Verdian Ag: Technical Brief & Methodology

1. Executive Summary

Verdian Ag provides agricultural intelligence infrastructure. Our primary objective is to bridge the data gap that prevents institutional capital from reaching smallholder farmers.

Traditional agricultural lending relies on collateral and credit history. Assets that rural farmers often lack. Verdian Ag replaces this requirement with verifiable, continuous agronomic data. By processing multi-spectral satellite imagery and localized weather data, we model field-level crop stress, vigor, and compliance with recommended agronomic practices.

This processing pipeline outputs the Verdian Score, a structured performance metric designed to support risk assessment. For financial institutions, this translates opaque farming operations into transparent, monitorable assets. For farmers, it provides a data-backed record of their reliability and yield potential.

2. The Verdian Score Architecture

The Verdian Score is a composite metric (0-100) intended to help lenders and partners make informed decisions. It is not a financial credit score; it is an agronomic performance and behavioral compliance score.

The score is synthesized from three primary indices:

1. Vigor & Biomass Index: Derived from continuous multi-spectral satellite observation, measuring the physiological health and growth stage of the crop.
2. Moisture & Stress Index: Evaluated by estimating soil moisture patterns, evapotranspiration rates, and historical drought stress penalties.
3. Behavioral Compliance Ratio: A measure of a farmer's adherence to localized agronomic advisories (e.g., timeliness of planting, responsiveness to irrigation alerts), acting as a proxy for operational reliability.

3. Data Ingestion & Processing

Verdian Ag operates entirely remotely, relying on publicly available and commercially accessible datasets rather than expensive on-site IoT sensors.

3.1 Satellite Analytics Engine

We ingest continuous optical and synthetic-aperture radar (SAR) data. This multi-modal approach ensures consistent data collection regardless of cloud cover.

Our preprocessing pipeline includes:

• Atmospheric Correction: Removing cloud shadows and atmospheric interference from optical imagery.
• Incidence Angle Normalization: Standardizing radar backscatter coefficients to ensure accurate structural modeling across different satellite passes.
• Spatial Aggregation: Processing pixel-level data precisely within the geofenced boundaries of individual smallholder farms.

3.2 Climate and Operations Data

The satellite baseline is enriched with:

• Meteorological APIs: Ingesting historical rainfall data, 14-day forecasts, and temperature extremes.
• Ground-Truth Operational Logs: Data submitted by farmers or extension agents via our mobile interfaces, including planting dates, fertilizer application, and soil type assessments.

4. Interactive Field Mapping

The Verdian dashboard incorporates a responsive, interactive satellite mapping interface built on Leaflet with Google Maps tile layers. This component provides real-time visualization of monitored field boundaries, health status, and geographic context.

4.1 Dynamic Zoom Behavior

The mapping interface implements intelligent zoom management:

• Initial View: On dashboard load, the map automatically fits all registered fields within the viewport using Leaflet's fitBounds method with 20px padding.
• Field Selection: When a user selects a field from the bottom tab interface or clicks directly on a field polygon, the map animates to that field's precise geofenced boundaries.
• Zoom Constraints: Field-specific zoom operations use 50px padding with a maximum zoom cap of level 16, ensuring small fields remain visually contextualized rather than filling the entire screen.
• Adaptive Scaling: Zoom levels automatically adapt to field geometry and distribution—clustered fields maintain higher zoom for detail examination, while geographically dispersed operations trigger wider contextual views.

4.2 Health Visualization Layer

Field boundaries render with dynamic styling based on continuous NDVI monitoring:

• Color Coding: Emerald for healthy (NDVI > 0.7), amber for warning (NDVI 0.5-0.7), orange for stress (NDVI 0.3-0.5), rose for critical (NDVI < 0.3).
• Selection State: Selected fields receive enhanced border weight (4px vs 2px) and slate-800 coloring for clear visual distinction.
• Layer Controls: Users can toggle between satellite, hybrid, terrain, and street map baselines, with optional NDVI heatmap overlay.

5. Agronomic Modeling

Rather than simply displaying raw satellite indices (like NDVI), Verdian Ag processes this data through specialized agronomic models to estimate specific stress factors.

5.1 Water Stress Modeling

We model soil moisture depletion by tracking precipitation inputs against estimated crop evapotranspiration (ETc). When the moisture level drops below the crop's specific permanent wilting point, the model applies a cumulative water stress penalty to the baseline yield forecast.

5.2 Biomass and Yield Estimation

We estimate biomass accumulation using established algorithms (such as adapting principles from the FAO AquaCrop model). By tracking the actual canopy cover growth curve against an idealized logistic S-curve for that specific crop and climate zone, we can estimate discrepancies in vigor and adjust end-of-season yield projections accordingly.

6. Technology Access & Delivery

To ensure our intelligence reaches the environments where it is needed most, Verdian Ag utilizes low-bandwidth, high-accessibility delivery channels.

6.1 The Offline-First PWA

Extension agents and cooperative managers utilize our Progressive Web App (PWA). This application is heavily optimized with Service Workers and IndexedDB, allowing agents to map geofenced field boundaries and log baseline data in areas with zero internet connectivity. Data synchronizes automatically when the device reconnects to a network.

6.2 Conversational Advisory (WhatsApp AI Agent)

Farmers receive localized agronomic advisories directly via WhatsApp. By integrating natural language processing (NLP), the AI Agent allows farmers to interact conversationally—including via audio voice notes for low-literacy users. This channel delivers timely alerts (e.g., "High heat stress predicted in 3 days, irrigate if possible") and allows farmers to easily log critical events like planting or harvesting.

7. Representative Output Schema

To support institutional integration, Verdian Ag structures its field intelligence into standardized data payloads.

(Note: This is a representative schema intended to illustrate the structure and types of data our models generate for pre-verification and risk assessment purposes.)

{
  "field_id": "uuid-v4-string",
  "assessment_date": "2026-02-22T00:00:00Z",
  "crop_type": "maize_white",
  "metrics": {
    "verdian_score": 82,
    "confidence_level": 0.88,
    "vigor_index": {
      "current_value": 0.65,
      "7_day_trend": "stable"
    },
    "water_stress": {
      "status": "nominal",
      "days_since_rain": 4,
      "estimated_depletion_percent": 35
    },
    "compliance_ratio": 0.90
  },
  "flags": [
    {
      "type": "heat_stress_warning",
      "severity": "medium",
      "predicted_onset": "2026-02-25"
    }
  ]
}