UN Tech Over Hackathon

June 16 & 17, 2025

The UN Tech Over is a call to action taking place at United Nations Headquarters from 16 to 17 June 2025, in advancing the UN Sustainable Development Goals (SDGs).

The event promotes collaboration, builds awareness, and deepens engagement with the open source community.

You are invited to select between participation in one of the three following UN Tech Over events, and submit your preference using the following link.

PARTICIPATION FORM
Child-Centric Weather Intelligence
Challenge 1

Child-Centric Extreme Weather Intelligence

Develop methodologies for child-focused impact assessments to enable proactive disaster response before storms strike.

Learn More →
Geo-Puzzle Solutions
Challenge 2

Solving the "Geo-Puzzle"

Harmonize diverse spatial data formats to overlay multi-hazard data layers for accurate child-centered analysis.

Learn More →
GeoSight Platform Enhancement
Challenge 3

Make Risk Data Visible & Actionable

Enhance GeoSight's functionality with new features to support effective disaster preparedness and response.

Learn More →

Brief Summary of Challenges

Challenge 1

Title: Unlocking Child-Centric Extreme Weather Intelligence: From Hindcasting to Forecasting, from Reaction to Proaction

Context: Understanding how extreme weather events impact children is fundamental to UNICEF's short-term emergency preparedness and response efforts. Traditionally, impact reports reach our country offices days after a disaster has struck, forcing a reactive stance and delivering aid that might arrive too late for some. However, new data sources and analytical methods offer a powerful alternative: conducting analyses ahead of the storm.

You will tackle the fundamental challenges of understanding weather-related impacts, especially on children, enabling UNICEF country offices to make more informed, timely decisions to protect children from extreme weather threats. Current weather forecasting and risk assessments often focus on asset damage, overlooking the specific vulnerabilities and needs of children during such crises.

Objectives: The core objective is to develop innovative methodologies and code scripts for child-focused short-term impact and risk assessments.

Participants should aim to quantify children's specific vulnerabilities to extreme weather events (e.g., hurricanes and their associated hazards like floods and landslides) and develop solutions that provide actionable insights over a short-term horizon. This could include, but is not limited to:

Bonus Objective: Participants looking for an additional challenge are encouraged to explore methods for scenario analysis (e.g., developing worst-case, best-case, and most-likely impact scenarios based on different forecast possibilities or assumptions) to further enhance preparedness planning under uncertainty.

Key skills: Data analysis, GIS and spatial analysis, statistical analysis, risk assessment, data visualization, scenario modelling (optional)

Visuals:

A child sitting on the ground with a map

Challenge 2

Title: Solving the “Geo-Puzzle”: Developing methods for overlaying multi-hazard data layers with children vulnerability factors

Context: Protecting children from hurricanes and storms requires timely analyses of how weather threats intersect with population vulnerability, especially children. This means combining hazard data with child exposure information (e.g., population density, access to health or education) in ways that are readily accessible and actionable by decision makers to strengthen resilience and implement appropriate emergency preparedness, disaster risk reduction, and climate and environmental action plans.

However, the critical data needed often comes in mismatched formats, e.g., weather forecasts and population distribution in grid (raster) format, child data as indicators linked to administrative units, and infrastructure data (e.g., health sites or schools) as point locations. While various tools exist to tackle parts of this issue, there is currently no open-source, one-stop solution that provides an end-to-end method for harmonizing heterogeneous spatial datasets into streamlined and actionable workflows, making it difficult to answer urgent questions like "How many children with limited access to health services will face dangerous storm surges?" Solving these spatial mismatches – or "geo-puzzles" – is essential for creating reliable, child-centered early warning systems that can trigger proactive protection measures before storms make landfall.

UNICEF is leading several initiatives aimed at addressing these challenges. The Children’s Climate Risk Index – Disaster Risk Model (CCRI-DRM) initiative plays a critical role in filling the gap on child-responsive climate, environmental, and weather-related disaster risk information. The CCRI-DRM initiative analyzes children’s exposure to climate and environmental hazards and their vulnerability across key child-critical social sectors (such as health, nutrition, education, and social protection), and makes information available to inform national and subnational policies and programmes through country-specific dashboards. CCRI-DRM has been developed in Cambodia, Saint Kitts and Nevis, Kenya, and Tajikistan. These countries have already demonstrated the usefulness of the models within their countries, using them to inform policy and programming decisions. However, the initiative currently faces limitations with gaps in data coverage as well as the need to establish reliable pipelines for the timely updating of climate and environmental hazards.

Separately, Giga—a joint initiative between ITU and UNICEF—developed Giga Spatial, an open-source Python library that streamlines the ingestion, cleaning, and alignment of geospatial data. With built-in data handlers and view generators, Giga Spatial offers scalable, reusable workflows that can be applied to support CCRI-DRM’s geospatial data processing needs as well as other similar challenges involving multi-source spatial data integration.

Objectives: Develop, validate, and extend methods for harmonizing diverse spatial datasets across varying formats (geospatial file formats and data types) , resolutions, and projections to enable accurate, child-centered natural hazard impact analyses. Participants are encouraged to utilize the Giga Spatial Python library to improve data processing pipelines, spatial reconciliation, and visualization techniques, specifically addressing data integration across administrative boundaries, spatial aggregation, and uncertainty quantification. Solutions will be evaluated using practical scenarios from the UNICEF CCRI-DRM initiative, aiming to support anticipatory action systems that clearly identify where and how children will be affected by natural hazards, ensuring timely, informed decision-making for emergency preparedness and resilience planning.

Ideally, we will be looking at the following outputs:

However, this is an exploratory sprint – pick any idea, hack, and show what you learned! Feel free to mix, match, or only half-finish. We're interested in your insights and approaches, not just complete solutions. We also encourage you to think out of the box. If you have an idea for a feature or improvement that could make Giga Spatial and/or CCRI-DRM more powerful, user-friendly, or impactful, bring it to life!

Key skills: Geospatial analysis, Python (or desktop GIS tools, e.g., QGIS), Pandas, GeoPandas, GDAL

Visuals:

A map of different colors

Challenge 3

Title: Make the Risk Data Visible & Actionable: Develop new GeoSight features to support effective disaster preparedness and response

Context: GeoSight is UNICEF's open-source geospatial web-based data visualization and analysis platform designed to make geospatial data easily accessible and shareable in support of risk-informed programming. It utilizes administrative reference datasets from GeoRepo to display data. Projects in GeoSight combine different elements such as reference datasets, indicators, and context layers, enabling comprehensive data visualization and analysis for end-users.

Objectives: This hackathon aims to enhance GeoSight's functionality by implementing new features that improve data analysis, visualization, and user interaction. Participants will collaborate to develop these features, contributing to a more robust and user-friendly platform.

Key skills: Python, Django, React, HTML, JavaScript, PostgreSQL

Visuals:

A map of the country

Useful Links:

Challenge 1

Basic Information

Title of Challenge: Unlocking Child-Centric Extreme Weather Intelligence: From Hindcasting to Forecasting, from Reaction to Proaction

Contact Person:

Name: Yves Jaques, Daniel Alvarez, Felix Schwebel

Email: yjaques@unicef.org, dalvarez@unicef.org, fschwebel@unicef.org

Organization/ Department: UNICEF, Data & Analytics Section, Computational Analytics and Geospatial Intelligence Unit

Challenge background

Understanding how extreme weather events impact children is fundamental to UNICEF's short-term emergency preparedness and response efforts. Traditionally, impact reports reach our country offices days after a disaster has struck, forcing a reactive stance and delivering aid that might arrive too late for some. However, new data sources and analytical methods offer a powerful alternative: conducting analyses ahead of the storm.

By leveraging weather forecast data and predictions of hurricane paths, we can proactively assess: What hazards are children exposed to? Which children are most vulnerable? What are the potential consequences? And how can we measure and respond to these impending impacts? This foresight allows for the proactive mobilization of resources and the engagement in emergency preparedness planning, transforming a potential disaster into a managed event.

This challenge invites participants to develop child-centered impact assessment capabilities for UNICEF's "Ahead of the Storm" initiative. This initiative was created based on requests from multiple country offices, meaning your work in this hackathon could directly contribute to enhancing our preparedness and potentially saving lives. You will tackle the fundamental challenges of understanding weather-related impacts, especially on children, enabling UNICEF country offices to make more informed, timely decisions to protect children from extreme weather threats. Current weather forecasting and risk assessments often focus on asset damage, overlooking the specific vulnerabilities and needs of children during such crises.

Challenge description

The core task is to develop methodologies and code scripts for child-focused short-term impact and risk assessments. These should quantify the specific vulnerabilities of children to extreme weather events (like hurricanes and their associated effects) to enable proactive, targeted interventions in high-risk areas before disasters occur, rather than reactive responses afterwards.

Below are some key areas we've identified to spark ideas, but we also participants to bring their own creative solutions and focus areas:

  • Advanced Hazard Exposure Analysis (Single or Multi-Hazard):
    • Focus on a specific primary hazard (e.g., hurricanes). Develop refined methods to assess child exposure, moving beyond traditional metrics (like solely relying on wind speed categories for hurricanes) to better understand its direct and nuanced impacts on children and child-critical infrastructure.
    • Analyze how a primary forecasted event can trigger a cascade of subsequent hazards (e.g., hurricanes leading to storm surges, widespread flooding, and landslides), and map the combined exposure for children.
  • Child-Centric Impact on Critical Infrastructure & Services:
    • Assess the exposure of critical child-related infrastructure (e.g., schools, hospitals, water points) to various hazard levels.
    • Evaluate the potential loss of functionality of this infrastructure (due to direct damage or disruption of essential services like power and transport).
    • Analyze the accessibility of these critical locations for children post-event (e.g., are roads usable?).
    • Consider the secondary impacts of infrastructure disruption on children (e.g., loss of school meals, lack of safe shelter).
  • Secondary Hazard & Compounding Vulnerabilities Analysis:
    • Identify and model potential secondary hazards that could arise post-event, such as disease outbreaks (e.g., cholera due to contaminated water, vector-borne diseases from standing water).
    • Explore how these secondary hazards might be exacerbated by pre-existing vulnerabilities in child populations (e.g., low vaccination rates, malnutrition, displacement).
  • Bonus Objective: Scenario-Based Impact Analysis:
    • As an additional exploration, develop methods for conducting scenario analysis based on different forecast possibilities (e.g., varying storm tracks, intensities, or impact footprints leading to worst-case, best-case, and most-likely scenarios for child impact). This helps in understanding the range of potential outcomes and preparing for different eventualities.

Key skills: Data analysis, GIS and spatial analysis, statistical analysis, risk assessment, data visualization, scenario modelling (optional)

Proposed Datasets

Explore a curated collection of datasets here:

https://opensource.unicc.org/open-source-united-initiative/un-tech-over/challenge-1/ahead-of-the-storm-challenge1-datasets

Check out the README for a quick overview and an Excel file for a summary of the available datasets.

These are completely optional — feel free to use your own sources, combine them, or explore external ones that best fit your approach!

Back to the top

Challenge 2

Basic Information

Title of Challenge: Solving the “Geo-Puzzle”: Developing methods for overlaying multi-hazard data layers with children vulnerability factors

Contact Person:

Name: Yves Jaques

Email: yjaques@unicef.org

Organization/ Department: UNICEF, Data & Analytics Section, Computational Analytics and Geospatial Intelligence Unit

Challenge background

Protecting children from hurricanes and storms requires timely analyses of how weather threats intersect with population vulnerability. This means combining hazard data with exposure information (e.g., population density, access to health or education) in ways that are readily accessible and actionable for decision makers to boost preparedness and strengthen resilience. At UNICEF, we face several interconnected gaps in our ability to protect children from weather and climate threats: the need of accessible, automated pipelines that transform data from different sources and formats (e.g., complex meteorological data) into usable and timely intelligence; analytical methods that reveal how hazards specifically impact children; and reliable, well-structured data on the impacts of past events to support innovative analysis (e.g., ML models).

Much of the critical data needed often comes in mismatched formats – weather forecasts and population distribution in grid (raster) format, child data as indicators linked to administrative boundaries, and infrastructure data (e.g., health sites or schools) as point locations. Currently, no standardized methods exist to harmonize these datasets, making it difficult to answer urgent questions like: "How many children with limited access to health services will face dangerous storm surges?" Solving these spatial mismatches – or "geo-puzzles" – is essential to building reliable, child-centered early warning systems that enable proactive protection measures before storms make landfall.

UNICEF is leading a number of initiatives that aim to address these challenges. The Children’s Climate Risk Index – Disaster Risk Model (CCRI-DRM) addresses the gap in child-responsive climate, environmental, and weather-related disaster risk information. It combines data on population exposure and vulnerability across key child-critical social sectors (e.g., health, nutrition, education, and social protection) to climate and environmental hazards, shocks, and stresses at the subnational level. The results are presented through an intuitive and accessible public platform to inform decision-making. While deployed in several countries, the CCRI-DRM model faces limitations in data coverage and lacks automated, scalable pipelines for regularly updating the climate and environmental hazards and shock data, an effort that currently relies on manual, time-consuming processes.

On the other end, Giga—a joint initiative between ITU and UNICEF—has developed Giga Spatial, an open-source Python library designed to automate the ingestion, standardization, and integration of geospatial data. Built from Giga’s work mapping schools and modeling infrastructure needs, the tool streamlines the ingestion, cleaning, and standardization of key datasets. Its two core components include data handlers, which prepare raw data into analysis-ready formats, and view generators, which align data to various geometry sets, such as subnational administrative boundaries. These flexible, reusable workflows help reduce manual effort and improve consistency in geospatial pipelines—capabilities that could meaningfully support CCRI-DRM’s climate and risk modeling efforts.

The same methodologies can also be applied to many other use cases, where there is a need to overlay, compare, and cross-analyze spatial datasets at different spatial resolutions or using different boundary references.

Challenge description

Develop and test methods to extract, process, deliver, and harmonize spatial data in diverse spatial formats (file formats and data types), resolutions, and projections to support accurate, child-centered natural hazard impact analyses. Participants are encouraged to work within the context of the Giga Spatial Python package and the real-world application use cases proposed here, but are welcome to explore alternative applications if deemed relevant and useful to the scope of the challenge. In alignment with the Ahead of the Storm theme, the focus will be on weather-related hazards (hurricanes, storm surges, floods, landslides).

  • Data download and processing pipelines: Extend the existing handlers module to support downloading and processing of additional geospatial datasets relevant to hazard analysis, such as meteorological data and weather forecasts
  • Spatial reconciliation methods: Develop, validate or extend existing workflows and methods for spatial reconciliation across differing geographic granularities, formats, and timeframes
    • Extend and validate the generators functionality to support integration and reconciliation of datasets across multiple resolutions and formats
    • Expand and refine spatial aggregation and attribution methods (e.g., sampling, small area estimation, summation, averaging) to summarize data values across spatial scales
    • Strengthen capabilities to overlay and visualize multiple spatial layers, accommodating diverse use cases and complex aggregation or attribution rules
    • Introduce functionalities to analyze time-series or real-time data, specifically focusing on high-frequency, short-term forecasts for anticipatory action
    • Develop and integrate approaches for quantifying, visualizing, and communicating uncertainty within spatial data layers and analytical outputs, improving transparency and trust in decision-making tools

The methods described above will be evaluated through their application in real-world use cases that support anticipatory action and child-centered risk analysis ahead of the storm, including (but not limited to) those listed below. Participants are free to tackle one or more of the following use cases:

  • Facilitate alignment of natural hazards exposure data with child vulnerability indicators by automating reconciliation of natural hazard spatial dataset across differing administrative and/or reporting boundaries to match child vulnerability indicators
  • Enable integrated risk analysis by combining meteorological forecast grids, flood extent rasters, landslide risk zones, spatial accessibility maps (e.g., travel time to the nearest school) with child-specific point data (schools, health centers) and administrative boundary data (vaccination rates, poverty) into unified analytical layers
  • Quantify and visualize uncertainty in weather impact predictions for children when combining forecasts with demographic data
  • Develop a micro-service that ingests forecast data (e.g., from ECMWF or DWD) and publish it as live web-accessible map layers (e.g., XYZ tiles, WMTS, or Cloud-Optimized GeoTIFFs) for integration into platforms such as MapLibre or as context layers in CCRI-DRM Geosight dashboards. For an additional challenge, consider the possibility of building animated open-source weather forecast web layers (similar to windy.com) compatible with open-source web mapping frameworks (e.g., MapLibre)
  • Using existing CCRI-DRM dashboards as reference cases (Cambodia, Saint Kitts and Nevis, Kenya, and Tajikistan) devise a methodology for processing data products from multiple satellite sensors for quantities requiring daily measurement frequency (e.g. surface temperatures) combining them into a reliable timeseries of hazard -specific information readily available for overlay with CCRI-DRM children vulnerability information, including detecting and where possible automatically addressing gaps in adequate spatial coverage (country wide), uniform resolution (better than 0.05 degrees), temporal consistency (no timeseries gaps) and large discrepancy between information provided by different products for same areas. The solution will ideally be designed for global scale replication
  • Develop reliable data pipelines and GIGA spatial data handlers to enhance the regular replication of the processing at global scale for CCRI-DRM models, including pulling of satellite data products
  • Other use cases related to anticipatory action systems to identify where and how children will be affected by incoming storms, with enough lead time to implement protective measures

Ideally, we will be looking at the following outputs:

  • Open-source code addressing at least one of the main objectives, ideally implemented as an extension of the Giga Spatial package.
  • A demonstrative application of the library with one of the proposed real-world use cases

However, this is an exploratory sprint – pick any idea, hack, and show what you learned! Feel free to mix, match, or only half-finish. We are interested in your insights and approaches, not just complete solutions. We also encourage you to think out of the box. If you have an idea for a feature or improvement that could make Giga Spatial and/or CCRI-DRM more powerful, user-friendly, or impactful, bring it to life!

Skills & Tools

Skills:

Essential:

  • Vector and raster data handling (e.g. GeoJSON, shapefiles, rasters, GeoTIFFs etc)
  • Spatial joins and overlays
  • Geostatistical analysis

Optional:

  • H3 spatial indexing and hex binning
  • Spatial accessibility analysis
  • Network analysis and routing (e.g. shortest path)
  • Familarity with Satellite data products

Tools:

Essential:

  • Python
  • Desktop GIS (e.g. QGIS)
  • Pandas – for tabular data manipulation
  • GeoPandas – for vector data analysis

Optional:

  • Shapely – for geometric operations
  • Pyproj – for CRS transformations
  • Fiona – for file I/O of geospatial data
  • Rasterio – for raster data handling
  • GDAL – for low-level spatial data processing
  • PostGIS – for spatial SQL queries (via psycopg2 or SQLAlchemy)
  • Matplotlib / Plotly – for plotting
  • Folium / Kepler.gl / Leafmap – for interactive map visualizations
  • Jupyter Notebook – for exploratory development
  • H3-py – H3 geospatial indexing in Python
  • pgRouting – routing and network analysis in PostGIS

Proposed Datasets

Open Source Package

Natural hazards

Demographic and vulnerability layers

Infrastructure

Multimedia Material

N/A

Further Information

Back to the top

Challenge 3

Basic Information

Title of Challenge: Make the Risk Data Visible & Actionable: Develop new GeoSight features to support effective disaster preparedness and response

Contact Person:

Name: Yves Jaques, Jan Burdziej

Email: yjaques@unicef.org, jburdziej@unicef.org

Organization/ Department: UNICEF, Data & Analytics Section, Computational Analytics and Geospatial Intelligence Unit

A map of the country

Challenge background

GeoSight is UNICEF's open-source geospatial web-based data visualization and analysis platform designed to make geospatial data easily accessible and shareable in support of risk-informed programming. It utilizes administrative reference datasets from GeoRepo to display data. Projects in GeoSight combine different elements such as reference datasets, indicators, and context layers, enabling comprehensive data visualization and analysis for end-users.

Effective disaster preparedness requires not only access to data but also intuitive visualization and scenario planning capabilities. While GeoSight provides a solid foundation for geospatial analysis, it lacks specific functionality for working with natural hazard datasets, such as hurricanes and interactive "what-if" scenario modelling that would enable decision-makers to anticipate impacts on children and plan interventions before disasters strike.

This hackathon aims to enhance GeoSight's functionality by implementing new features that improve data analysis, visualization, and user interaction. Participants will collaborate to develop these features, contributing to a more robust and user-friendly platform.

Skills & Tools

Skills:

Essential:

  • Experience in Python web development
  • Managing data models with Django ORM
  • Strong knowledge of React for building dynamic and responsive UI
  • HTML, CSS, and JavaScript to create and style modern web applications
  • Using Git and GitHub for version control, branching, and collaborative development

Optional:

  • Proficiency in Django for developing RESTful APIs and managing server-side architecture
  • Experience with PostgreSQL / PostGIS, including writing queries and managing relational databases
  • GeoDjango for spatial models and queries
  • Basic understanding of GIS concepts
  • Familiarity with web mapping frameworks (preferably MapLibre)
  • CI/CD pipelines or GitHub Actions

Tools:

Essential:

  • Django – backend framework
  • React – frontend library
  • Git & GitHub – version control and collaboration

Optional:

  • PostgreSQL + PostGIS – spatial database
  • GeoDjango – Django's built-in module for managing spatial data and performing geospatial queries using the ORM
  • Django REST Framework – API development
  • MapLibre – for displaying interactive maps
  • H3 – hexagonal spatial indexing (via h3-py)

Useful Links:

Challenge description

There are several documented enhancements ready for development during the hackathon, for example:

  1. Improvements to Summary Group Widget
  2. Add another compare mode: swipe
  3. Implement Dynamic H3 Binning for Cloud Native Point Layers
  4. Add Support for H3 Hexagonal Indexing of Indicator Data
  5. Improve dashboard UI for mobile screens

You will find more feature requests which you can find on our GitHub Repository: https://github.com/unicef-drp/GeoSight-OS/issues?q=is%3Aissue%20state%3Aopen%20label%3AUN-OS-Week-2025

We also encourage participants to think outside the box. If you have an idea for a feature or improvement that could make GeoSight more powerful, user-friendly, or impactful, bring it to life!

This is your opportunity to:

  • Brainstorm new features or tools that enhance geospatial data analysis
  • Solve challenges users might face in the field
  • Build integrations or UI/UX improvements that increase accessibility and usability

Whether you're refining the core or imagining something entirely new—your contribution matters. Innovation often comes from fresh perspectives, so don't hesitate to explore ideas beyond the existing issues!

Feature Request 1: Enhancements to Summary Group Widget

Objective: Enhance the Summary Group Widget in GeoSight dashboards to support more flexible, insightful data aggregation and quick data insights—empowering emergency response and preparedness actions by quickly identifying and prioritizing areas of greatest need during disasters or crisis situations.

Estimated time: 8-12 hours

Issue: https://github.com/unicef-drp/GeoSight-OS/issues/136

Tasks:

  • Add additional aggregation functions: MIN, MAX, AVG, COUNT, COUNT_UNIQUE.
  • Enable grouping values by administrative names.
  • Implement sorting options by geographical names and allow ascending/descending order.
  • Introduce a "Top N" filter to display only the top N records based on specific indicators.

Impact: These enhancements to the Summary Group Widget will significantly improve the analytical power and usability of GeoSight dashboards, especially during emergencies where rapid, targeted insights are essential. Emergency managers will be able to instantly surface critical patterns—like the most affected districts or areas with the lowest access to services—without needing to manually sift through raw data. This empowers teams to make faster, data-informed decisions, prioritize interventions, and communicate key findings clearly to both field teams and stakeholders.

Feature Request 2: Add another compare mode: swipe

Objective: Introduce a Swipe Comparison Mode in GeoSight to enhance the ability to visually analyze and compare multiple hazard-related layers, such as hurricanes, floods, droughts, or landslides.

This feature will allow users to interactively swipe between two overlaid map layers, making it easier to identify spatial relationships and overlaps—such as regions affected by multiple hazards or correlations between hazard impact and vulnerability indicators (e.g. access to schools, health services).

By enabling side-by-side comparisons within a single map view, this tool will support more nuanced analysis of complex emergency contexts, facilitate better communication of risk scenarios, and strengthen decision-making for preparedness, response, and recovery planning.

Estimated time: 8-12 hours

Issue: https://github.com/unicef-drp/GeoSight-OS/issues/37

Tasks:

  • UI/UX Enhancements
    • Add a dropdown menu under the existing "Compare Mode" button to switch between:
      • Outline & Fill Mode (existing)
      • Swipe Mode (new)
    • Design and implement a split-view interface with a vertical swipe bar that separates two layers.
  • Layer Selection Logic
    • Allow users to select up to two indicator layers for comparison in Swipe Mode.
    • Ensure consistent logic with existing compare mode for selecting and displaying indicator layers.
  • Map Integration
    • Integrate the MapLibre GL Compare plugin (maplibre-gl-compare) or similar tool for enabling swipe functionality.
    • Ensure both sides of the swipe view render independently and clearly (left = layer A, right = layer B).
  • Legend & Widget Updates
    • Display both legends side by side or toggle between them as needed.
    • Ensure relevant widgets reflect and synchronize with both selected indicators.
  • Performance Optimization
    • Optimize rendering for large datasets to maintain smooth interaction with the swipe tool.
    • Ensure mobile responsiveness and cross-browser compatibility.
  • Accessibility & UX Polishing
    • Ensure swipe handle is intuitive and easily draggable.
  • Testing & Validation
    • Conduct functional testing with different hazard datasets (e.g., hurricane and flood layers).
    • Validate accuracy of layer rendering and interactivity in various scenarios.
  • Documentation & Support
    • Update user documentation to explain how to use Swipe Mode.
    • Provide example use cases, especially for hazard layer comparisons.

Impact: The addition of the Swipe Comparison Mode will significantly enhance GeoSight's utility for hazard visualization and analysis, particularly in emergency response, risk assessment, and preparedness planning. By enabling users to visually compare two geospatial layers side by side, this feature will:

  • Improve situational awareness by allowing users to easily identify overlaps between hazard layers (e.g. hurricane paths vs. flood-prone zones).
  • Support multi-dimensional analysis, such as comparing hazard exposure with indicators of vulnerability (e.g. population density, access to health or education).
  • Enable rapid decision-making in crisis contexts by offering an intuitive, visual way to interpret complex spatial relationships.
  • Enhance storytelling and communication, especially for non-technical stakeholders who need to understand and act on data quickly.
  • Increase user engagement and insight generation by offering a more interactive and explorative approach to data comparison.
  • Ultimately, this feature empowers users to gain deeper insights from hazard data, helping humanitarian and development teams make more informed, timely, and targeted decisions.

Feature Request 3: Implement Dynamic H3 Binning for Cloud Native Point Layers

Objective: Enable users to dynamically aggregate and visualize point-based data—such as facilities, incidents, or population data—into H3 hexagonal grids within GeoSight, allowing for clearer spatial pattern analysis, improved performance with large datasets, and more actionable insights in emergency and hazard response scenarios.

Estimated time: at least 12 hours

Issue: https://github.com/unicef-drp/GeoSight-OS/issues/447

Tasks:

  • Add support for the H3 spatial indexing library (e.g., Uber's H3) in the platform's backend.
  • Implement logic to compute H3 indices for each point in a Cloud Native Point Layer based on dynamic resolution adjusted to support different levels of spatial granularity.
  • Enable on-the-fly aggregation of point attributes within H3 cells using common statistical functions (e.g., count, sum, average, max).
  • Auto-generate and update the Mapbox Style JSON to reflect selected aggregation and color scheme (e.g., continuous or stepped).
  • Create a user-friendly UI panel where users can select aggregation method, choose H3 resolution, pick color style (continuous or stepped) and see real-time preview of changes.

Impact: The addition of dynamic H3 binning for point layers will be a powerful tool in emergency contexts, especially when working with natural hazard data such as hurricanes. During emergencies, decision-makers need to quickly interpret large volumes of geospatial data—such as the location of shelters, clinics, damaged infrastructure, or population clusters. By aggregating these point datasets into H3 hexagonal grids, GeoSight can provide a clearer and more scalable visual representation of impact areas and resource distribution, even in densely populated or data-heavy regions. Unlike administrative boundaries, H3 offers a uniform spatial unit, ensuring consistent analysis across regions and zoom levels.

Feature Request 4: Implement H3 Hexagonal Indexing for Indicator Data

Objective: Enable GeoSight to support indicator data using H3 hexagonal spatial indexing to provide more flexible and precise mapping of critical indicators—such as population vulnerability, infrastructure access, and hazard exposure—in support of real-time emergency preparedness and disaster response.

Estimated time: at least 12 hours

Issue: https://github.com/unicef-drp/GeoSight-OS/issues/447

Tasks:

  • Enable Storage of Indicator Data by H3 Tile ID
    • Modify data schema to support importing and storing indicator values linked to H3 indexes rather than admin boundaries.
  • Implement H3-Based Data Import Workflow
    • Allow users to upload datasets where each value is georeferenced to an H3 tile, including validation and tile resolution selection.
  • Develop Visualization Engine for H3 Tiles
    • Render indicator layers using H3 tiles on the map, supporting existing styling options (such as dynamic or classified color schemes).
  • Integrate H3 into Project/Dashboard Workflows
    • Allow users to create dashboards using H3-referenced indicator layers.
  • Optimize Performance for On-the-Fly Tiling
    • Ensure scalable rendering and data handling across varying zoom levels without preloading large shapefiles or geometries.

Impact: Integrating H3 support into GeoSight will dramatically enhance the platform's ability to analyze and visualize data in emergency scenarios, especially where events like hurricanes, floods, or earthquakes do not align with administrative boundaries. It allows for real-time, area-agnostic spatial analysis, offering uniform spatial units that enable consistent and scalable insights across regions. This reduces reliance on potentially outdated or politically sensitive boundaries, improves clarity for decision-makers, and enables faster, more precise targeting of resources and response efforts when every minute counts.

Feature Request 5: Improve dashboard UI for mobile screens

Objective: Redesign the GeoSight dashboard UI to be fully responsive on mobile devices, ensuring that field teams, emergency responders, and decision-makers can seamlessly access, interpret, and act on geospatial data from any device during critical situations.

Estimated time: at least 12 hours

Issue: https://github.com/unicef-drp/GeoSight-OS/issues/38

Tasks:

  • Implement Responsive Dashboard Layout
    • Redesign layout components to adapt gracefully to small screens (e.g., stacked views, collapsible panels, fluid grids).
  • Create Tabbed or Menu-Based Navigation for UI Components
    • Move key dashboard sections—such as Context Layers, Indicator Layers, Filters, Map, and Widgets—into mobile-friendly tabs or a hamburger menu.
  • Optimize Map and Widget Views for Mobile Interaction
    • Ensure the map and data widgets are touch-friendly and clearly legible on smaller screens, with responsive resizing and scroll support.
  • Test Across Devices and Screen Sizes
    • Conduct usability testing across a range of mobile devices and browsers to ensure consistent behavior and performance.

Impact: A mobile-responsive GeoSight dashboard dramatically increases the reach and utility of the platform in emergency and field settings, where laptops may not be readily available. By ensuring intuitive and readable layouts on smartphones, this enhancement allows emergency teams, program managers, and frontline workers to access critical geospatial insights on the go—whether assessing damage, identifying high-risk zones, or coordinating resources in real time. It enhances agility, improves accessibility, and ensures that data-driven decisions can be made anywhere, anytime.

Feature Request 6: Expand Support for Uploading Cloud Optimized GeoTIFFs (COG)

Objective: Expand the file upload and processing pipeline to support Cloud Optimized GeoTIFFs (COGs) context layers, which are a common raster format that enables fast, scalable access to large geospatial datasets – such as satellite imagery and climate grids – directly over HTTP, making them ideal for web-based Earth observation and GIS applications.

Estimated time: 8-12 hours

Issue: https://github.com/unicef-drp/GeoSight-OS/issues/481

Tasks:

  • Add support for COG files for Cloud Native GIS context layers
  • Enable upload of COG files
  • Validate file format and contents on upload
  • Provide user feedback on supported formats and size limits
  • Auto-generate layer previews and metadata summaries (e.g. CRS – coordinate reference systems, geometry type, bounds, layer type)
  • Store COGs on object storage (or file system), with metadata indexed in database
  • Use STAC-like metadata schema for previewing and querying
  • Enable preview of COG layers along with styling (color ramp, pixel value classification)
  • Enable generating tiles for faster performance and serving as tile layer using a tile server (e.g. titiler, GDAL WMS, or dynamic endpoint)

Impact: Supporting modern cloud-native raster formats such as COGs will:

  • Improve interoperability and user experience
  • Enable native raster layer uploads (e.g. satellite imagery, elevation models, high resolution population rasters) which are essential datasets for weather and climate related analysis
  • Future-proof our platform for scalable, cloud-native workflows
  • Maximize developer efficiency by building on top of the existing cloud-native layer infrastructure

Proposed Datasets

Sample and test datasets are provided in respective GitHub tickets.

Multimedia Material

N/A

Further Information

Participants are strongly encouraged to familiarize themselves with the GeoSight platform and its documentation ahead of the hackathon. Understanding how the platform works will give you a significant head start and allow you to dive straight into building and collaborating during the event.

To make the most of your time at the hackathon, we recommend:

  • Deploying a local instance of GeoSight on your machine to identify and troubleshoot any deployment issues and to explore GeoSight features hands-on
  • Reviewing the codebase on GitHub to understand the architecture, components, and development workflow
  • Identifying the areas you're most interested in contributing to—whether it's UI/UX, data visualization, back-end services, or integrations

We'll also be organizing optional prep sessions before the hackathon to walk through:

  • How GeoSight is structured
  • How to set up the platform locally
  • Where and how to contribute code
  • Tips for working with the existing tech stack

These sessions are a great chance to ask questions, connect with fellow participants, and get comfortable with the tools before the main event. Collaboration and innovation are key—let's work together to enhance geospatial insights and build solutions that drive meaningful impact!

Back to the top