Table of Contents

1. The Industry Challenge

Over 2,600 GW of solar and storage capacity sits in US interconnection queues. Yet the tools used to evaluate these sites haven’t kept pace — developers still rely on manual field surveys, consultant-dependent risk ratings, and 3–6 month timelines costing $50K–$150K per site. By the time traditional assessments are complete, interconnection positions have been lost and capital deployment windows have closed.

The risks are quantifiable and physics-driven — expansive clay requiring $3M in foundation upgrades, FEMA flood zones tripling insurance, hail events driving 400% premium spikes (54% of all solar claims), wildfire smoke reducing yield by 6%, and erosion triggering $50K/day EPA fines. These are knowable facts that the traditional process discovers too late. Earthflow delivers comprehensive, physics-based site intelligence in minutes rather than months, with full data provenance and reproducible methodology.

1.1 Before & After: Traditional vs. Earthflow

2. Solutions Across the Solar Lifecycle

Earthflow addresses the entire solar development lifecycle — from site screening through construction, operations, and monitoring. All 7 solutions share the same Physics AI™ engine, 70+ data sources, and single data model, so screening data flows into construction planning, operational monitoring, and portfolio management without fragmentation.

3. The 3-Minute Site Assessment

Enter coordinates. In under 3 minutes, 13 parallel physics modules query 70+ data sources and generate 1,000+ fields across 7 intelligence domains. Each module operates independently with graceful fallbacks — a single source outage never blocks the other 12. The following pages detail the output using Solar Star (579 MW, Kern County, CA) as example.

3.1 The 13 Physics Modules

Each module runs independently and in parallel, querying its designated data sources, applying physics-based calculations, and producing structured output fields. The result is a comprehensive site intelligence profile that no single consultant or data provider can match.

The field counts shown below are representative. The actual number of fields generated depends on data availability and site characteristics — sites in areas with rich historical data (e.g., extensive NOAA storm event records) may generate additional fields. Every field includes its value, unit, data source, confidence level, and provenance metadata.

3.2 Sample Output Data

The following tables show representative output from an Earthflow analysis of the Solar Star facility (34.818°N, 118.397°W) in Kern County, California — one of the largest solar installations in the United States at 579 MW. These are a small subset of the 1,000+ fields generated for every site, selected to illustrate the depth and specificity of Earthflow’s output across four key intelligence domains.

Understanding the Output Scale

To put 1,000+ fields in perspective: a typical Phase I Environmental Site Assessment (ESA) under ASTM E1527-21 generates approximately 30–50 discrete data points about a property. A comprehensive geotechnical investigation adds another 20–30. A solar resource assessment contributes 10–15. Combined, the full traditional assessment suite produces roughly 80–100 data points — after months of field work and $100K+ in consulting fees. Earthflow generates 10× this volume in under 3 minutes, with every data point traceable to its source.

Cross-Module Intelligence

The true power of Earthflow’s 13-module architecture is not in the individual modules — it is in the cross-module intelligence that emerges when all 1,000+ fields are analyzed together. Consider the following example from the Solar Star analysis:

• The Soil Module identifies sandy loam with K-factor 0.28 (moderately erodible)
• The Precipitation Module calculates R-factor of 28.5 (low rainfall erosivity)
• The Satellite Module measures slope at 2.4 degrees (gentle terrain)
• The Erosion Module combines these into a RUSLE soil loss estimate of 0.8 tons/acre/year
• The Physics AI™ KPI Engine rates Erosion Management Priority as “Low” and estimates BMP costs at $15K/MW

No single data source could produce this conclusion. It requires the integration of soil chemistry (SSURGO), climate data (PRISM), topography (SRTM), and physics-based modeling (RUSLE) — all automated, all reproducible, all completed in under 3 minutes. This is what Physics AI™ delivers that traditional approaches cannot match at scale.

3.3 Heatmap Visualization (24 Types)

Beyond tabular data, Earthflow generates 24 spatial heatmaps that provide cell-by-cell intelligence across the entire site footprint. Each heatmap is computed at configurable resolution (30–60m cells) and rendered as an interactive overlay on satellite imagery. For large sites spanning thousands of acres, spatial variation is the difference between a successful project and a costly surprise — the northwest corner of a 2,000-acre site may have fundamentally different soil, slope, and vegetation conditions than the southeast corner.

3.4 Advanced Vegetation Analysis

Earthflow computes 40+ vegetation metrics from 7 satellite sources, providing the most comprehensive automated vegetation assessment available for solar site evaluation. This goes far beyond simple NDVI — it includes spectral indices calibrated for different vegetation conditions, tree canopy analysis from LiDAR, AI-powered land cover classification, and management-oriented risk assessments.

10 Vegetation Indices

Tree Cover & Canopy Metrics

Eight specialized metrics characterize tree cover and canopy structure using GEDI LiDAR, Hansen Global Forest Change, and AI classification:

• AI Tree Cover % — Machine learning classification of tree coverage per cell
• Tree Height — GEDI LiDAR + SAR fusion for canopy height estimation
• Hansen Forest Change — Annual forest loss/gain since 2000 at 30m resolution
• Canopy Density — Vertical canopy closure percentage
• Canopy Height Model — Digital surface model minus digital terrain model
• Spectral Tree Scoring — Multi-band spectral signature analysis for tree identification
• Tree Removal Cost — Estimated clearing cost based on density and species
• Clearing Difficulty Index — Composite score accounting for density, height, and access

Land Cover Classification (22 Classes)

Two complementary classification systems provide comprehensive land cover characterization:

• Primary Classification (8 classes, 10m) — Water, Trees, Grass, Flooded Vegetation, Crops, Shrub & Scrub, Built, Bare Ground — with per-class probability scores and a diversity index
• Extended Classification (14 categories) — Evergreen Needleleaf, Evergreen Broadleaf, Deciduous Needleleaf, Deciduous Broadleaf, Mixed Forest, Closed Shrublands, Open Shrublands, Woody Savannas, Savannas, Grasslands, Wetlands, Croplands, Urban, Barren

Vegetation Risk & Management (12 Fields)

Beyond classification, Earthflow generates actionable management intelligence:

• Growth Risk, Fire Risk, Erosion Risk — Per-zone severity ratings (Low/Moderate/High)
• Management Frequency — Recommended mowing/clearing interval based on growth rate
• Clearing Cost — Estimated $/acre for initial site preparation
• Optimal Clearing Season — Best window for vegetation removal (regulatory + weather)
• Invasive Species Risk — Probability of invasive encroachment based on regional data
• Revegetation Timeline — Expected regrowth rate for restoration compliance

3.5 Site Monitor & Change Detection

Earthflow’s temporal analysis system maintains a rolling 12-month satellite archive for every monitored site, enabling automated change detection, construction progress tracking, and vegetation compliance monitoring without dispatching field crews.

3.6 Satellite Data Sources

Earthflow integrates data from 12+ satellite constellations and Earth observation systems, providing multi-resolution, multi-spectral coverage for any location in the United States. This diversity ensures redundancy — if one source experiences gaps or latency, alternative sources provide coverage.

3.7 Data Fabric Integration

Earthflow’s open data fabric architecture supports integration with field-collected data, IoT sensors, and third-party systems — overlaying ground-truth measurements onto satellite-derived baselines for enhanced accuracy and operational insight.

• Drone & Aerial Imagery — Orthomosaic imagery, LiDAR point clouds, and thermal scans with automatic georeferencing and overlay against satellite baselines
• Weather & Environmental Sensors — On-site weather stations, pyranometers, and anemometers with automated QA/QC and satellite backup validation
• IoT & Field Devices — Soil moisture probes, tilt sensors, structural monitors, and panel-level trackers with time-series ingestion and anomaly alerting
• Third-Party Data & APIs — Enterprise GIS platforms, SCADA systems, and asset management software via RESTful API connectors with standardized field mapping

4. From Data to Decisions — The Physics AI™ KPI Engine

Earthflow’s Physics AI™ KPI Engine synthesizes 1,000+ data fields into 200+ actionable Key Performance Indicators organized into 7 decision categories. Each KPI is computed from first-principles equations with full metadata: calculation methodology, input variables with sources, confidence scores, and threshold definitions. These are quantitative outputs of physics-based models — reproducible, auditable, and directly comparable across sites.

4.1 Seven KPI Categories

Complete KPI Inventory

The following table enumerates the primary KPIs generated for every site. Each KPI includes full calculation methodology, input variable attribution, confidence scoring, and threshold definitions.

4.2 KPI Thresholds & Interpretation

Every KPI includes a threshold framework for interpretation. Earthflow’s KPI metadata includes:

• Calculation methodology: The exact formula and input variables used to compute the KPI
• Input sources: Every data field that contributes to the KPI, with its source and confidence level
• Threshold definitions: What each rating level means in practical terms (e.g., “High erosion risk indicates RUSLE soil loss >5 tons/acre/year, requiring engineered BMPs”)
• Confidence score: How reliable the KPI is based on the availability and quality of input data
• Comparable benchmarks: How this value compares to regional averages and portfolio-level statistics

When an investment committee sees a Total Cost of Risk of $142,000/MW, they can drill into exactly which risk factors contribute (flood: $18K, erosion: $15K, fire: $22K, seismic: $45K, hail: $12K, wind: $30K) and understand the sensitivity of each component.

Decision Thresholds

The following table shows representative threshold bands for key KPIs — the green/amber/red framework that enables rapid portfolio screening and triage.

5. Cirra AI — Autonomous Intelligence

Cirra is not a chatbot. It is an autonomous AI agent system with 25 specialized tools that plans, reasons, and executes multi-step workflows on your behalf. Where traditional chatbots answer questions, Cirra does the work — autonomously orchestrating site assessments, generating reports, comparing portfolios, and delivering actionable intelligence in natural language.

Ask Cirra to “compare our top 5 Texas sites and generate a PDF for the board meeting” and it resolves site names, retrieves analysis data, compares 200+ KPIs, generates a branded PDF with satellite imagery, and returns a download link — all autonomously, in a single interaction.

Cirra’s domain expertise spans 9 specialized knowledge domains: site assessment, environmental compliance, construction engineering, financial modeling, weather analysis, vegetation management, interconnection, detection/monitoring, and regulatory requirements.

5.1 Five Agent System

5.2 How Agents Work Together

When you ask Cirra a question, five agents collaborate in a coordinated pipeline:

• Ask: You submit a natural-language query via any channel (Web, Slack, Teams, Email, Signal)
• Parse: Query Intelligence extracts entities, identifies intent, and creates an execution plan
• Route: Orchestration selects the optimal cost path — cached response, vector database lookup, or specialist agent delegation
• Analyze: Specialist agents (Decision Support, Predictive, Financial) work in parallel, each invoking the tools they need
• Deliver: Results are synthesized with confidence scores, source attribution, and actionable recommendations

5.3 Agent Advantages

5.4 What Agents Enable

5.5 Twenty-Five Specialized Tools

Available on 5 live channels: Web, Slack, Microsoft Teams, Email, and Signal. Cirra meets your team where they already work — no new interfaces to learn, no context switching required.

Cirra’s cost architecture is designed for enterprise-scale usage. Through intelligent query routing, selective knowledge base injection, and aggressive caching, Cirra delivers AI capabilities at 1/2,500th the cost of naïve LLM approaches. This means your team can ask hundreds of questions per day without concern about API costs — the kind of usage pattern that makes AI adoption practical rather than theoretical.

Security and data governance are built into the architecture. All queries are authenticated, all data access is permissioned, and all interactions are logged for audit. Cirra operates within the same security boundary as the Earthflow platform — no data leaves the environment, no third-party services have access to your site data, and all AI processing occurs within Earthflow’s infrastructure.

Cirra also supports workflow automation through predefined chains — multi-step sequences that execute common analytical patterns. Examples include the “New Site Evaluation” chain (geocode → analyze → assess → report), the “Environmental Compliance” chain (vegetation analysis → species classification → compliance check → environmental report), and the “Portfolio Rebalancing” chain (search → compare → predictive insights → executive summary). These chains can be triggered by a single natural-language instruction, reducing complex multi-tool workflows to a single interaction.

6. Detection, Monitoring & Grid Intelligence

🤖6.1 AI-Powered Asset Detection

Earthflow Detect uses multiple concurrent AI models to analyze satellite imagery and detect solar infrastructure — panels, substations, and inverters — with confidence scoring and GeoJSON export. The system combines 12-band multispectral analysis for large-scale detection with high-resolution native mask segmentation for precise polygon boundaries. Running multiple detection engines concurrently provides cross-validated results; when models diverge, the system flags discrepancies for human review.

Earthflow Detect — Resolution Tiers

Earthflow Detect operates at multiple resolution tiers, each optimized for different use cases. All run concurrently via parallel execution, with independent confidence scoring and color-coded GeoJSON layers (green/blue/amber) for visual overlay.

📡6.2 Site Monitoring & Vegetation Compliance

Monthly NDVI-based change detection replaces quarterly manual inspections. Earthflow tracks construction progress, vegetation encroachment, and compliance status at portfolio scale — automatically, every month. Overgrown vegetation reduces output by 5–15% in affected areas; satellite-based monitoring provides the objective, time-stamped evidence that compliance programs require.

• Monthly Sentinel-2 composites — cloud-filtered, atmospherically corrected imagery at 10m resolution
• 12-month trend analysis — track NDVI trajectories and detect anomalies with Mann-Kendall statistics
• 3-zone compliance mapping — clearing zone, buffer zone, and restoration zone status
• Species classification — identify vegetation types from spectral signatures using AI-powered analysis
• 10 vegetation indices — NDVI, EVI, SAVI, NDWI, NDMI, GNDVI, RENDVI, NBR, BSI, RVI
• Change detection metrics — vegetation loss/gain %, area cleared/stable/gained, Z-score anomaly detection
• Interactive playback — timeline slider with 0.5x–4x speed, split-screen before/after comparison

3-Zone Compliance Detail

⚡6.3 Interconnection Queue Intelligence

Proprietary Queue Pressure Score (QPS) analysis across all 7 US ISO/RTO regions (CAISO, ERCOT, MISO, PJM, SPP, ISO-NE, NYISO). With 2,600+ GW in queues and 4+ year average wait times, interconnection is the single largest bottleneck in solar development. Earthflow’s QPS integrates queue position, substation capacity, regional congestion, and historical processing rates to quantify this risk before capital is committed.

• Queue Pressure Score (0–100) — composite measure of congestion and competition
• MW ahead at nearest POI — quantified competition at the point of interconnection
• Active/total project counts — pipeline density by region and substation
• Average wait time — historical processing times by ISO and voltage class
• Solar/storage competition — technology mix analysis at competing queue positions
• Cost implications — estimated interconnection costs based on queue position and upgrades

🏆6.4 Competitive Intelligence

Detection capabilities extend beyond your own assets. Identify competitor installations, estimate capacity from detected panel area, and assess market saturation at specific interconnection points. The combination of detection data (what’s built), queue data (what’s planned), and environmental data (what’s viable) provides near-real-time market intelligence — existing capacity, planned capacity, grid constraints, and environmental suitability — before committing to land acquisition.

👁️6.5 Computer Vision Roadmap

Earthflow’s detection capabilities are expanding beyond satellite imagery into aerial and close-range computer vision for operational asset management:

• Aerial Intelligence — Drone imagery and LiDAR point cloud processing for high-resolution site surveys and topographic mapping
• Solar Panel Detection — AI-powered automatic identification and counting of individual panels from aerial imagery
• Defect Identification — Automated quality inspection detecting hot spots, micro-cracks, and soiling from thermal and RGB imagery
• Installation Quality — Visual assessment of installation alignment, torque tube straightness, and racking integrity

7. Reports, Exports & Deliverables

Every analysis produces outputs ready for clients, lenders, boards, and regulatory submissions. Deliverable generation is integrated into every workflow — ask Cirra for a report and it generates one from live data with full provenance metadata.

The Earthflow Data Browser provides searchable, filterable access to all 1,000+ fields across 16 analysis modules. Browse by category, search by field name, filter by data source, and export instantly in any format. Every field displays its value, unit, source, confidence level, and fallback status — complete transparency into how each data point was obtained.

All PDF reports are branded with Orbyfy identity, include Mapbox satellite imagery of the site, and contain full methodology sections that explain how each score and rating was derived. These are not generic templates — they are site-specific, data-driven documents designed to withstand the scrutiny of lenders, investors, and regulatory reviewers.

The Excel export deserves special mention. Each workbook contains 16 worksheets — one for each analysis module — with every field, its value, unit, source, confidence level, and fallback status. This format is designed for analysts who need to perform their own calculations, build custom models, or integrate Earthflow data into existing financial tools. The structured layout makes it immediately usable without any data cleaning or reformatting.

For GIS teams, GeoJSON exports include detection polygons, site boundaries, and spatial analysis results that can be loaded directly into QGIS, ArcGIS, or any standards-compliant mapping platform. Coordinate reference systems, feature properties, and geometry types follow OGC standards for maximum interoperability.

Report Customization

Earthflow’s reporting system supports four distinct report types, each tailored to a specific audience:

• Full Site Report (20–30 pages) — Comprehensive analysis covering all 13 modules, all KPIs, satellite imagery, methodology documentation, and data provenance. Designed for due diligence packages and lender submissions.
• Executive Summary (3–4 pages) — Key KPIs, risk flags, and recommendations in a format optimized for board presentations and executive reviews. Concise enough to read in 10 minutes, comprehensive enough to support decision-making.
• Geotechnical Report — Focused on soil conditions, bearing capacity, seismic hazard, erosion risk, and foundation recommendations. The format that structural engineers and EPC contractors need for bid preparation.
• Environmental Report — Focused on flood risk, vegetation classification, fire hazard, and compliance status. Formatted for regulatory submissions and permitting applications.

All reports can be generated on-demand through the dashboard, through Cirra AI (just ask — “Generate a full site report for Solar Star”), or programmatically through the API for batch operations. Reports are generated as PDFs with embedded satellite imagery and uploaded to secure cloud storage with time-limited download links (7-day default, configurable per organization).

For organizations that require custom branding or formatting, Earthflow’s report templates can be configured to match corporate identity guidelines — colors, logos, disclaimer language, and section ordering. This ensures that deliverables to external stakeholders maintain brand consistency while preserving the rigorous analytical content that makes them valuable.

8. Data Provenance & Scientific Foundation

Every data point in Earthflow is traceable from its authoritative source through the analysis pipeline to the final KPI. Lenders, tax equity investors, and insurance underwriters require data sourced from recognized authorities, processed through documented methodologies, and presented with confidence intervals. Earthflow was architected to meet this bankability-grade standard.

🗃️8.1 Authoritative Data Sources

These are the same authoritative datasets used by the USDA, FEMA, USGS, NREL, and the scientific community. Earthflow’s innovation is in the automated ingestion, physics-based processing, cross-module synthesis, and KPI generation that transforms raw geospatial data into decision-ready intelligence. The 70+ source architecture ensures every dimension of site risk is captured — from below-grade soil chemistry to atmospheric hail patterns.

🔬8.2 Physics AI™ Methodology

Earthflow’s Physics AI™ engine is built on a Physics-Informed Neural Network (PINN) architecture — a fundamentally different approach from black-box machine learning. Instead of learning patterns from data alone, PINNs embed the governing equations of physics directly into the network as hard constraints. The neural network cannot violate the laws of physics, because those laws are encoded in its loss function and architecture.

PINN Architecture: Inputs → Physics Constraints → Outputs

Traditional ML learns correlations from data — it might predict negative erosion rates or capacity factors above 100%. A PINN cannot, because the governing equations are baked into the architecture itself. The physics constrains the solution space, meaning the model only explores physically valid predictions.

Why PINNs > Traditional ML for Site Assessment

🛡️8.3 Data Integrity Architecture

A 5-level fallback hierarchy ensures complete data coverage for every US location. Not every API responds for every coordinate. Network outages, coverage gaps, and service limitations are inevitable. Earthflow’s fallback architecture handles these gracefully:

Every field carries provenance metadata identifying which source and fallback level was used. For data-sparse areas, the fallback system derives estimates from alternative sources (e.g., MERIT Hydro for flood risk where FEMA mapping is limited) — clearly noting the derivation and confidence adjustment. Every site gets a complete analysis regardless of data source availability.

9. Recognition & Partnerships

Earthflow’s approach has been validated by leading industry organizations, strategic partners, and technology programs that recognize the platform’s potential to transform how the solar industry evaluates and manages site risk. These recognitions span the full spectrum of the energy innovation ecosystem — from the Edison Awards (the longest-running innovation prize in the world) to NREL (the US government’s premier renewable energy laboratory) to NVIDIA and Google (the world’s leading AI infrastructure providers).

Industry Context

No other platform combines physics-based environmental analysis, autonomous AI agents, multi-model satellite detection, and interconnection queue intelligence in a single integrated system — delivering the depth engineers require, the transparency investors demand, and the speed development timelines necessitate.

10. Getting Started

Web-based, no installation required. Most users run their first site analysis within minutes. Enter coordinates, click “Analyze,” and receive 1,000+ data fields in 2–5 minutes. For enterprise deployments, Earthflow integrates via API, webhooks, Slack, Microsoft Teams, Email, and Signal.

Portfolio onboarding: import your site list, run batch analysis, and compare results the same day. Import GeoJSON boundaries for precise extents, or let Cirra geocode site descriptions into coordinates automatically.

Three Steps to Site Intelligence

Contact & Next Steps

To schedule a personalized demo, discuss enterprise integration, or explore partnership opportunities, contact our team. We will walk you through a live analysis of one of your sites — so you can see exactly what 1,000+ data fields and 200+ KPIs look like for a location you care about.

Why Now

2,600 GW in US interconnection queues. $100M+ capital allocation decisions being made with incomplete data. Development teams need physics-based analysis at the speed the market demands. Satellite constellations, cloud infrastructure, and AI have finally matured to make this possible.

Earthflow brings it together — physics-based environmental analysis, autonomous AI agents, multi-model detection, and queue intelligence in a single platform. Purpose-built for solar, calibrated against real-world data, validated by leading industry organizations. The decisions being made today will shape the energy landscape for decades. Those decisions deserve the best available data and the most rigorous analysis. That is what Earthflow delivers.