County Intelligence Platform
A database can tell you Terrebonne Parish's flood risk score.
This can answer something harder.

"Which counties have Very High wildfire risk, a majority of ALICE households, and no trauma center within 60 miles — and have already received a federal disaster declaration in the last five years?"

A knowledge graph doesn't just store data — it stores the relationships between data. That's what makes that question answerable. This is a research prototype connecting every US county's risk profile, vulnerability measures, and history in a single queryable graph.

Independent Research · Not an official Red Cross application

County Intelligence Platform

3,232 county documents. Every hazard. Every risk index. Queryable with plain English.
3,232
County docs
18
Hazard types
5
Risk indices
All
Counties
Hazards
Indices
Click · Zoom · Drag
· connections

A knowledge graph of every US county — queryable with plain English

Every US county has been encoded into a graph database connecting hazard risks, social vulnerability, economic hardship, health outcomes, and disaster declaration history. Instead of searching spreadsheets, you ask a question and get a synthesized answer grounded in that data.

Layer 1 — Raw data

Source Documents

FEMA NRI, CDC SVI, CDC PLACES, Census ACS, FEMA Disaster Declarations — one structured document per county

Layer 1 — Raw data

Hazard Coverage

18 hazard types per county: wildfire, hurricane, flood, tornado, earthquake, drought, hail, ice storm, strong wind, and more

Layer 2 — Intelligence engine

LightRAG on Railway

Extracts entities and relationships. Builds the graph: county → hazard → risk index → health indicator → declaration history. Running live at explorer.jbf.com

Layer 3 — Query interface

AI Query Layer

Plain English question → graph traversal + vector search → synthesized answer with citations. Any app can call the API.

The Knowledge Graph tab is a live visualization of the graph structure. Each node is an entity. Each edge is a relationship. Click any node to explore its connections.

The technology stack, layer by layer

01
Document Ingestion
3,232 county profiles loaded into LightRAG

Each county document contains structured data from five federal sources: FEMA National Risk Index (risk scores for 18 hazard types), CDC Social Vulnerability Index, CDC PLACES health outcomes, Census ACS socioeconomic indicators, and FEMA Major Disaster Declaration history. LightRAG ingests these as text and extracts entities and relationships automatically using an LLM.

02
Graph Construction
Entities and relationships extracted and indexed

LightRAG reads each document and extracts named entities (county names, hazard types, index names) and the relationships between them. These become nodes and edges in a graph database. When you see Terrebonne → Hurricane in the graph, that's a stored relationship with associated risk data. "Degree 19" means a county has 19 direct connections — one for each hazard type and risk index it's linked to.

03
Hybrid Retrieval
Graph traversal + vector similarity search combined

When you ask a question, LightRAG runs two searches simultaneously: (1) traverses the graph to find relevant entities and their neighbors, and (2) runs vector similarity search over the raw document chunks. Combining both results is what allows it to answer specific factual questions ("what is Terrebonne Parish's hurricane risk?") and complex comparative questions ("which Louisiana parishes have both high flood risk and low healthcare access?").

04
AI Synthesis
Claude generates grounded answers from retrieved context

The retrieved graph context is passed to Claude with your question. The model synthesizes a natural language answer from the graph data — not from general training knowledge. This is Retrieval Augmented Generation (RAG): the AI is grounded in your specific dataset. Answers include citations back to source counties and federal datasets.

05
API + Spatial RAG
Any app can query the graph — including ArcGIS maps

LightRAG exposes a REST API. Any application can send a natural language question and receive a structured answer. The next evolution is Spatial RAG: connecting this graph to ArcGIS maps so a disaster planner can click a county on a map and ask "what's this county's compounded risk profile?" — answered in seconds by the knowledge graph, displayed directly in the map interface.

What you can actually ask

Select a query type to see a representative answer grounded in the county graph.

Query answer
Identifying counties where FEMA NRI composite risk score exceeds 70 and CDC SVI score exceeds 0.75: the platform surfaces 147 counties meeting both thresholds. Top clusters appear in the Mississippi Delta, Appalachian coalfields, and Rio Grande Valley — areas with elevated hazard exposure and populations with limited capacity to recover. These are priority counties for pre-positioning resources and community resilience investment.
FEMA NRI CDC SVI 2022 Census ACS 2022

These are illustrative examples, not live query results. Try it yourself at explorer.jbf.com — Retrieval tab.

What this unlocks for disaster response

Chapter boundary decisions

Compare reorganization scenarios by the aggregate risk and vulnerability profile of counties in each scenario. Data-driven, not politics-driven.

Pre-positioning resources

Identify counties with high compound risk entering storm season. Get a ranked list with reasoning, not just a map of dots.

Gap analysis

Find counties with repeated major disaster declarations but no nearby chapter capacity. Surface blind spots before a disaster hits.

Mutual aid planning

When disaster strikes, find neighboring counties with similar risk profiles that may be co-impacted — and which have capacity to send aid.

Leadership briefings

"Which counties in our region are most at risk this hurricane season?" — 30-second answer instead of a week of analyst work.

Spatial RAG (next)

Connect this graph to ArcGIS maps. Click a county, ask a question, get an AI answer grounded in the graph — directly in the map interface.

Why a knowledge graph — not a database

This is a knowledge graph of every US county — queryable with plain English. A database can tell you Terrebonne Parish's flood risk score. This knowledge graph can answer: "Which counties have Very High wildfire risk, a majority of ALICE households, and no trauma center within 60 miles — and have already received a federal disaster declaration in the last five years?"

A knowledge graph doesn't just store data — it stores the relationships between data. That's what makes the question above answerable. This is a research prototype exploring what becomes possible when every US county's risk profile, vulnerability, and history are connected in a single queryable graph. The data types below represent the architecture — not a product claim.

Structured data — the numbers

Federal indices encode risk as scores, ranks, and thresholds. Each county document contains these as structured fields — precise, comparable, and directly queryable. When the graph connects FEMA NRI to CDC SVI, it's revealing that the same counties that score high on hazard risk also score high on social vulnerability — a compounding effect invisible in either dataset alone.

FEMA National Risk Index
18 natural hazard risk scores per county — the authoritative federal baseline

FEMA's NRI scores every US county on 18 natural hazards — wildfire, hurricane, flood, tornado, earthquake, drought, and more. Each score combines expected annual loss with social vulnerability and community resilience. It's the primary federal answer to "how dangerous is this county?" and directly informs Red Cross regional planning and grant allocation decisions.

CDC Social Vulnerability Index
Which communities have the least capacity to recover

The CDC SVI measures a community's ability to withstand and bounce back from external stresses. It combines 16 Census variables across socioeconomic status, household composition, minority status, and housing type. High SVI + high FEMA NRI = the counties where disasters hit hardest and recovery takes longest. That intersection is where Red Cross resources matter most.

CDC PLACES + Census ACS
Health outcomes and socioeconomic conditions at the county level

CDC PLACES provides chronic disease prevalence for every county — asthma, diabetes, heart disease. A county with high wildfire risk and high asthma rates has a population acutely vulnerable to smoke. Census ACS adds income, housing age, vehicle access, and language isolation — the factors that determine whether a household can evacuate, shelter-in-place, or self-recover.

ALICE (United Way)
The working poor — invisible to most federal metrics

ALICE households earn above the poverty line but cannot afford basic necessities. They don't qualify for most aid programs. They can't absorb a week without income. They can't self-evacuate without a car or savings. United Way's ALICE data fills the gap between "in poverty" and "doing fine" — revealing the hidden fragility of working households in high-risk counties. This is the dataset that changes the resource allocation argument.

Narrative data — the context

Numbers tell you a county's score. Narrative tells you why it scores that way — and what it means on the ground. LightRAG's graph-enhanced retrieval is specifically designed to connect structured and narrative data, enabling queries that neither a spreadsheet nor a traditional search engine can answer.

Wikipedia County Articles INGESTION IN PROGRESS
History, economy, geography, and notable events for all 3,232 counties

A FEMA score doesn't tell you that Terrebonne Parish has lost 500 square miles of coastline since 1930, that Butte County's Camp Fire destroyed the entire town of Paradise, or that one Iowa county is home to the Meskwaki Nation's Settlement. Wikipedia county articles add the narrative layer that transforms risk scores into contextual intelligence — and demonstrate why graph-enhanced RAG outperforms pure vector search for complex humanitarian questions.

On the roadmap

Each addition compounds the graph's intelligence. The goal is not more data — it's more meaningful connections between the data that already exists.

Future risk

NOAA Climate Projections

Where hazard risk is expanding. Planning against tomorrow's profile, not yesterday's.

Real-time demand

211 Call Data

Where ALICE households are in crisis right now. Ground-truth need signals.

Medical access

HRSA Shortage Areas

Counties with high risk and no nearby hospital. Triple-threat disaster scenarios.

Fire incidents

NFIRS Fire Data

National fire incident history. Structure fires disproportionately hit ALICE households.