml:mn mathvariant="normal">1</mml:mn><mml:mo>,</mml:mo><mml:mn mathvariant="normal">4</mml:mn></mml:mrow></mml:msub></mml:mrow></mml:math></inline-formula></oasis:entry>
         <oasis:entry colname="col5"><inline-formula><mml:math id="M81" display="inline"><mml:mi>p</mml:mi></mml:math></inline-formula></oasis:entry>
         <oasis:entry colname="col6">RMSE</oasis:entry>
       </oasis:row>
       <oasis:row>
         <oasis:entry colname="col1">SD</oasis:entry>
         <oasis:entry colname="col2">0.014</oasis:entry>
         <oasis:entry colname="col3">0.987</oasis:entry>
         <oasis:entry colname="col4">300.379</oasis:entry>
         <oasis:entry colname="col5"><inline-formula><mml:math id="M82" display="inline"><mml:mo>&lt;</mml:mo></mml:math></inline-formula>0.001</oasis:entry>
         <oasis:entry colname="col6">0.001</oasis:entry>
       </oasis:row>
       <oasis:row>
         <oasis:entry colname="col1">IQR</oasis:entry>
         <oasis:entry colname="col2">0.019</oasis:entry>
         <oasis:entry colname="col3">0.986</oasis:entry>
         <oasis:entry colname="col4">289.531</oasis:entry>
         <oasis:entry colname="col5"><inline-formula><mml:math id="M83" display="inline"><mml:mo>&lt;</mml:mo></mml:math></inline-formula>0.001</oasis:entry>
         <oasis:entry colname="col6">0.001</oasis:entry>
       </oasis:row>
       <oasis:row>
         <oasis:entry colname="col1">Min. peak</oasis:entry>
         <oasis:entry colname="col2">0.026</oasis:entry>
         <oasis:entry colname="col3">0.948</oasis:entry>
         <oasis:entry colname="col4">72.658</oasis:entry>
         <oasis:entry colname="col5">0.001</oasis:entry>
         <oasis:entry colname="col6">0.002</oasis:entry>
       </oasis:row>
       <oasis:row>
         <oasis:entry colname="col1">Max. peak</oasis:entry>
         <oasis:entry colname="col2">0.037</oasis:entry>
         <oasis:entry colname="col3">0.882</oasis:entry>
         <oasis:entry colname="col4">29.764</oasis:entry>
         <oasis:entry colname="col5">0.005</oasis:entry>
         <oasis:entry colname="col6">0.005</oasis:entry>
       </oasis:row>
       <oasis:row>
         <oasis:entry colname="col1">Mean peak</oasis:entry>
         <oasis:entry colname="col2">0.033</oasis:entry>
         <oasis:entry colname="col3">0.936</oasis:entry>
         <oasis:entry colname="col4">58.070</oasis:entry>
         <oasis:entry colname="col5">0.002</oasis:entry>
         <oasis:entry colname="col6">0.003</oasis:entry>
       </oasis:row>
       <oasis:row>
         <oasis:entry colname="col1">84th percentile of peaks</oasis:entry>
         <oasis:entry colname="col2">0.038</oasis:entry>
         <oasis:entry colname="col3">0.935</oasis:entry>
         <oasis:entry colname="col4">57.587</oasis:entry>
         <oasis:entry colname="col5">0.002</oasis:entry>
         <oasis:entry colname="col6">0.003</oasis:entry>
       </oasis:row>
     </oasis:tbody>
   </oasis:tgroup></oasis:table></table-wrap>

<sec id="Ch1.S5.SS2.SSS1">
  <label>5.2.1</label><?xmltex \opttitle{RPA-measured ice surface Manning's $n$ vs. Nezhikhovskiy-predicted subsurface ice Manning's $n$}?><title>RPA-measured ice surface Manning's <inline-formula><mml:math id="M84" display="inline"><mml:mi>n</mml:mi></mml:math></inline-formula> vs. Nezhikhovskiy-predicted subsurface ice Manning's <inline-formula><mml:math id="M85" display="inline"><mml:mi>n</mml:mi></mml:math></inline-formula></title>
      <?pagebreak page4041?><p id="d1e2318">The <inline-formula><mml:math id="M86" display="inline"><mml:mi>p</mml:mi></mml:math></inline-formula> statistic found that all relationships were significant (<inline-formula><mml:math id="M87" display="inline"><mml:mrow><mml:mi mathvariant="italic">α</mml:mi><mml:mo>≤</mml:mo><mml:mn mathvariant="normal">0.05</mml:mn></mml:mrow></mml:math></inline-formula>) in this comparison. The <inline-formula><mml:math id="M88" display="inline"><mml:mrow><mml:msub><mml:mi>F</mml:mi><mml:mrow><mml:mn mathvariant="normal">1</mml:mn><mml:mo>,</mml:mo><mml:mn mathvariant="normal">4</mml:mn></mml:mrow></mml:msub></mml:mrow></mml:math></inline-formula> statistic exceeded the <inline-formula><mml:math id="M89" display="inline"><mml:mrow><mml:msub><mml:mi>F</mml:mi><mml:mi mathvariant="normal">crit</mml:mi></mml:msub></mml:mrow></mml:math></inline-formula> value of 0.0045 in all cases. All metrics illustrated strong correlation with high <inline-formula><mml:math id="M90" display="inline"><mml:mrow><mml:msup><mml:mi>R</mml:mi><mml:mn mathvariant="normal">2</mml:mn></mml:msup></mml:mrow></mml:math></inline-formula> values and low RMSE. The SD and IQR performed the best of all metrics, with the highest <inline-formula><mml:math id="M91" display="inline"><mml:mrow><mml:msup><mml:mi>R</mml:mi><mml:mn mathvariant="normal">2</mml:mn></mml:msup></mml:mrow></mml:math></inline-formula> values and lowest <inline-formula><mml:math id="M92" display="inline"><mml:mi>p</mml:mi></mml:math></inline-formula> and RMSE results. These results indicate the ice roughness values derived from RPA photogrammetry were proportional to roughness values predicted using the Nezhikhovskiy equation.</p>

      <?xmltex \floatpos{t}?><fig id="Ch1.F9" specific-use="star"><?xmltex \currentcnt{9}?><?xmltex \def\figurename{Figure}?><label>Figure 9</label><caption><p id="d1e2399">Comparison of RPA-measured surface ice Manning's <inline-formula><mml:math id="M93" display="inline"><mml:mi>n</mml:mi></mml:math></inline-formula> calculated from various statistical metrics or roughness height to Nezhikhovskiy-predicted subsurface ice Manning's <inline-formula><mml:math id="M94" display="inline"><mml:mi>n</mml:mi></mml:math></inline-formula>.</p></caption>
            <?xmltex \igopts{width=341.433071pt}?><graphic xlink:href="https://tc.copernicus.org/articles/15/4031/2021/tc-15-4031-2021-f09.png"/>

          </fig>

</sec>
<sec id="Ch1.S5.SS2.SSS2">
  <label>5.2.2</label><?xmltex \opttitle{RPA-measured ice surface Manning's $n$ vs. ice thickness}?><title>RPA-measured ice surface Manning's <inline-formula><mml:math id="M95" display="inline"><mml:mi>n</mml:mi></mml:math></inline-formula> vs. ice thickness</title>
      <p id="d1e2438">Since the only input parameter for the Nezhikhovskiy equation is ice thickness, the RPA-measured ice surface Manning's <inline-formula><mml:math id="M96" display="inline"><mml:mi>n</mml:mi></mml:math></inline-formula> values were compared directly to their associated ice thickness measurements (and estimates) as shown in Fig. <xref ref-type="fig" rid="Ch1.F10"/>. All metrics of roughness height taken from the DEMs were found to generate RPA-measured ice surface Manning's <inline-formula><mml:math id="M97" display="inline"><mml:mi>n</mml:mi></mml:math></inline-formula> values that were strongly correlated with ice thickness. The corresponding regression statistics are reported in Table <xref ref-type="table" rid="Ch1.T4"/>. The IQR and SD again performed the best, with the highest <inline-formula><mml:math id="M98" display="inline"><mml:mrow><mml:msup><mml:mi>R</mml:mi><mml:mn mathvariant="normal">2</mml:mn></mml:msup></mml:mrow></mml:math></inline-formula> value and lowest RMSE, although all metrics performed exceptionally well. The <inline-formula><mml:math id="M99" display="inline"><mml:mi>p</mml:mi></mml:math></inline-formula> statistic showed that all relationships were significant (<inline-formula><mml:math id="M100" display="inline"><mml:mrow><mml:mi mathvariant="italic">α</mml:mi><mml:mo>≤</mml:mo><mml:mn mathvariant="normal">0.05</mml:mn></mml:mrow></mml:math></inline-formula>), and the <inline-formula><mml:math id="M101" display="inline"><mml:mrow><mml:msub><mml:mi>F</mml:mi><mml:mrow><mml:mn mathvariant="normal">1</mml:mn><mml:mo>,</mml:mo><mml:mn mathvariant="normal">4</mml:mn></mml:mrow></mml:msub></mml:mrow></mml:math></inline-formula> statistic exceeded the <inline-formula><mml:math id="M102" display="inline"><mml:mrow><mml:msub><mml:mi>F</mml:mi><mml:mi mathvariant="normal">crit</mml:mi></mml:msub></mml:mrow></mml:math></inline-formula> value of 0.0045 in all cases. Further research would be required to determine if this relationship may be used to estimate ice thickness based on observed surface roughness.</p>

      <?xmltex \floatpos{t}?><fig id="Ch1.F10" specific-use="star"><?xmltex \currentcnt{10}?><?xmltex \def\figurename{Figure}?><label>Figure 10</label><caption><p id="d1e2519">Comparison of RPA-measured ice surface Manning's <inline-formula><mml:math id="M103" display="inline"><mml:mi>n</mml:mi></mml:math></inline-formula> calculated from various statistical metrics to site ice thickness.</p></caption>
            <?xmltex \igopts{width=341.433071pt}?><graphic xlink:href="https://tc.copernicus.org/articles/15/4031/2021/tc-15-4031-2021-f10.png"/>

          </fig>

</sec>
</sec>
<sec id="Ch1.S5.SS3">
  <label>5.3</label><title>Statistical properties of ice surface roughness heights</title>
      <p id="d1e2544">All observations at site DRLL06 were found to belong to cluster “1”, while observations at site DRLL03b, DRLL05, DRLL08, and DRLL08a belonged to cluster “2”. The cluster results were enforced by observations taken at the time of sample collection. The samples were broadly separated into two groups corresponding to the visual extent of rough or smooth ice, as defined earlier in this section. Samples in cluster 1 were observed to be rough, while samples in cluster 2 were observed to be smooth. Considering the SD of the LMFFT-full data, which was the best performing metric in Sect. <xref ref-type="sec" rid="Ch1.S5.SS2"/>, Table <xref ref-type="table" rid="Ch1.T5"/> provides the minimum, mean, and maximum values of the sites within each identified cluster. Cluster 1, the rough cluster, corresponded to ice surfaces with higher SD values, while cluster 2, the smooth cluster, has lower SD values. The range of values are mutually exclusive, indicating a strong divide between the two groups.</p>

<?xmltex \floatpos{t}?><table-wrap id="Ch1.T5"><?xmltex \currentcnt{5}?><label>Table 5</label><caption><p id="d1e2554">Minimum, mean, and maximum values of surface SD for clusters.</p></caption><oasis:table frame="topbot"><oasis:tgroup cols="4">
     <oasis:colspec colnum="1" colname="col1" align="left"/>
     <oasis:colspec colnum="2" colname="col2" align="center"/>
     <oasis:colspec colnum="3" colname="col3" align="center"/>
     <oasis:colspec colnum="4" colname="col4" align="center"/>
     <oasis:thead>
       <oasis:row rowsep="1">
         <oasis:entry colname="col1">Cluster</oasis:entry>
         <oasis:entry colname="col2">Min.</oasis:entry>
         <oasis:entry colname="col3">Mean</oasis:entry>
         <oasis:entry colname="col4">Max.</oasis:entry>
       </oasis:row>
     </oasis:thead>
     <oasis:tbody>
       <oasis:row>
         <oasis:entry colname="col1">1 (rough)</oasis:entry>
         <oasis:entry colname="col2">0.06</oasis:entry>
         <oasis:entry colname="col3">0.08</oasis:entry>
         <oasis:entry colname="col4">0.11</oasis:entry>
       </oasis:row>
       <oasis:row>
         <oasis:entry colname="col1">2 (smooth)</oasis:entry>
         <oasis:entry colname="col2">0.02</oasis:entry>
         <oasis:entry colname="col3">0.03</oasis:entry>
         <oasis:entry colname="col4">0.05</oasis:entry>
       </oasis:row>
     </oasis:tbody>
   </oasis:tgroup></oasis:table></table-wrap>

      <p id="d1e2622">For each site, the LMFTT-full data as well as the LMFTT-peaks data subset were analysed in detail. Figure <xref ref-type="fig" rid="Ch1.F11"/> presents typical examples of rough, smooth, and ridged ice. The rough and smooth samples corresponded to sites from each of the rough and smooth clusters. The ridged samples were those identified in Sect. <xref ref-type="sec" rid="Ch1.S4.SS2.SSS1"/>. To the authors' knowledge this is the first time that roughness of a river ice cover's surface has been quantified in such detail. The LMFTT-full histograms were observed to have distributions that appeared Gaussian in a qualitative sense, as shown in Fig. <xref ref-type="fig" rid="Ch1.F11"/>. Despite the appearance, all distributions failed the Shapiro–Wilk normality test, with <inline-formula><mml:math id="M104" display="inline"><mml:mrow><mml:mi>p</mml:mi><mml:mo>≪</mml:mo><mml:mn mathvariant="normal">0.05</mml:mn></mml:mrow></mml:math></inline-formula>, using a randomized subsample (<inline-formula><mml:math id="M105" display="inline"><mml:mrow><mml:mo>≈</mml:mo><mml:mn mathvariant="normal">5000</mml:mn></mml:mrow></mml:math></inline-formula> points) of the data. Considering the distribution of peak<?pagebreak page4042?> values only, all sites exhibited clearly non-normal distributions that were heavily biased to the extreme left of the chart. This was interpreted to mean that the majority of peaks were small compared to the maximum peak values.</p>

      <?xmltex \floatpos{t}?><fig id="Ch1.F11" specific-use="star"><?xmltex \currentcnt{11}?><?xmltex \def\figurename{Figure}?><label>Figure 11</label><caption><p id="d1e2656">Comparison of representative roughness height histograms across three identified ice types.</p></caption>
          <?xmltex \igopts{width=426.791339pt}?><graphic xlink:href="https://tc.copernicus.org/articles/15/4031/2021/tc-15-4031-2021-f11.png"/>

        </fig>

      <p id="d1e2665">It was noted that the rough and smooth samples differed primarily in the width and height of their distributions. The rough samples tended to have wider distributions with a lower peak count, while the smooth distributions had a higher peak count and a narrower distribution of roughness heights. The distributions of peak roughness height also differed, with the rough samples having more larger peaks than the smooth samples. The ridged samples exhibited more irregular distributions than the rough or smooth cases but were more similar to rough distributions in being wider and having lower peak counts than smooth distributions.</p>
</sec>
<sec id="Ch1.S5.SS4">
  <label>5.4</label><title>Alternative uses for RPA photogrammetry</title>
      <p id="d1e2676">A common problem in river ice elevation surveys is the selection of a single, representative value to define the average ice surface elevation in a given area, especially when the ice cover is quite rough. It is up to the practitioner to use their best judgement to visually select a single representative point to survey while walking along the river bank, and the simple choice of resting the survey rod on the top of a piece of ice or the bottom of that same piece of ice will cause the local ice elevation measurement to vary considerably. During massive ice jam events, this field task is dangerous and frequently impossible. The above research shows that RPA photogrammetry can be used to accurately survey ice areas for the purpose of observing local topography, with much lower risks to field personnel than traditional ice surveying methods. Once<?pagebreak page4043?> a geo-rectified DEM has been established, a representative ice elevation value of the entire rough ice surface can be determined using the LM approach introduced in Sect. <xref ref-type="sec" rid="Ch1.S3.SS2.SSS3"/>. This analysis could be extended to examine shear walls and determine maximum ice elevation of a freeze-up jam after the fact.</p>
</sec>
</sec>
<sec id="Ch1.S6" sec-type="conclusions">
  <label>6</label><title>Conclusions</title>
      <p id="d1e2690">The research presented in this work has developed novel methods for the capture and analysis of fluvial ice surface elevation and roughness data. Through field trials and controlled land-based experiments, it was determined that RPA photogrammetry produced an accurate digital representation of rough or smooth ice covers, with a maximum vertical error of 0.06 m if using three or more ground control points over a 200 m by 100 m area. For the sole purpose of roughness characterization, it was determined that geo-rectification was unnecessary using our equipment. The relatively inexpensive consumer-grade RPA was able to operate in harsh winter field conditions, with an approximately 50 % reduced battery life. For the first time in the river ice engineering field, the top-of-ice surface roughness has been quantified in detail using high-resolution RPA photogrammetry. A combination of linear processing and Fourier filtering were proposed, and bulk statistical properties of the ice roughness samples have been computed. Histograms of different ice types have been presented for the first time, and <inline-formula><mml:math id="M106" display="inline"><mml:mi>k</mml:mi></mml:math></inline-formula>-means clustering analysis has identified two distinct classes of surface ice roughness. Through evaluation of the statistical properties of the distribution of DEM heights observed via RPA photogrammetry, several interesting patterns were found. All distributions were found to be non-normal when evaluated with the<?pagebreak page4044?> Shapiro–Wilk normality test; however, they display a qualitatively normal appearance.</p>
      <p id="d1e2700">The hypothesis of this research, i.e. that the surface roughness of a newly frozen fluvial ice cover is indicative of subsurface roughness, was tested. When comparing the Nezhikhovskiy-predicted values to the RPA-measured values, all roughness height metrics produced significant relationships, many with <inline-formula><mml:math id="M107" display="inline"><mml:mrow><mml:msup><mml:mi>R</mml:mi><mml:mn mathvariant="normal">2</mml:mn></mml:msup><mml:mo>&gt;</mml:mo><mml:mn mathvariant="normal">0.9</mml:mn></mml:mrow></mml:math></inline-formula>. In this comparison IQR and SD had the lowest <inline-formula><mml:math id="M108" display="inline"><mml:mi>p</mml:mi></mml:math></inline-formula> values, highest <inline-formula><mml:math id="M109" display="inline"><mml:mrow><mml:msup><mml:mi>R</mml:mi><mml:mn mathvariant="normal">2</mml:mn></mml:msup></mml:mrow></mml:math></inline-formula> values, and lowest RMSE values. It can be concluded that even though the RPA-measured ice surface Manning's <inline-formula><mml:math id="M110" display="inline"><mml:mi>n</mml:mi></mml:math></inline-formula> was not equivalent to Nezhikhovskiy-predicted subsurface ice Manning's <inline-formula><mml:math id="M111" display="inline"><mml:mi>n</mml:mi></mml:math></inline-formula>, the correlations were strong enough to suggest that RPA-measured surface ice roughness and subsurface ice roughness increase in tandem.</p>
      <p id="d1e2750">Very strong positive correlation was observed between the RPA-measured surface ice roughness and ice thickness across all tested metrics. Once again the SD and IQR of the roughness height DEM were found to be the best metrics for roughness height, with the highest <inline-formula><mml:math id="M112" display="inline"><mml:mrow><mml:msup><mml:mi>R</mml:mi><mml:mn mathvariant="normal">2</mml:mn></mml:msup></mml:mrow></mml:math></inline-formula> and lowest RMSE and <inline-formula><mml:math id="M113" display="inline"><mml:mi>p</mml:mi></mml:math></inline-formula> values. This relationship also supports the hypothesis that surface ice roughness is related to subsurface ice roughness, since the data show that RPA-measured ice surface roughness increases with ice thickness and a widely used empirical formula predicts the same relationship. Future work should focus on increasing the number of field observations taken using these methods, across a wider range of ice roughness values and on different rivers of various size and flow conditions.</p>
      <p id="d1e2771">The methods presented in this research can conceivably be applied to further uses in the field of fluvial ice monitoring. For example, the high-resolution DEMs produced by this method can be used to determine a more representative spot measurement of ice elevation, which has the potential to significantly increase the accuracy of river ice jam profile plots. This could potentially improve ice jam numerical model development since model performance is consistently evaluated by comparing simulated jam profiles to measured ice jam profiles. Furthermore, shear walls may be captured and analysed in their entirety, even immediately after or during break-up. This research also presents a possible link between surface ice roughness and ice thickness, which may provide for a method of estimating ice thickness using RPA photogrammetry. The use of RPA photogrammetry for the monitoring of fluvial ice covers offers a quicker, safer, and cheaper alternative to any previous method of high-resolution topographic data collection, and its applications are open for development.</p>
</sec>

      
      </body>
    <back><notes notes-type="dataavailability"><title>Data availability</title>

      <p id="d1e2778">Data related to this study, including raw images, point clouds, and summary data, are available upon request from the corresponding author.</p>
  </notes><notes notes-type="authorcontribution"><title>Author contributions</title>

      <p id="d1e2784">JE, SC, and AW developed the research plan and conducted fieldwork. JE processed all field data and developed all data analysis tools. JE and SC drew conclusions based on the results.</p>
  </notes><notes notes-type="competinginterests"><title>Competing interests</title>

      <p id="d1e2790">The authors declare that they have no conflict of interest.</p>
  </notes><notes notes-type="disclaimer"><title>Disclaimer</title>

      <p id="d1e2796">Publisher's note: Copernicus Publications remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.</p>
  </notes><notes notes-type="financialsupport"><title>Financial support</title>

      <p id="d1e2802">This research has been supported by the Natural Sciences and Engineering Research Council of Canada (grant no. IRCPJ472185-13) and Manitoba Hydro (grant no. G274/P274).</p>
  </notes><notes notes-type="reviewstatement"><title>Review statement</title>

      <p id="d1e2808">This paper was edited by Nicholas Barrand and reviewed by two anonymous referees.</p>
  </notes><ref-list>
    <title>References</title>

      <ref id="bib1.bibx1"><?xmltex \def\ref@label{{Aberle and Nikora(2006)}}?><label>Aberle and Nikora(2006)</label><?label Aberle2006?><mixed-citation>Aberle, J. and Nikora, V.: Statistical properties of armored gravel bed
surfaces, Water Resour. Res., 42, W11414, <ext-link xlink:href="https://doi.org/10.1029/2005WR004674" ext-link-type="DOI">10.1029/2005WR004674</ext-link>, 2006.</mixed-citation></ref>
      <ref id="bib1.bibx2"><?xmltex \def\ref@label{{Alfredsen et~al.(2018)Alfredsen, Haas, Tuhtan, and
Zinke}}?><label>Alfredsen et al.(2018)Alfredsen, Haas, Tuhtan, and
Zinke</label><?label Alfredsen2018?><mixed-citation>Alfredsen, K., Haas, C., Tuhtan, J. A., and Zinke, P.: Brief Communication: Mapping river ice using drones and structure from motion, The Cryosphere, 12, 627–633, <ext-link xlink:href="https://doi.org/10.5194/tc-12-627-2018" ext-link-type="DOI">10.5194/tc-12-627-2018</ext-link>, 2018.</mixed-citation></ref>
      <ref id="bib1.bibx3"><?xmltex \def\ref@label{{Ashton(1986)}}?><label>Ashton(1986)</label><?label Ashton1986?><mixed-citation>
Ashton, G. D. (Ed.): River and lake ice engineering, Water Resources
Publication, Chelsea, Michigan, USA,  261–361, ISBN 0-918334-59-4,
1986.</mixed-citation></ref>
      <ref id="bib1.bibx4"><?xmltex \def\ref@label{{Beltaos(2013)}}?><label>Beltaos(2013)</label><?label Beltaos2013?><mixed-citation>
Beltaos, S. (Ed.): River Ice Formation, The Committee on River Ice Processes and
the Environment, Edmonton, Alberta, Canada, 218–232, ISBN 978-0-9920022-0-6, 2013.</mixed-citation></ref>
      <ref id="bib1.bibx5"><?xmltex \def\ref@label{{Buffin-Belanger et~al.(2015)Buffin-Belanger, Demers, and
Olsen}}?><label>Buffin-Belanger et al.(2015)Buffin-Belanger, Demers, and
Olsen</label><?label Buffin2015?><mixed-citation>
Buffin-Belanger, T., Demers, S., and Olsen, T.: Quantification of under ice
cover roughness, in: 18th Workshop on the Hydraulics of Ice Covered Rivers,
CGU HS Committee on River Ice Processes and the Environment, Quebec City, QC,
Canada, 2015.</mixed-citation></ref>
      <ref id="bib1.bibx6"><?xmltex \def\ref@label{{Chudley et~al.(2019)Chudley, Christoffersen, Doyle, Abellan, and
Snooke}}?><label>Chudley et al.(2019)Chudley, Christoffersen, Doyle, Abellan, and
Snooke</label><?label Chudley2019?><mixed-citation>Chudley, T. R., Christoffersen, P., Doyle, S. H., Abellan, A., and Snooke, N.: High-accuracy UAV photogrammetry of ice sheet dynamics with no ground control, The Cryosphere, 13, 955–968, <ext-link xlink:href="https://doi.org/10.5194/tc-13-955-2019" ext-link-type="DOI">10.5194/tc-13-955-2019</ext-link>, 2019.</mixed-citation></ref>
      <ref id="bib1.bibx7"><?xmltex \def\ref@label{{Colomina and Molina(2014)}}?><label>Colomina and Molina(2014)</label><?label Colomina2014?><mixed-citation>Colomina, I. and Molina, P.: Unmanned aerial systems for photogrammetry and
remote sensing: A review, ISPRS Jounral of Photogrammetry and Remote Sensing,
92, 79–97, <ext-link xlink:href="https://doi.org/10.1016/j.isprsjprs.2014.02.013" ext-link-type="DOI">10.1016/j.isprsjprs.2014.02.013</ext-link>, 2014.</mixed-citation></ref>
      <ref id="bib1.bibx8"><?xmltex \def\ref@label{{Crance and Frothingham(2008)}}?><label>Crance and Frothingham(2008)</label><?label Crance2008?><mixed-citation>
Crance, M.-J. and Frothingham, K. M.: The Impact of Ice Cover Roughness on
Stream Hydrology, in: 65th Eastern Snow Conference, pp. 149–165, Fairlee,
Vermont, USA, 2008.</mixed-citation></ref>
      <ref id="bib1.bibx9"><?xmltex \def\ref@label{{Dammann et~al.(2018)Dammann, Eicken, Mahoney, Saiet, Meyer, and
George}}?><label>Dammann et al.(2018)Dammann, Eicken, Mahoney, Saiet, Meyer, and
George</label><?label Dammann2018?><mixed-citation>Dammann, D. O., Eicken, H., Mahoney, A. R., Saiet, E., Meyer, F. J., and
George, J. C.: Traversing Sea Ice – Linking Surface Roughness and Ice
Trafficability Trhough SAR Polarimetry and Interferometry, IEEE Journal of
Selected Topics in Applied Earth Observations and Remote Sensing, 11,
416–433, <ext-link xlink:href="https://doi.org/10.1109/JSTARS.2017.2764961" ext-link-type="DOI">10.1109/JSTARS.2017.2764961</ext-link>, 2018.</mixed-citation></ref>
      <ref id="bib1.bibx10"><?xmltex \def\ref@label{{Eisenbeiss et~al.(2005)Eisenbeiss, Lambers, and
Sauerbrier}}?><label>Eisenbeiss et al.(2005)Eisenbeiss, Lambers, and
Sauerbrier</label><?label Eisenbeiss2005?><mixed-citation>
Eisenbeiss, H., Lambers, K., and Sauerbrier, M.: Photogrammetric recording of
the archaeological site of Pinchango Alto (Palpa, Peru) using a mini
helicopter (UAV), in: The world is in your eyes: CAA 2005: Computer
Applications and Quantitative Methods in Archaeology, proceedings of the 33rd
Conference, edited by: Figueiredo, A., pp. 175–184, 2005.</mixed-citation></ref>
      <ref id="bib1.bibx11"><?xmltex \def\ref@label{{Eltner and Schneider(2015)}}?><label>Eltner and Schneider(2015)</label><?label Eltner2015?><mixed-citation>Eltner, A. and Schneider, D.: Analysis of different methods for 3D
reconstruction of natural surfaces from parallel-axes UAV images, The
Photogrammetric Record, 30, 279–299, <ext-link xlink:href="https://doi.org/10.1111/phor.12115" ext-link-type="DOI">10.1111/phor.12115</ext-link>, 2015.</mixed-citation></ref>
      <ref id="bib1.bibx12"><?xmltex \def\ref@label{{Eltner et~al.(2016)Eltner, Kaiser, Castillo, Rock, Neugirg, and
Abellan}}?><label>Eltner et al.(2016)Eltner, Kaiser, Castillo, Rock, Neugirg, and
Abellan</label><?label Eltner2016?><mixed-citation>Eltner, A., Kaiser, A., Castillo, C., Rock, G., Neugirg, F., and Abellán, A.: Image-based surface reconstruction in geomorphometry – merits, limits and developments, Earth Surf. Dynam., 4, 359–389, <ext-link xlink:href="https://doi.org/10.5194/esurf-4-359-2016" ext-link-type="DOI">10.5194/esurf-4-359-2016</ext-link>, 2016.</mixed-citation></ref>
      <ref id="bib1.bibx13"><?xmltex \def\ref@label{{Fitzpatrick et~al.(2019)Fitzpatrick, Radić, and
Menounos}}?><label>Fitzpatrick et al.(2019)Fitzpatrick, Radić, and
Menounos</label><?label Fitzpatrick2019?><mixed-citation>Fitzpatrick, N., Radić, V., and Menounos, B.: A multi-season investigation of glacier surface roughness lengths through in situ and remote observation, The Cryosphere, 13, 1051–1071, <ext-link xlink:href="https://doi.org/10.5194/tc-13-1051-2019" ext-link-type="DOI">10.5194/tc-13-1051-2019</ext-link>, 2019.</mixed-citation></ref>
      <ref id="bib1.bibx14"><?xmltex \def\ref@label{{Fraser and Cronk(2009)}}?><label>Fraser and Cronk(2009)</label><?label Fraser2009?><mixed-citation>
Fraser, C. S. and Cronk, S.: A hybrid measurement approach for close-range
photogrammetry, ISPRS Journal of Photogrammetry and Remote Sensing, 64,
328–333, 2009.</mixed-citation></ref>
      <ref id="bib1.bibx15"><?xmltex \def\ref@label{{Gadelmawla et~al.(2002)Gadelmawla, Koura, Maksoud, Elewa, and
Soliman}}?><label>Gadelmawla et al.(2002)Gadelmawla, Koura, Maksoud, Elewa, and
Soliman</label><?label Gadelmawla2002?><mixed-citation>Gadelmawla, E. S., Koura, M. M., Maksoud, T. M. A., Elewa, I. M., and Soliman,
H. H.: Roughness parameters, J. Mater. Proc. Technol., 123,
133–145, <ext-link xlink:href="https://doi.org/10.1016/S0924-0136(02)00060-2" ext-link-type="DOI">10.1016/S0924-0136(02)00060-2</ext-link>, 2002.</mixed-citation></ref>
      <ref id="bib1.bibx16"><?xmltex \def\ref@label{{Gerard and Andres(1984)}}?><label>Gerard and Andres(1984)</label><?label Gerard1984?><mixed-citation>
Gerard, R. and Andres, D.: Hydraulic roughness of freeze-up ice accumulations:
North Sasksatchewan River through Edmonton, 2nd Workshop on the Hydraulics of Ice Covered Rivers, Edmonton, Alberta, Canada, pp. 62–86, 1984.</mixed-citation></ref>
      <ref id="bib1.bibx17"><?xmltex \def\ref@label{{Gini et~al.(2013)Gini, Pagliari, Passoni, Pinto, Sona, and
Dosso}}?><label>Gini et al.(2013)Gini, Pagliari, Passoni, Pinto, Sona, and
Dosso</label><?label Gini2013?><mixed-citation>
Gini, R., Pagliari, D., Passoni, D., Pinto, L., Sona, G., and Dosso, P.: UAV
photogrammetry: Block triangulation comparisons, Internation Archives of the
Photogrammetry, Remote Sensing and Spatial Information Sciences, XL-1/W2,
157–162, 2013.</mixed-citation></ref>
      <ref id="bib1.bibx18"><?xmltex \def\ref@label{{Gomez(1993)}}?><label>Gomez(1993)</label><?label Gomez1993?><mixed-citation>
Gomez, B.: Roughness of Stable, Armored Gravel Beds, Water Resour. Res.,
29, 3631–3642, 1993.</mixed-citation></ref>
      <ref id="bib1.bibx19"><?xmltex \def\ref@label{{Hamshaw et~al.(2019)Hamshaw, Engel, Rizzo, O'Neil-Dunne, and
Dewoolkar}}?><label>Hamshaw et al.(2019)Hamshaw, Engel, Rizzo, O'Neil-Dunne, and
Dewoolkar</label><?label Hamshaw2019?><mixed-citation>Hamshaw, S. D., Engel, T., Rizzo, D. M., O'Neil-Dunne, J., and Dewoolkar,
M. M.: Application of unmanned aircraft system (UAS) for monitoring bank
erosion along river corridors, Geomatics, Nat. Haz. Risk, 10,
1285–1305, <ext-link xlink:href="https://doi.org/10.1080/19475705.2019.1571533" ext-link-type="DOI">10.1080/19475705.2019.1571533</ext-link>, 2019.</mixed-citation></ref>
      <ref id="bib1.bibx20"><?xmltex \def\ref@label{{Jain(2010)}}?><label>Jain(2010)</label><?label Jain2010?><mixed-citation>Jain, A. K.: Data clustering: 50 years beyond k-means, Pattern Recogn.
Lett., 31, 651–666, <ext-link xlink:href="https://doi.org/10.1016/j.patrec.2009.09.011" ext-link-type="DOI">10.1016/j.patrec.2009.09.011</ext-link>, 2010.</mixed-citation></ref>
      <ref id="bib1.bibx21"><?xmltex \def\ref@label{{Johnston et~al.(2009)Johnston, Masterdon, and Wright}}?><label>Johnston et al.(2009)Johnston, Masterdon, and Wright</label><?label Johnston2009?><mixed-citation>
Johnston, M., Masterdon, D., and Wright, B.: Multi-Year Ice Thickness: Knowns
and Unknowns, in: Proceedings of the International Conference on Port and
Ocean Engineering Under Arctic Conditions, pp. 969–981, 2009.</mixed-citation></ref>
      <ref id="bib1.bibx22"><?xmltex \def\ref@label{{Kirby(1991)}}?><label>Kirby(1991)</label><?label Kirby1991?><mixed-citation>
Kirby, R. P.: Measurement of Surface Roughness in Desert Terrain By Close Range
Photogrammetry, Photogrammetric Record, 13, 855–875, 1991.</mixed-citation></ref>
      <ref id="bib1.bibx23"><?xmltex \def\ref@label{{Nezhikhovskiy(1964)}}?><label>Nezhikhovskiy(1964)</label><?label Nez1964?><mixed-citation>
Nezhikhovskiy, R. A.: Coefficients of roughness of bottom surface on slush-ice
cover, Soviet Hydrology, Washington, American Geophysical Union, Selected Papers 2, pp. 127–150, 1964.</mixed-citation></ref>
      <ref id="bib1.bibx24"><?xmltex \def\ref@label{{Niethammer et~al.(2012)Niethammer, James, Rothmund, Travelletti, and
Joswig}}?><label>Niethammer et al.(2012)Niethammer, James, Rothmund, Travelletti, and
Joswig</label><?label Niethammer2012?><mixed-citation>Niethammer, U., James, M., Rothmund, S., Travelletti, J., and Joswig, M.:
UAV-based remote sensing of the Super-Sauze landslide: Evaluation and
results, Eng. Geol., 128, 2–11, <ext-link xlink:href="https://doi.org/10.1016/j.enggeo.2011.03.012" ext-link-type="DOI">10.1016/j.enggeo.2011.03.012</ext-link>,
2012.</mixed-citation></ref>
      <?pagebreak page4046?><ref id="bib1.bibx25"><?xmltex \def\ref@label{{Nikora et~al.(1998)Nikora, Goring, and Biggs}}?><label>Nikora et al.(1998)Nikora, Goring, and Biggs</label><?label Nikora1998?><mixed-citation>
Nikora, V. I., Goring, D. G., and Biggs, B. J. F.: On gravel-bed roughness
characterization, Water Resour. Res., 34, 517–527, 1998.</mixed-citation></ref>
      <ref id="bib1.bibx26"><?xmltex \def\ref@label{{Nikuradse(1950)}}?><label>Nikuradse(1950)</label><?label Nikuradse1950?><mixed-citation>
Nikuradse, J.: Laws of Flow in Rough Pipes, Technical Memorandum 1292, National
Advisory Committee for Aeronautics, 1950.</mixed-citation></ref>
      <ref id="bib1.bibx27"><?xmltex \def\ref@label{{Razali and Wah(2011)}}?><label>Razali and Wah(2011)</label><?label Razali2011?><mixed-citation>
Razali, N. M. and Wah, Y. B.: Power Comparisons of Shapiro0Wilk,
Komolgorov-Smirnov, Lilliefors and Anderson-Darling tests, J.
Stat. Model. Anal., 2, 21–33, 2011.</mixed-citation></ref>
      <ref id="bib1.bibx28"><?xmltex \def\ref@label{{Strickler(1923)}}?><label>Strickler(1923)</label><?label Strickler1923?><mixed-citation>
Strickler, A.: Contributions to the Question of a Velocity Formula and
Roughness Data for Streams, Channels and Closed Pipelines, Tech. Rep. 16,
Swiss Department of the Interior, Bureau of Water Affairs, 1923.</mixed-citation></ref>
      <ref id="bib1.bibx29"><?xmltex \def\ref@label{{Sturm(2001)}}?><label>Sturm(2001)</label><?label Sturm2001?><mixed-citation>
Sturm, T. W.: Open Channel Hydraulics, McGraw Hill, 2001.</mixed-citation></ref>
      <ref id="bib1.bibx30"><?xmltex \def\ref@label{{Watts et~al.(2012)Watts, Ambrosia, and Hinkley}}?><label>Watts et al.(2012)Watts, Ambrosia, and Hinkley</label><?label Watts2012?><mixed-citation>Watts, A. C., Ambrosia, V. G., and Hinkley, E. A.: Unmanned Aircraft Systems in
Remote Sensing and Scientific Research: Classification and Considerations of
Use, Remote Sens., 4, 1671–1692, <ext-link xlink:href="https://doi.org/10.3390/rs4061671" ext-link-type="DOI">10.3390/rs4061671</ext-link>, 2012.</mixed-citation></ref>
      <ref id="bib1.bibx31"><?xmltex \def\ref@label{{Wazney et~al.(2018)Wazney, Clark, and Wall}}?><label>Wazney et al.(2018)Wazney, Clark, and Wall</label><?label Wazney2018?><mixed-citation>Wazney, L., Clark, S. P., and Wall, A. J.: Field monitoring of secondary
consolidation events and ice cover progression during freeze-up on the Lower
Dauphin River, Manitoba, Cold Reg. Sci. Technol., 148, 159–171,
<ext-link xlink:href="https://doi.org/10.1016/j.coldregions.2018.01.014" ext-link-type="DOI">10.1016/j.coldregions.2018.01.014</ext-link>, 2018.
</mixed-citation></ref><?xmltex \hack{\newpage}?>
      <ref id="bib1.bibx32"><?xmltex \def\ref@label{{Wazney et~al.(2019)Wazney, Clark, and Malenchak}}?><label>Wazney et al.(2019)Wazney, Clark, and Malenchak</label><?label Wazney2019?><mixed-citation>Wazney, L., Clark, S. P., and Malenchak, J.: Effects of freeze-up consolidation
event surges on river hydraulics and ice dynamics on the Lower Dauphin River,
Cold Reg. Sci. Technol., 158, 264–274,
<ext-link xlink:href="https://doi.org/10.1016/j.coldregions.2018.09.003" ext-link-type="DOI">10.1016/j.coldregions.2018.09.003</ext-link>, 2019.</mixed-citation></ref>
      <ref id="bib1.bibx33"><?xmltex \def\ref@label{{Weber et~al.(2014)Weber, Ramachandran, and Henikoff}}?><label>Weber et al.(2014)Weber, Ramachandran, and Henikoff</label><?label Weber2014?><mixed-citation>
Weber, C. M., Ramachandran, S., and Henikoff, S.: Nucleosomes are
context-specific, H2A.Z-modulated barriers to RNA polymerase, Molecular Cell,
53, 819–830, 2014.</mixed-citation></ref>
      <ref id="bib1.bibx34"><?xmltex \def\ref@label{{Westoby et~al.(2012)Westoby, Brasington, Glasser, Hambrey, and
Reynolds}}?><label>Westoby et al.(2012)Westoby, Brasington, Glasser, Hambrey, and
Reynolds</label><?label Westoby2012?><mixed-citation>Westoby, M. J., Brasington, J., Glasser, N. F., Hambrey, M. J., and Reynolds,
J. M.: 'Structure-from-Motion' photogrammtery: A low-cost, effective tool for
geoscience applications, Geomorphology, 179, 300–314,
<ext-link xlink:href="https://doi.org/10.1016/j.geomorph.2012.08.021" ext-link-type="DOI">10.1016/j.geomorph.2012.08.021</ext-link>, 2012.</mixed-citation></ref>
      <ref id="bib1.bibx35"><?xmltex \def\ref@label{{Yitayew et~al.(2018)Yitayew, Dierking, Divine, Eltoft, Ferro-Famil,
Rösel, and Negrel}}?><label>Yitayew et al.(2018)Yitayew, Dierking, Divine, Eltoft, Ferro-Famil,
Rösel, and Negrel</label><?label Yitayew2018?><mixed-citation>Yitayew, T. G., Dierking, W., Divine, D. V., Eltoft, T., Ferro-Famil, L.,
Rösel, A., and Negrel, J.: Validation of Sea-Ice Topographic Heights Derived
From TanDEM-X Interferometric SAD Data With Results From Laser Profiler and
Photogrammetry, IEEE T. Geosci. Remote, 56,
6504–6520, <ext-link xlink:href="https://doi.org/10.1109/TGRS.2018.2839590" ext-link-type="DOI">10.1109/TGRS.2018.2839590</ext-link>, 2018.</mixed-citation></ref>

  </ref-list></back>
    <!--<article-title-html>Ice roughness estimation via remotely piloted aircraft and photogrammetry</article-title-html>
<abstract-html><p>The monitoring of fluvial ice covers can be time intensive, dangerous, and costly if detailed data are required. Ice covers on a river surface cause resistance to water flow, which increases upstream water levels. Ice with a higher degree of roughness causes increased flow resistance and therefore even higher upstream water levels. Aerial images collected via remotely piloted aircraft (RPA) were processed with structure from motion photogrammetry to create a digital elevation model (DEM) and then produce quantitative measurements of surface ice roughness. Images and surface ice roughness values were collected over 2 years on the Dauphin River in Manitoba, Canada. It was hypothesized that surface ice roughness would be indicative of subsurface ice roughness. This hypothesis was tested by comparing RPA-measured surface ice roughness values to those predicted by the Nezhikhovskiy equation, wherein subsurface ice roughness is proportional to ice thickness. Various statistical metrics were used to represent the roughness height of the DEMs. Strong trends were identified in the comparison of RPA-measured ice surface roughness to subsurface ice roughness values predicted by the Nezhikhovskiy equation, as well as with comparisons to ice thickness. The standard deviation and interquartile range of roughness heights were determined to be the most representative statistical metrics and several properties of the DEMs of fluvial ice covers were calculated and observed. No DEMs were found to be normally distributed. This first attempt at using RPA-derived measurements of surface ice roughness to estimate river ice flow resistance is shown to have considerable potential and will hopefully be verified and improved upon by subsequent measurements on a wide variety of rivers and ice covers.</p></abstract-html>
<ref-html id="bib1.bib1"><label>Aberle and Nikora(2006)</label><mixed-citation>
Aberle, J. and Nikora, V.: Statistical properties of armored gravel bed
surfaces, Water Resour. Res., 42, W11414, <a href="https://doi.org/10.1029/2005WR004674" target="_blank">https://doi.org/10.1029/2005WR004674</a>, 2006.
</mixed-citation></ref-html>
<ref-html id="bib1.bib2"><label>Alfredsen et al.(2018)Alfredsen, Haas, Tuhtan, and
Zinke</label><mixed-citation>
Alfredsen, K., Haas, C., Tuhtan, J. A., and Zinke, P.: Brief Communication: Mapping river ice using drones and structure from motion, The Cryosphere, 12, 627–633, <a href="https://doi.org/10.5194/tc-12-627-2018" target="_blank">https://doi.org/10.5194/tc-12-627-2018</a>, 2018.
</mixed-citation></ref-html>
<ref-html id="bib1.bib3"><label>Ashton(1986)</label><mixed-citation>
Ashton, G. D. (Ed.): River and lake ice engineering, Water Resources
Publication, Chelsea, Michigan, USA,  261–361, ISBN 0-918334-59-4,
1986.
</mixed-citation></ref-html>
<ref-html id="bib1.bib4"><label>Beltaos(2013)</label><mixed-citation>
Beltaos, S. (Ed.): River Ice Formation, The Committee on River Ice Processes and
the Environment, Edmonton, Alberta, Canada, 218–232, ISBN 978-0-9920022-0-6, 2013.
</mixed-citation></ref-html>
<ref-html id="bib1.bib5"><label>Buffin-Belanger et al.(2015)Buffin-Belanger, Demers, and
Olsen</label><mixed-citation>
Buffin-Belanger, T., Demers, S., and Olsen, T.: Quantification of under ice
cover roughness, in: 18th Workshop on the Hydraulics of Ice Covered Rivers,
CGU HS Committee on River Ice Processes and the Environment, Quebec City, QC,
Canada, 2015.
</mixed-citation></ref-html>
<ref-html id="bib1.bib6"><label>Chudley et al.(2019)Chudley, Christoffersen, Doyle, Abellan, and
Snooke</label><mixed-citation>
Chudley, T. R., Christoffersen, P., Doyle, S. H., Abellan, A., and Snooke, N.: High-accuracy UAV photogrammetry of ice sheet dynamics with no ground control, The Cryosphere, 13, 955–968, <a href="https://doi.org/10.5194/tc-13-955-2019" target="_blank">https://doi.org/10.5194/tc-13-955-2019</a>, 2019.
</mixed-citation></ref-html>
<ref-html id="bib1.bib7"><label>Colomina and Molina(2014)</label><mixed-citation>
Colomina, I. and Molina, P.: Unmanned aerial systems for photogrammetry and
remote sensing: A review, ISPRS Jounral of Photogrammetry and Remote Sensing,
92, 79–97, <a href="https://doi.org/10.1016/j.isprsjprs.2014.02.013" target="_blank">https://doi.org/10.1016/j.isprsjprs.2014.02.013</a>, 2014.
</mixed-citation></ref-html>
<ref-html id="bib1.bib8"><label>Crance and Frothingham(2008)</label><mixed-citation>
Crance, M.-J. and Frothingham, K. M.: The Impact of Ice Cover Roughness on
Stream Hydrology, in: 65th Eastern Snow Conference, pp. 149–165, Fairlee,
Vermont, USA, 2008.
</mixed-citation></ref-html>
<ref-html id="bib1.bib9"><label>Dammann et al.(2018)Dammann, Eicken, Mahoney, Saiet, Meyer, and
George</label><mixed-citation>
Dammann, D. O., Eicken, H., Mahoney, A. R., Saiet, E., Meyer, F. J., and
George, J. C.: Traversing Sea Ice – Linking Surface Roughness and Ice
Trafficability Trhough SAR Polarimetry and Interferometry, IEEE Journal of
Selected Topics in Applied Earth Observations and Remote Sensing, 11,
416–433, <a href="https://doi.org/10.1109/JSTARS.2017.2764961" target="_blank">https://doi.org/10.1109/JSTARS.2017.2764961</a>, 2018.
</mixed-citation></ref-html>
<ref-html id="bib1.bib10"><label>Eisenbeiss et al.(2005)Eisenbeiss, Lambers, and
Sauerbrier</label><mixed-citation>
Eisenbeiss, H., Lambers, K., and Sauerbrier, M.: Photogrammetric recording of
the archaeological site of Pinchango Alto (Palpa, Peru) using a mini
helicopter (UAV), in: The world is in your eyes: CAA 2005: Computer
Applications and Quantitative Methods in Archaeology, proceedings of the 33rd
Conference, edited by: Figueiredo, A., pp. 175–184, 2005.
</mixed-citation></ref-html>
<ref-html id="bib1.bib11"><label>Eltner and Schneider(2015)</label><mixed-citation>
Eltner, A. and Schneider, D.: Analysis of different methods for 3D
reconstruction of natural surfaces from parallel-axes UAV images, The
Photogrammetric Record, 30, 279–299, <a href="https://doi.org/10.1111/phor.12115" target="_blank">https://doi.org/10.1111/phor.12115</a>, 2015.
</mixed-citation></ref-html>
<ref-html id="bib1.bib12"><label>Eltner et al.(2016)Eltner, Kaiser, Castillo, Rock, Neugirg, and
Abellan</label><mixed-citation>
Eltner, A., Kaiser, A., Castillo, C., Rock, G., Neugirg, F., and Abellán, A.: Image-based surface reconstruction in geomorphometry – merits, limits and developments, Earth Surf. Dynam., 4, 359–389, <a href="https://doi.org/10.5194/esurf-4-359-2016" target="_blank">https://doi.org/10.5194/esurf-4-359-2016</a>, 2016.
</mixed-citation></ref-html>
<ref-html id="bib1.bib13"><label>Fitzpatrick et al.(2019)Fitzpatrick, Radić, and
Menounos</label><mixed-citation>
Fitzpatrick, N., Radić, V., and Menounos, B.: A multi-season investigation of glacier surface roughness lengths through in situ and remote observation, The Cryosphere, 13, 1051–1071, <a href="https://doi.org/10.5194/tc-13-1051-2019" target="_blank">https://doi.org/10.5194/tc-13-1051-2019</a>, 2019.
</mixed-citation></ref-html>
<ref-html id="bib1.bib14"><label>Fraser and Cronk(2009)</label><mixed-citation>
Fraser, C. S. and Cronk, S.: A hybrid measurement approach for close-range
photogrammetry, ISPRS Journal of Photogrammetry and Remote Sensing, 64,
328–333, 2009.
</mixed-citation></ref-html>
<ref-html id="bib1.bib15"><label>Gadelmawla et al.(2002)Gadelmawla, Koura, Maksoud, Elewa, and
Soliman</label><mixed-citation>
Gadelmawla, E. S., Koura, M. M., Maksoud, T. M. A., Elewa, I. M., and Soliman,
H. H.: Roughness parameters, J. Mater. Proc. Technol., 123,
133–145, <a href="https://doi.org/10.1016/S0924-0136(02)00060-2" target="_blank">https://doi.org/10.1016/S0924-0136(02)00060-2</a>, 2002.
</mixed-citation></ref-html>
<ref-html id="bib1.bib16"><label>Gerard and Andres(1984)</label><mixed-citation>
Gerard, R. and Andres, D.: Hydraulic roughness of freeze-up ice accumulations:
North Sasksatchewan River through Edmonton, 2nd Workshop on the Hydraulics of Ice Covered Rivers, Edmonton, Alberta, Canada, pp. 62–86, 1984.
</mixed-citation></ref-html>
<ref-html id="bib1.bib17"><label>Gini et al.(2013)Gini, Pagliari, Passoni, Pinto, Sona, and
Dosso</label><mixed-citation>
Gini, R., Pagliari, D., Passoni, D., Pinto, L., Sona, G., and Dosso, P.: UAV
photogrammetry: Block triangulation comparisons, Internation Archives of the
Photogrammetry, Remote Sensing and Spatial Information Sciences, XL-1/W2,
157–162, 2013.
</mixed-citation></ref-html>
<ref-html id="bib1.bib18"><label>Gomez(1993)</label><mixed-citation>
Gomez, B.: Roughness of Stable, Armored Gravel Beds, Water Resour. Res.,
29, 3631–3642, 1993.
</mixed-citation></ref-html>
<ref-html id="bib1.bib19"><label>Hamshaw et al.(2019)Hamshaw, Engel, Rizzo, O'Neil-Dunne, and
Dewoolkar</label><mixed-citation>
Hamshaw, S. D., Engel, T., Rizzo, D. M., O'Neil-Dunne, J., and Dewoolkar,
M. M.: Application of unmanned aircraft system (UAS) for monitoring bank
erosion along river corridors, Geomatics, Nat. Haz. Risk, 10,
1285–1305, <a href="https://doi.org/10.1080/19475705.2019.1571533" target="_blank">https://doi.org/10.1080/19475705.2019.1571533</a>, 2019.
</mixed-citation></ref-html>
<ref-html id="bib1.bib20"><label>Jain(2010)</label><mixed-citation>
Jain, A. K.: Data clustering: 50 years beyond k-means, Pattern Recogn.
Lett., 31, 651–666, <a href="https://doi.org/10.1016/j.patrec.2009.09.011" target="_blank">https://doi.org/10.1016/j.patrec.2009.09.011</a>, 2010.
</mixed-citation></ref-html>
<ref-html id="bib1.bib21"><label>Johnston et al.(2009)Johnston, Masterdon, and Wright</label><mixed-citation>
Johnston, M., Masterdon, D., and Wright, B.: Multi-Year Ice Thickness: Knowns
and Unknowns, in: Proceedings of the International Conference on Port and
Ocean Engineering Under Arctic Conditions, pp. 969–981, 2009.
</mixed-citation></ref-html>
<ref-html id="bib1.bib22"><label>Kirby(1991)</label><mixed-citation>
Kirby, R. P.: Measurement of Surface Roughness in Desert Terrain By Close Range
Photogrammetry, Photogrammetric Record, 13, 855–875, 1991.
</mixed-citation></ref-html>
<ref-html id="bib1.bib23"><label>Nezhikhovskiy(1964)</label><mixed-citation>
Nezhikhovskiy, R. A.: Coefficients of roughness of bottom surface on slush-ice
cover, Soviet Hydrology, Washington, American Geophysical Union, Selected Papers 2, pp. 127–150, 1964.
</mixed-citation></ref-html>
<ref-html id="bib1.bib24"><label>Niethammer et al.(2012)Niethammer, James, Rothmund, Travelletti, and
Joswig</label><mixed-citation>
Niethammer, U., James, M., Rothmund, S., Travelletti, J., and Joswig, M.:
UAV-based remote sensing of the Super-Sauze landslide: Evaluation and
results, Eng. Geol., 128, 2–11, <a href="https://doi.org/10.1016/j.enggeo.2011.03.012" target="_blank">https://doi.org/10.1016/j.enggeo.2011.03.012</a>,
2012.
</mixed-citation></ref-html>
<ref-html id="bib1.bib25"><label>Nikora et al.(1998)Nikora, Goring, and Biggs</label><mixed-citation>
Nikora, V. I., Goring, D. G., and Biggs, B. J. F.: On gravel-bed roughness
characterization, Water Resour. Res., 34, 517–527, 1998.
</mixed-citation></ref-html>
<ref-html id="bib1.bib26"><label>Nikuradse(1950)</label><mixed-citation>
Nikuradse, J.: Laws of Flow in Rough Pipes, Technical Memorandum 1292, National
Advisory Committee for Aeronautics, 1950.
</mixed-citation></ref-html>
<ref-html id="bib1.bib27"><label>Razali and Wah(2011)</label><mixed-citation>
Razali, N. M. and Wah, Y. B.: Power Comparisons of Shapiro0Wilk,
Komolgorov-Smirnov, Lilliefors and Anderson-Darling tests, J.
Stat. Model. Anal., 2, 21–33, 2011.
</mixed-citation></ref-html>
<ref-html id="bib1.bib28"><label>Strickler(1923)</label><mixed-citation>
Strickler, A.: Contributions to the Question of a Velocity Formula and
Roughness Data for Streams, Channels and Closed Pipelines, Tech. Rep. 16,
Swiss Department of the Interior, Bureau of Water Affairs, 1923.
</mixed-citation></ref-html>
<ref-html id="bib1.bib29"><label>Sturm(2001)</label><mixed-citation>
Sturm, T. W.: Open Channel Hydraulics, McGraw Hill, 2001.
</mixed-citation></ref-html>
<ref-html id="bib1.bib30"><label>Watts et al.(2012)Watts, Ambrosia, and Hinkley</label><mixed-citation>
Watts, A. C., Ambrosia, V. G., and Hinkley, E. A.: Unmanned Aircraft Systems in
Remote Sensing and Scientific Research: Classification and Considerations of
Use, Remote Sens., 4, 1671–1692, <a href="https://doi.org/10.3390/rs4061671" target="_blank">https://doi.org/10.3390/rs4061671</a>, 2012.
</mixed-citation></ref-html>
<ref-html id="bib1.bib31"><label>Wazney et al.(2018)Wazney, Clark, and Wall</label><mixed-citation>
Wazney, L., Clark, S. P., and Wall, A. J.: Field monitoring of secondary
consolidation events and ice cover progression during freeze-up on the Lower
Dauphin River, Manitoba, Cold Reg. Sci. Technol., 148, 159–171,
<a href="https://doi.org/10.1016/j.coldregions.2018.01.014" target="_blank">https://doi.org/10.1016/j.coldregions.2018.01.014</a>, 2018.

</mixed-citation></ref-html>
<ref-html id="bib1.bib32"><label>Wazney et al.(2019)Wazney, Clark, and Malenchak</label><mixed-citation>
Wazney, L., Clark, S. P., and Malenchak, J.: Effects of freeze-up consolidation
event surges on river hydraulics and ice dynamics on the Lower Dauphin River,
Cold Reg. Sci. Technol., 158, 264–274,
<a href="https://doi.org/10.1016/j.coldregions.2018.09.003" target="_blank">https://doi.org/10.1016/j.coldregions.2018.09.003</a>, 2019.
</mixed-citation></ref-html>
<ref-html id="bib1.bib33"><label>Weber et al.(2014)Weber, Ramachandran, and Henikoff</label><mixed-citation>
Weber, C. M., Ramachandran, S., and Henikoff, S.: Nucleosomes are
context-specific, H2A.Z-modulated barriers to RNA polymerase, Molecular Cell,
53, 819–830, 2014.
</mixed-citation></ref-html>
<ref-html id="bib1.bib34"><label>Westoby et al.(2012)Westoby, Brasington, Glasser, Hambrey, and
Reynolds</label><mixed-citation>
Westoby, M. J., Brasington, J., Glasser, N. F., Hambrey, M. J., and Reynolds,
J. M.: 'Structure-from-Motion' photogrammtery: A low-cost, effective tool for
geoscience applications, Geomorphology, 179, 300–314,
<a href="https://doi.org/10.1016/j.geomorph.2012.08.021" target="_blank">https://doi.org/10.1016/j.geomorph.2012.08.021</a>, 2012.
</mixed-citation></ref-html>
<ref-html id="bib1.bib35"><label>Yitayew et al.(2018)Yitayew, Dierking, Divine, Eltoft, Ferro-Famil,
Rösel, and Negrel</label><mixed-citation>
Yitayew, T. G., Dierking, W., Divine, D. V., Eltoft, T., Ferro-Famil, L.,
Rösel, A., and Negrel, J.: Validation of Sea-Ice Topographic Heights Derived
From TanDEM-X Interferometric SAD Data With Results From Laser Profiler and
Photogrammetry, IEEE T. Geosci. Remote, 56,
6504–6520, <a href="https://doi.org/10.1109/TGRS.2018.2839590" target="_blank">https://doi.org/10.1109/TGRS.2018.2839590</a>, 2018.
</mixed-citation></ref-html>--></article>
