Property:CSDMS meeting abstract
From CSDMS
This is a property of type Text.
A
A number of two- and three-dimensional models are currently available to calculate sediment transport and channel change in rivers. These three-dimensional models rely on time-averaging and parameterization of the turbulence. Available depth-averaged, two-dimensional models also rely on simple boundary stress closures. In relatively simple channels these models have predictive capability, but they often perform poorly when there is large-scale flow separation or when secondary circulation is strong. Sharp meanders, channel constrictions, many engineering structures, vegetation, and certain types of bedforms all cause flow separation, secondary circulation, and free shear layers. Turbulence-resolving flow and sediment transport models may do better at predicting channel change in complex channels, but at a substantially larger computational cost. With parallelization, turbulence-resolving models can now be developed and applied to refractory fluvial morphodynamic problems.
Detached-Eddy Simulation (DES) is a hybrid large eddy simulation (LES) and Reynolds-averaged Navier Stokes (RANS) method. RANS is applied to the near-bed grid cells, where grid resolution is not sufficient to fully resolve wall turbulence. LES is applied further from the bed and banks. A one equation turbulence closure model with a wall-distance dependence, such as that of Spalart and Allmaras (SA), is ideally suited for the DES approach. The rough wall extension of the SA model is utilized herein. Our river DES numerical modeling system was developed in OpenFOAM. The model resolves large-scale turbulence using DES and simultaneously integrates the suspended sediment advection-diffusion equation, wherein advection is provided by the DES velocity field minus particle settling, and diffusion is provided by the sub-grid or RANS eddy viscosity. As such, turbulent suspension throughout most of the flow depth results from resolved turbulent motions. A two-dimensional, depth-averaged flow model, also written in OpenFOAM, determines the local water surface elevation. A separate program was written to automatically construct the block-hexagonal, computational grid between the calculated water surface and a triangulated surface of a digital elevation model of the given river reach. Domain decomposition of the grid is employed to break up the integration between multiple processors, and Open MPI provides communication between the processors. The model has shown very good scalability up to at least 128 processors.
Results of the modeling system will be shown of flow and suspended sediment model in lateral separation eddies in the Colorado River in Grand Canyon. The eddy recirculation zones exist downstream of channel constrictions from tributary debris fans. The modeling system is currently being developed and validated to be used in designing discharges from Glen Canyon Dam for the preservation of sandbar beaches, which are critical habitat for endangered fish.
Keywords: fluvial geomorphology, sediment transport, lateral separation zones.
Movie at:
https://csdms.colorado.edu/mediawiki/images/GrandCanyonDES.avi
A series of controlled laboratory experiments were conducted at the St. Anthony Falls laboratory of the University of Minnesota to study the effect of changing precipitation patterns on landscape evolution over long-time scales. High resolution digital elevation (DEM) both in space and time along with instantaneous sediment transport rates were measured over a range of rainfall and uplift rates. These experiments were designed to develop a complete drainage network by growth and propagation of erosional instabilities in response to tectonic uplift. We focus our study to the investigation of how changes in the frequency and magnitude of large-scale rainfall patterns (e.g. monsoonal variability) might influence the development of mountainous landscapes. Preliminary analysis suggests that the statistics of topographic signatures, for example, evolution of drainage network, slopes, curvatures, etc., show dependence on both rainfall patterns and uplift rate. The implications of these results for predictive modeling of landscapes and the resulting sediment transport are discussed. +
2
A series of controlled laboratory experiments were conducted to study the effect of changing precipitation patterns on landscape evolution at the short and long-time scales. High resolution digital elevation (DEM) both in space and time were measured for a range of rainfall patterns and uplift rates. Results from our study show distinct signatures of extreme climatic fluctuations on the statistical and geometrical structure of landscape features. These signatures are evident in widening and deepening of channels and valleys, change in drainage patterns within a basin and change in the probabilistic structure of erosional events, such as, landslides and debris flows. Our results suggest a change in scale-dependent behavior of erosion rates at the transient state resulting in a regime shift in the transport processes in channels from supply-limited to sediment-flux dependent. This regime shift causes variation in sediment supply, and thus in water to sediment flux ratio (Qs/Qw), in channels of different sub-drainage basins which is further manifested in the longitudinal river profiles as the abrupt changes in their gradients (knickpoints), advecting upstream on the river network as the time proceeds. +
A
A two-dimensional numerical model was developed for simulating free surface flow. The model is based on the solutions of two-dimensional depth averaged Navier-Stokes equations. A finite volume method is applied such that mass conservation is satisfied both locally and globally. The model adopted the two-step, high resolution MUSCL-Hancock scheme. This Godunov type scheme is used together with the approximate Riemann solver. The boundary cells are treated as cut-cells in order to accommodate arbitrarily geometries of natural rivers. There are sixteen types of cut-cells depending on the slope of the boundary intersection with the cell. A cell merging technique is incorporated in the model that combines small cells with neighboring cells to create a larger cell, helps keeping the CFL condition. The cut-cells approach permits a fully boundary-fitted mesh without implementing a complex mesh generation procedure for irregular geometries. The model is verified by several laboratory experiments including unsteady flow passing through cylindrical piers and dam break flow in a rectangular channel. The model is also applied to simulate a 100-year flood event occurred at the Huron Island reach of the Mississippi River, where flow paths in the island formed a complex channel network as flood propagates. +
2
Accurate tsunami forecasts often rely on direct measurements of the waves, which are only available at sparse locations, and only after the tsunami has passed the sensors. By contrast, we are investigating the use of a convolutional neural network (CNN) that can forecast tsunamis based only on Global Navigation Satellite System (GNSS) data, which is available within minutes at many existing stations in earthquake-prone regions. Training the model requires a large set of hypothetical earthquakes that are generated using a Karhunen-Loeve expansion, as implemented in the MudPy software, which provides synthetic GNSS data. It also provides seafloor motions that are used for GeoClaw simulations of the tsunami resulting from each event. We consider forecasting both multi-hour time series at select locations and inundation maps for target communities. Successes and limitations of this approach will be discussed. +
Acoustic sediment monitoring technology provides a practical means to obtain high resolution estimates of suspended sand flux. However, bed bedload flux can be a significant component of total load and remains difficult to measure directly. In most cases, bedload is treated as a power-law function of water discharge, a constant fraction of suspended sand flux, or ignored. However, bedload flux may vary independently from water discharge or suspended sediment flux in supply-limited rivers due to systematic grain size and reach-geometric effects. We propose a model for bedload flux that enables improved prediction using variables that are routinely measured at acoustic sediment monitoring stations.<br><br>Though this model is rooted in causal physical theory, it contains several scaling parameters that must be constrained empirically. To this end, we propose a Bayesian statistical procedure that facilitates propagation of uncertainty from multiple sources of information. Application of this procedure is demonstrated at one monitoring station on the Colorado River in Grand Canyon National Park. Repeat bathymetric surveys of dune migration in the reach adjacent to the monitoring station are used to estimate bedload flux, providing an observational basis for statistical analysis. Parameter estimation and prediction are also informed by data from other rivers, which is incorporated in a hierarchical framework.<br><br>We find that conventional methods of estimating bedload flux fail to capture fluctuations driven by the interaction between flow strength and sediment supply, and can introduce large persistent biases to estimates of total load. Our model is applicable in a wide range of scenarios, and substantially reduces the uncertainty associated with estimating bedload flux in sand bed rivers. +
Addressing society's water and energy challenges requires sustainable use of the Earth's critical zones and subsurface environment, as well as technological innovations in treatment and other engineered systems. Reactive transport models (RTMs) provide a powerful tool to inform engineering design and provide solutions for these critical challenges. In this keynote, I will showcase the flexibility and value of RTMs using real-world applications that focus on (1) assessing groundwater quality management with respect to nitrate under agricultural managed aquifer recharge, and (2) systematically investigating the physical, chemical and biological conditions that enhance CO2 drawdown rates in agricultural settings using enhanced weathering. The keynote will conclude with a discussion of the possibilities to advance the use of reactive transport models and future research opportunities therein. +
Advances in the measurement of environmental DNA (eDNA) have enhanced our ability to monitor both the occurrence and distribution of species in aquatic ecosystems. However, linking eDNA concentrations in streams and rivers with species abundance remains challenging due to the combination of unidirectional flow, streambed retention, and decay hindering identification of the source location. While numerous biotic and abiotic factors can cause reductions in eDNA concentration, their relative impact on overall concentration variations remains unclear. A mechanistic framework capturing the fate and transport of eDNA in the presence of both biological and physical heterogeneities still requires development. In this study, we use a continuous time random walk model to explore how eDNA removal is controlled by water column and hyporheic processes. We fit the model to concentration profiles produced from pulse conservative (Rhodamine WT) and continuous eDNA continuous tracer injections in the Big Piney river (Missouri), using multiple concentration profiles to capture evolving hydrodynamics across the reach. Regardless of location, modeled hyporheic removal rates are 2 to 4 orders of magnitude greater than laboratory based estimates of the water column reaction rate. This suggests that using only laboratory estimates of water column decay to predict eDNA concentrations with distance from the source may lead to significant prediction overestimates, as the hyporheic zone is likely the primary driver of eDNA removal in rivers. +
Aeolian sediment transport in river corridors remains an underexplored process despite its significant implications for landscape evolution, ecosystem management, and cultural resource preservation. Therefore, predictive tools for forecasting and hindcasting aeolian transport dynamics can improve the management of these resources. While progress has been made in numerical models capturing aeolian dynamics in coastal environments, their application to fluvial-aeolian interactions has been limited. In this study, we adapt the AeoLiS coastal aeolian model to a dryland river corridor, where wind-driven sediment transport is closely linked to sandbar dynamics. We apply AeoLiS to a fluvial sandbar along the Colorado River corridor, through Grand Canyon National Park, where sediment exchange between fluvial sandbars and adjacent aeolian dunes is influenced by regulated flow regimes, climatic forcing, and vegetation dynamics. Using 7.5 months of field data—including meteorological measurements, river gauge records, sediment traps, and topographic surveys—we parameterize and validate the model. We compare observed and modeled estimates of topographic change and aeolian sediment flux at a sandbar-dune complex to assess model performance. Our findings demonstrate that AeoLiS effectively simulates aeolian transport dynamics in this setting, supporting its broader applicability beyond coastal systems. This research advances understanding of coupled fluvial-aeolian geomorphic processes and highlights the potential for modeling to inform management strategies in dynamic sedimentary systems. +
After Glacial Lake Agassiz drained ~8.5 ka, the Red River (North Dakota, USA) formed, flowing northward into Lake Winnipeg and incising into paleolacustrine sediment as it meandered. The Red River provides a natural experiment to interrogate the role of slope change on river meandering and morphologic evolution as it is characterized by shallow bed slopes (~0.0001), which have been controlled by crustal deformation due to glacial isostatic adjustment (GIA) in response to ice sheet unloading since the river’s inception. GIA has changed Red River channel elevation by 10s of meters, reducing slopes by up to 60% in the downstream reaches. We isolate the role of slope in order to explore its importance to lateral migration rate relative to other factors such as bank strength, sediment supply, and fluid flow.
We quantified the impact of GIA-induced slope changes on the Red River’s morphology by performing an analysis of river meanders and cutoffs (Kodama et al., 2023). We constructed a dataset that quantified the number of meanders, cutoffs, and modeled change in slopes caused by GIA along the Red River. Notably, the abundance of cutoffs normalized for channel width (a proxy for time-averaged meander rate) statistically significantly correlates with changes in slope, with far fewer cutoffs in the downstream reaches of the river, where the largest slope reduction occurred. We expanded this analysis to two tributaries of the Red River, and found that this relationship holds in all three river systems regardless of sign (negative or positive) of the GIA-induced slope change. We infer that slope drives changes in lateral migration rate for these detachment-limited systems by modulating the magnitude of shear stress on riverbanks.
We next developed a modeling framework by modifying a simple kinematic model of meander migration (Howard & Knutson, 1984) to explore the impact of GIA-induced slope change on the temporal trends of meander migration rate along the Red River. Previous work showed that the meander rate of two Red River meander scrolls exponentially decayed over the Holocene (Brooks 2003), which we are able to simulate with our GIA forced meandering model. Our study isolates the role of slope on river lateral migration and highlights how rivers near former ice sheets can respond to changes in slope that occur over thousands of years.
After Superstorm Sandy impacted the New Jersey coastline in 2012, the state’s primary coastal resiliency plan was to fortify the entire shoreline by constructing large-scale berm-dune systems along the beach. These large artificial dunes, funded entirely by Congress, were constructed with the goal of mitigating future storm damage to houses and infrastructure. Two long-term management questions are 1) is it feasible for a beachfront community to maintain these projects over the long term?; and 2) if not, what fraction of the cost would need to be subsidized? To tackle these questions, we use a “geo-economic” model that captures the natural processes of beach and dune erosion and migration via storm overwash coupled with engineering interventions of beach nourishment and dune construction. The economic portion of the model accounts for the relationship between property values and berm-dune geometry. Previous work suggests that due to their protective and recreational value, higher dunes and wider beaches increase that property values. However, it is unclear whether this relationship holds true for dune protection some years after a storm has occurred as lags in major storm events may lead to perceptions of lower risks. Thus, beachfront communities may place greater value upon viewership and private property, rather than on protection by artificial dunes. By deriving mathematical expressions for optimal berm and dune size as a function of geologic and economic parameters, our model suggests that changes in risk perception can lower property values and therefore reduce the ability of a community to keep up with the costs of maintaining these structures. We are currently testing this hypothesis by analyzing past and present LiDAR imagery (i.e. 2010, 2014, and 2018) and real-estate data from Long Beach Island, NJ. +
Agricultural expansion has led to high rates of deforestation and land-use change in tropical ecosystems, relegating many of the remaining native forests to networks of fragmented patches. As a result, large forest-dwelling ungulates may alter movement and habitat-use patterns to accommodate for the changed spatial orientation of essential resources. In turn, some native patches may be subjected to increased ungulate impacts (e.g. trampling, bioturbation, and seed dispersal/ predation), while others may be devoid of these influences. We created an individual-based model utilizing empirical ungulate movement data from white-lipped peccaries (WLP) in the Brazilian Cerrado to evaluate variations in habitat use with degree of fragmentation (e.g. connectivity and number of patches) and percent of native forest cover (FC). In the model, a peccary herd moves across a landscape with a percent FC between 10% and 100% and one to four native forest patches. We then quantified the distribution of habitat-use intensity and percent of unused native habitat after five years. To empirically quantify impacts of white-lipped peccary habitat use, we measured seedling density in 72 1x1 plots in the Cerrado, 44 with and 28 without WLP.
Results indicate that in a fully-connected landscape (one-patch simulations), as percent FC decreases, the frequency distribution of habitat use goes from narrow and left-skewed (low use in the majority of the habitat) to widely and evenly distributed (no use to high use in distinct parts of the habitat), reflecting a more heterogeneous use of the habitat with less FC. In a fragmented landscape (two-four patch simulations) below 30% FC, habitat use is driven by the degree of connectivity between forest patches. However, above 60% FC, the percent of unused forest is negligible (similar to one-patch simulations), indicating that patch spatial configuration is no longer the driving factor of habitat use past a 60% FC threshold. Between 40% and 60% FC, habitat use is a function of both connectivity and percent FC. Preliminary empirical results suggest riparian forests have the greatest difference in mean seedling density between areas with or without WLP, while palm swamps have the least. Collectively, these results suggest conservation measures in agricultural landscapes should emphasize percent FC, connectivity, or both, depending on the amount of forest remaining and that riparian zones may be most adversely affected by the loss of large ungulates.
Alluvial megafans are sensitive recorders of landscape evolution: the influence of both autogenic processes and allogenic forcing and of the coupled dynamics of the fan with its mountainous catchment can often be deciphered from the megafan sediment record and the system’s morphometric characteristics. The Lannemezan megafan in the northern Pyrenean foreland was abandoned by its mountainous feeder stream during the Quaternary and subsequently incised. During the incision, a flight of alluvial terraces was left along the stream network. We use numerical models (CIDRE model, Carretier et al. 2015) to explore the relative roles of autogenic processes and external forcing in the building, abandonment and incision of a foreland megafan. We then compare the results with the inferred evolution of the Lannemezan megafan. We conclude that autogenic processes are sufficient to explain the building of a megafan and the long-term entrenchment of its feeding river at the time and space scales that match the Lannemezan setting. In the case of the Lannemezan megafan, climate, through temporal variations in precipitation rate, may have played a second-order role in the pattern of incision at a shorter time-scale. In contrast, base-level changes, tectonic activity in the mountain range or tilting of the foreland through flexural isostatic rebound do not appear to have played a role in the abandonment of the Lannemezan megafan. +
Alluvial rivers record the external drivers of change, such as climate, tectonics and anthropogenic disturbances, and they code their dynamics in their bankfull channel geometry and planform geometries and longitudinal profiles. The key to understanding the past and predicting the future alluvial rivers is learning to interpret the language of the grains of sediment and to decode the responses they record. Sand-bed river long-profile evolution model (SRLP) is a complete mechanistic model, describing both transient and steady-state long-profile evolution of a transport-limited sand-bed river by linking sediment transport and river morphodynamics, including planform (width) adjustment as a function of excess shear stress (following Parker, 1978, and Dunne et al., 2018), thereby linearizing the sediment-transport response to changing river discharge. Through quantifying the changes in the river's bed elevation, cross section, and channel slope, this one-dimensional, physics-based model captures the internal dynamics of this inherently complicated system and ultimately provides the critical information about how the river is responding to both natural and anthropogenic disturbances. We now further aim to understand and model how sediment is transported and where it is deposited within sand-bed alluvial river networks, how different portions of the alluvial river network respond and how the behavior of those network components are impacted over both human and geological time scales. In order to do this, we use an updated network model approach (by Wickert A., GRLP v2.0.0-alpha). This latest release works as a core network engine, allows integration of different process modules and ultimately provides a new network-model platform where we can include SRLP model. Through the addition of SRLP module, we present examples of long-profiles of sand-bed alluvial river networks under a variety of base level and water- and sediment-supply boundary conditions and investigate the mainstem river and both upstream and downstream tributary responses over time. Finally, we compare the response of the model of linked sand-bed tributaries to the one of gravel-bed rivers to further discuss the effects of variations in grain- and reach-scale dynamics on the longitudinal evolution of these two classes.
Along Andean-type convergent margins, the preserved stratigraphic successions in retroarc foreland basins record complex interactions between oceanic plate subduction, overriding lithosphere deformation, and surface processes. Modeling their interactions and their impacts on basin stratigraphy helps to distinguish the geological footprint of the operating processes. We use a source-to-sink landscape evolution model, Badlands, to investigate the basin stratigraphic formation in response to changes in subduction morphology, hinterland orogenic uplift, overriding lithosphere strength, and surface erosional efficiency. Our modeling results reveal distinguishable responses of basin sedimentation to the imposed tectonic and surface forcings. Firstly, with sufficient sediment supply (i.e., the basin is filled with sediments), subduction at higher slab dip leads to development of shallower and narrower basins, with increasing volume of fluvial and shallow-water deposits accumulation. For mechanically thicker overriding plates, a deeper foreland basin tends to develop, though the basin width does not show consistent changes with increasing lithosphere strength. When sediment supply is further enhanced by either increasing orogenic uplift rate or surface erodibility, the basin sedimentation extends horizontally while the basin depth changes in an opposite way. Secondly, our basin subsidence analysis reveals strong impact of flexural rebound at the foredeep on modifying the basin morphology and strata dipping. We further found positive correlations between the flexural rebound and the progradation of fluvial deposits at the foredeep. Lastly, by normalizing the basin width to orogenic belt width and basin depth to maximum foreland flexure, we categorize the basins to be accommodation-dominant and supply-dominant, which helps to evaluate the impact of varying each contributing process on the basin development. Overall, our source-to-sink models reveal the complex interactions between surface and tectonic forcings, and highlight the huge potential of extracting their signals from the geological record.
Along a quarter of the Beaufort Sea coast, back-barrier estuaries modulate the transport and transformation of nitrogen and carbon, impacting food webs and carbon budgets. These estuaries are adjacent to permafrost, a large carbon reservoir that contains ~1700 Gt of organic carbon that is thawing from rapid Arctic warming. Thawed dissolved organic matter and nutrients may be transported to the coastal ocean by groundwater and rivers, adding nutrients to the coast that may impact production and biogeochemical cycles. It is unclear what effect permafrost thaw will have on Arctic estuarine biogeochemistry, partly because present-day spatial and temporal variability of residence time and export in Arctic back-barrier estuaries is unknown and complicates efforts to predict future change. To investigate the residence time of water, as well as estuary-shelf fluxes, this study uses a numerical modeling approach. Specifically, a hydrodynamic model, the Regional Ocean Modeling System (ROMS), is being implemented for Arey, Kaktovik, and Jago Lagoons along the Beaufort Sea coast of northern Alaska. The model accounts for processes including local winds, rivers, and larger scale circulation. Analysis will focus on variations in circulation dynamics within the ice break-up and open water season of 2019. +
Along wave-influenced deltas, wave-driven longshore currents usually interact with the fluvial jet at the river mouth, creating a sharp gradient alongshore in sediment transport/deposition and hence a corresponding change in coastline morphology. When multiple channels intersect a delta coastline, morphological changes can take on a complex outlook driven by the multiplicity of the river channels and their ‘hydraulic groyne effect’, whereby the overall effect of the river jets is to limit the loss of sediment within the coastal littoral system and ensure shoreline stability. This study explores the dynamic relationships between waves and fluvial discharge along a coastline intersected by river mouths by employing a numerical model of an idealized delta coastline containing two river mouths. The modelling is undertaken using Delft3D, in order to simulate both sediment transport and wave propagation along with the accompanying changes in coastline morphology. Analysis focuses on the relative change in coastline morphology, updrift, and downdrift of the river channels, in relation to varying scenarios of the incident wave climate, fluvial input, and river channel geometry. Specifically, water discharge entering the basin is set temporally constant during a model run but is varied in the range of 500 - 2000 m3/s between runs. Further, 3 scenarios of fluvial sediment discharge corresponding to low, medium, and high sediment discharges, are incorporated into each fluvial discharge scenario. Finally, waves approach the coastline from an incident angle of <45o, generating longshore sediment transport proportional to its significant height and approach angle, which are varied between model runs in the ranges of 1.0 - 1.5 m and 15 - 42 degrees, respectively. The study is set to provide new insights into the morphodynamics of wave-influenced deltas resulting from the interaction of waves with fluvial discharges at interannual timescales. +
Although groundwater and lakes are vital freshwater reservoirs, little is known about how their storage volumes change in the long term. Here, we present a transient simulation of the global water table over the past 21,000 years, including both groundwater and lake surfaces. We obtain this using the Water Table Model (WTM), which solves the 2D horizontal groundwater equation and uses Fill-Spill-Merge to route surface water into lake depressions. The model is forced using Trace-21K climate-simulation outputs, the ICE-6G ice-sheet reconstruction, and time-variable topography as a result of Glacial Isostatic Adjustment (GIA). Our results highlight the impact of well-known climatic intervals – such as the Younger Dryas and the Bølling–Allerød – on the global water table, and indicate that changes in groundwater and lake storage can modify global sea level by several decimeters at a millennial time scale. +
Although mangroves provide several beneficial ecosystem services, such as blue carbon storage, coastal protection, and nursery habitats, they rapidly decline due to human development and climate change. In particular, in areas in the Caribbean, such as Puerto Rico, climate change
will likely lead to an increase in evaporation over precipitation. Such an increase in drought-like conditions will drive porewater salinity to increase exceeding the threshold beyond which mangroves can survive. To improve our understanding of this interplay, we developed a numerical model using the Landlab Python library that describes the spatial distribution of vegetation and die-back in low-lying and undeveloped mangrove islands where freshwater inputs come solely from precipitation. We apply the model to a series of islands with elongated and asymmetric die-backs in La Parguera, a bay environment in southern Puerto Rico. Our model can explain the die-back shape and location for all islands as a function of the average net evaporation rate (i.e. evaporation – precipitation), the island's edge water salinity, and the mangrove soil dispersion coefficient, or the porewater exchange through tidal flushing. We gathered evaporation data from the Woods Hole Oceanographic Institute's OAFLUX project and precipitation data from the Tropical Rain Monitoring Mission, and quantified the soil dispersion as a function of the area of red mangroves, which was calculated via satellite imagery analysis. Additionally, we infered the outer edge salinity from the maximum canopy heights, gathered from Goddard's LiDAR, Hyperspectral, and Thermal Imager. In our model results, some islands presented a subtle bayward shift of the die-back. This can be explained by a higher island's edge water salinity on the landwards side, where bay depths are shallow and mixing with the rest of the bay is low. This spatial difference in salinity was consistent with the differences in canopy heights derived from LiDAR, and fell within the range of values reported in the literature. This portable modeling framework can be applied to other low lying mangrove carbonate islands with complex geometries.
Although numerous approaches for deriving water depth from bands of remotely-sensed imagery in the visible spectrum exist, digital terrain models for remote tropical carbonate landscapes remain few in number. The paucity is due, in part, to the lack of in situ measurements of pertinent information needed to tune water depth derivation algorithms. In many cases, the collection of the needed ground-truth data is often prohibitively expensive or logistically infeasible. We present an approach for deriving water depth from multi-spectral satellite imagery without the need for direct measurement of water depth, bottom reflectance, or water column properties within the site of interest. The reliability of the approach is demonstrated for five satellite images, each at a different study site, with overall RMSE values ranging from 0.84 m to 1.56 m when using chlorophyll concentrations equal to 0.05 $\text{mg m}^{-3}$ and a generic seafloor spectrum generated from a spectral library of common benthic constituents. Sensitivity analyses show that the model is robust to selection of bottom reflectance inputs and errors in the atmospheric correction and sensitive to parameterization of chlorophyll concentration. +
