JP

Incorporating Below Ground City Textures into Urban Flood Modeling for Enhanced Flood Prediction and Risk Assessment: A Case Study of the MIT CAMPUS

Abstract

Pluvial (rain-driven) flooding poses a significant threat to urban areas worldwide, necessitating accurate flood prediction for effective flood risk adaptation and damage mitigation. This research explores the impact of incorporating below ground city textures, such as basements, garages, and tunnels, into urban flood models on the extent and propagation of surface and subsurface flooding and damages. A high-resolution 1&2D Rain On Mesh (ROM) hydrodynamic flood model for the city of Cambridge was developed. The model integrated various layers of geospatial data and the city's DEM. An extended version of the model, the Basement Model, applies an approach to incorporate below ground city textures within a surface flooding model and was developed for the MIT campus neighborhood, approximately 10% of the case study area. The Campus has a unique feature that 33 of the buildings of the main campus are connected via their basements. A recent crowd-sources activity revealed that there are over 1000 potential points where water can enter the buildings.

Our simulation results showed significant basement flooding that altered the spatial-temporal pattern of surface flooding in the study area as compared to the surface model. The basement model identified significant water inflow to the basement reducing the volume of surface flood waters and reducing much of the surface flooding observed in the surface model. Moreover, this model detected flooding within buildings that lacked directed surface flooding due to flow between the basements.

These models were compared in a flood risk study. Using the basement model the expected damages from the 100-year flood were 33% of the expected damages using state of the art USACE surface flood damage methods. Additionally, there was significant differenced in which buildings were at risk, the total amount of flooding volume, the spatial-temporal propagation, and the damage assessments within structures.

The incorporation of basements and below ground city textures into flood modeling proves to be invaluable in accurately predicting flood extent and propagation. These findings contributed to MIT’s Climate Resilience Pathway to develop of a resilient flood management plan incorporating climate change into campus design processes.

Plain-language Summary

Incorporating the flood storage capacities of basement in urban flood modeling can have a significant impact on the potential flood damage in at building and neighborhood scale. A case study of the MIT Campus shows expected damages can be over 70% less if basements are modeled.

Urban Flood Modeling: The Influence of surface run-off modeling and Geospecificity in estimating Flood Extent and Propagation

Abstract

This research investigates the critical factors governing the extent and propagation of urban flooding, and their risks on urban systems. A drainage-catchment-based -2D (Catch) and a 1&2D Rain on Mesh (ROM) FE hydrodynamic flood models were constructed, validated, and compared for case studies of Cambridge and Cleveland. The Catch model directs the volume of run-off in a catchment directly into the pipe system bypassing any localized surface flooding in the catchment. Flooding would occur only when a pipe surcharges and that volume is routed over the surface. In contrast the ROM model distributes the rainfall directly on a 2D mesh, calculates the infiltration at each cell and utilizes the city terrain, road, green-zones and building textures geospatial data to route the surface flow via the shallow flow equations to manholes. The ROM model has coupled 1D pipe inflow/outflow with the surface 2D flow and is able to account for the interaction of surcharge flows as well as surface rainfall flows.

Spatial results of the ROM simulations, see Figure 1, revealed for a range of design storms flood extents 7-9.5 times greater and depths 1.2-7 times deeper across the city during peak flooding, as compared to the Catch model. Additionally, the ROM model effectively identified surface flooding in areas where drainage-based model predicted no flooding. The temporal flood propagation was vastly different between the two models, with the ROM model showing peak flooding that more than 5% of the city remained flooded at the conclusion as compared to the 0% for the drainage-based model. The underestimation of duration and extent of flooding greatly impacts the expected damages on urban infrastructure and current drainage-based models dramatically underestimate this simulation output. Our findings demonstrate that ROM mechanisms, when introduced in pluvial flooding simulations, along with the granular details of city texture incorporated within the model, are critical determinants of the extent and propagation of highly dynamic urban flooding. The results highlight the significance of incorporating ROM approaches and granular city information for more informed and refined pluvial and fluvial regional flood modeling, offering valuable insights into urban flooding impacts.

Plain-language Summary

Catchment-based surface runoff models underestimate the duration and extent of flooding as compared to models that incorporate of rain-on-mesh surface runoff modeling in urban drainage analysis. This underestimation of flood extent and duration leads to underestimation of the expected damages and impacts on lifespan of urban infrastructure. Case studies of Cambridge, Massachusetts and Cleveland, Ohio are presented.

Maximizing Air Quality Benefits of Organizational Climate Action: Evidence from Electricity Purchase and Personnel Business Travel

Abstract: Organizational decisions to mitigate climate change are often focused solely on reducing greenhouse gas emissions, but also can have multiple sustainability-related impacts. A substantial area of impact from reducing greenhouse gases relates to air quality, where reductions in fossil fuel use can cause health damages locally and regionally. While much research has quantified the air quality benefits of large-scale strategies to reduce greenhouse gas emissions, information about the different impacts of organizational Scope 1–3 emissions on air quality is lacking. We use data from two universities and one multinational corporation based in the northeast U.S. to examine the magnitude and location of air quality changes associated with reducing carbon emissions under two different strategies: replacing purchased fossil-based electricity with renewable energy (scope 2), and reducing personnel business travel by air (scope 3). We estimate the marginal climate response and spatially-resolved air quality impacts associated with these two strategies. To do this, we first use an energy system model (US Energy Grid Optimization, US-EGO) to simulate electricity grid responses and related emissions (CO2, NOx and SO2) due to organizational electricity consumption. We calculate business travel emissions (principally NOx, SOx and non-volatile particulate matter) with the Aviation Emission Inventory Code (AEIC) based on detailed flight data provided by the organizations. For both sectors, we run GEOS-Chem High Performance (GCHP), an atmospheric chemistry-transport model, to simulate the ground-level concentration of ozone and fine particulate matter (PM2.5). We estimate the health damages from these two pollutants with Concentration Response Functions (CRFs). We calculate the social costs using the Social Cost of Carbon (SCC) for carbon emissions, and use a Value of Statistical Life (VSL) to quantify costs for air-pollution-induced health damages. We explore how marginal estimates of damages vary depending on system-level assumptions. Finally, we compare our estimates from detailed modeling with results from reduced form air quality estimators (InMAP, AP2 and EASIUR) to identify how these tools might assist organizations in prioritizing emissions reductions to maximize overall air quality benefits.

Emulating Spatial Patterns of Surface Temperature Responses to CO2 Emissions

Abstract 

The ability to rapidly simulate the local climate implications of a large number of future climate policy scenarios is important for planning and implementing adequate climate mitigation and adaptation policies. Given the societal impacts from rising local temperature, we build an emulator of spatially resolved near-surface air temperature responses to carbon dioxide (CO2) emissions. Near surface air temperature is approximately proportional to cumulative carbon dioxide (CO2) emissions, allowing for linearization of the temperature response to emissions scenarios. This linearity enables us to diagnose Green’s Functions for the spatial temperature response to CO2 emissions from pulse simulations conducted as part of the CDRMIP experiments. We then apply this emulator across a wide range emissions scenarios to estimate local temperature responses.

We evaluate this emulator with two CMIP6 experiments: 1) a 1% increase in CO2 concentration, and 2) an experiment that branches from this after concentrations of 1000 PgC are reached. We find that this emulation approach captures the spatial temperature response to CO2 emissions within one standard deviation of the CMIP6 range, with some limited accuracy in polar regions where nonlinearities in climate feedbacks dominate and internal variability may influence the Green’s Function. This approach incorporates emissions path dependency, accounting for various timescales of warming due to CO2 emissions. It is useful for evaluating large ensembles of policy scenarios that are otherwise prohibitively expensive to simulate using earth system models, as it takes less than one second to emulate 90 years of temperature response. We apply this emulator to quantify differing local temperature responses when a global mean of 2ºC is reached, showing that some locations (such as Lagos and Buenos Aires) warm slower than the global mean, while others warm faster (such as Boston and Shanghai). We also evaluate varying CO2 emissions trajectories with the same cumulative emissions, showing that the resulting temperature changes are path dependent.

Plain-language Summary

The ability to quantify the climate impacts of changes in future emissions due to various climate policies is important for planning and implementing adequate climate mitigation and adaptation. Given the societal impacts from rising local temperature, we build a rapid model that can quantify local temperature responses to changes in carbon dioxide (CO2) emissions, a key greenhouse gas responsible for climate change. This simple approach takes advantage of a proportional relationship between total CO2 emissions and temperature, and it takes only one second to quantify temperature impacts over 90 years. We test this approach against the Climate Model Intercomparison Project Phase 6 (CMIP6) experiments, finding some limited accuracy in polar regions. We then use this approach to quantify the local temperature impacts in different cities when a global mean increase of 2 ºC is reached, showing that some cities warm faster than the global mean and others warm slower. We also show that the emissions pathway matters, even if the same total CO2 emissions are reached. Importantly, this approach can be used to estimate local temperature response to dozens of policy scenarios without the computational power and time needed for running a climate model.

Application of Scenario Discovery Techniques to Probabilistic Ensembles in Exploration of Energy Futures

Abstract: Energy-economic and coupled human-natural system models are often used to explore potential energy futures and their implications for climate. There are many uncertain assumptions in the human system models that drive those futures, and in previous work we used a traditional Monte Carlo approach to explore socio-economic uncertainties in a multi-sector, multi-region energy-economic-emissions model of the global economy and generate probabilistic ensembles. The amount of data and information generated from these large ensembles is immense and it can be difficult to sort through and extract relevant insights. The goal of this work is to apply a variety of scenario discovery techniques to the probabilistic ensembles in order to extract insights related to energy futures, with a particular focus on the penetration of renewable energy. We apply Classification and Regression Trees (CART) with Random Forest Classifier (RFC) and Time Series Clustering (TSC) to explore key input drivers of the share of renewable generation, how those drivers can vary over time and across regions, different types of pathways for renewables, and relationships among model outputs. We find that the key drivers of renewables can vary significantly based on the policy scenario, region and time period. In particular, the time series clustering revealed interesting dynamics that are missed by looking at individual years. Through this work we demonstrate the value of scenario discovery techniques in drawing insights from large ensembles of energy-projecting models by facilitating the identification of drivers, relationships among variables and areas of the uncertainty space that are particularly interesting or relevant.

Advancing Complex Human-Earth Systems Science through Natural Language Processing, Graph Machine Learning, and Artificial Intelligence

Abstract: The interconnected risks to interdependent infrastructure, environmental, and socioeconomic systems posed by climate change, energy transitions, and sustainable development require transdisciplinary perspectives to understand the involved complex dynamics and interdependencies. The breadth and diversity of systems, processes, and risks require a synthesis of an extremely diverse set of research fields, literatures, and operational expertise. Recent breakthroughs in artificial intelligence (AI) present promising opportunities to accelerate progress in transdisciplinary synthesis in MultiSector Dynamics (MSD) research. AI can potentially help to clarify connections across scientific communities and accelerate the translation of insights across domains. Here we demonstrate a systematic approach using a combination of modern natural language processing (NLP), graph-based and other machine learning approaches to gain on-demand topical access to, and insight from, a corpus of over 100,000 scientific publications and other ancillary data sources that are representative of the relevant literature landscape for the field of MSD. These insights help us identify stable and emerging communities of researchers and research topics that align with advancing the aspirations of the MSD community. We identify and describe advances in the cross-domain bodies of literature addressing the interconnected sustainability and climate change risks across scales, sectors, and systems. Our analysis seeks to quickly understand gaps and opportunities that currently exist for MSD researchers. We provide these state-of-the-art AI/ML/NLP workflows to the MSD community as we believe that cross-disciplinary training and teaming is critical for advancing complex adaptive human-Earth systems science in a world of deeply uncertain and interconnected risks.

Disentangling the Changing Nature of Water Quality, Availability, and Equity in the United States

Abstract: Threats to future water “security” are increasingly assessed through not just the lens of water and water quality, but how these may unequally expand across sociodemographic and ethnic landscapes. Evidence to date indicates that low-carbon, climate mitigation policies and targets provide marginal benefits to water scarcity trends. Therefore, effective measures require integrated solutions to co-evolving system-wide features of supply, demand, nutrient loading, and conveyance, and avoid inequities and unjust transitions. Based on our current assessment with the MIT System for the Triage of Risks from Environmental and Socioeconomic Stressors (STRESS) platform, we find co-existing areas of water stress, water quality, poverty, and minority populations are extensive – particularly in the south and southeast United States – but with important granular hotspots in populated areas. Therefore, an underlying question and scientific challenge is to understand and quantify the extent that natural and human-forced drivers affect (or benefit) these landscapes – and what are the salient response patterns amidst climatic and human-forced uncertainties?

In view of these considerations, we have conducted a suite of simulations with a linked model system that resolves the contiguous United States at over 2,100 basins and includes a water management module as well as a parsimonious water-quality model. The experimental simulations combine altered landscapes of water supply, demand, nutrient loadings, and conveyance landscapes sequentially and successively. These altered landscapes reflect plausible changes in human-forced climate patterns, land use and management (cultivated for food, agriculture, and bioenergy), water demands (domestic, industrial, energy, and agriculture), as well as water-system efficiencies. Overall, we find that uniform and large-scale patterns of these drivers produce heterogenous and complex responses across U.S. basins, but with important exceptions. These heterogenous response features, however, can be ascribed to precursory conditions of the basins’ environments, and thus indicate potentially predictable consequences. We demonstrate these predictable features through a series of future scenarios generated by our multi-sector dynamical prediction framework.

Future Patterns of US Agricultural Land Use under Multiple Stressors

Abstract: Climate change, income and population growth, and changing diets are major stressors for global agricultural markets with implications for land use change. US land use at regional and local scales is directly affected by domestic forces and indirectly through international trade. In order to investigate the effects of several potential forces on land use changes in the US at multiple spatial scales, we advanced the capabilities in representing the interactions between natural and human system through a collaborative effort between two MSD teams. This effort couples a multi-sectoral and multi-regional socio-economic model of the world economy with detailed representation of land use and agricultural systems to an open-source downscaling model which enables translating regional projections of future land use into high-resolution representations of time-evolving land cover. We exemplify the framework over the Mississippi river basin and consider the effects of a range of global drivers and stressors, such as: high or low economic and population growth, more negative or more positive impacts of climate change, and more or less dietary change. The resulting regional land use changes are further translated into more detailed projections of land use changes through the downscaling model. In addition, we examine assumptions, including 1) how global stressors might, in combination, affect regional land use change and 2) how alternative rules and constraints spatialize the regional projections. Our results help better understand the implications of land use change on carbon storage, soil erosion, chemical use, hydrology, and water quality. The employed downscaling model facilitates interoperability among models and across various spatial scales. The presented framework can be readily applied to other basins with little effort.

Logs or permits? Forestry land use decisions in an emissions trading scheme

Abstract: Negative carbon emissions options are required to meet long-term climate goals in many countries. One way to incentivize these options is by paying farmers for carbon sequestered by forests through an emissions trading scheme (ETS). New Zealand has a comprehensive ETS, which includes incentives for farmers to plant permanent exotic forests.

This research uses an economy-wide model, a forestry model and land use change functions to measure the expected proportion of farmers with trees at harvesting age that will change land use from production to permanent forests in New Zealand from 2014 to 2050. We also estimate the impacts on carbon sequestration, the carbon price, gross emissions, GDP and welfare.

When there is forestry land use change, the results indicate that the responsiveness of land owners to the carbon price has a measured impact on carbon sequestration. For example, under the fastest land use change scenario, carbon sequestration reaches 29.93 Mt CO2e by 2050 compared to 23.41 Mt CO2e in the no land use change scenario (a 28% increase). Even under the slowest land use change scenario, carbon sequestration is 25.89 Mt CO2e by 2050 (an 11% increase compared with no land use change). This is because, if foresters decide not to switch to permanent forests in 1 year, carbon prices and ultimately incentives to convert to permanent forests will be higher in future years.

Part of this research was completed while co-author Dominic White was visiting the MIT Joint Program.

Pages

Subscribe to JP