Property:CSDMS meeting abstract
From CSDMS
This is a property of type Text.
2
In the nearshore environment, bedforms are prominent small-scale morphological features controlling hydrodynamic dissipation and sediment transport. Vortices are typically regarded as the primary mechanism driving sediment transport over vortex ripples with a steepness larger than 0.1 due to boundary layer separation. As these bedforms induced vortices must cascade into small turbulent coherent structures and eventually get dissipated, a research gap exists concerning the role of coherent structures on bedform evolutions. Here, we investigate the mechanism driving bedform evolution from an initially flat bed, for medium and fine sand ripples in oscillatory flows. Due to much lower steepness, the ripple-induced vortices are much weaker, and we hypothesize that turbulent coherent structures play a crucial role in sediment transport to initiate small bed features. In this numerical study, the Eulerian two-phase flow model, SedFoam, is utilized. The laboratory experimental scenarios from Perillo et al. (2014, Sedimentology) are modeled for medium sand ripple (d_50=0.25 mm) at mobility numbers varied from 10 to 60. By conducting a three-dimensional (3D) large-eddy simulation (LES), the generation and evolution of energy-containing turbulent coherent structures are resolved, as well as their effect on sediment transport. The simulation results demonstrate that the turbulent coherent structures, generated by shear flow in the turbulent boundary layer, are the dominant mechanism driving the initial formation of 3D small bed features, which eventually evolve into more organized symmetrical small ripples (SSR). The simulated time-dependent bedform characteristics, including ripple length and height, are further validated against measurements. +
In the same way that watersheds filter precipitation signals into a time series of flow response, watersheds also filter sediment production signals into a time series of bedload transport. Here, we describe the Mass Wasting Router, a new watershed-scale sediment production and transport model written for Landlab that couples an existing shallow landslide hazard model (LandslideProbability) with an existing network-scale bedload transport model (NetworkSedimentTransporter) by (1) delineating hillslope scale landslides from maps of landslide probability, (2) routing the landslides through the watershed using a “precipiton” or “agent” style model and (3) fluvially eroding the mass wasting deposits and creating parcels for the NetworkSedimentTransporter. Preliminary model runs indicate that variation in soil cohesion and precipitation intensity drive landslide-derived hillslope sediment production rates but valley storage processes, driven by debris flow deposition patterns, modulate bedload transport rates at the basin outlet. +
In the southern San Andreas Fault zone, the San Gorgonio Pass (SGP) stands as a region of intricate structural complexity, pivotal for the assessment of seismic hazards due to its potential role in modulating earthquake rupture propagation. This investigation delves into the SGP's crucial function in earthquake dynamics amid ongoing discussions on slip partitioning among its fault strands, aiming to fill a substantial knowledge gap concerning fault activity spanning the last 1 to 100 thousand years. The challenge of estimating slip rates, exacerbated by a dearth of datable materials within the SGP's challenging terrain, calls for innovative methodologies to assess uplift rates along previously overlooked fault segments. In our study, we use thermoluminescence (TL) thermochronology to evaluate differential uplift by analysing bedrock erosion rates. Although AHe dating sheds light on thermal histories and erosion rates across millions of years, it falls short in detailing the recent uplift history vital for grasping Quaternary fault dynamics. In contrast, cosmogenic 10Be dating proves effective in measuring surface erosion rates over millennial timescales, providing insights into contemporary geological activities. TL dating, with its capacity to discern bedrock exhumation over 10-100 ka, acts as a bridge between the temporal scales of AHe thermochronology (Ma) and cosmogenic 10Be denudation rates (ka). By juxtaposing erosion rates across different faults within the SGP, our research aims to pinpoint active fault segments, thereby enriching our understanding of fault dynamics and seismic risk in the southern San Bernardino Mountains. +
A
In this study, a methodology based on a multi-resolution wavelet analysis is introduced to extract a regularized topographic index (TI) distribution from a high-resolution DEM (digital elevation model). The methodology is a promising method to deal with common problems in hydrological applications of high-resolution DEMs, which usually contain noise, pits and redundant information. Formation of several unconnected saturated zones is a particular case of such problems when TOPMODEL is employed for simulation of hydrological processes within a basin. The proposed method includes four steps. The first two steps are used for smoothing and de-noising purposes and include decomposition of the original DEM into multi-level sub-signals by 2-dimentional discrete wavelet transform (DWT) and thresholding of the wavelet coefficients. In the next step, the original smoothed and filtered DEM is reconstructed using inverse DWT. Finally, the TI distribution and its information content are computed. The computed information content is used as a metric to identify an optimal TI distribution which contains reasonable topography information in the absence of noise and redundancy. Application of the procedure to 1-m resolution LiDAR (light detection and ranging) DEM of the Elder Creek River watershed via the TOPMODEL framework indicates its filtering ability to smooth and connect the saturated areas during the hydrological process. In addition to the rainfall-runoff modeling, the proposed pre-processing technique may be applied wherever a high-resolution DEM is employed for distributed simulation of hydro-environmental processes. +
2
In this study, implicit and explicit spectral solutions are considered for solving the linear diffusion term of a simple 2D loosely coupled landscape evolution model. Spectral methods are powerful tools for solving elliptical partial differential equations and are widely used in other fields, though they have received comparatively little attention in landscape evolution modelling. In the LEM considered, the land surface elevation is altered by three processes: regional uplift, fluvial incision, and linear hillslope diffusion. In the simplest case, these processes act in an undifferentiated way across the entire landscape. While a recent algorithm has provided a powerful implicit solution to for the fluvial incision term, explicit formulations of diffusion remain standard. However, when the desired grid is large, an explicit method may be restricted by stability to a time step too small for the timescales of interest. To solve this problem implicitly, I transform the problem into the spectral domain, solve the 2D diffusion equation with a Crank-Nicholson method, and compare the results to explicit finite difference and explicit spectral methods. In its most simple formulation, the spectral methods require periodic boundary conditions in both dimensions. Resulting from these conditions, I show a tessellating solution where the landscape takes the form of a flat torus. +
Information on flood inundation extent is important for understanding societal exposure, water storage volumes, flood wave attenuation, future flood hazard, and other variables. A number of organizations now provide flood inundation maps based on satellite remote sensing. These data products can efficiently and accurately provide the areal extent of a flood event, but do not provide floodwater depth, an important attribute for first responders and damage assessment. Here we present a new methodology and a GIS-based tool, the Floodwater Depth Estimation Tool (FwDET), for estimating floodwater depth based solely on an inundation map and a digital elevation model (DEM). We compare the FwDET results against water depth maps derived from hydraulic simulation of two flood events, a large-scale event for which we use medium resolution input layer (10 m) and a small-scale event for which we use a high-resolution (LiDAR; 1 m) input. Further testing is per- formed for two inundation maps with a number of challenging features that include a narrow valley, a large reservoir, and an urban setting. The results show FwDET can accurately calculate floodwater depth for diverse flooding scenarios but also leads to considerable bias in locations where the inundation extent does not align well with the DEM. In these locations, manual adjustment or higher spatial resolution input is required. +
Infrequent, large-magnitude discharge (>10^6 m^3/s) outburst floods—megafloods—can play a major role in landscape evolution. Prehistoric glacial lake outburst megafloods transported and deposited large boulders (≥4 m), yet few studies consider their potential lasting impact on river processes and form. We use a numerical model, constrained by observed boulder size distributions, to investigate the fluvial response to boulder deposition by megaflooding in the Yarlung-Siang River, eastern Himalaya. Results show that boulder deposition changes local channel steepness (ksn) up to ∼180% compared to simulations without boulder bars, introducing >100 meter-scale knickpoints to the channel that can be sustained for >20 kyr. Simulations demonstrate that deposition of boulders in a single megaflood can have a greater influence on ksn than another common source of fluvial boulders: incision-rate-dependent delivery of boulders from hillslopes. Through widespread boulder deposition, megafloods leave a lasting legacy of channel disequilibrium that compounds over multiple floods and persists for millennia. +
Inhabitants of Bangladesh and West Bengal rely significantly on groundwater for drinking water. Estimates suggest that 7 to 11 million drinking water wells are contaminated by high concentrations of naturally occurring arsenic. The arsenic likely derives from the pyrite-rich sediment of the Ganges basin, however, the cause and timing of mobilization of the arsenic have been difficult to determine. Generally, shallow and deep aquifers contain low concentrations of arsenic, while mid-depth aquifers (20 to 100 m) are often contaminated. The Ganges river is extremely active and has dissected large portions of previously deposited sediments, introducing significant subsurface heterogeneity and complicating the search for safe drinking water. Here, we have aggregated a variety of datasets into a PostgreSQL database, which we use to build predictive models of arsenic concentration in groundwater. We use the Bangladesh Arsenic Mitigation Water Supply Project (BAMWSP) dataset of ~4.5 million wells to train our models. The predictors for our models are largely based on ~15,000 stratigraphic sediment samples from ~10 transect and ~400 total cores. Approximately 5,000 of these samples have been analyzed for grain size, magnetic susceptibility, chemical composition, and organic matter content. We use elevation and population density as additional predictors. With this database, we will create a regional statistical model that may lead to better prediction of arsenic contaminated wells. By compiling and analyzing these data, we hope to improve water security in this rapidly developing region. +
A
Integrated hydrologic models are growing in application and show significant promise in unraveling connections between the surface, subsurface, land-surface and lower atmospheric systems. Recent advances in numerical methods, coupled formulation and computing power have all enabled these simulation advances. Here, I will discuss the modeling platform ParFlow, an integrated hydrologic model that has been coupled to land surface and atmospheric models. I will then discuss a recent application of this model to a large, Continental-Scale domain in North America at high resolution that encompasses both the Mississippi and Colorado watersheds. Details will include techniques for model setup and initialization, in addition to results that focus on understanding fluxes, feedbacks and systems dynamics. Additional anthropogenic complications such as the effects of pumping, irrigation and urbanization will be discussed and a path forward for integrated simulations of the hydrologic cycle will be presented. +
2
Integration of humans within landscape evolution models (LEM) as responsive actors in complex human-environmental systems, is still in its infancy. LEMs that included human decision-making have done so either entirely within an agent-based model (ABM) (e.g., CYBEROSION (Wainwright 2008)) or by coupling an ABM with a LEM (e.g., MedLanD (Barton et al. 2012)). These LEM-ABM examples have analyzed the effects of land use and tillage decisions on landscape evolution, but other ways in which humans interact with geomorphic systems have yet to be explored. Our research expands human-environment interaction modeling to landscapes modified by agricultural terraces to explore the long-term geomorphic evolution in these regions.
Agricultural terraces are anthropogenic landforms that have been constructed for centuries in many parts of the world. Despite their widespread distribution and well-known reduction of sediment transport, terraces have rarely been included within LEMs (cf. Lesschen, Schoorl, and Cammeraat 2009). Recent research on agricultural terraces has revealed that terrace abandonment often increases soil erosion and landscape degradation, reversing landscape evolution patterns modified by terrace construction (Tarolli, Preti, and Romano 2014/6; Arnáez et al. 2015/5). We present the Agricultural Terraces Model (AgrTerrModel), which is a coupled LEM-ABM system for analyzing long-term human-environment interactions in terraced landscapes. The LEM component is implemented using the Landlab library and features adjustments to governing landscape evolution equations to reflect changes to geomorphic processes after terrace construction, such as the impact of stone terrace walls that block sediment movement downslope. The ABM component is implemented using the Mesa ABM framework and includes mechanisms for terrace wall collapse and maintenance, as well as agents who determine cultivation and maintenance practices for terraced land. Using the AgrTerrModel, we simulate landscape evolution in Vernazza, Liguria, Italy near Cinque Terre to analyze how the timing and amount of terrace wall maintenance affects sediment transport. The interaction between seasonal precipitation and the timing of terrace wall maintenance is of special interest due to the Mediterranean climate of the study area. This project provides new insights into the evolution of terraced landscapes and an avenue for further research into the complexity of human-environment systems.
References Cited:
Arnáez, J., N. Lana-Renault, T. Lasanta, P. Ruiz-Flaño, and J. Castroviejo. 2015/5. “Effects of Farming Terraces on Hydrological and Geomorphological Processes. A Review.” Catena 128: 122–34.
Barton, C. Michael, Isaac I. T. Ullah, Sean M. Bergin, Helena Mitasova, and Hessam Sarjoughian. 2012. “Looking for the Future in the Past: Long-Term Change in Socioecological Systems.” Ecological Modelling 241 (August): 42–53.
Lesschen, J. P., J. M. Schoorl, and L. H. Cammeraat. 2009. “Modelling Runoff and Erosion for a Semi-Arid Catchment Using a Multi-Scale Approach Based on Hydrological Connectivity.” Geomorphology 109 (3–4): 174–83.
Tarolli, Paolo, Federico Preti, and Nunzio Romano. 2014/6. “Terraced Landscapes: From an Old Best Practice to a Potential Hazard for Soil Degradation due to Land Abandonment.” Anthropocene 6: 10–25.
Wainwright, John. 2008. “Can Modelling Enable Us to Understand the Rôle of Humans in Landscape Evolution?” Geoforum; Journal of Physical, Human, and Regional Geosciences 39 (2): 659–74.
It has been hypothesized that large, rare flooding events in semi-arid to arid climate regimes may do more erosive work than the frequent storm events that occur in humid or temperate climates. Previous work has demonstrated that added variability in modeled climate or water discharges may be linked to changes in landscape form or channel characteristics. Many landscape evolution models do not capture hydrograph dynamics, so they may miss critical aspects linking flood events and erosion. To explore how different climates shape landscapes, this work uses a hydrodynamic model to simulate flooding and erosion processes. Precipitation time series, based on observed event frequency data from NOAA, are used to differentiate modeled wet and dry regimes. The drier regime is characterized by a heavy-tailed flood probability distribution, where the rarest events have a greater magnitude than storms of a similar recurrence in wetter regions. Hydrographs driven by these precipitation time series are used to erode the topography of a synthetic watershed. Simulations are run with and without an incision threshold. After 10^4 modeled years, landscape characteristics such as relief and channel concavity can be compared. Total eroded depths are evaluated for the different storm frequencies to explore how individual floods and the cumulative work of all floods sculpt landscapes. We propose when an incision threshold is considered, the higher magnitude events in arid regimes will be more effective at shaping watersheds than events of the same frequency in temperate climates. These results inform the discussion of how fluvial erosion may change if anthropogenic climate change leads to the aridification of presently temperate regimes. Additionally, this study will illustrate how hydrograph shape and duration impact modeled landforms, processes not captured in traditional landscape evolution models. +
It has been well documented that climate warming was greater in the Arctic than elsewhere. However, it is still poorly understood how climate changed over different permafrost zones and its potential impacts on permafrost thermal dynamics. In this study, we investigated changes in air temperatures, especially seasonal air temperatures, over different permafrost regions in the Northern Hemisphere using the Climate Research Unit (CRU) gridded datasets from 1976-2016. The primary results indicated that permafrost regions as a whole experienced a warming at 0.36, 0.41, and 0.46 °C/decade in mean annual maximum, mean, and minimum air temperature, respectively, which are 16%, 32%, and 44% higher than the corresponding trend in non-permafrost regions. More importantly, strong increases occurred in cold months and nighttime over continuous permafrost zone, exceeding 0.72 °C/decade in Spring and Autumn; while summer air temperature had a relatively small increase or no statistically significant trends. As a result, the decrease of air freezing index by 529 °C-day would result in permafrost temperature increase by 1.43 °C in continuous permafrost zone over the past four decades. This may explain the observed evidence that increase of cold permafrost temperature was greater than that of warm permafrost, while active layer thickness had little or no change during the past several decades. These results suggest that predicted reduction of permafrost area by previous studies might be overestimated. +
Lago Cachet Dos (LC2) is a glacially-dammed lake adjacent to the Northern Patagonian Ice Field (NPIF), formed by the blockage of Cachet Basin (CB) by the Colonia Glacier. This glacier has experienced rapid (~1-2 km) retreat of its terminus as well as ~1-2 m/yr of thinning, documented over the past several decades. Furthermore, the glacier has exhibited a change in hydrologic regime and the frequency of high energy glacial lake outburst flood (GLOF) events since 2008. These historical changes appear to be coupled with regional climate change; summer mean maximum and minimum temperatures in nearby Cochrane show a steady increase since 1971, whereas winter mean maximum temperatures show cooling in the 1970s and 1980s, followed by gradual warming with rapid acceleration in the 2000s-present. Preliminary correlations with a recently installed weather station at Sol de Mayo (~12 km downstream of the Colonia Glacier terminus) show a strong positive correlation with the Cochrane data, indicating these climate changes are regional and not local and thereby have implications for the evolution of other alpine basins of the NPIF and perhaps glaciers on a global scale. Recent observations from unmanned aerial vehicle (UAV) flights, satellite imagery, and geologic mapping suggest unprecedented glacier deterioration near the southern limit of CB. An UAV flight in January 2016 revealed that during GLOF events, the lake drained through a large hole at the base of the glacier. Upon entering this chasm, the water made a sharp east turn (towards the bedrock abutting the glacier’s eastern margin) and appeared to flow beneath the ice at this point. Subsequently, a large (~2km long x 100 m wide) supra-glacial channel has opened directly above the drainage hole, effectively separating the glacier from bedrock. Ice elevation data reveal that healing of this channel may not be possible under the current climate regime, suggesting the basin could be experiencing a long-term (over human timescales) shift to fluvial deposition from a dominantly lacustrine environment, corresponding to an inability to impound water associated with the glacier's retreat. Basin stratigraphy indicates these oscillations between lacustrine and fluvial conditions have occurred repeatedly throughout the Holocene, but the timing of these changes are poorly constrained. Optically stimulated luminescence (OSL) dating of CB sediments will be applied to identify the timing and periodicity of these depositional shifts, with the broader goal of linking these oscillations with local and regional climate and stability of the Colonia Glacier.
A
Lake records provide a long-term record of climate events and transitions, earthquakes in tectonically active regions, landscape response during and following deglaciation and recent human influenced land use changes. In order to unravel the story preserved in lake sediments, it is necessary to understand the dynamics of the lake system and the source of the sediment coming into the lake. Our study focuses on Lake Ohau, New Zealand, which occupies a fault controlled glacial valley and contains a high resolution sedimentary record of the last ~17 ka. It is presently the focus of a multi-disciplinary studied which aims to recover a long core encompassing the whole ~17 ka record in the next several years.
We use two CSDMS codes: HydroTrend, a climate-driven hydrological model, and Sedflux, a basin filling model, to model sediment flux into Lake Ohau. Using measured climate parameters from the last 60 years, we model water and sediment discharge into the lake and the distribution of sediment through the lake basin. Using a simple conceptual model of the lake dynamics, we produce a series of simulations to examine sediment accumulation at different positions across the lake basin. We then compare these modelled accumulation records to short cores from a number of locations within the lake basin. +
2
Landlab is python software framework for the creation of surface dynamics and process models. It provides grid structures, stable and intercompatible process components, and utilities for data input, output and visualization. Here we present two new types of components within the Landlab framework: the FlowAccumulator and the FlowDirectors. These components have been designed to implement one of the basic functions of surface dynamics modeling, the routing and accumulation of water over a surface. These components split up the functionality of the previously implemented FlowRouter component in order to make it easier for the addition of new algorithms for flow direction to Landlab. As part of these components, we include a new algorithm for efficient flow accumulation when flow is routed to multiple neighboring nodes.
Routing of water over a surface can be split into two steps: direction and accumulation. Before outlining these steps, it is useful to state the terminology used to describe the grid. In Landlab the physical processes operate on a model grid which stores information about spatial location and properties that may vary in space (e.g. soil thickness, surface water discharge). The model grid is a dual plane graph in which quantities such as topographic elevation are defined at node points. Between neighboring node points are lines called links on which water, sediment, or other quantities can flow.
To route flow over the surface, flow directions at a given node must first be assigned to indicate which, if any, of the neighboring nodes will receive any flow that arrives in that node. This is typically done using the relative elevations of a node’s neighbors. Previously Landlab supported the steepest descent (or D4) algorithm for both rectilinear and non-rectilinear grids and the D8 algorithms for rectilinear grids. As part of the presented improvement, Landlab includes the Multiple Flow Direction and D infinity algorithms. Each algorithm for directing flow is its own component, but all share core functionality of the FlowDirector class. This shared functionality includes attributes necessary for interacting with other Landlab components, including the FlowAccumulator. This design permits easy addition of new flow direction algorithms while maintaining interoperability with other Landlab components.
Once flow directions have been assigned, surface water discharge and drainage area can be calculated through flow accumulation. This functionality is provided by the FlowAccumulator component which is compatible with all FlowDirector components. Depending on the algorithm chosen, flow accumulation can be computationally inefficient, scaling at a rate greater than O(N). We present a new algorithm for accumulating flow for in the case where flow is directed to more than one receiver that scales with the number of links that flow is directed over.
Landscape evolution is driven by tectonic processes that build up topography and erosional processes that work to tear down that topography and move material out of the landscape. Climate or more specifically water accumulated in rivers as discharge, is an important driver of these erosional processes. Despite its influence in shaping landscapes, there remains much to be learned concerning the relationships between climate, topography and discharge variability in forcing erosion. One reason for this is that climate itself, as well as the relationship between climate and erosion, is difficult to measure. Climate data typically consists of temporally averaged modern measurements of rainfall, which often miss important variability; or spatially inconsistent discharge data from stream gauges. Further, fluvial incision occurs during flood events that overcome a threshold for erosion, the frequency of exceeding this threshold is influenced by both discharge magnitude and variability. In this work we employ a numerical modeling approach to build upon previous research concerning the importance of discharge variability in driving erosion. We couple a landscape evolution model, Landlab, with a high-resolution atmospheric model, WRF (Weather Research and Forecasting), in order to generate high resolution discharge data. In these experiments, we run the WRF model over two artificial topographies with low (100m) and high (4km) elevations and extract discharge data from five drainage basins across the study domain at 10 to 40S. With this experimental setup we are able to analyze how discharge distributions change with topography and across several climate regimes. We find that the presence of high topography results in higher mean discharge across the five domains, particularly at 10, 20, and 30S. Additionally, discharge variability is greatest at 20 and 30S and less variable at 0, 10, and 40S for both the low- and high-elevation experiments. We then use these discharge distributions to drive a 1D river incision model to explore the relationship between discharge variability and erosion between high and low elevation domains and among different climate regimes, particularly in the presence of erosion thresholds. With this modeling approach we aim to gain insight into the influence of discharge variability and erosion thresholds on the relationships between topography, climate and erosion.
Landscape evolution models (LEMs) are a virtual representation of geomorphic processes as observed in the field or in experimental settings. LEMs offer the flexibility to evaluate a range of interactions between surface processes at timescales which cannot be observed. Notwithstanding the added value of LEMs in unravelling the tectonic-climate-erosion enigma at geological timescales, the use of LEMs to explain real-world earth surface processes remains challenging. For a LEM to be representative for a specific area, field data should be used to calibrate and validate the simulated processes. Notwithstanding the continuously growing database on erosion rates at different spatial and temporal scales, the number of datasets and the area they cover is inversely correlated with the timescale considered. Although more data is thus available at shorter timescales, including them into LEMs is not straightforward as short-term observations are known to reflect the stochasticity of earth surface processes.
In this contribution, we focus on the role landslides, a stochastic hillslope processes in steep mountainous mostly not included in long term LEMs, but strongly reflected in short term field data. We first integrate the formation of landslides and the transport of the thereby generated sediments in a previously developed LEM (TTLEM). Landslide initiation is implemented as a stochastic process depending on a landslide failure index whereas landslide size depends on the slope stability calculated using the Cullman index. The updated model (TTLEM_Sed) is thereafter applied to the New Zealand-Alps where long term erosion measurements and landslide inventories allow to calibrate model parameters. Landslide inventories are traditionally analyzed using statistical relationships between slope stability and conditioning factors such as distance to rivers and distance to active faults. However, the use of conditioning factors in landslide hazard maps is based on empirical observations and lacks physical grounding. Indeed, hillslopes closer to rivers should automatically become more prone to landslides as rivers incise and undercut hillslope foots. The integration of landslides in a LEM now allows to simulate this dynamic interplay over different timescales. By varying the simulated timescale over which the model is run, we identify critical timescales at which a-priori imposed statistical relations between landscape characteristics and landslide occurrence are no longer required and represented by the internal dynamics of the evolving landscape.
With TTLEM_Sed, we present an open-source model tool that allows to simulate landslides and sediment propagation. A modelling approach to study landslides is different from classical landslide hazard mapping approaches as it allows to simulate landscapes over longer timescales therefore allowing to identify physical drivers of landslide formation and landslide initiation. Moreover, the explicit integration of landslides in TTLEM_Sed potentially allows for the integration of widely available short-term field data in future model applications.
A
Landscape evolution models use mass transport rules to simulate the development of topography over timescales too long for humans to observe. The ability of models to reproduce various attributes of real landscapes must be tested against natural systems in which driving forces, boundary conditions, and timescales of landscape evolution can be well constrained over millennia. We test and calibrate a landscape evolution model by comparing it with a well-constrained natural experiment using a formal inversion method to obtain best-fitting parameter values.
Our case study is the Dragon's Back Pressure Ridge, a region of elevated topography parallel to the south central San Andreas Fault that serves as a natural laboratory for studying how the timing and spatial distribution of uplift affects topography. We apply an optimization procedure to identify the parameter ranges and combinations that best account for the observed topography. Direct-search inversion models can be used to convert observations from such natural systems into inferences of the processes that governed their
formation through the use of repeat forward modeling. Simple inversion techniques have been used before in landscape evolution modeling, but these are imprecise and computationally expensive. We present the application of a more efficient inversion technique, the Neighborhood Algorithm (NA), to optimize the search for the model parameters values that are most consistent with the formation of the Dragon's Back Pressure Ridge through repeat forward modeling using CHILD.
Inversion techniques require the comparison of model results with direct observations to evaluate misfit. For our target landscape, this is done through a series of topographic metrics that include hypsometry, slope-area curves, and channel concavity. NA uses an initial Monte Carlo simulation for which misfits have been calculated to guide a second iteration of forward models. At each iteration, NA uses n-dimensional Voronoi cells to explore the parameter space and find the zones of best-fit, from which it selects new parameter values for the forward models. As it proceeds, the algorithm concentrates sampling around the cells with the best-fit models. The resulting distribution of forward models and misfits in multi-parameter space can then be analyzed to obtain probability density distributions for each parameter.
Preliminary results suggest that, when combined with robust algorithms for the calculation of the misfit, NA quickly centers the parameter search around values that capture the key features of the observed topography. The ability of NA to provide probability distributions for parameter values gives an indication of uncertainty in each, and can be used to guide field measurements for model testing. This application of advanced inversion techniques for landscape evolution modeling is a significant step towards the use of more formal mathematical methods in geomorphology that are already applied by other disciplines in the geosciences.
2
Landscape evolution studies typically simulate long-term soil development by varying production rates as a function of the local thickness of soil. Whether this function monotonically decreases with increasing soil depth remains an active source of debate. In modest relief landscapes, the presence of isolated bedrock outcrops called tors are often used to argue for the so-called ‘humped’ soil production function, which hypothesizes that there is an optimal, non-zero soil depth for soil production. Furthermore, in many steep landscapes, the fraction of bedrock exposed at the surface can be very high where cliffs bands form in response to high erosion rates. Yet, numerical models of hillslope evolution struggle to reproduce the continuous transition from fully soil-mantled to bedrock-dominated hillsides within a single framework. To address this, we present a new Agent-Based Model (ABM) of forest dynamics and soil production that is coupled to continuum-based models of hillslope sediment transport (i.e., linear and nonlinear creep). In this model, tors and bedrock cliffs are emergent features that manifest even when maximum possible soil production rates occur at zero soil depth. The intermittency of seed germination and sapling recruitment on bare bedrock surfaces combined with the rapid downslope transport of newly developed soil facilitates the persistence of exposed bedrock. By linking soil development to plant functional type, this work also shows how soil depths and bedrock exposure patterns that evolve over millennia may be mechanistically linked to forest properties that stabilize over shorter timescales. To illustrate model behavior, we use plant parameters inspired by Pinus ponderosa, the dominant species observed at low to intermediate montane forests in the Colorado Front Range. Using this as a baseline, we then test model sensitivity to a variety of tree functional parameters including seed fecundity, seed dispersal distance, maximum rooting depth, and maximum individual lifespans. Our work is the first to couple a NetLogo ABM to Python-based Landlab via the pyNetLogo library. As such, this use-case serves as a template for other landscape evolution studies targeting feedbacks among biological agents, substrate properties, and sediment transport.
Landscape morphology is the fingerprint of natural processes and human activities and is commonly represented by Digital Elevation Models (DEMs). The spatial structure of a landscape encodes essential information, making it fundamental for analyzing and understanding their responses to various influencing factors. In this study, we utilize the hierarchical relations of topographic depressions—features once dismissed as artifacts or errors in DEMs—to characterize landscape structure. We extract the hierarchical relationships using the Depression Hierarchy algorithm applied to the DEMs.
In this study, we extract depression hierarchies from the SRTM DEM at 30-meter resolution for the continental United States, initially representing them as binary trees. To enhance real-world applicability, we convert these binary trees into general trees. In the depression hierarchy general tree, the top-level depressions are the depressions traditionally defined by DEM filling algorithms, while the bottom-level depressions generally are the small and independent pits in the DEM. We extract the top three and bottom three depression levels and analyze the relationship between depression parameters, such as median depression volume, and various environmental variables, including land use, lithology, soil type, climate classification, glacier zones, aridity index, and mean annual precipitation at a HUC-8 watershed scale.
We use ridge regression to analyze the correlation between median depression volumes and environmental variables. The strongest correlation occurs at the bottom hierarchical level (leaf depression). The three variables most strongly correlated with depression volume are land use, climate classification, and aridity index. The R² values range from 0.30 to 0.45. When all variables, which include land use, lithology, soil type, climate classification, glacier zones, aridity index, and mean annual precipitation, are used in the ridge regression, the model has an R² of approximately 0.6. These results highlight the significance of depression structures in shaping the landscape and reveal the connections between depressions and both natural processes and human activities.
