Lawrence Berkeley National Lab
Modeling approaches for providing water and energy solutions to the world 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.
IHE Delft Institute for Water Education; Delft University of Technology; Universitas Brawijaya
Simulating mangrove-mudflat dynamics with a hybrid eco-hydro-morphodynamic model There is global recognition to push forward mangrove restoration and conservation for climate mitigation and adaptation. Unfortunately, although our understanding of mangrove processes has significantly improved, 80-90% of the reported restoration projects have experienced failures. The main reasons are related to a poor understanding of the eco-geomorphological dynamics and mangrove species-specific ecological requirements. Mangrove restoration guidelines exist; however, they may be site-specific and cannot be easily replicated in other restoration cases. Hence, it emphasizes the need for a system understanding of mangrove ecosystem physical and ecological interactions. We developed a hybrid model by coupling the process-based hydro-morphodynamic model Delft3D-FM (DFM) and the individual-based mangrove model MesoFON (MFON). The model (DFMFON) allows us to resolve spatiotemporal processes, including tidal, seasonal, and decadal environmental changes with full-life-cycle mangrove interactions. The DFMFON model successfully reproduced observed spatiotemporal (seasonal-decadal) mangrove development, like the age-height relationship and morphodynamic delta features in a prograding Porong Delta, Indonesia.
Getting meaning from models: lessons from salt marsh geomorphic modeling Salt marshes are biogeomorphic features that are under increasing pressure from sea level rise, land use change, and other external stressors. Modeling of salt marshes has traditionally been “stovepiped” into three general disciplines: ecology, geomorphology, and engineering, resulting in contrasting approaches and relative rigor. I will highlight successes and failures across these efforts, and identify how the three disciplines can move forward using advances from each other.
Adapting Urban Coastlines to Climate Change: using models to understand the potential and pitfalls of nature-based solutions Urban areas located along the coastline face critical choices in the coming decades to respond effectively to climate change, especially with regards to sea level rise (SLR) and intensified ocean storms. These choices include adaptation to let the water in, retreat to avoid new flooded areas, or resilient infrastructure to keep the water out. Nature-based solutions (NBS), which range from restoration of existing ecosystems to infrastructure inspired by natural ecosystems, have the potential to soften the consequences of choosing either hard infrastructure or adaptation. However, in urban environments the lack of available land space may reduce the efficacy of traditional NBS (e.g. living shorelines). Here, we present work to understand and alleviate the problem of NBS efficacy in an urban area with little space to give back to the natural environment. We use coastal hydrodynamic models of the Boston Harbor to show the potential for a range of NBS to protect against storms and SLR with the available area for these kinds of infrastructure projects. We further show how these models can be simplified and used as tools to understand trade-offs between NBS, hard infrastructure, and retreat, which may be as likely to come from an adaptation strategy as from SLR. Finally, we discuss our models of combinations of these solutions, and the current potential for NBS to protect an urban area from climate change.
Flatiron Institute at the Center for Computational Mathematics
Measuring the Change in Coastal Flooding Risk from a Multi-Hazard Perspective Coastal flooding and related hazards have increasingly become one of the most impactful events as climate change continues to change the risk due to these events. Measuring the change in the risk of a particular flood level has therefore taken on a greater urgency, as historic measurements and statistics are no longer sufficient to measure the risk to coastal communities. Enabling our ability to compute these changes has become the focus as adaptation strategies due to the changing climate become increasingly critical. This talk will outline some of these challenges and ways we are attempting to address the problem in a multi-hazard aware way.
University of California, Santa Cruz
Rivers and Glacial Isostatic Adjustment TBD
On the Practical Importance of Theory in Landscape Evolution Models Landscape evolution involves manifold processes from different disciplines, including geology, geomorphology and ecohydrology, often interacting nonlinearly at different space-time scales. While this gives rise to fascinating patterns of interconnected networks of ridges and valleys, it also challenges Landscape Evolution Models (LEMs), which typically rely on long-term numerical simulations and mostly have only current topographies for comparison. While adding process complexity (and presumably realism) is certainly useful to overcome some of these challenges, is also exacerbates issues related to proper calibration and simulation.</br>This talk advocates more focus on the theoretical analysis of LEMs to alleviate some of these issues. By focusing on the essential elements that distinguish landscape evolution, the resulting minimalist LEMs become more amenable to dimensional analysis and other methods of nonlinear field equations, used for example in fluid mechanics and turbulence, offering fertile ground to sharpen model formulation (i.e., the stream-power erosion term), unveil distinct dynamic regimes (e.g., unchannelized, from incipient valley formation, transitional and statistically self-similar fractal regime), and properly formulate questions related to the existence of steady state solution (as opposed to a situation of space time chaos, similar to a geomorphological turbulence). We also discuss benchmarks for evaluating numerical simulation and novel avenues for numerical methods, as well as ways to bridge between spatially discrete models (i.e., river networks) and continuous, partial-differential-equation models.
University of Wisconsin-Madison
Designing and applying a landscape evolution model infused with cosmogenic nuclides for geomorphic insights Terrestrial cosmogenic nuclides (TCN) are commonly used to assess denudation rates in soil-mantled uplands. The estimation of an inferred denudation rate (Dinf) from TCN concentrations typically relies on the assumptions of steady denudation rates during TCN accumulation and negligible impact from soil chemical erosion on soil mineral abundances. However, in many landscapes, denudation rates are not steady, and the composition of soil is markedly affected by chemical erosion, adding complexity to the analysis of TCN concentrations. We introduce a landscape evolution model that computes transient changes in topography, soil thickness, soil mineralogy, and soil TCN concentrations. With this model, we explored TCN responses in transient landscapes by imposing idealized perturbations in tectonically (bedrock uplift rate) and climatically sensitive parameters (soil production efficiency, hillslope transport efficiency, and mineral dissolution rate) on synthetic, steady-state landscapes. The experiments on synthetic landscapes delivered important insights about TCN responses in transient landscapes. Results showed that responses of Dinf to tectonic perturbations differ from those to climatic perturbations, indicating that spatial and temporal trends in Dinf serve as indicators of perturbation type and magnitude. Also, if soil chemical erosion is accounted for, basin-averaged Dinf inferred from TCN in stream sediment closely tracks actual basin-averaged denudation rate, showing that Dinf is a reliable representation of actual denudation rate, even in many transient landscapes. In addition, we demonstrate how this model can be applied to a real landscape in the Oregon Coast Range and how model predictions can be compared to field measurements of cosmogenic nuclides and chemical depletion in sediments. Overall, landscape evolution models infused with cosmogenic nuclides can be used to scrutinize methodological assumptions, reveal potential real-world patterns in transient landscapes, and deepen the comprehension of field data.
Colorado State University
Tidy and... tedious: empowering large-scale watershed science through inclusive, welcoming data harmonization Understanding and predicting large scale watershed-ecosystem dynamics requires datasets that empower research at both the local and continental scale. Yet, creating, maintaining and delivering diverse harmonized datasets to researchers and decision-makers is costly and a relatively rare endeavor. In our lab, we have been working on two different projects meant to make it easier for anyone to better understand and predict the hydrobiogeochemical behavior of watersheds, big and small. In Macrosheds, we have harmonized all of the small watershed-ecosystem datasets in the LTER, CZO, USFS, and other programs where there is, at a minimum, data on streamflow and concentration of at least one dissolved constituent (e.g. Nitrate). This dataset provides a critical complement to datasets from larger watersheds like CAMELS and CAMELS-Chem, enabling more focused interrogation of watershed behavior at the scale of small streams. Second, we are actively rebuilding and improving on AquaSat - a dataset built to empower broader use of remote sensing for water quality. This data is focused on large rivers and lakes, visible to LandSat satellites (typically wider than 60 meters). Through both of these projects, we have learned critical lessons about what data end-users actually need, how to make their lives easier, the limits of data portals, and the community required to maintain open source software.
Pennsylvania State University
Differentiable modeling to unify neural networks and process-based modeling for global geosciences under global change Process-based modeling offers interpretability and physical consistency in many domains of geosciences but struggles to leverage large datasets efficiently. Machine-learning methods, especially deep networks, have strong predictive skills yet are not easily interpretable and are unable to answer specific scientific questions. A recently proposed genre of physics-informed machine learning, called “differentiable” modeling (DM, https://t.co/qyuAzYPA6Y), trains neural networks (NNs) with process-based equations (priors) together in one stage (so-called “end-to-end”) to benefit from the best of both NNs and process-based paradigms. The NNs do not need target variables for training but can be indirectly supervised by observations matching the outputs of the combined model, and differentiability critically supports learning from big data. We propose that differentiable models are especially suitable as global- or continental-scale geoscientific models because they can harvest information from big earth observations to produce state-of-the-art predictions (https://mhpi.github.io/benchmarks/), enable physical interpretation naturally, extrapolate well (due to physical constraints) in space and time, enforce known physical laws and sensitivities, and leverage progress in modern AI computing architecture and infrastructure. Differentiable models can also synergize with existing process-based models in terms of providing to them parameters or identifying optimal processes, learning from the lessons of the community. Differentiable models can answer pressing societal questions on water resources availability, climate change impact assessment, water management, and disaster risk mitigation, among others. We demonstrate the power of differentiable modeling using computational examples in rainfall-runoff modeling, river routing, ecosystem and water quality modeling, and forcing fusion. We discuss how to address potential challenges such as implementing gradient tracking for implicit numerical schemes and addressing process tradeoffs. Furthermore, we show how differentiable modeling can enable us to ask fundamental questions in hydrologic sciences and get robust answers from big global data.
North Carolina State University
Parameterizing human dynamics in geomorphic models: learning from coastal barrier evolution models Exploratory models that simulate landscape change incorporate only the most essential processes that are hypothesized to control a behavior of interest. These “rule-based” models have been used successfully to examine behaviors in natural landscapes over large spatial (many kms) and temporal scales (decades to millennia). In many geomorphic systems, the dynamics of developed landscapes differ significantly from natural landscapes. For example, humans can alter the physical landscape through the introduction of hard infrastructure and removal of vegetation. Humans can also modify the internal and external forces that naturally change landscapes, including flows of water, wind, and sediment as well as climatic factors. As with natural processes, in exploratory models human behavior must be parameterized. However, the level of detail to which human behavior can be reduced while still accurately reproducing feedbacks across the coupled human-natural landscape is a complex, user-based decision. </br></br>In this clinic, we will work in small groups and through a Jupyter Notebook to parameterize a new human behavior within a modular coastal barrier evolution model (Barrier3D, within the CASCADE modeling framework). The clinic will incorporate discussions and prompts about how to broadly identify important model “ingredients” and reduce model complexity, and will therefore be generalizable to other geomorphic landscapes.
Lamont-Doherty Earth Observatory, Columbia University
Solving the sea level equation: Earth’s response to ice and ocean load changes In this clinic we will first review concepts of glacial isostatic adjustment and the algorithm that is used to solve the sea level equation. We will then provide an overview of the sea level code, which calculates the viscoelastic response of the solid Earth, Earth’s gravity field, and rotation axis to changes in surface load while conserving water between ice sheets and oceans. Participants will run the code, explore manipulating the input ice changes, and investigate its effect on the predicted changes in sea level, solid Earth deformation, and gravity field.
A Hands-On Workshop on GPU-Based Landscape Evolution Modeling This workshop introduces Swiftscape, a CPU/GPU-hybrid landscape evolution library with C++ and Python interfaces that can run hundreds of times faster than previous models. Participants will gain hands-on experience in both interfaces, offering flexibility and accessibility for diverse applications. Special focus will be given to the model's ability to run many simulations in parallel as well as its utility for solving inverse problems.
The Ohio State University
Introduction to agent-based modeling for socio-environmental systems Agent-based modeling (ABM) is a powerful simulation tool with applications across disciplines. ABM has also emerged as a useful tool for capturing complex processes and interactions within socio-environmental systems. This workshop will offer a brief introduction to ABM for socio-environmental systems modeling including an overview of opportunities and challenges. Participants will be introduced to NetLogo, a popular programming language and modeling environment for ABM. In groups, participants will have the hands-on opportunity to program different decision-making methods in an existing model and observe how outcomes change. We will conclude with an opportunity for participants to raise questions or challenges they are experiencing with their own ABMs and receive feedback from the group.
University of Illinois Chicago
Using Fill-Spill-Merge to understand and analyze landscape depressions Fill-Spill-Merge (FSM) is an algorithm that distributes runoff on a landscape to fill or partially fill depressions. When a depression fills, excess water can overflow into neighbouring depressions or the ocean. In this clinic, we will use FSM to assess changes in a landscape’s hydrology when depressions in a DEM are partially or fully filled with water. We will discuss why it may be important to consider depressions more closely than just with removal. I will describe the design of the FSM algorithm, and then we will use FSM on a DEM to look at how landscape hydrology changes under different hydrologic conditions. </br></br>This clinic may be helpful to those interested in topics such as landscape hydrology, landscape evolution, flow routing, hydrologic connectivity, and lake water storage.
NC State University, Center for Geospatial Analytics
Coastal evolution analysis and inundation modeling with GRASS GIS This clinic will introduce participants to GRASS GIS tools with a focus on applications for coastal hazards analysis including flooding and coastal evolution. We will explain and practice GRASS GIS data management and concepts, and demonstrate them on examples of efficient LiDAR point cloud, raster, and vector data processing.The clinic will begin with a brief introduction to the GRASS GIS software and continue with a hands-on tutorial exploring coastal evolution through a LiDAR timeseries of Bald Head Island in North Carolina, USA. Finally, we will explore some of the inundation and flood modeling tools available in GRASS GIS. The tutorial will be formatted in a series of Jupyter Notebooks executed in a cloud-based (or locally installed) JupyterLab environment, taking advantage of the latest GRASS GIS Python features for Jupyter, including 2D, 3D, webmap, and temporal visualizations. By the end of the clinic, participants will have hands-on experience with:<ul> <li>Setting up GRASS GIS Projects and importing data</li> <li>Creating high-quality DEMs from LiDAR point clouds and computing topographic parameters</li> <li>Deriving shorelines from the DEMs</li> <li>Animating changes in topography over time and computing erosion rates</li> <li>Generating simplified storm surge inundation timeseries</li></ul>
University of Washington
New features and basic usage of the GeoClaw software for depth-averaged flow GeoClaw (http://www.geoclaw.org) is an open-source software package for solving two-dimensional depth-averaged equations over general topography using high-resolution finite volume methods and adaptive mesh refinement. Wetting-and-drying algorithms allow modeling inundation and overland flows. The primary applications where GeoClaw has been used are tsunami modeling and storm surge, although it has also been applied to dam break and other overland flooding problems.</br></br>The first part of this clinic will present an overview of the capabilities of GeoClaw, including a number of new features have been added in the past few years. These include:</br> </br> - Depth-averaged Boussinesq-type dispersive equations that better model short-wavelength tsunamis, such as those generated by landslides or asteroid impacts. Solving these equations requires implicit solvers (due to the higher-order derivatives in the equations). This is now working with the adaptive mesh refinement (AMR) algorithms in GeoClaw, which are critical for problems that require high-resolution coastal modeling while also modeling trans-oceanic propagation, for example.</br> </br> - Better capabilities for extracting output at frequent times on a fixed spatial grid by interpolation from the AMR grids during a computation. The resulting output can then be use for making high-resolution animations or for post-processing (e.g. the velocity field at frequent times can be used for particle tracking, as needed when tracking tsunami debris, for example).</br> </br> - Ways to incorporate river flows or tidal currents into GeoClaw simulation.</br></br> - Better coupling with the D-Claw code for modeling debris flows, landslides, lahars, and landslide-generated tsunamis. (D-Claw is primarily developed by USGS researchers Dave George and Katy Barnhart).</br> </br>The second part of the clinic will be a hands-on introduction to installing GeoClaw and running some of the examples included in the distribution, with tips on how best to get started on a new project.</br></br>GeoClaw is distributed as part of Clawpack (http://www.clawpack.org), and is available via the CSDMS model repository. For those who wish to install the software in advance on laptops, please see http://www.clawpack.org/installing.html. We will also go through this briefly and help with any issues that arise on your laptop (provided it is a Mac or Linux machine; we do not support Windows.) You may need to install some prerequisites in advance, such as Xcode on a Mac (since we require "make" and other command line tools), a Fortran compiler such as gfortran, and basic scientific Python tools such as NumPy and Matplotlib. See https://www.clawpack.org/prereqs.html.
Arizona State University
The Last FAIR Clinic You'll Ever Need* Are you tired of hearing about the FAIR Principles? This clinic is for you then, because after you participate you’ll never need to attend another one!*</br></br></br>Good science depends on the careful and meticulous management and documentation of our research process. This includes our computational models, the datasets we use, the data transformation, analysis, and visualization scripts and workflows we build to evaluate and assess our models, and the assumptions and design decisions we make while writing our software.</br></br>Join us for a Carpentries-style interactive clinic with hands-on exercises where we will provide concrete guidance and examples for how to approach, conceptualize, and transform your computational models of earth systems into FAIR contributions to the scientific record whether they are greenfield projects or legacy code with a focus on existing, open infrastructure (GitHub / GitLab / Zenodo). We’ll also cover containerization (Docker, Apptainer) as a way to transparently document system and software dependencies for your models, and how it can be used to support execution on the Open Science Grid Consortium’s Open Science Pool fair-share access compute resources. Big parallel fun! https://osg-htc.org </br></br>∗ individual results may vary, this statement is provably false
Chang Liao & Matthew Cooper
Pacific Northwest National Laboratory
Mesh independent flow direction modeling using HexWatershed 3.0 HexWatershed is a hydrologic flow direction model that supports structured and unstructured meshes. It uses state-of-the-art topological relationship-based stream burning and depression-filling techniques to produce high-quality flow-routing datasets across scales. HexWatershed has substantially improved over the past two years, including support for the DGGRID discrete global grid system (DGGS). </br></br>This presentation will provide an overview of HexWatershed, highlighting its capabilities, new features, and improvements. Through hands-on tutorials and demonstrations, attendees will gain insights into the underlying philosophy of the HexWatershed model, and how to use HexWatershed products to run large-scale hydrologic models in watersheds worldwide. Specifically, this tutorial will cover major components in the HexWatershed ecosystem, including the computational mesh generation process, river network representation, and flow direction modeling.</br>We will provide participants with resources to extend the tutorial problems and gain additional familiarity with the tools and workflows introduced. </br></br>Attendees are encouraged to bring their laptops with internet access and a functional web browser. Tutorials will involve scripting operations in the Python language, such as Jupyter Notebook. We will use the Conda utility to install dependency libraries and Visual Studio Code to run the notebooks.
Coupling biological and surface processes in landscape evolution models It is now well established that the evolution of terrestrial species is highly impacted by long term topographic changes (e.g., high biodiversity in mountain ranges globally). Recent advances in landscape and biological models have opened the gate for deep investigation of the feedback between topographic changes and biological processes over millions of years timescale (e.g., dispersal, adaptation, speciation). In this clinic, we will use novel codes that couple biological processes with FastScape, a widely used landscape evolution model, and explore biological processes and speciation during and after mountain building under different magnitudes of tectonic rock uplift rates. We will explore and deduce how the magnitude and pace of mountain building impact biodiversity and how such interactions can be tracked in mountain ranges today. Python and Jupyter Notebook will be used in the clinic, and basic knowledge in python is desirable.
Introduction & Building with Google Earth Engine: Batteries Included Google Earth Engine(GEE) is a multi-petabyte catalog of satellite imagery and geospatial datasets with planetary-scale analysis capabilities. Now imagine all you need to work on it is a browser and an internet connection. This hands-on workshop will introduce you to and showcase cloud-native geospatial processing. </br></br>We will explore the platform’s built-in catalog of 100+ petabytes of geospatial datasets and build some analysis workflows. Additional topics will also include uploading & ingesting your own data to Google Earth Engine, time series analysis essential for change monitoring, and data and code principles for effective collaboration. The hope is to introduce to cloud native geospatial analysis platform and to rethink data as we produce and consume more.</br></br>If you want to follow along, bring your laptops, and register for an Earth Engine account here https://signup.earthengine.google.com</br></br>P.S I recommend using a personal account :) you get to keep it
Vegetation as ecogeomorphic features: incorporating vegetation into Earth Surface Models Vegetation is a critical ecogeomorphic agent within landscapes and is instrumental to many physical, biochemical, and ecological processes that can vary across spatial and temporal scales (e.g., erosion, sediment deposition, primary productivity, nutrient cycling, etc.). Modeling vegetation dynamics can be challenging, not only because of these scale-dependent variations, but also because of the breadth of existing approaches. The purpose of this clinic is to provide a technical overview for incorporating or developing vegetation models for earth surface dynamics modeling questions. The instructors will introduce vegetation processes commonly modeled, existing types of vegetation models, and how to choose an appropriate level of complexity for your system. Attendees will gain hands-on experience with existing vegetation components within and outside the Landlab system. These models will include the Cellular Automaton Tree Grass Simulator (CATGraSS), a mechanistic, photosynthesis-driven generalized vegetation model as well as how to incorporate vegetation models from Netlogo into Landlab. While active developers in the Landlab community will find this clinic useful, advanced programming experience is not needed.
Fora.ai: A participatory modeling platform to reshape how we collaborate for climate and social impact Participatory modeling (PM) is a collaborative approach to formalize shared representations of a problem and design and test solutions through a joint modeling process. PM is well-suited for addressing complex social and environmental problems like climate change, social and economic injustice, and sustainable resource management. This workshop will introduce and test a prototype version of Fora.ai, a new PM platform developed at Northeastern University. Fora.ai is a simple digital environment that enables groups to collaboratively understand real world problems and create novel solutions. Stakeholders interact through this digital representation with input from other stakeholders, then iteratively revise and test solutions until diverse needs are addressed. Fora.ai provides quick simulation results for data-driven proof of concepts that are ready to be presented, designed, and implemented in the real world, giving everyone in a team the power to share their unique perspective and build the world they want to live in together.
Interested in providing a clinic during the next annual meeting? Contact CSDMS@Colorado.EDU.