Property:CSDMS meeting abstract
From CSDMS
This is a property of type Text.
2
During the 21st century, anthropogenically modulated changes in climate and land cover will drive variations in sediment dynamics throughout rivers, reservoirs, and coastlines. These changes threaten the integrity of dams, levees, and riparian ecosystems, necessitating strategies to help mitigate their associated hazards and to detect and prevent adverse consequences of engineering solutions. To optimize these strategies, geomorphologists require calibrated, watershed-scale numerical simulations of sediment transport that can predict how fluvial networks will respond to different forcings throughout their catchments.
We aim to develop watershed-scale landscape evolution models of several U.S. rivers to explore how climate and land-use change over the coming decades to centuries will influence sediment delivery to reservoirs, locks, harbors, and coasts. The models will be calibrated by historical sediment flux data, allowing them to predict how the fluvial systems will respond to plausible scenarios of future climatic and anthropogenic forcings. The Chattahoochee River in the southeastern U.S. is an ideal catchment to begin this work due to its recent urban development and sedimentation records near its outlet at Lake Seminole. We devise procedures for processing USGS NHDPlus HR datasets (Moore et al., 2019) at the HU4 and HU8 scale for compatibility with the fluvial process components of Landlab (Hobley et al., 2017; Barnhard et al., 2020). Using NLCD land cover products (Wickham et al., 2021), NRI erosion rate estimates (USDA, 2020), and historical streamflow and sediment load data (USGS, 2021), we will leverage Landlab to construct models of the Chattahoochee catchment and test their ability to replicate sedimentation records at Lake Seminole. Here, we present preliminary results obtained by applying these procedures to the Chestatee branch of the Chattahoochee River and its outlet at Lake Lanier in northern Georgia. Future versions of this workflow will use a range of projected 21st century precipitation and land cover changes to predict potential variations in future sediment generation, transport, and storage throughout the Chattahoochee watershed and other U.S. rivers.
Earthquakes can trigger the failure of thousands or even tens of thousands of landslides throughout tectonically active landscapes. Short (<10 years) term studies of these events reveal their important place in a hillslope and fluvial hazard cascade. However, it remains unclear if these widespread catastrophic landslide events leave a long lasting impact on landscape forms, and what that impact would look like. We present landscape evolution model experimental design and some preliminary results exploring the impact of earthquakes or other widespread simultaneous landsliding events on landscapes at timescales much longer than the event return intervals. We use the Hylands Landlab component to test landslide, hillslope, and river sediment interactions, and probe landscape metrics like hilltop concavity, drainage density, slope-area relationships, and possibly valley width to test and validate our models. +
Eastern oysters (Crassostrea virginica) are reef-building organisms that occupy tidal and subtidal zones along the eastern coasts of the Americas. They provide key ecosystem services by improving water quality, providing habitat, providing food, and adding to local economies. At the population level, eastern oysters also form reefs which protect coastal habitats from storms and tidal erosion by attenuating waves. The decline of eastern oyster populations coupled with increased coastal storm intensity and rising sea level is exposing coastal habitats to higher levels of risk. One potential avenue to increase coastal protection is to use artificial reef structures that can also boost eastern oyster populations. Yet, there is little research on how oyster population dynamics influence the structure of the reef–artificial or not, and in turn, how the reef structure influences wave attenuation. Our research aims to address this gap by developing a model to simulate oyster populations in St. Augustine, Florida using an agent-based model coded in the Mesa Python framework. This will be coupled with the Landlab TidalFlowCalculator component, to simulate how reef structures affect tidal velocity and water depth. This model represents the first phase of a larger research effort, which aims to investigate the effects of climate change on the evolution of reef structures and estimate their wave attenuation performance over time. +
Ecohydrological modeling capacity of Landlab is introduced and illustrated using examples that couple components for local soil moisture and plant dynamics with spatially explicit cellular automaton-based (CA) plant establishment, mortality, fire and grazing. Several key features of arid and semiarid ecosystems are discussed. Coexistence of tree-grass cover on north facing slopes (NFS) and shrub cover on south facing slopes (SFS) in central New Mexico is attributed to the competitive advantage of trees due to their longer seed dispersal range against shrubs in cooler and more moist NFS. Incorporating a rule on the inhibitory effects of shrubs on grasses enhance modeled shrub cover, while both trees and grasses are favored when runon is included in the local soil moisture model. Feedbacks among livestock grazing, grassland fire frequency and size, resource redistribution and woody plant encroachment are investigated using different ecohydrologic model configurations. These feedbacks are manifested in a three-phase woody plant expansion processes in the model, with rates of encroachment controlled by the state transition probabilities in relation to plant susceptibility to fires, grazing, and age-related mortality. A critical area of woody plant emerges in the model with which a negative feedback between fire size and woody plant expansion begins. Our results underscore the need for developing models that emphasize local and non-local plant interactions for modeling transient ecosystems. +
Environmental change interacts with human migration in complex ways and across multiple scales. This complexity makes agent-based modeling (ABM) a powerful tool to investigate environment-migration dynamics. Here, we present results from an original ABM of environmental migration in Bangladesh. The model simulates an origin community and how a stylized environmental shock to the community impacts labor opportunities and household decisions surrounding migration. Pattern-oriented modeling is a useful approach for evaluating ABM’s by assessing a model’s ability to reproduce multiple observed patterns of phenomena. We use a pattern-oriented approach to test our model’s ability to reproduce multi-level patterns of environmental migration from the literature. Previous work used machine learning methods to calibrate our ABM by identifying regions in parameter space that successfully reproduced the observed patterns. We demonstrated that a strictly income-based migration decision method was able to reproduce patterns of interest, but inconsistently. However, the pattern-oriented approach allows us to implement more complex, behaviorally driven decision-making methods of migration and evaluate their success. In this work, we will present preliminary results implementing and comparing different decision-making methods in our ABM based on existing theories including Theory of Planned Behavior, Protection Motivation Theory, and a mobility potential framework. Ultimately, we hypothesize that a hybrid framework of migration decision-making that includes community norms, social networks, and place attachment will most successfully be able to replicate known patterns of environmental migration. +
Environmental migration is an example of a complex coupled human and natural system with dynamics that operate across multiple spatial and temporal scales. Agent-based modeling (ABM) has demonstrated potential for studying such complex systems, especially where individual decision-making is an important component. In this work, we use an original ABM of environmental shock, livelihood opportunities, and migration decisions to study dynamics of environmental migration in rural Bangladesh. As ABMs are sensitive to the decision-making methods used, we present results utilizing multiple plausible decision-making methods for households deciding whether or not to send an internal migrant. We present results using both a simple economic method based on utility maximization as well as a more behaviorally complex method based on the Theory of Planned Behavior. We hypothesized that a more behaviorally complex decision method which incorporates social networks and community norms would more successfully reproduce the patterns of migration. However, using a pattern-oriented approach to reproduce two key patterns of migration from the empirical literature, we demonstrate that an economic model can reproduce our patterns of interest with high levels of success. For both decision methods, the level of community inequality in distribution of land ownership, which impacts the number of agricultural jobs available within the community, is critically important for patterns of migration outcomes. In this way, our model suggests that community-level inequality is has significant implications of migration dynamics in this study area. +
Ephemeral, steep-side channels (known as gullies and arroyos) are fundamental elements of soil erosion that threaten agricultural lands worldwide with the associated expectation that landscape degradation will accelerate due to anthropogenic climate change. Gullies are also central to landscape evolution as they are dynamic features that intensively altered between infilling and incision phases in the recent geological past. Yet, exogenic (e.g., due to climate or land-use change) and autogenic (e.g., due to natural oscillations between erosion/deposition phases) drivers of gully formation and of changes in their widespread occurrence are incompletely understood and quantified. This is, in part, because erosional dynamics of gully landforms are complex and hard to capture due to: (1) episodic and discontinuous sediment movement in response to discrete rain events, (2) unstable channel walls with frequent mass wasting, and (3) soil and vegetation properties that vary dynamically thus altering both the hydrology and slope stability.
In this work, we focus on developing a new catchment-scale gully erosion model that enables quantification of soil erosion rates and topographic evolution in response to changes in rainfall patterns and vegetation cover over historical and longer timescales. The model includes explicit representation of rainstorm runoff and erosion over sub-minute time scales. During simulation, soil particles are transported both in suspension and as bedload in accordance with their size. Episodic bank failures and headcuts evolve based on local stability criteria derived from soil properties and failure geometry. This poster presents the model configuration, its main components, and the general modeling approach that aims to bridging gaps between event-scale hydrology, sediment dynamics and longer-term landscape evolution models using new and existing components in the Landlab modeling library. We also present preliminary results of model validation against runoff and sediment data from a field site and a sensitivity analysis on how sediment flux and landform development respond to plausible changes in rainstorm properties, landcover, and vegetation dynamics.
Eroding coasts make up the majority of the coastlines on Earth, including the west coast of the United States, and host critical infrastructure like roads, railways, and residential structures. The precarious siting of infrastructure is particularly true for Del Mar, California, where a major railway between Los Angeles and San Diego sits within just a few meters of a cliff edge that is closely backed by dense housing subdivisions. Coastal cliff retreat presents a danger to these communities that is potentially amplified under rising sea level conditions, among other factors, yet constraints on retreat rates are most often limited to those derived from historical imagn ery and maps dating back 10-100 years. These modern retreat rates are then used, in conjunction with multi-model ensembles, for forecasting cliff retreat over the next 50-100 years in order to gauge future impacts to coastal communities. Managers and policymakers make decisions for mitigation efforts based on these results, however they may not capture the full picture of cliff retreat, and the factors that influence it, over time. While nearly all of the existing forecasting models explicitly account for projected sea level rise, the majority of them ignore other factors (e.g. subtidal and subaerial weathering) that may also play a large role.
A recently developed combination of in situ-produced cosmogenic 10Be surface exposure dating in conjunction with a new numerical model of shore platform profile development that takes into account sea level rise, intertidal weathering, and wave attack on cliff retreat provides quantification of cliff retreat histories over hundreds to thousands of years via cliff-normal 10Be sample transects. Here, we use a shore-perpendicular transect of cosmogenic 10Be concentrations from the surface of a sandy claystone shore platform exposed along a narrow and sandy beach backed by a near vertical ~20-meter-tall cliff in Del Mar, California to present a long-term cliff retreat rate of 5.5 - 8 cm a-1 over the last two millennia for this site. This is the first long term cliff retreat rate for any coast in North America determined by this new methodology. Existing decadal retreat rates at and proximal to this site range from 5-20 cm a-1, suggesting that cliff retreat here may be accelerating towards the present. Preliminary modeling results suggest that uplift-corrected sea level rise in Southern California, which remained constant during the late Holocene (0.8 mm a-1) but doubled in the last century, cannot alone explain this potential increase, as modeled platform geometries and associated development rates show a dependence on the imposed weathering rate as well as wave erosion efficacy. Recent investigation into the relative influence of weathering and wave attack on observed cliff retreat at this same location also shows a roughly equal contribution for both drivers. We further explore this and other potential drivers (e.g. land use change) for this potential increase, and speculate on the implications of these results for future cliff retreat forecasting efforts.
Exchange of material across the nearshore region, extending from the shoreline to a few kilometers offshore, determines the concentrations of pathogens and nutrients near the coast and the transport of larvae, whose cross-shore positions influence dispersal and recruitment. Here, we describe a framework for estimating the relative importance of cross-shore exchange mechanisms, including winds, Stokes drift, rip currents, internal waves, and diurnal heating and cooling (Moulton et al., 2023). For each mechanism, we define an exchange velocity as a function of environmental conditions. The exchange velocity applies for organisms that keep a particular depth due to swimming or buoyancy. A related exchange diffusivity quantifies horizontal spreading of particles without enough vertical swimming speed or buoyancy to counteract turbulent velocities. This framework provides a way to determine which processes are important for cross-shore exchange for a particular study site, time period, and particle behavior. I will also describe approaches we've used to communicate the framework to different audiences, including an interactive tools developed by undergraduates.
Moulton M, Suanda S, Garwood J, Kumar N, Fewings M, Pringle J. Exchange of Plankton, Pollutants, and Particles Across the Nearshore Region. Annual Review of Marine Science. 2023 January 16; 15(1):167-202. DOI:10.1146/annurev-marine-032122-115057 +
Extensive overwash occurred on North Core Banks during Hurricane Florence (September 2018). The washover deposits were partially revegetated when, a year later, sound-side inundation and outwash caused substantial erosion during Hurricane Dorian (September 2019). Repeat aerial mapping shows that reestablishment of vegetation on deposits that partially filled the washout channels is slow to non-existent. We suggest washout site revegetation is delayed by lack of organic material, slowing dune growth, and extending vulnerability to overwash. Washout channels often erode several meters of beach, berm, and back-barrier platform and are later filled with inorganic marine sands deposited as spits, bars, and overwash. By contrast, washover deposits can contain ripped-up vegetation and (partially) bury pre-existing vegetation, providing seeds, rhizomes, and plant fragments to generate new growth. We propose a heuristic model of vegetation growth following a sigmoidal curve that depends on an initial (seed) concentration and show that simulations with this model using realistic overwash recurrence, reproduces our observations of slow revegetation in deposits on former washout channels. +
Extreme drought events are becoming more frequent and severe. For example, since the flash drought of 2012 that ravaged the central United States, 2019 was the only year that has not experienced a billion dollar drought disaster. Examining how vegetation-atmosphere interactions change during extreme drought events can improve our understanding of how resilient different plants are at dealing with water stress during drought. We couple a prognostic phenology routine to a 1-D version of the Duke Coupled surface-subsurface Hydrology Model with dynamic vegetation (DCHM-V) to simultaneously simulate changes in plant life stage with water, energy, and carbon fluxes. The predictive phenology model simulates daily changes in canopy greenness and density based on the current meteorological conditions within the DCHM-V. We run the DCHM-V at a 4 km spatial resolution and hourly time step for pixels encompassing three AmeriFlux sites in the Midwestern United States. Modeling phenological changes and resulting land-atmosphere interactions allows us to investigate physical processes governing vegetation water use strategies in response to flash drought. Results show that vegetation under average water-use scenarios experience smaller reductions in growth as compared to isohydric or anisohydric water-use strategies. Transpiration dominates evapotranspiration with ample precipitation but is nearly cut in half during extreme drought resulting in reduced plant water use efficiency. These findings demonstrate the importance of incorporating dynamic phenological when investigating how vegetation modulates water, energy, and carbon under different water stress conditions, and have implications for improving predictions of drought impacts on the land surface. +
Features of landscape morphology including slope, curvature, and drainage dissection are important controls on runoff generation in upland landscapes, while over long timescales runoff plays an essential role in shaping these same features through surface erosion. Many hydrologists have speculated about the importance of this coevolution and its potential for generating hydrological insights; however, observational and computational limits have long prevented direct study of coupled hydro-geomorphic systems over long timescales. What kinds of hydrological features do landscapes exhibit when their runoff is `in-tune' with the form of the landscape? Here we answer this question using a new coupled hydro-geomorphic model that is sophisticated enough to capture saturated and unsaturated zone storage and water balance partitioning between surface flow, subsurface flow, and evapotranspiration, but efficient enough to drive a landscape evolution model over millions of years. We nondimensionalize the model to arrive at a minimal set of dimensionless numbers that provide insight into how hydrologic and geomorphic parameters together affect the ultimate state. Model results show a diverse array of behaviors observed in real watersheds, including the presence of variable source areas and nonperennial streams. We also found some results that were unique and surprising, such as non-dendritic drainage networks. We hope that these results will inspire hydrologists to consider the role that landscape history plays in the hydrological processes observed today and inspire geomorphologists to consider the role of more nuanced hydrological processes in long-term landscape evolution. +
Field-based observations and numerical models of strike-slip faults indicate that the regional footprint and preservation of the landscape response depends on fault slip rates, climatic conditions, and surface erosional activity. Arid desert environments, on one end of the climate spectrum, are especially sensitive to climate changes and tend to provide an excellent record of fault-slip histories and landscape modification in response to faulting. For example, the Salar Grande strike-slip fault slips at slow to moderate rates (~1 mm/yr) across the Atacama Desert of Chile and is characterized by long periods of hyper aridity with the absence of fluvial activity, but still preserves dextral offset geomarkers evidencing past humid periods and faulting. Conversely, wet environments are intensively affected by constant fluvial erosion and mass wasting. For example, in Aotearoa New Zealand, complex systems of parallel right-lateral faults in the Tararua Mountains, North Island, interact with each other with neighboring rivers flowing across and along fault branches that slip at different rates (< 1 mm/yr to > 10 mm/yr) and juxtapose different scale high-relief topography (shutter ridges). Inspired by the complexities of these real-world contrasting strike-slip fault settings, we create analog numerical simulations in Landlab to observe the role of climate variability, sediment, and the interaction between multiple structures affecting the topography. Model results are compared with field observations, focusing on channels, ridges, and mountain range scale observations. +
Field-based studies, remote sensing analyses, and model development inform our understanding of patterns and processes in incipient delta formation and continued progradation. Rarely, however, have the three approaches been used together to understand the entire progradation history of a real-world system. This may be largely due to the paucity of appropriate sites: an accessible location where a delta has been growing for just a few decades. One site that meets these criteria is Mamawi Creek delta, located within the larger Peace Athabasca Delta ecosystem in northeastern Alberta, Canada. Mamawi Creek delta began forming in 1982, just two years before the beginning of Landsat TM observations over boreal Canada. Due to strong scientific interest and human presence in the region, data on sediment, flow, water levels, and elevations have been collected for decades. Here, we use these available field data to create a site-specific morphodynamic model of Mamawi Creek Delta in Delft3D. Leveraging discoveries from previous modeling studies that have examined how inputs, such as percent of cohesive sediment and median grain size, affect resulting delta forms, we approach the inverse question to see how observed delta characteristics in the Landsat record can constrain model setup and inputs. By comparing annual model outputs against remotely sensed observations of true delta form over the last 40 years, we learn the limitations of our field data and consequences of model simplifications. We then iteratively use these comparisons to inform model parameter adjustments and changes to boundary conditions that enhance agreement between these two methods, ultimately producing a simulated delta that acceptably mimics observed progradation. +
First-order delta morphology is primarily governed by the dominant driver of sediment transport (river, waves, or tides), which dictates the overall sediment balance of the delta. Vegetation affects morphology through its influence on hydraulic roughness, introducing flow resistance, altering velocity distributions, and modulating sediment deposition. In turn, morphodynamic changes affect vegetation, creating a bio-geomorphic feedback-loop that results in a dynamic equilibrium, where the delta’s morphology and dynamic vegetation co-evolve over decadal timescales. This study examines how dynamic saltmarsh vegetation influences delta morphology in a river-dominated delta. We used Delft3D coupled with an ecomorphodynamic model to simulate the feedback-loop between dynamic vegetation and an idealized river-dominated delta, inspired by Wax Lake Delta, Louisiana. The ecomorphodynamic model simulates vegetation establishment, growth, and mortality resulting in dynamic variations of vegetation height and density that produce realistic hydraulic roughness distributions. Our findings demonstrate how vegetation-induced roughness variations can influence sediment redistribution and changes in bed elevation, affecting the morphology. This research highlights the importance of considering vegetation in morphodynamic studies, improving the understanding of delta systems. +
A
Flash floods are among the most devastating natural hazards, which cause loss of life and severe economic damages. Modeling flash floods to provide warnings to the public to prevent/mitigate the impacts of this type of disaster is still challenging. A coupled model which consists of the currently used Hydrology Laboratory - Research Distributed Hydrologic Model (HL-RDHM) at NWS and a high resolution hydraulic model (BreZo) has been developed for flash flood modeling purposes. The model employs HL-RDHM as a rainfall-runoff generator in coarse resolution to produce surface runoff which will be zoned into point source hydrographs at the sub-catchment outlets. With point source input, BreZo simulates the spatial distributions of water depth and velocity of the flow in the river/channel and flood plain.
The model was utilized to investigate the historical flash flood event in the Upper Little Missouri River watershed, Arkansas. This event occurred on June 11th, 2010 and had killed 20 people and caused severe property damages. The catchment was divided into 55 sub-catchments based on Digital Elevation Model (DEM) at 10m resolution from USGS. From HL-RDHM surface runoff, 55 hydrographs can be derived, which then become 55 point sources as input in BreZo. The system was calibrated by tuning the roughness parameter in BreZo to best match the USGS discharge observation at the catchment outlet. The simulation results show the system performed very well not only for the total discharge at the catchment outlet (Nash-Sutcliffe efficiency = 0.91) but also the spatial distribution of the flash floods. +
2
Flood hazards can increase or decrease as a result of changes in the frequency of high flows and changes in the geometry of river channels, through aggradation, incision, or widening. Across the US, Slater et al. (2015) found that a statistically significant majority of studied sites saw increases in the frequency of flooding over the past several decades. Notably, the magnitude of channel response and hydrologic non-stationarity varied between channels within a region. Here, we focus in on a single region, the Pacific Northwest, and ask 1) can the geomorphic characteristics of a basin explain historical changes in flood hazard? And, 2) how will flood risk change with climate change in relation to source-to-sink sediment dynamics? As a first step in understanding the sensitivity of different basins to future climate change, we look at historical records of both channel geometry change and discharge records at ~60 USGS gage sites across Washington state. We find substantial variation among the studied sites in the magnitude of channel change (quantified in terms of changes in the stage-discharge relationship) over the past 3 decades. Some channels have maintained a steady stage-discharge relationship over 30 years, while others change dramatically on an annual basis. Many, but not all, of these unstable channels drain basins with retreating alpine glaciers. Inspecting the discharge records, we find substantial variation as well, likely driven by the differences in hydrologic regime. In the future, we will use this understanding of historical channel sensitivity to inform our predictive models of both channel geometry change and non-stationarity in high flows. +
A
Floodplain deposition maintains and builds up low-lying lands along rivers and in deltas. Floodplain aggradation processes and patterns determine how vulnerability of low-lying land changes over timescales of decades to hundreds of years. Over the longterm, floodplain deposition and channel migration determine the depositional architecture with impacts on groundwater and hydrocarbon reservoirs.
We build and enhanced a 3D floodplain architecture model, AquaTellUs. AquaTellUs uses a nested model approach; a 2D longitudinal profile, embedded as a dynamical flowpath in a 3D grid-based space. A main channel belt is modeled as a 2D longitudinal profile that responds dynamically to changes in discharge, sediment load and sea level. Sediment flux is described with a modified Exner equation by separate erosion and sedimentation components. Erosion flux along the main flowpath depends on river discharge and channel slope, and is independent of grain-size. Depositional flux along the channel path as well as in the lateral direction into the floodplain depends on the local stream velocity, and on grainsize-dependant settling rates. Multiple grainsize classes are independently tracked.
Floodplain deposition is an event-driven system, only peak discharge events cause overbanking, flooding and perhaps channel avulsion. The computational architecture of AquaTellUs preserves stratigraphy by event, allowing for preservation of information of depositional layers of variable thickness and composition.
We here present experiments that show the pronounced effect of different probability density functions for river discharge and sediment load, i.e. flooding recurrence times, on the stratigraphic architecture. +
2
Floods can be devastating to society and the environment. Recent flood events around the globe, such as Harvey and Irma for instance, have been disastrous and broke records in damage and loss of life. Flood disasters often operate at spatial and temporal scales that far exceed local and regional, or even national, assessment and response capabilities. There is no doubt that remote sensing observations of floods, particularly from satellites, can be of great value. Earth observation (EO) data of floods can either be used directly through numerous services providing flood maps and other datasets, or indirectly through integration with hydrodynamic models simulating events continuously in time and space. In this project, we demonstrate the value of satellite flood maps for Harvey 2017 and Twitter feeds during the event for integration with a forecast inundation model (LISFLOOD-FP). Initial results are illustrated and we discuss current challenges and next steps. +
Flow network models are commonly used to study the formation and evolution of karst conduit systems and subglacial conduit systems. Such models involve: 1) numerical solution of flow within the network, and 2) calculation of the rate of change of conduit or fracture size within each segment of the network. Solution of flow and conduit growth is alternated to simulate long-term evolution of the system. Head loss equations, such as the Darcy-Weisbach or Hagen Poiseuille Equations, and a prescription of flow conservation at conduit junctions, are used to iteratively solve for flow within each segment of the network. In the case of karst development codes, discharges within the network are used along with kinetic rate equations to calculate transport and dissolution rates within every conduit segment. For subglacial systems, pressure head and frictional energy dissipation determined from the flow solution are utilized to calculate conduit growth by ice melting and closure due to ice creep. The Landlab modeling environment and associated gridding library greatly ease the development of a flow solver. Here we present the first stages of development of a conduit evolution code within Landlab, with applications both to subglacial and karst systems. Future work will focus on coupling landscape evolution models with network growth models to examine interactions between surface and subsurface processes. +
