Concept Note- E4DRR

Aug 31, 2024 · 5 min read

Activites of project:

Outcome 1. Enhanced Risk Knowledge and Awareness Through the Creation of Event-Based Climate Storylines

Work package 1 Establish Hazard and Impact Modelling for the Region: Setting up of hazard and impact modelling for the region

  • Activity 1.1.1 EPS Data API for storylines

    • Organize and implement cost-effective programming routines for Ensemble Prediction System (ERA5 and GEFS) to be used for storylines creation as an Application Programming Interface (API).

  • Activity 1.1.2 Hazard Models Input Data API

    • Collect static data related to parsimonious hazard models along with dynamic input data as an Application Programming Interface (API).

  • Activity 1.1.3 Hazard Models Drought DevOps WPS

    • Set up the hazard model for Drought (CWALID) for validation and training schemes for operational purposes on cloud computing as a Development Operations (DevOps) Web Processing Services (WPS).

  • Activity 1.1.4 Hazard Models Flood DevOps WPS

    • Set up the hazard model for flood (RIM2D) for validation and training schemes on GPU hardware and for operational purposes on cloud computing with Web Processing Services (WPS).

  • Activity 1.1.5 Impact Modeling Input Data API

    • Collect static data required for impact calculations in CLIMADA, including exposure data, vulnerability defined in terms of impact functions, and hazard model outputs as an Application Programming Interface (API).

  • Activity 1.1.6 Impact Modeling DevOps WPS

    • Set up the CLIMADA impact calculation tool using the hazard model outputs as input as a Development Operations (DevOps) Web Processing Services (WPS).

Work package 2 Implement Co-development Method for Event-Based Climate Storylines: Implement a co-development method for event-based climate storylines

  • Activity 1.2.1 Collect Documented Hazard Data and select events

    • Collect available documented hazard datasets related to flood and drought disasters in the 11 East African countries for the period 2000-2024. Select 11 flood and 11 drought events for each country in EAHW.

  • Activity 1.2.2 Simulation of Hazard-Impact models

    • Perform the hazard and impact modeling for the selected events using observed weather conditions as references for factual storylines based on documented hazards.

  • Activity 1.2.3 Story Map Development

    • Design and develop story maps based on the reference storylines datasets.

  • Activity 1.2.4 Co-production Initiatives

    • Conduct preliminary coproduction activities to scope and select representative events for drought and flood in the respective countries.

Work package 3 Consolidate Composite Dataset and Storylines into Story Maps Linked with EAHW: Consolidate the composite dataset and storylines as story maps and link with East Africa Hazard Watch (EAHW) web application

  • Activity 1.3.1 Multiple Simulations of Hazard-Impact Models

    • Run the hazard and impact models in storylines mode for each selected event and obtain counterfactual initiation data from the rainfall forecast members of the EPS.

  • Activity 1.3.2 Storylines Organization in Story maps

    • Organize the outputs of the hazard and impact models into storylines by extending the story maps from WP2 and link them with EAHW.

  • Activity 1.3.3 Bayesian Network Development

    • Develop causal and Bayesian networks to represent the storylines, along with a web interface for coproduction.

  • Activity 1.3.4 Co-production Validation

    • Conduct a second-stage coproduction process to validate and optimize the storylines, utilizing the Bayesian network web interface.

  • Activity 1.3.5 Experiment Gen AI for event-based climate storylines

    • Explore the role of Generative AI Large Language Models (LLMs/chatGPT) in Bayesian and causal network creation and provide chat support with storylines narrative.

Outcome 2: Exploration of Storylines creation processes and applications in Ensemble Prediction System (EPS) Impact-Based Forecasting (IBF)

Work package 4 Establish an Analysis and Forecast Validation Facility Utilizing Storylines Datasets and Processes for Anticipatory Action: Develop analysis and forecast validation facility utilizing storyline dataset and processes for anticipatory action

  • Activity 2.1.1 Ensemble Prediction System (EPS) forecast verification and Extreme Value Analysis (Thresholds and Triggers) WPS

    • Utilize the EPS used for storylines, GEFS for flood and SEAS5 for drought, their last 20 years hindcast for extreme value analysis on cloud computing as a Development Operations (DevOps) Web Processing Services (WPS).

  • Activity 2.1.2 Web interface for Anticipatory Action

    • Following anticipatory action analysis routines on extreme value analysis for thresholds, forecast verification to set triggers utilizing impact data from storylines. Provide an analytical platform in a web interface through a Jupyter notebook-based setup.

Work package 5 Demonstrate the Application of Storylines as Bayesian Networks in IBF Risk and Decision Analysis

  • Activity 2.2.1 Storylines Synthesis

    • Synthesize the storylines into a single Bayesian network for use in IBF.

  • Activity 2.2.2 Bayesian Network Backend Development

    • Develop backend code for real-time EPS data download for flood and drought, operationalize semi-automated hazards and impact modeling, and perform risk and decision analysis.

Work package 6 Operationalize EPS IBF with Hazard and Impact Modelling for the Region Operationalize EPS IBF with hazards and impact modeling for the region

  • Activity 2.3.1 Cloud-native EPS IBF DevOps WPS

    • Set up programming routines for interoperable data consumption and dissemination of project datasets, following the EPS IBF routines in cloud-native operations (CNO) and Analysis Ready, Cloud Optimized data formats (ARCO) as a Development Operations (DevOps) Web Processing Services (WPS).

  • Activity 2.3.2 IBF dissemination in EAHW layer/API

    • Create a web application that shows backend processes with integration on Bayesian networks for operational Early Warning Early Action (EWEA) IBF, including workflow visualization for auditability of evidence for non-public operation center consumption. End product from IBF workflow in ICPAC climate services and EAHW dissemination in Application Programming Interface (API).

  • Activity 2.3.3 Experiment generative AI for EPS

    • Explore the application of Generative AI (cGAN) in bias correction and downscaling of EPS data in an operational setup.

Outcome 3: Capacity Development in the Use of Methods and Tools for Storyline Creation

Work package 7 Develop Documentation and Training Materials on Methods and Tools Used in Storyline Creation Activities Prepare documentation and training materials on methods and tools used in storyline creation

  • Activity 3.1.1 Open Sourcing the software programs used for storylines tools and methods

    • Open source the scripts and methodology for hazard modeling, impact estimation, Bayesian networks, operational downloading of EPS, story maps, interactive Bayesian networks.

  • Activity 3.1.2 Documentation on the software programs used for storylines tools and methods

    • Create documentation for hazard modeling, impact estimation, Bayesian networks, operational downloading of EPS, story maps, interactive Bayesian Networks.

  • Activity 3.3.3 Tutorials on storylines tools and methods

    • Tutorial development on hazard modeling, impact estimation, Bayesian networks, EPS in Jupyter notebooks.

Work package 8 Conduct Capacity Development Activities in Co-development, Workshops, and Tutorials Capacity development activities in co-development, workshops, and tutorials

  • Activity 3.2.1 Conduct Capacity Development
    • Conduct workshops and tutorials for capacity development by utilizing the ongoing activities at ICPAC and along with co-development events.