EcoPlanar-Chain: Autonomous Geospatial AI & dMRV

Autonomous geospatial AI compiling natural language into urban eco-plans & on-chain carbon dMRV.

Comment

πŸ’‘ Inspiration

EcoPlanar-Chain was engineered directly to resolve four documented systemic failures exposed by statutory audits, investigative journalism, and peer-reviewed climate research:

  1. The Carbon Credit Integrity Crisis (The Guardian / Nature Communications): A 2023 investigation revealed that 94% of rainforest offset credits issued by Verra had zero empirical climate benefit due to unverified counterfactual baselines. A 2024 Nature Communications study corroborated systemic baseline overcrediting.
  2. Climate-Risk Telemetry Infrastructure Collapse (CAG Karnataka Audit): An official performance audit by the Comptroller and Auditor General of India (CAG) revealed that 49 out of 100 installed water-level sensors on Bengaluru stormwater drains were non-functional during audit inspection in December 2023, causing predictive flood models to collapse.
  3. Municipal DPR Forfeiture & Ghost Plantations (CAG Report No. 2 of 2021 & Maharashtra Audit): CAG audit records confirm that BBMP forfeited β‚Ή83.59 crore in central grant assistance under JNNURM due to missing or deficient Detailed Project Reports (DPRs) and cadastral unclarity. Simultaneously, a CAG audit on Maharashtra's Green India Mission (Indian Express, Oct 2026) revealed that despite 49,500 saplings recorded on paper, physical audit found only 116 saplings on ground (a 99.76% ghost plantation rate).
  4. Rapid Impervious Urbanization (Elsevier ScienceDirect / Catena Study): Peer-reviewed research in Mumbai's Oshiwara River Basin proved that uncontrolled urban encroachment drove a 74.84% surge in built-up area and a 42.8% loss in natural open space, causing a 22.27% expansion in extreme flood hazard zones.

We built EcoPlanar-Chain to solve both crises simultaneously: an autonomous Neuro-Symbolic GIS Compiler that translates natural language goals into physically feasible, budget-bounded spatial interventions (GeoJSON) in under 45 seconds, while binding every ecological plan to on-chain cryptographic evidence, anti-double-claiming centroid nullifiers, and automated parametric green bonds.

⚑ What It Does

EcoPlanar-Chain provides an end-to-end platform for municipal town planners, carbon auditors, and Web3 green bond underwriters:

  1. Global Location Geocoding & Boundary Delineation:

    • Planners search any global location or select built-in disaster presets (Thane Naupada, Bengaluru Bellandur, Mumbai Kurla, Istanbul KadΔ±kΓΆy).
    • Draw custom site polygons on an open-source MapLibre GL WebGL canvas with real-time hectare and perimeter calculation.

  2. Automated Multi-Source Satellite & Telemetry Ingestion (Track 4):

    • Instantly ingests Copernicus Sentinel-2 NDVI (vegetation health), NASA SRTM 30m DEM (elevation and slope percent), Open-Meteo ERA5 (monsoon precipitation and heat island delta), and OpenStreetMap (OSM) (existing building footprints and road networks).

  3. Neuro-Symbolic Ecological Plan Compilation (Zero Geometric Hallucination):

    • Users express complex objectives in natural language: "Reduce monsoon flooding first, mitigate heat island effects, increase native biodiversity without demolishing existing structures under a $600,000 USD budget."
    • A dual-engine pipeline pairs Gemini 3.8 Flash + Groq 600ms failover (intent extraction) with Turf.js computational geometry to generate candidate intervention geometries (bioswales, Miyawaki pocket forests, permeable pavements, green roofs, and pollinator corridors) that mathematically subtract building footprints (turf.difference), ensuring zero building collision.

  4. Multi-Objective Scenario Pareto Optimization:

    • Compiles and compares three optimized masterplans:
      • Flood-First: Maximizes stormwater retention capacity ($m^3/\text{storm}$).
      • Biodiversity-First: Maximizes native allometric biomass carbon sequestration and pollinator connectivity.
      • Balanced: Optimal multi-criteria equilibrium under strict budgetary bounds.

  5. Monsoon Stress & Cloudburst Simulator (40mm - 250mm):

    • Real-time hydrodynamic stress slider allowing engineers to simulate 40mm normal storms up to 250mm extreme cloudbursts, calculating live stormwater infiltration (millions of liters absorbed) and peak discharge rate attenuation ($m^3/s$).

  6. Statutory Municipal Detailed Project Report (Form DPR-ENV-2026):

    • Generates an official, publication-grade municipal DPR with executive summaries, site diagnoses, an itemized Bill of Quantities (BOQ) using PWD Schedule of Rates (SOR) unit pricing, and zero demolition guarantees pre-qualifying for AMRUT 2.0 (Component 2.1) and Nagar Van Yojana (MoEFCC).

  7. Web3 Ledger, Cryptographic Provenance & Parametric Green Bonds (Track 1):

    • RFC-8785 Canonical Serialization & Keccak-256 Hashing: Deterministic JSON Canonicalization Scheme (JCS) producing identical Keccak-256 digests across all runtimes.
    • Anti-Double-Claiming Geographic Centroid Nullifier: Cryptographic nullifier (keccak256(latE6, lngE6)) in EcoEvidenceRegistry.sol preventing fraudulent duplicate carbon credit registrations on the same land parcel.
    • Pinata IPFS Pinning: Permanently anchors GeoJSON and DPR artifacts to decentralized storage (CIDv1).
    • Tokenized Carbon Offsets (UrbanCarbonCredit.sol): Programmatically mints ERC-20 UCC tokens (1 UCC = 1 metric ton $\text{CO}_2\text{e}$ sequestered) upon satellite verification, with on-chain retirement mechanisms.
    • Parametric Green Bonds (ParametricGreenBond.sol): Escrows municipal capital and programmatically releases funding in tranches (30% site prep, 40% installation, 30% performance) tied to empirical satellite proof.

πŸ› οΈ How We Built It

  • Frontend & GIS Engine: Next.js 16.3 (Turbopack), React 19.2, TypeScript 5, Tailwind CSS 4, MapLibre GL 6.10, and Turf.js 7.4.
  • Cognitive Intent Parser: Dual-Tier AI (Gemini 3.8 Flash + Groq 600ms failover) with structured JSON schema enforcement.
  • Environmental Telemetry: Open-Meteo ERA5 Reanalysis API, NASA SRTM 30m elevation rasters, Copernicus Sentinel-2 NDVI, and OpenStreetMap Overpass API.
  • Smart Contract Protocol (Solidity 0.8.20 on Polygon Amoy β€” Chain ID: 80002):
    • EcoEvidenceRegistry.sol: Canonical hash anchoring, plan metadata, and geographic centroid collision nullifiers (0x35688536f87C414c77c6a9c1A3c847d066922F26).
    • UrbanCarbonCredit.sol: ERC-20 tokenized carbon offsets with retirement logging (0x8842Af529D7E842187b5a5B49d863f6C5110E9e1).
    • ParametricGreenBond.sol: 3-tranche milestone release escrow contract (0x497eA73Ff52b61D1537e2D48c7A5Ac387D344B60).
  • Cryptographic Provenance: RFC-8785 Canonical JSON Serialization, Keccak-256 preimages, and Pinata IPFS CIDv1 pinning.

πŸ§— Challenges We Ran Into

  1. Eliminating LLM Geometric Hallucinations: Standard LLMs generate distorted GeoJSON with overlapping polygons or trees placed on top of commercial buildings. We overcame this by implementing a Neuro-Symbolic architecture: Gemini extracts high-level semantic intent and weights, while deterministic Turf.js spatial algorithms synthesize and clip candidate geometries against real OpenStreetMap cadastre with zero structural collision.
  2. Preventing Carbon Credit Double-Claiming on Public Land: In urban carbon projects, multiple entities could attempt to register the same public park or municipal corridor. We designed a cryptographic Centroid Nullifier Registry in EcoEvidenceRegistry.sol: each site's coordinate centroid is scaled by $10^6$ and hashed into a unique nullifier key. Any attempt to register an overlapping plan reverts on-chain.
  3. Deterministic RFC-8785 Serialization Across Runtimes: Standard JSON.stringify does not guarantee key ordering across different browser engines and server runtimes, resulting in divergent hashes. We built a custom recursive RFC-8785 canonicalizer that sorts all keys in strict lexicographical Unicode order prior to Keccak-256 digesting.

🌟 Accomplishments That We're Proud Of

  • Sub-45 Second End-to-End Planning: From a natural language prompt to a multi-scenario, budget-bounded GeoJSON plan and municipal DPR in under 45 seconds (replacing 6-month consulting cycles).
  • 10/10 Automated Verification Tests Passing: Complete mathematical and cryptographic verification executed via automated acceptance suites with zero failures.
  • Zero TypeScript Errors & Clean Build: Clean production compilation across Next.js 16 and React 19 (npm run build).
  • Dual-Track Synergy: Seamlessly bridges Track 4 (Satellite & Environmental Telemetry) into Track 1 (On-Chain Carbon Markets & Parametric Green Bonds).

πŸ“š What We Learned

  • Designing parametric smart contracts that directly consume environmental telemetry requires robust nullifier schemes to prevent replay and double-spending of real-world ecological assets.
  • Pairing LLMs with deterministic topological constraint solvers (Neuro-Symbolic design) is vastly superior to prompt-only pipelines for mission-critical geospatial engineering.

πŸš€ What's Next for EcoPlanar-Chain

  1. Municipal Ward Pilots: Partnering with municipal development authorities across Maharashtra and Karnataka for local ward sponge-city pilots.
  2. Automated Drone LiDAR & Drone-Based MRV: Integrating drone point-cloud photogrammetry for sub-centimeter canopy height verification.
  3. Decentralized Carbon Retirement Marketplace: Enabling multinational corporations subject to SEBI BRSR and EU CSDDD mandates to retire verified Urban Carbon Credits (UCC) directly from municipal treasuries.

Built With

Share this project:

Updates

Submission history