Research

The Sound of Flowing Water

Hydrologic processes are typically measured with sensors that offer spatially- and, in some cases, temporally- discrete data. For example, the United States Geological Survey streamflow network is one of the largest and most comprehensive worldwide, however, it covers less than 1% of the country’s streams and rivers and flow monitoring is biased toward larger, perennial streams and rivers . The growth of Distributed Acoustic Sensing (DAS) with fiber optic cables in the geosciences over the last decade offers a unique and timely opportunity to evaluate the potential of DAS to monitor diverse hydrologic processes and to develop next-generation methods and algorithms to quantify hydrologic variables continuously and more rapidly than currently possible. The long-term goal is to develop the necessary tools and guidelines that will enable agencies such as the USGS to have continuous and reliable Discrete Acoustic Sensing measurements of hydrologic processes (fluvial, groundwater, groundwater/surface water exchange), augmenting more traditional monitoring methods. As an initial step to achieve that goal, the main objective of this project is to evaluate the potential of Distributed Acoustic Sensing (DAS) to quantify different hydrologic processes in the surface and the subsurface through field deployments and laboratory experiments.

Collaborators: Negar Soltani, Tieyuan Zhu, Martin Briggs

MLIV - Machine-Learning Enhanced Image Velocimetry for Streams 

Quantifying streamflow is vital for water resource management, habitat monitoring, and emergency response. Streamflow measurements typically use direct or indirect approaches that depend on site and flow conditions. Extreme events and/or flashy systems (remote, urban) are difficult to measure because of the need for timely access to the sites, sensor maintenance issues, and safety concerns. Image-based techniques can be relatively low cost and have the advantage of not needing to be in the water to conduct the measurements. Nevertheless, they may face limitations depending on sunlight conditions, surface reflections, presence of fog, rain and camera angle or field of view. Under such conditions, flow field reconstruction from image-based velocimetry may become sparse or have lower quality.  This project will enhance image-based streamflow measurements (STIV and LSPIV) by combining proven machine learning approaches with image-based velocimetry methods to improve the reliability of these measurements and make the technique accessible to users worldwide.

Collaborators: Frank Engel, Xiaofeng Liu, Alejandro Tenorio

                                                                                                                                                                                                                                                                                                                                                                                Funding: USGS

Streambank Erosion in Cold Regions

Ten percent of our planet is covered by ice, and eleven percent of it is underlain by permafrost (i.e., ground that is frozen year-round). Thawing induced by increasing temperatures makes these regions extremely vulnerable to large landscape and ecosystem changes and will affect the health and wellbeing of people that inhabit and rely on them. Moreover, key infrastructure, including key U.S. energy infrastructure, built in these regions is at risk of being destroyed. At the root of these forthcoming landscape changes lie small-scale erosive processes. As ice melts and permafrost thaws, they promote the movement of sediment grains that would otherwise be fixed in place. To provide resilient solutions to the challenges imposed by climate-change in cold regions, we desperately need to quantify those small-scale erosive processes driving most of the large-scale landscape changes. 

This project has been featured in the IEE website: Cold Regions Research at PSU and in the Growing Impact podcast: Investigating Thawing Permafrost

Collaborators: Anastasia Piliouras, Talley Fisher, Daryl Branford, Elena Halmi, Navneet Singh

                                                                                                                                                                                                                                                                                                  Funding: Institute of Energy and the Environment

Natural sculptures carved by melting and dissolution

Erosion is nature’s most successful artist. Its artwork is found everywhere on our planet and across many extra-terrestrial environments. It sculpts a full range of landscapes, including urban desire paths, created by human and animal footfall; meandering river canyons, created by water flows; and coastal cliffs worn away by repeated wave action. Erosion operates at many scales, often creating repeating patterns, which can be produced through different processes such as: erosion and deposition of sediment (e.g. Amazon River), mechanical wear and abrasion of bedrock (e.g. Colorado River), dissolution (e.g. meandering channels over limestone) and melting (meltwater channels over icecaps or glaciers). 

This cross-disciplinary project at the science-arts interface will allow me to move beyond communicating an idea (scientific outputs) to make people experience it (artistic outputs). 

This project has been featured in The Leverhulme Trust website: Grants in Focus

 Funding: The Leverhulme Trust

Meltwater Meandering Channels: (Images) Self-formed laboratory mm-scale meltwater meandering channels. Flow direction indicated by arrows. Black lines on right two images are @1mm. (Animation) ~30 minutes of meander bend growth and migration in a cm-scale meltwater meandering channel. Flow from left to right (-->). 

Laboratory observations on meltwater meandering rivulets on ice   

Rivers develop meandering patterns over many different media and across many different scales. One particular case is that of meltwater meandering channels, which form and evolve over glaciers and other ice surfaces. These channels show remarkable similarities with meandering channels in other media such as alluvium and bedrock. 

I conducted laboratory experiments on meltwater meandering channels at two scales, i.e. width ~1cm and ~1mm,  to quantify their planform morphologies and similarity with meandering channels in other media. The results, which are being prepared for publication, answer the following questions: 

Results from this study have been published in Earth Surface Dynamics.  Paper: Fernández and Parker (2021)

 Funding: NSF and The Leverhulme Trust

The influence of sediment composition and granulometry over coastal morphology: a case study in three distinct Brazilian rivers. 

Our coasts, estuaries, and low-land river environments are some of the most sensitive systems to environmental change. In order to manage these systems, and adapt to future changes, we desperately need to be able to predict how they will alter under various scenarios. However, our knowledge of these environments is not yet enough to predict, with confidence, far into the future. One of the main reasons our models and geological interpretations of coastal environments are not yet good enough, is because these models have formulations that are based on assumptions that these systems are composed of only non-cohesive sands. However, mud is the most common sediment on Earth and many of these systems are actually dominated by biologically-active muds and complex sediment mixtures. This project will provide answers to the following questions:

Collaborators: Marco Martins, Lívia Carrera, Daniel Parsons. 

Project-Funding: ERC-Confap-CNPq 

Different coastal morphologies in Northern-Brazil (a) Muddy-sand beach in Marudá, Brazil near the Tocantins river estuary. (b) Satellite image showing muddy and sandy coastlines in Northern-Brazil. (c) Sandy beach in Parnaíba, Brazil near the Parnaiba River estuary.  

Morphodynamic Signatures of Clay Content on Combined-Flow Ripples  

Sediment beds with mixtures of cohesive and non-cohesive material exhibit different behaviors than those of purely non-cohesive sediment under similar hydrodynamic forcings. I am currently running experiments that will provide answers to the following questions:

Experiments are being conducted in the Total Environment Simulator (TES), University of Hull. 

Collaborators: Xuxu Wu, Ellen Pollard, Elena Bastianon, Anne Baar, Jonathan Malarkey, Jaco Baas, Andrew Manning, Daniel Parsons. 

Project-Funding: GEOSTICK - ERC

Clay/sand mixing

Clay was soaked in water before opening the bags to avoid issues with dust in the lab. It was sticky and slippery!

Combined-Flow Ripples

Three channels exposed to the same flow conditions (waves plus current). Flow and waves from top of image. Left channel had an initial clay concentration of 20%; middle channel 15%; and right channel 10%. Image shown was obtained after 20 hours (all channels were flat at t=0 hr).

Microplastic and Cohesive Sediment Settling Properties 

Flocculation of cohesive sediments occurs in both natural and engineered environments, and is affected by biotic and abiotic factors alike. Microplastics are commonly find in waterbodies throughout the world. The settling properties of mixtures of these materials are yet unknown. 

Experiments are being conducted using the LabSFLOC system. Image shows the equipment and a picture acquired with it. 

Collaborators: Freija Mendrik, Catherine Waller,  Daniel Parsons. 

Project-Funding: GEOSTICK - ERC

Sediment transport on sticky substrates

Sediment transport mechanics for non-cohesive materials have been studied extensively. However, the mechanics change when the bed is under the influence of sticky agents (cohesion/adhesion). The mechanics are different when cohesion comes from the presence of clay or when it comes from the presence of extra polymeric substances (EPS) secreted by organisms. I am beginning to study the problem, with the collaboartors listed below, by using some simple visulaization techniques. An initial test is shown below (images courtesy of Jonathan Malarkey). The bed contains a mixture of sand, kaolin clay, and xanthan gum (EPS surrogate). Winnowed clay appears in suspension and large clumps of sand/clay stuck to an EPS matrix saltate and roll over the bed. 

Collaborators: Elena Bastianon, Jonathan Malarkey, Jaco Baas, Daniel Parsons. 

Project-Funding: GEOSTICK - ERC

Hydraulic resistance in mixed bedrock-alluvial meandering channels 

Hydraulic roughness is typically described in terms of a reach-averaged friction factor. In the case of alluvial rivers, this coefficient depends on the size of the material on the bed (skin friction) and, if present, on the size of bedforms (form drag). In the case of mixed bedrock-alluvial rivers, the definition of an appropriate roughness coefficient is more challenging since it also depends on the size of the roughness of the bedrock elements, i.e. the macro-roughness, and the percentage of areal cover of alluvium, defined as the ratio of area covered with sediment to total area.

I ran experiments to learn:

This study was published in the Journal of Hydraulic Research: Fernández et al. (2020)

Funding: NSF

(a) Kinoshita flume sketch with water elevation sensor locations and a few cross sections indicating streamwise distance along the Kinoshita shape; (b) Kinoshita flume side view image with rectangle indicating approximate area shown in (c); (c) partially alluviated bed inside the Kinoshita flume, eTape (water level sensor) is shown in the back. Shaded areas indicate bedrock reach in experiments. Darker shade indicates the region of interest (ROI) used to measure alluvial cover 

Temporal fluctuations of alluvial cover and water surface slope under steady flow conditions: Temporal series of alluvial cover for runs with (a)79% alluvial cover and (b) 21% alluvial cover; and temporal series of water surface slope for same runs (c) and (d).

Role of Alluvial Cover in Mixed Bedrock-Alluvial Meandering Channels

Bedrock erosion by abrasion is driven by sediment particles that strike bare bedrock while traveling downstream with the flow. If sediment deposits and forms patches of alluvial cover, this mode of erosion is no longer possible due to the protection offered by the grains themselves. Bedrock erosion by abrasion is related to the amount of alluvial cover which in turn is a function of sediment load and hydraulic conditions. 

I conducted experiments in the Kinoshita meandering flume at the Ven Te Chow Hydrosystems Laboratory at the University of Illinois at Urbana-Champaign and the results, available as a pre print (Fernández et al., 2019), answer the following questions:

This study was published in Earth Surface Dynamics: Fernández et al. (2019)

Videos from experiments

Funding: NSF

Spatio-temporal Average of Alluvial Cover in a Bend of the Kinoshita Meandering Flume: Time lapse images acquired every 10 seconds provide quasi-instantaneous alluvial cover maps. The image shows the spatio-temporal average of alluvial cover after three hours. Black indicates channel bed locations with permanent alluvial cover and white indicates locations of permanently exposed bedrock. Incision is expected to occur within the colored areas. 

Meandering by Dissolution  

Channels created by dissolution into bedrock can also meander; such channels are observed at very small scale (∼10 cm wavelength) in a surface karst phenomenon known as meanderkarren. Some of the best examples in the field may be found at The Burren, Irleand. They lie in decantation runnels several meters long draining meter-scale solution ponds or kamenitzas. Typical runnel widths and wavelengths are centimeters to decimeters and their channel gradients are of the order of a few percent. The Burren meanders occasionally exhibit high sinuosity (>2), terracing, loop cutoffs, and outer-bend undercutting, and there is little doubt they have evolved in-situ through flow-mediated calcium carbonate dissolution. 

To better understand such processes in the laboratory, I tested different materials and ran trial experiments which contributed to the making of the video shown below.  

Funding: NSF

Meanderkarren Example: Image from The Burren, Ireland; courtesy of Stephen Marshak. Flow from right to left (<--).

Make your own dissolutional meandering channel: Video showing three different materials that may be used to learn about meandering by dissolution at a small scale in a lab or class setting. 


Dissolutional Meandering: This channel was fromed by dissolution. The picture shows the channel after 19 hours. Check the video for more details. 

Particle Tracking Velocimetry

 - PTV - 

I have expanded, sped-up, improved and developed MatLab routines to do Particle Tracking Velocimetry. I learned and used this technique during my MSc degree. I conducted measurements in the Kinoshita meandering flume at the Ven Te Chow Hydrosystems Laboratory at the University of Illinois at Urbana-Champaign. The experiments looked at the differences and similarities of surface flow velocities for three different width-to-depth ratios.  

PTV Image Example: Experimental image overlain with PTV results. The arrows indicate the direction and magnitude of the displacements of the particles between succesive frames. Displacement magnitudes are also shown with color. 

RVR Meander - Graphical User Interface - 

I developed the graphical user interface (GUI) for RVR Meander, a river planform migration software that previously ran only in stand-alone mode through the command prompt. The interface was developed in C# and was implemented as a toolbar within ArcGIS-ArcMap. It provides the same capabilities as the stand‐alone version of RVR Meander with the added benefits of the GIS environment. 

The toolbar has five different buttons. The first one (Text to Shapefile) allows creating a river centerline shapefile from its cartesian coordinates stored within a text file. The second button (RVR Meander) opens the main GUI which contains different tabs and menus which allow the user to input or import/export the required parameters and data. Once the simulation is set up, the user can run it. 

The third button (1D Output) allows the user to import the planform migration results as a shapefile into a map. The fourth button (2D Output) imports the results from the hydrodynamics and bed morphodynamics for as many time steps as specified by the user. The fifth button (Curvature Mapper) allows the user to assess the local curvature values for either an input river centerline or for any of the resulting centerlines. 

More details here: Langendoen et al. (2015)

Funding: USDA

RVR Meander Output Example: Top panel shows basic input used by the software. Middle panel shows modeled river centerline planform evolution and bottom panel shows 2D simulation results.

Riverine Harbor Siltation  

Siltation inside a harbor built on the Mississippi River downstream of St. Louis, MO has been a problem since the harbor started operating in 2007. The study was conducted in order to answer the following questions: 

1. What is the source of the sediment responsible for siltation inside the harbor? 

2. When are these sediments most likely to be deposited?  

3. How does siltation relate to the hydraulic conditions in the river?  

A comparison between the characteristics of the sediment found in the harbor and the suspended and bed material in the river showed that the wash load of the Mississippi River enters the harbor and deposits due to the lack of flow-through velocities. Characteristic discharges in the river and associated suspended sediment concentrations were computed to determine the thresholds above which significant siltation could be expected.  

This study was published in the Journal of Hydraulic Engineering: Fernández at al. (2018)
The full report is also available: Fernández et al. 2012

Funding: Holcim US

Harbor Bathymetric Surveys: The volume of sediment deposited inside the harbor increases rapidly between March and July. 

Sensitivity of Sediment Transport Formulas to Input Variables

I proposed the use of a methodology to assess input-variable sensitivity for sediment transport formulas, namely, the Mean Value First Order Second Moment Method (MVFOSM). I applied it to two bedload transport equations showing that it may be used to rank all input variables in terms of how their specific variance affects the overall variance of the sediment transport estimation. The method proposed can be easily extended to any other sediment transport formulas. 

In sites where data are readily available, the results would allow quantifying the effect that the variance associated with each input variable has on the variance of the sediment transport estimates. In sites where data are scarce or nonexistent, the results obtained may be used to:

1. Determine what variables would have the largest impact when estimating sediment loads in the absence of field observations. 

2. Design field campaigns to specifically measure those variables for which a given transport formula is most sensitive. 

This study was published in Water Resources Research: Fernández and Garcia (2017) 


MVFOSM Example: Sensitivity of Recking (2013) equation to variance in grain size and slope. For the same coefficient of variation in both variables, slope contributes less than 40% to the overall variance of the result.  

Cyclic Steps Due to Pulsed Turbidity Currents

I conducted research with Japanese visitors and colleagues from the University of Illinois looking at the morphologies of cyclic steps formed by pulsed turbidity currents. The goal of the experiment was to understand how the duration of the pulses affects the characteristics of the cyclic steps and the deposits. 

Researchers involved:

Kazunori Fujita, Isamu Mori, Miwa Yokokawa (Osaka Institute of Technology); Roberto Fernández, Matt Czapiga, John Berens, Jeffrey Kwang, Kensuke Naito, Gary Parker(Univ. of Illinois); Norihiro Izumi (Hokkaido Univ.); Hajime Naruse (Kyoto Univ.)

Cyclic Steps: Top panel shows turbidity current over deposit created after 20 pulses. The run shown had turbidity current pulses of 20 seconds. Bottom panel shows deposit after 40 pulses. Velocities were measured with ADVs and grain size distributions of the material analyzed with a Mastersizer.