Property:CSDMS meeting abstract presentation
From CSDMS
This is a property of type Text.
P
Decision framing is a key, early step in any effective decision support engagement in which modelers aim to inform decision and policy making. In this clinic participants will work through and share the results of decision framing exercises for a variety of policy decisions. We will organize the exercise using the XLRM elicitation, commonly used in decision making under deep uncertainty (DMDU) stakeholder engagements. The XLRM framework is useful because it helps organize relevant factors into the components of a decision-centric analysis. The letters X, L, R, and M refer to four categories of factors important to RDM analysis: outcome measures (M) that reflect decision makers’ goals; policy levers (L) that decision makers use to pursue their goals; uncertainties (X) that may affect the connection between policy choices and outcomes; and relationships (R), often instantiated in mathematical simulation models, between uncertainties and levers and outcomes. +
Deep-learning emulators permit to reduce dramatically the computational times for solving physical models. Trained from a state-of-the-art high-order ice flow model, the Instructed Glacier Model (IGM, https://github.com/jouvetg/igm) is an easy-to-use python code based on the Tensorflow library that can simulate the 3D evolution of glaciers several orders of magnitude faster than the instructor model with minor loss of accuracy. Switching to Graphics Processing Unit (GPU) permits additional significant speed-ups, especially when modeling large-scale glacier networks and/or high spatial resolutions. Taking advantage of GPUs, IGM can also track a massive amount of particles moving within the ice flow, opening new perspectives for modeling debris transportation of any size (e.g., erratic boulders). Here I give an overview of IGM, illustrate its potential to simulate paleo and future glacier evolution in the Alps together with particle tracking applications, and do a quick live demo of the model. +
Delta morphology +
Deltas are highly sensitive to local human activities, land subsidence, regional water management, global sea-level rise, and climate extremes. In this talk, I’ll discuss a recently developed risk framework for estimating the sensitivity of deltas to relative sea level rise, and the expected impact on flood risk. We apply this framework to an integrated set of global environmental, geophysical, and social indicators over 48 major deltas to quantify how delta flood risk due to extreme events is changing over time. Although geophysical and relative sea-level rise derived risks are distributed across all levels of economic development, wealthy countries effectively limit their present-day threat by gross domestic product–enabled infrastructure and coastal defense investments. However, when investments do not address the long-term drivers of land subsidence and relative sea-level rise, overall risk can be very sensitive to changes in protective capability. For instance, we show how in an energy-constrained future scenario, such protections will probably prove to be unsustainable, raising relative risks by four to eight times in the Mississippi and Rhine deltas and by one-and-a-half to four times in the Chao Phraya and Yangtze deltas. This suggests that the current emphasis on short-term solutions on the world’s deltas will greatly constrain options for designing sustainable solutions in the long term. +
Developed barriers are tightly-coupled systems driven by feedbacks between natural processes and human decisions to maintain development. Coastal property markets are dynamically linked to the physical environment: large tax revenues and high-value infrastructure necessitate defensive coastal management through beach nourishment, dune development, overwash removal, and construction of hard structures. In turn, changes to environmental characteristics such as proximity to the beach, beach width, and the height of dunes influence coastal property values. In this talk I will use a new exploratory model framework – the CoAStal Community-lAnDscape Evolution (CASCADE) model – to explore the coupled evolution of coastal real estate markets and barrier landscapes. The framework couples two geomorphic models of barrier evolution (Barrier3D and BRIE) with an agent-based real estate model – the Coastal Home Ownership Model (CHOM). CHOM receives information about the coastal environment and acts on that information to cause change to the environment, including decisions about beach nourishment and dune construction and maintenance. Through this coupled model framework, I will show how the effects of dune and beach management strategies employed in the wake of extreme storms cascade through decades to alter the evolution of barriers, inadvertently inhibiting their resilience to sea level rise and storms, and ultimately unraveling coastal real estate markets. +
Developers of solvers for PDE-based models and other computationally intensive tasks are confronted with myriad complexity, from science requirements to algorithms and data structures to GPU programming models. We will share a fresh approach that has delivered order of magnitude speedups in computational mechanics workloads, minimizing incidental complexity while offering transparency and extensibility. In doing so, we'll examine the PETSc and libCEED libraries, validate performance models, and discuss sustainable architecture for community development. We'll also check out Enzyme, an LLVM-based automatic differentiation tool that can be used with legacy code and multi-language projects to provide adjoint (gradient) capabilities. +
Digital twins are increasingly important in many domains, including for understanding and managing the natural environment. Digital twins of the natural environment are fueled by the unprecedented amounts of environmental data now available from a variety of sources from remote sensing to potentially dense deployment of earth-based sensors. Because of this, data science techniques inevitably have a crucial role to play in making sense of this complex, highly heterogeneous data. This webinar will reflect on the role of data science in digital twins of the natural environment, with particular attention on how resultant data models can work alongside the rich legacy of process models that exist in this domain. We will seek to unpick the complex two-way relationship between data and process understanding. By focusing on the interactions, we will end up with a template for digital twins that incorporates a rich, highly dynamic learning process with the potential to handle the complexities and emergent behaviors of this important area. +
Do you have code that you'd like to share with others—maybe you've written a model for your thesis, or perhaps you're required to do so by a journal—but you're not sure of the best way to go about it? We'll try to address this problem in this clinic. We'll use Python because it's the standard language of CSDMS; however, much of what we show can be translated to other languages.
This clinic has two parts. First, we'll show how to properly package Python code so that it can easily be used by others. Second, we'll configure a GitHub repository with files and services that will help make the code FAIR and sustainable over time—a "community-ready" repository.
The following topics address what could be included in such a repository. While we probably won't be able to cover all of these in the clinic, we list them for reference.
* Writing an informative README (and adding status badges!)
* Choosing a software license
* Packaging, using guidance from the Python Packaging Authority (PyPA)
* Automating repository tasks with nox
* Linting with black, flake8, and pre-commit
* Unit testing with pytest
* Continuous integration with GitHub Actions
* Building documentation with sphinx
* Adding a Digital Object Identifier (DOI) with Zenodo
* Creating a citation file with cffinit
* Including instructions for contributors, and a code of conduct
* Crediting contributors
* Acknowledging funding support
References will be provided for each topic for further exploration.
Participants will leave with a clear, practical template for sharing scientific software in a way that supports reuse, citation, and long-term community engagement. +
Does permafrost impart topographic signatures, and how does subsequent warming affect hillslope and channel form? Permafrost controls the depth to immobile soil, and tundra vegetation influences infiltration and erosion thresholds. I will use high-resolution maps of arctic landscapes to examine morphometric properties like hillslope length, curvature and drainage density as functions of climate and vegetation. I will then compare these data to existing models of climate-modulated sediment flux and channel incision in Landlab, exploring the effect of more nuanced representations of permafrost flux laws and hydrology. I will also compare modeled landscapes forced with Pleistocene-Holocene climate to mid-latitude landscape form. +
Drainage divides, the topographic boundaries of river basins, are dynamic landscape features. Asymmetric erosion rates across drainage divides can cause gradual divide migration and occasionally, the flow of water may be redirected towards a neighboring basin in via ‘river capture’. Geomorphologists often use topographic evidence to infer erosional disequilibrium across drainage divides. Similarly, freshwater biologists infer a history of river capture using biological lines of evidence including the presence of disjunct populations of freshwater organisms found on the ‘wrong’ side of the drainage divide and/or phylogenetic relationships that reflect paleo-river networks. Yet, these lines of evidence may be challenging to interpret when there is a history of multiple river capture events in the same basin that each cause transient erosional responses and repeatedly transfer freshwater organisms across the drainage divide. Here, I explore the topographic and biological signatures of repeated river captures using coupled population-genetic simulations and landscape evolution models. I also present an empirical case-study across the Eastern Continental Divide, USA. Population genetic analysis of the Saffron Shiner (Hydrophlox rubricoceus) and topographic evidence (knickpoints and windgaps) suggest that the Linville River (Atlantic-draining) has repeatedly captured area from the upper tributaries of the Tennessee River (Gulf-draining). The results highlight the challenges and promise of integrating biological and geologic datasets to investigate the history of river network reorganization. +
During a clinic session in the 2013 CSDMS annual meeting, the OpenFOAM®, an open source computational fluid dynamics (CFD) platform, was first introduced by Dr. Xiaofeng Liu (now at Penn State University) for modeling general earth surface dynamics. OpenFOAM® provides various libraries, solvers and toolboxes for solving various fluid physics via finite volume method. The objective of this clinic is to further discuss its recent development and applications to coastal sediment transport. The clinic will start with an overview of a range of coastal applications using OpenFOAM®. We will then focus on a recently released solver, SedFOAM, for modeling sand transport by using an Eulerian two-phase flow methodology. Specifically, we will focus on applying the model to study wave-driven sheet flows and the occurrence of momentary bed failure. The code can be downloaded via CSDMS code repository and participants will receive a hands-on training of the coding style, available numerical schemes in OpenFOAM®, computational domain setup, input/output and model result analysis. Knowledge of C++, object-oriented programming, and parallel computing is not required but will be helpful. +
During the clinic we'll introduce the new Delft3D Flexible Mesh modeling environment. We'll discuss the basic features and set up a simple 2D morphological model. The ongoing developments and the possibility to use BMI for runtime interaction will be presented as well.
The user interface runs on Windows, so make sure that you have a Windows computer or virtual machine available during the meeting. The user interface will be provided precompiled; the computational kernels you'll have to compile yourself. We'll provide instructions on how to compile the FORTRAN/C kernels before the clinic. +
Earth scientists face serious challenges when working with large datasets. Pangeo is a rapidly growing community initiative and open source software ecosystem for scalable geoscience using Python. Three of Pangeo’s core packages are 1) Jupyter, a web-based tool for interactive computing, 2) Xarray, a data-model and toolkit for working with N-dimensional labeled arrays, and 3) Dask, a flexible parallel computing library. When combined with distributed computing, these tools can help geoscientists perform interactive analysis on datasets up to petabytes in size. In this interactive tutorial we will demonstrate how to employ this platform using real science examples from hydrology, remote sensing, and oceanography. Participants will follow along using Jupyter notebooks to interact with Xarray and Dask running in Google Cloud Platform. +
Earth surface processes are modulated by fascinating interactions between climate, tectonics, and biota. These interactions are manifested over diverse temporal and spatial scales ranging from seconds to millions of years, and microns to thousands of kilometers, respectively. Investigations into Earth surface shaping by biota have gained growing attention over the last decades and are a research frontier. In this lecture, I present an integration of new observational and numerical modeling research on the influence of vegetation type and cover on the erosion of mountains. I do this through an investigation of millennial timescale catchment denudation rates measured along the extreme climate and ecologic gradient of the western margin of South America. +
Earthquakes are the most frequent source of classic tsunami waves. Other processes that generate tsunami waves include, landslides, volcanic eruption and meteorite impacts. Furthermore, atmospheric disturbances can also generate tsunami waves or at least tsunami-like waves, but we are just at the beginning of understanding their physics and frequency. Classic tsunami waves long waves with wavelength that are much longer than the water depth. For earthquake-generated tsunami waves that is true. However, landslides and meteorite impacts generate tsunami waves that are shorter which has a profound effect on the tsunami evolution, but no less dangerous.<br>Fortunately, tsunamis do not occur frequently enough in any given region to make meaningful prediction of the future tsunami hazard based only on recorded history. The geologic record has to be interrogated. The inversion of meaningful and quantitative data from the geologic record is the main goal of my research. However, there are problems with the geologic record. The most important problem is that we often have trouble to identify tsunami deposits. Second, it is very often difficult to separate the tsunami record from the storm record in regions where storms and tsunamis are competing agents of coastal change. Other problems are concerned with he completeness of the deposits, but also the fact that sedimentary environment before the tsunami hit most likely was eroded is no longer part of the record makes inversion especially tricky. In my research, I assume that the tsunami deposit is identified, but perhaps not complete and what we know about the pre-event conditions is limited.<br>My talk will cover how the geologic record is used to invert quantitative information about the causative process. We are going to look at grain sizes from sand to boulders and what we can learn from the transport of these very different grain sizes about tsunamis and their impacts along respective coastal areas. The models that are employed to invert flow characteristics from deposits are based on Monte-Carlo simulations to overcome the issue of not knowing the pre-tsunami conditions with great confidence. If time permits, we also see how sea-level change affects tsunami impact at the coast.
Earth’s surface is the living skin of our planet – it connects physical, chemical, & biological systems. Over geological time, this surface evolves with rivers fragmenting the landscape into environmentally diverse range of habitats. These rivers not only carve canyons & form valleys, but also serve as the main conveyors of sediment & nutrients from mountains to continental plains & oceans. Here we hypothesise that it is not just geodynamics or climate, but their interaction, which, by regulating topography and sedimentary flows, determines long-term evolution of biodiversity. As such, we propose that surface processes are a prime limiting factor of diversification of Life on Earth before any form of intrinsic biotic process.
To test this hypothesis, we use reconstructions of ancient climates & plate tectonics to simulate the evolution of landscape & sedimentary history over the entire Phanerozoic era, a period of 540 million years. We then compare these results with reconstructions of marine & continental biodiversity over geological times. Our findings suggest that biodiversity is strongly influenced by landscape dynamics, which at any given moment determine the carrying capacity of continental & oceanic domains, i.e., the maximum number of different species they can support at any given time.
In the oceans, diversity closely correlates with the sedimentary flow from the continents, providing the necessary nutrients for primary production. Episodes of mass extinctions in the oceans have occurred shortly after a significant decrease in sedimentary flow, suggesting that a nutrient deficit destabilizes biodiversity & makes it particularly vulnerable to catastrophic events.
On the continents, it took the gradual coverage of the surface with sedimentary basins for plants to develop & diversify, thanks to the development of more elaborate root systems. This slow expansion of terrestrial flora was further stimulated during tectonic episodes.
Ecological Network Analysis (ENA) enables quantitative study of ecosystem models by formulating system-wide organizational properties, such as how much nutrient cycling occurs within the system, or how essential a particular component is to the entire ecosystem function. EcoNet is a free online software for modeling, simulation and analysis of ecosystem network models, and compartmental flow-storage type models in general. It combines dynamic simulation with Ecological Network Analysis. EcoNet does not require an installation, and runs on any platform equipped with a standard browser. While it is designed to be easy to use, it does contain interesting features such as discrete and continuous stochastic solutions methods. +
Ecology is largely considered to have its foundations in physics, and indeed physics frames many of the constraints on ecosystem dynamics. Physics has its limitations, however, especially when dealing with strongly heterogeneous systems and with the absence of entities. Networks are convenient tools for dealing with heterogeneity and have a long history in ecology, however most research in networks is dedicated to uncovering the mechanisms that give rise to network types. Causality in complex heterogeneous systems deals more with configurations of processes than it does with objects moving according to laws. Phenomenological observation of ecosystems networks reveals regularities that the laws of physics are unequipped to determine. The ecosystem is not a machine, but rather a transaction between contingent organization and entropic disorder. +
Economic losses and casualties due to riverine flooding increased in past decades and are most likely to further increase due to global change. To plan effective mitigation and adaptation measures and since floods often affect large areas showing spatial correlation, several global flood models (GFMs) were developed. Yet, they are either based on hydrologic or on hydrodynamic model codes. This may lower the accuracy of inundation estimates as large-scale hydrologic models often lack advanced routing schemes, reducing timeliness of simulated discharge, while hydrodynamic models depend on observed discharge or synthesized flood waves, hampering the representation of intra-domain processes.<br>To overcome this, GLOFRIM was developed. Currently, it allows for coupling one global hydrologic model, producing discharge and runoff estimates, with two hydrodynamics which perform the routing of surface water. By employing the Basic Model Interface (BMI) concept, both online and spatially explicit coupling of the models is supported. This way the coupled models remained unaffected, facilitating the separate development, storage, and updating of the models and their schematizations. Additionally, the framework is developed with easy accessibility and extensibility in mind, which allows other models to be added without extensive re-structuring. <br>In this presentation, the main underlying concepts of GLOFRIM as well as its workflow will be outlined, and first results showing the benefit of model coupling will be discussed. Besides, current limitations and need for future improvements will be pointed out. Last, current developments in code development, applications, and integrations with other research fields will be presented and discussed. +
Ecosystems are in transition globally with critical societal consequences. Global warming, growing climatic extremes, land degradation, human-introduced herbivores, and climate-related disturbances (e.g., wildfires) drive rapid changes in ecosystem productivity and structure, with complex feedbacks in watershed hydrology, geomorphology, and biogeochemistry. There is a need to develop models that can represent ecosystem changes by incorporating the role of individual plant patches. We developed ecohydrologic components in Landlab that can be coupled to create models to simulate local soil moisture dynamics and plant dynamics with spatially-explicit cellular automaton plant establishment, mortality, fires, and grazing. In this talk, I will present a model developed to explore the interplay between ecosystem state, change in climate, resultant grass connectivity, fire frequency, and topography. A transition from a cool-wet climate to a warm-dry climate leads to shrub expansion due to drought-induced loss of grass connectivity. Shrubs dominate the ecosystem if dry conditions persist longer. The transition back to a tree or grass-dominated ecosystem from a shrub-dominated ecosystem can only happen when climate shifts from dry to wet. The importance of the length of dry or wet spells on ecosystem structure is highlighted. Aspect plays a critical role in providing topographical refugia for trees during dry periods and influences the rate of ecosystem transitions during climate change. +
