Plot Validation¶

The /validate_plots and /validate_plot API endpoints perform comprehensive integrity checks on agricultural plot geometries to ensure they meet quality standards for sustainability analysis. These endpoints validate geometric properties, detect common issues, and provide detailed feedback for data quality improvement.

Validation Checks¶

The system performs comprehensive integrity checks to ensure plot geometries are suitable for:

Sustainability Analysis: Accurate area calculations and boundary definitions
Satellite Analysis: Proper geometric structure for Earth Engine processing
Compliance Reporting: Valid geometries for regulatory submissions
Data Quality: Clean, consistent data for downstream processing

Supported Commodities¶

The system has dedicated detection models for these commodities:

Supported: shrimp, rubber, palm, palm oil, oil palm, cocoa, coffee, cattle, beef, leather, soy, soybean, soya, soja

Unsupported Commodities: If you provide a commodity not in the list above, the system will: - Use a plantation forest mask as an umbrella commodity detection model (appropriate for most agro-forestry commodities) - Include a warning in the response indicating the fallback behavior - Still perform all geometric and protected area checks normally

This allows validation to proceed for unlisted commodities while making it transparent that a generic plantation detection model is being used.

Endpoints¶

POST /validate_plots¶

Validates multiple plot geometries from an uploaded GeoJSON file. Supports GeoJSON, Shapefile, CSV, and Excel (with optional AI normalization).

Parameters:

Parameter	Type	Required	Default	Description
`file`	File	Yes	-	GeoJSON file containing plot geometries
`country`	string	No	-	Country for commodity and tenure validation; used for commodity detection and Canada/US deforestation tenure exemptions
`commodity_type`	string	No	-	Commodity type for validation (optional)
`include_deforestation`	boolean	No	false	Include deforestation check (plot-level + vicinity); enables GEE deforestation mask and cherry-picking (vicinity) risk
`production_volume`	float	No	-	Claimed total annual production (tonnes) for circumvention pillar (production volume vs. yield)

Response:

Returns a streaming GeoJSON FeatureCollection with:

Collection Properties: Aggregate integrity statistics
Feature Properties: Individual validation results for each plot

Example Request:

curl -X POST "https://api.epoch.eco/validate_plots" \
  -H "Authorization: Bearer YOUR_TOKEN" \
  -F "file=@plots.geojson" \
  -F "country=Thailand" \
  -F "commodity_type=rubber" \
  -F "include_deforestation=false" \
  -F "min_area=0.5" \
  -F "distance=1000" \
  -F "angle=30"

Usage Examples¶

Python¶

import requests
import json

# Set up the request
url = "https://api.epoch.eco/validate_plots"
headers = {
    "Authorization": "Bearer <your_firebase_token>"
}

# Prepare the file and parameters
files = {
    'file': ('my_plots.geojson', open('path/to/your/file.geojson', 'rb'), 'application/geo+json')
}

data = {
    'country': 'Thailand',  # Optional: for commodity and tenure validation
    'commodity_type': 'rubber',  # Optional: for commodity validation
    'include_deforestation': False,  # Set True for deforestation check
    'min_area': 0.5,  # Minimum area threshold
    'distance': 1000,  # Maximum distance between vertices
    'angle': 30,  # Minimum angle threshold
    'min_precision': 6,  # Minimum coordinate precision
    'min_length': 10,  # Minimum boundary segment length
    'overlap_threshold': 0.5,  # Overlap threshold
    'max_distance': 250000  # Maximum distance to nearest neighbor
}

# Make the request
response = requests.post(url, headers=headers, files=files, data=data)

# Check if request was successful
if response.status_code == 200:
    result = response.json()

    # Access the summary statistics
    properties = result['properties']
    print(f"Total plots: {properties.get('total_plots', 0)}")
    print(f"Total area (ha): {properties.get('total_plot_area_ha', 0)}")
    print(f"Overall risk: {properties.get('overall', {}).get('risk', 'N/A')}")
    print(f"Investigate: {properties.get('overall', {}).get('investigate', [])}")

    # Access individual features with validation results
    for feature in result['features']:
        props = feature['properties']
        risk = props.get('overall', {}).get('risk', 'N/A')
        print(f"Plot {props.get('plot_id', 'unknown')}: risk={risk}")
else:
    print(f"Error: {response.status_code} - {response.text}")

JavaScript¶

const axios = require('axios');
const FormData = require('form-data');
const fs = require('fs');

async function validatePlots(filePath, token, options = {}) {
    try {
        const formData = new FormData();

        // Add the file
        formData.append('file', fs.createReadStream(filePath));

        // Add optional parameters
        if (options.country) formData.append('country', options.country);
        if (options.commodity_type) formData.append('commodity_type', options.commodity_type);
        if (options.min_area) formData.append('min_area', options.min_area);
        if (options.distance) formData.append('distance', options.distance);
        if (options.angle) formData.append('angle', options.angle);

        const response = await axios({
            method: 'post',
            url: 'https://api.epoch.eco/validate_plots',
            headers: {
                'Authorization': `Bearer ${token}`,
                ...formData.getHeaders()
            },
            data: formData
        });

        // Process the response
        const result = response.data;

        // Access the summary statistics
        const properties = result.properties;
        console.log(`Total plots: ${properties.total_plots || 0}`);
        console.log(`Total area (ha): ${properties.total_plot_area_ha || 0}`);
        console.log(`Overall risk: ${properties.overall?.risk || 'N/A'}`);
        console.log(`Investigate: ${JSON.stringify(properties.overall?.investigate || [])}`);

        // Access individual features with validation results
        result.features.forEach(feature => {
            const props = feature.properties;
            const risk = props.overall?.risk || 'N/A';
            console.log(`Plot ${props.plot_id || 'unknown'}: risk=${risk}`);
        });

    } catch (error) {
        console.error('Error:', error.response?.data || error.message);
    }
}

// Usage example
validatePlots('./my_plots.geojson', 'your_token', {
    country: 'Thailand',
    commodity_type: 'rubber',
    min_area: 0.5,
    distance: 1000,
    angle: 30
});

cURL¶

curl -X POST "https://api.epoch.eco/validate_plots" \
  -H "Authorization: Bearer <your_firebase_token>" \
  -H "Accept: application/json" \
  -F "file=@path/to/your/file.geojson" \
  -F "country=Thailand" \
  -F "commodity_type=rubber" \
  -F "min_area=0.5" \
  -F "distance=1000" \
  -F "angle=30" \
  -F "min_precision=6" \
  -F "min_length=10" \
  -F "overlap_threshold=0.5" \
  -F "max_distance=250000" \
  --output validation_results.json

GET /validate_plot¶

Validates a single plot geometry using a WKT geometry string.

Parameters:

Parameter	Type	Required	Default	Description
`geometry`	string	Yes	-	WKT geometry (Polygon, Multipolygon, etc.)
`country`	string	No	-	Country for commodity and tenure validation
`commodity_type`	string	No	-	Commodity type for validation (optional)
`include_deforestation`	boolean	No	false	Include deforestation check

Response:

Returns a GeoJSON FeatureCollection with a single feature containing validation results.

Example Request:

curl -X GET "https://api.epoch.eco/validate_plot" \
  -H "Authorization: Bearer YOUR_TOKEN" \
  -G \
  -d "geometry=POLYGON((100.5018 13.7563, 100.5028 13.7563, 100.5028 13.7573, 100.5018 13.7573, 100.5018 13.7563))" \
  -d "country=Thailand"

POST /validate_locations¶

Universal validation endpoint that accepts any file format. Auto-detects whether input is facility or plot data and routes to the appropriate validation logic. Use normalize_input=true to enable AI schema detection and geocoding for CSV/Excel files.

Parameter	Type	Default	Description
`file`	File	-	Any input file (GeoJSON, SHP, CSV, Excel)
`normalize_input`	boolean	false	Enable AI/geocoding for non-GeoJSON formats
`location_type`	string	auto	`facility` or `plot` (auto-detected if omitted)
`limit`	int	100	Max ranked features to return
`country`	string	-	Country for commodity and tenure validation
`commodity_type`	string	-	Commodity type
`include_deforestation`	boolean	false	Include deforestation check (plot validation only)

POST /normalize_locations¶

Normalizes geospatial data (clean/geocode) without running full validation. Useful for preparing CSV/Excel files before validation. Returns normalized GeoJSON.

Parameter	Type	Description
`file`	File	CSV, Excel, or GeoJSON file
`location_type`	string	`facility` or `plot`
`commodity_type`	string	Commodity type for context

Validation Checks¶

Geometric Integrity Checks¶

1. Area Validation¶

Purpose: Ensures plots meet minimum size requirements
Check: area >= min_area (default: 0.1 hectares)
Issue: Plots too small for meaningful analysis
Fix: Combine small adjacent plots or increase minimum threshold

2. Self-Intersection Detection¶

Purpose: Identifies invalid polygon geometries
Check: No self-intersecting boundaries
Issue: Invalid geometry that can cause analysis errors
Fix: Use GIS tools to fix self-intersections

3. Duplicate Vertices¶

Purpose: Removes redundant coordinate points
Check: No consecutive identical coordinates
Issue: Inefficient geometry storage and potential analysis issues
Fix: Remove duplicate vertices using GIS tools

4. Angle Validation¶

Purpose: Ensures reasonable boundary angles (spike detection)
Check: All angles >= angle (default: 30°)
Issue: Extremely sharp angles may indicate digitization errors
Fix: Smooth boundaries or adjust digitization precision

5. Distance Validation¶

Purpose: Ensures reasonable distances between vertices
Check: All vertex distances <= distance (default: 1000m)
Issue: Extremely long segments may indicate missing vertices
Fix: Add intermediate vertices along long boundaries

6. Precision Validation¶

Purpose: Ensures adequate coordinate precision
Check: Coordinates have >= min_precision decimal places (default: 6)
Issue: Low precision may cause analysis inaccuracies
Fix: Increase coordinate precision in source data

7. Boundary Segment Length¶

Purpose: Ensures reasonable boundary segment lengths
Check: All segments >= min_length (default: 10m)
Issue: Very short segments may indicate digitization noise
Fix: Simplify geometry or adjust digitization settings

Spatial Relationship Checks¶

8. Overlap Detection¶

Purpose: Identifies overlapping plots within the dataset
Check: Overlap area <= overlap_threshold (default: 1% of plot area)
Issue: Overlapping plots can cause double-counting in analysis
Fix: Resolve overlaps by adjusting boundaries or removing duplicates

9. Nearest Neighbor Distance¶

Purpose: Identifies isolated plots that may be errors
Check: Distance to nearest neighbor <= max_distance (default: 250,000m)
Issue: Isolated plots may be digitization errors or require separate handling
Fix: Verify plot location or adjust analysis parameters

Geometry Type Handling¶

10. Geometry Collection Processing¶

Purpose: Handles complex geometry collections
Check: Separates and validates individual geometry components
Issue: Mixed geometry types in single features
Fix: Split into separate features by geometry type

11. MultiPolygon Flattening¶

Purpose: Processes MultiPolygon geometries
Check: Separates MultiPolygons into individual Polygons
Issue: Complex MultiPolygon structures
Fix: Flatten to individual Polygon features

Deforestation Checks (when `include_deforestation=true`)¶

The deforestation check is aligned with the icechunk zonal stats (Prefect) flow and uses the same exemptions so validation results match batch processing outputs.

12. Plot-Level Deforestation¶

Purpose: Detects deforestation alerts (GLAD, TMF, Hansen) within each plot, restricted to natural forest
Baseline: FDP natural forest + JRC TMF
Exemptions (pixels excluded from deforestation count):
GFC subtype 20: Managed forest (plantation) from JRC GFC2020 subtypes
Canada: Managed forest (values 11, 12, 40, 50) from Canada_MFv2020.tif
United States: Forest ownership (values 1, 2, 3, 4) from US_forest_ownership.tif
Country: Required for Canada/US tenure; TIFFs are loaded only when country matches (no overhead for other countries)

13. Cherry-Picking (Vicinity) Check¶

Purpose: Assesses geographic cherry-picking risk — whether plots are purposefully selected to outline only low-risk areas (no/minimal deforestation, high commodity presence) while the surrounding landscape has much higher risk.
What it detects: Plots drawn to avoid deforestation hotspots; the vicinity (extent of all plots) may have high deforestation or low commodity coverage compared to the broader region.
Metrics: Vicinity area, commodity area, natural forest area, protected area, deforestation area within the extent (bounding box) of all plots.
Risk: Low (≤5% deforestation in vicinity), medium (≤20%), or high (>20%).

Circumvention (Production Volume) Check¶

14. Circumvention — Production Volume vs. Yield¶

Purpose: Assesses production volume circumvention risk — whether the claimed annual production volume is plausible given the total plot area and typical yields.
What it detects: Suppliers declaring a much larger production volume than the plots can physically support (e.g., claiming 10,000 tonnes from 100 ha when typical yield would support ~2,000 tonnes).
Requires: production_volume parameter (claimed total annual production in tonnes).
Metrics: estimated_annual_production_tonnes (yield × area), production_volume_tonnes (claimed), deviation_pct (signed: positive = overstatement).
Risk: Low (claimed ≤ estimated or small overstatement), medium (moderate overstatement), or high (large overstatement).

Cherry-picking vs circumvention: Cherry-picking is about where plots are located (spatial selection bias). Circumvention is about how much production is claimed relative to plot area (volume overstatement).

Protected Area Checks¶

15. WDPA Overlap Detection¶

Purpose: Identifies plots that overlap World Database of Protected Areas
Check: Spatial intersection with WCMC/WDPA/current/polygons dataset
Issue: Plots in protected areas may have restrictions on agricultural activities
Information Extracted:
NAME: Protected area name (e.g., "Serengeti National Park")
DESIG: Designation type (e.g., "National Park", "Wildlife Reserve")
IUCN_CAT: IUCN management category (I, II, III, IV, V, VI)
Fix: Verify plot legality and check local regulations for protected areas

Risk Assessment¶

The validation system uses a risk-based model (low, medium, high) instead of numeric confidence scores. Lower risk indicates better data quality.

Cherry-Picking vs Circumvention (Distinct Concepts)¶

Concept	What it measures	Risk type
Cherry-picking	Plots purposefully outlining low-risk areas while the surrounding landscape has higher deforestation/risk	Geographic selection bias
Circumvention	Claimed production volume exceeds what the plot area can support based on typical yields	Production volume overstatement

Cherry-picking asks: Are these plots drawn to avoid bad areas? Circumvention asks: Can this area actually produce the claimed volume?

Pillar Structure¶

Pillar	Description
plot_validity	Shape validity, commodity presence, protected area overlap
deforestation	Plots with deforestation alerts (when `include_deforestation=true`)
cherry_picking	Geographic risk: Plots purposefully outlining low-risk areas while surrounding landscape has higher risk. Vicinity metrics (commodity, natural forest, deforestation in extent of all plots).
circumvention	Production volume risk: Claimed production vs. yield-based estimate. Detects overstatement of annual production relative to plot area. Requires `production_volume`.
overall	Weighted combination; `investigate` lists categories to review

Shape Validity Issue Weights¶

Used to compute plot_validity.shape_validity.risk:

Issue Type	Weight	Description
`intersects_another_polygon_count`	0.30	Overlaps - can cause double-counting
`self_intersection_count`	0.25	Invalid geometry
`distance_nearest_neighbor_count`	0.20	Isolated plots - may be errors
`commodity_absence_count`	0.25	Wrong commodity area
`protected_area_overlap_count`	0.20	Protected area overlap
`area_too_small_count`	0.10	Small areas
`max_distance_gt_min_distance_count`	0.10	Long edges
`max_angle_lt_min_angle_count`	0.10	Sharp angles
`bad_precision_count`	0.05	Precision issues
`short_boundary_segments_count`	0.05	Short segments
`triangular_geometry_count`	0.05	Triangular shapes
`duplicate_vertices_count`	0.05	Duplicate vertices
`nested_geometry_collection_count`	0.05	Nested collections
`simplifiable_collections_count`	0.05	Simplifiable collections

Risk Thresholds¶

low: Good data quality, minimal issues
medium: Some issues requiring attention
high: Significant issues requiring correction

Response Format¶

The response uses a pillar-based structure with nested risk assessments for plot validity, deforestation, circumvention, and cherry-picking (vicinity). Feature properties use flat top-level keys (no nested validation object); location_type identifies output as "plot".

Collection Properties (Aggregate Statistics)¶

{
  "type": "FeatureCollection",
  "properties": {
    "total_plots": 150,
    "total_plot_area_ha": 1250.5,
    "plot_validity": {
      "risk": "medium",
      "shape_validity": {
        "risk": "low",
        "issues": {
          "intersects_another_polygon_count": 4,
          "self_intersection_count": 2,
          "area_too_small_count": 3,
          "max_distance_gt_min_distance_count": 0,
          "max_angle_lt_min_angle_count": 1,
          "bad_precision_count": 2,
          "short_boundary_segments_count": 1,
          "distance_nearest_neighbor_count": 2,
          "triangular_geometry_count": 0,
          "duplicate_vertices_count": 5,
          "nested_geometry_collection_count": 0,
          "simplifiable_collections_count": 0
        }
      },
      "commodity_presence": {
        "risk": "low",
        "commodity_absence_count": 5,
        "description": "5 plots lack commodity presence"
      },
      "protected_area_overlap": {
        "risk": "medium",
        "protected_area_overlap_count": 12,
        "description": "12 plots overlap with protected areas"
      }
    },
    "deforestation": {
      "risk": "low",
      "plot_deforestation_pct": 2.5,
      "deforestation_area_ha": 31.26,
      "plots_with_deforestation": 3,
      "description": "3 plots have deforestation alerts (31.3 ha total)"
    },
    "circumvention": {
      "risk": "low",
      "estimated_annual_production_tonnes": 4500.0,
      "production_volume_tonnes": 4200.0,
      "deviation": 0.067
    },
    "cherry_picking": {
      "risk": "low",
      "vicinity_area_ha": 5000.0,
      "vicinity_commodity_area_ha": 3200.0,
      "vicinity_natural_forest_area_ha": 800.0,
      "vicinity_protected_area_ha": 150.0,
      "vicinity_deforestation_ha": 45.0,
      "deforestation_pct": 1.4
    },
    "overall": {
      "risk": "medium",
      "investigate": ["protected_area_overlap", "commodity_presence"]
    }
  },
  "features": [...]
}

Note: deforestation and cherry_picking are null/empty when include_deforestation=false or when no vicinity check is run. circumvention (production volume vs. yield) requires production_volume.

Feature Properties (Individual Results)¶

Each feature includes user-provided properties plus validation results as flat top-level keys (no nested validation object):

{
  "type": "Feature",
  "geometry": {
    "type": "Polygon",
    "coordinates": [[[100.5018, 13.7563], [100.5028, 13.7563], [100.5028, 13.7573], [100.5018, 13.7573], [100.5018, 13.7563]]]
  },
  "properties": {
    "location_type": "plot",
    "plot_id": "plot_001",
    "area": 1.25,
    "plot_validity": {
      "risk": "low",
      "shape_validity": {
        "risk": "low",
        "indicators": { "self_intersection_count": 0, "intersects_another_polygon_count": 0, ... }
      },
      "commodity_presence": {
        "risk": "low",
        "commodity_absence_count": 0
      },
      "protected_area_overlap": {
        "risk": "high",
        "protected_area_overlap_count": 1,
        "protected_area_metadata": [
          { "NAME": "Serengeti National Park", "DESIG": "National Park", "IUCN_CAT": "II" }
        ]
      }
    },
    "deforestation": {
      "risk": "low",
      "deforestation_present": 0,
      "deforestation_value": 0.0
    },
    "circumvention": {
      "risk": "low",
      "estimated_annual_production_tonnes": 45.2,
      "production_volume_tonnes": 42.0,
      "deviation": 0.07
    },
    "overall": {
      "risk": "low"
    },
    "primary_warning": null
  }
}

location_type: "plot" — identifies this as plot validation output
primary_warning: Optional string when critical issues exist (e.g. self-intersection, overlaps)
Risk values: "low" (good), "medium", "high" (bad). overall.investigate (in collection properties) lists categories to review, ordered by priority.

Validation Results¶

Low Risk¶

Good data quality with minimal issues
Critical checks pass
Suitable for sustainability analysis

Medium Risk¶

Some issues requiring attention
Most checks pass
Review overall.investigate for priority categories

High Risk¶

Significant issues
One or more critical checks fail
Correction recommended before analysis

Common Issues and Solutions¶

1. Small Plot Areas¶

Issue: Plots smaller than minimum threshold Solution:

Increase min_area parameter if appropriate
Combine adjacent small plots
Use different analysis approach for small plots

2. Self-Intersections¶

Issue: Invalid polygon boundaries Solution:

Use GIS software to fix geometries
Re-digitize problematic boundaries
Use buffer operations to clean boundaries

3. Overlapping Plots¶

Issue: Multiple plots occupy same area Solution:

Adjust plot boundaries to eliminate overlaps
Remove duplicate plots
Use spatial analysis to resolve conflicts

4. Low Coordinate Precision¶

Issue: Insufficient decimal places in coordinates Solution:

Increase precision in source data
Adjust min_precision parameter if appropriate
Re-digitize with higher precision

5. Isolated Plots¶

Issue: Plots far from nearest neighbors Solution:

Verify plot location accuracy
Adjust max_neighbor_distance parameter
Handle isolated plots separately in analysis

6. Protected Area Overlaps (NEW)¶

Issue: Plots overlap with World Database of Protected Areas Solution:

Check local regulations for the specific protected area (IUCN category in response)
Verify plot legality within the protected area
Consider alternative plot locations outside protected areas
Review the protected_area_metadata for the protected area name and designation
IUCN Categories: I (Strict Protection), II (National Park), III (Monument), IV (Habitat Management), V (Protected Landscape), VI (Sustainable Use)

Execution Architecture¶

Validation runs three parallel streams for maximum throughput:

GEE plot task: Commodity presence and deforestation per plot (Earth Engine)
GEE cherry-picking task: Vicinity metrics (convex hull of all plots)
Local geometric task: Angles, WDPA, distances, overlaps, area, precision, etc. (CPU-bound)

The shared deforestation mask is built once and reused by both GEE tasks (instruction caching). Geometric checks run in a thread pool while the network requests are in flight, so CPU and I/O overlap.

Performance Considerations¶

Large Files: Use streaming for files with >1000 plots; 24k+ plots may use 2–4GB RAM during serialization
Complex Geometries: MultiPolygons and GeometryCollections require more processing time
Precision: Higher precision requirements increase processing time
Spatial Checks: Overlap and neighbor distance checks are computationally intensive
Deforestation: Set include_deforestation=true only when needed; adds GEE calls and smaller batch sizes

Best Practices¶

1. Parameter Tuning¶

Start with default values and adjust based on your data characteristics
Lower angle thresholds (0.5-1.0°) for more sensitive spike detection
Lower distance thresholds (500-800m) for more sensitive edge detection
Adjust area thresholds based on your expected plot sizes
Increase precision requirements for high-accuracy applications

2. Data Preparation¶

Clean geometries before uploading using GIS software like QGIS
Fix self-intersections and topological issues
Ensure adequate coordinate precision (6+ decimal places)
Remove duplicate vertices and simplify complex geometries
Validate coordinate systems and ensure proper projection

Plot Validation¶

Validation Checks¶

Supported Commodities¶

Endpoints¶

POST /validate_plots¶

Usage Examples¶

Python¶

JavaScript¶

cURL¶

GET /validate_plot¶

POST /validate_locations¶

POST /normalize_locations¶

Validation Checks¶

Geometric Integrity Checks¶

1. Area Validation¶

2. Self-Intersection Detection¶

3. Duplicate Vertices¶

4. Angle Validation¶

5. Distance Validation¶

6. Precision Validation¶

7. Boundary Segment Length¶

Spatial Relationship Checks¶

8. Overlap Detection¶

9. Nearest Neighbor Distance¶

Geometry Type Handling¶

10. Geometry Collection Processing¶

11. MultiPolygon Flattening¶

Deforestation Checks (when include_deforestation=true)¶

12. Plot-Level Deforestation¶

13. Cherry-Picking (Vicinity) Check¶

Circumvention (Production Volume) Check¶

14. Circumvention — Production Volume vs. Yield¶

Protected Area Checks¶

15. WDPA Overlap Detection¶

Risk Assessment¶

Cherry-Picking vs Circumvention (Distinct Concepts)¶

Pillar Structure¶

Shape Validity Issue Weights¶

Risk Thresholds¶

Response Format¶

Collection Properties (Aggregate Statistics)¶

Feature Properties (Individual Results)¶

Validation Results¶

Low Risk¶

Medium Risk¶

High Risk¶

Common Issues and Solutions¶

1. Small Plot Areas¶

2. Self-Intersections¶

3. Overlapping Plots¶

4. Low Coordinate Precision¶

5. Isolated Plots¶

6. Protected Area Overlaps (NEW)¶

Execution Architecture¶

Performance Considerations¶

Best Practices¶

1. Parameter Tuning¶

2. Data Preparation¶

Deforestation Checks (when `include_deforestation=true`)¶