Objective/Mission
The Earth Signals and Systems Group (ESSG) quantifies cyclone-induced risks in Bangladesh under climate change to strengthen community resilience. It assesses hazards including wind, rainfall, coastal storm tides, compound flooding, cascade flooding, and salinity impacts. ESSG operates a dynamic soil observatory in southwest Bangladesh combining remote and in-situ data for soil-salinity mapping and forecasting. It also integrates hazard-, exposure-, vulnerability-, and agricultural-impact assessments into community-based frameworks to guide optimal agricultural practices. ESSG works at national and sub-national scales for hazard assessment, sub-nationally for community-driven observatories, andlocally for agriculture-focused resilience. ESSG’s objective under the Climate Grand Challenges is to advance national adaptation, improve sub-national awareness, and fundamentally develop local resilience to cyclone-induced risks in Bangladesh.
Approach in Year 3
The ESSG team developed the authoritative cyclone-risk methodology central to adaptation and mitigation planning in Bangladesh. Our physics-based method rapidly downscales climate models to generate large synthetic cyclone ensembles for present and future mid-to-late-century climates. We pioneered Statistical-Physical Adversarial Learning (Dn-SPAL) to efficiently downscale extreme rainfall events and conducted Bangladesh’s first comprehensive storm-tide risk quantification using detailed hydrodynamic modeling. Our rapid-inundation model integrates upstream riverine and downstream oceanic conditions to quantify cascade flooding from compounded cyclone and seasonal rainfall extremes at national, subnational, and local scales. Our observatory produces southwest Bangladesh’s first dynamic root-zone, soil-salinity maps utilizing adaptive in-situ sampling, satellite observations, and machine learning for short-term forecasting. By coupling hazard assessments with satellite-derived exposure and survey-based vulnerability, ESSG supports communities in optimizing local livelihood resilience amid disruptions from extreme events. Our comprehensive approach operates in collaboration with academia, NGOs (including BRAC), and government agencies. We anticipate direct uptake of our hazard products by flagship initiatives, integration of salinity data into YRISE activities,and application of detailed local mapping to enhance agricultural resilience.
Progress in Year 3
ESSG leverages diverse datasets to drive its cyclone-risk analyses. Reanalysis and CMIP6 climate models under IPCC AR6 are primary inputs for cyclone downscaling, incorporating new 1-km surface roughness maps developed via machine learning. For coastal flood modeling, we acquired improved bathymetric datasets for the Northern Bay of Bengal and Bengal Delta, along with long-term water-level observations, including GPS coordinates and gauge data (Hiron Point, Khepupara) through collaboration with Dr. Michael Steckler. Large-scale inland flood modeling relies on open-source and collaborative data for topography, bathymetry, and upstream flows informed by the greater GBM basin.At the polder scale, we complement boundary condition data with our surveys of river bathymetry,polder topography, and infrastructure monitoring. The dynamic soil observatory integrates satellite data and extensive in-situ measurements collected by students and community members, coordinated alongside other teams, including the proposed Permanent Automated Soil Observatory (PASO) withBRAC. Using Sentinel satellite data, we perform automated semantic classification of polder-levelstructures (e.g., Ghers). Socio-economic vulnerability and community decision-making data were significantly enriched by extensive community games initiated in Year 3.ESSG develops and deploys innovative models critical for cyclone risk analysis in Bangladesh. Our original modeling includes cyclone downscaling and Dn-SPAL methods for extreme rainfall, using large-ensemble approaches to quantify uncertainty. We employ state-of-the-art 2D and 3D floodmodels and have developed new coupled inundation-salinity models for inland (pluvial and fluvial)flooding. In Year 3, we integrated improved wind modeling into coastal flood simulations. Our dynamic observatory utilizes informative observation strategies to optimally and densely gather adaptive soil-salinity data. Additionally, we introduced novel “informative learning” methods in machine learning to improve model accuracy and interpretability. We developed new satellite-driven methods to classify polder-level structures, particularly ghers, and piloted agent-based, agricultural-yield modeling in collaboration with IIT Bombay. Our original computational games engage communities in decision-making, forming a basis for livelihood-resilience strategies. Overall, ESSG’s modeling advancedhazard assessments at national and sub-national scales, mapping and observations at sub-national and local scales, and resilience-driven decision-making at the local level.The outcomes include cyclone-induced wind hazards, cyclone and seasonal rain hazards, and flood-hazard assessments in a large ensemble uncertainty quantifying setting. The storm-tide assessment is the first such result for Bangladesh for a changing climate under a wide variety of scenarios and models. The dynamic observatory generates high-resolution (30-m) root-zone, soil-salinity maps, andis poised to enter into a self-sustaining operational model. It informatively gathers new salinity samples and processes them at ESSG’s soil laboratory currently housed at KUET. The new PASO system will offer a permanent predictive solution to soil -salinity monitoring in southwest Bangladesh, and our work also directly informs YRISE’s efforts with water entrepreneurship.Our cyclone-induced, flood-hazard work attracted significant attention, leading to extensive collaborations with Bangladeshi universities, IIT Bombay, and Kanpur. Interest from national planners has grown significantly, paving the way for engagement in Y4. Regionally, the first dynamic soil observatory and PASO have been recognized as important advances. They use a self-sustaining,community-business model to supply dense maps—at present, root-zone soil salinity. We directly operate in Assasuni, Polder 7/1, and the adjacent Munshiganj areas, closely engaging with three communities. Here, hazard modeling, exposure and vulnerability assessment, and impact/yield analysis are coupled with community-based participation for agricultural resilience, reported to the Jameel Observatory CREWSnet audience.
Data Hub and Decision Support Tool
Our wind, rain, and flood hazard products are publicly accessible in the repository, and access to the salinity products is also provided. We anticipate releasing additional data on exposure and vulnerability,as well as inferences drawn from livelihood resilience at the local-scale in the data hub. Our primary contributions include cyclone-induced hazards, large-ensemble seasonal extremes for national and sub-national scales, dynamic observatories for sub-national scales, and hazard-exposure-vulnerability-impact-decision-making for cyclone-induced livelihood resilience at the local scale.