Goals

• Estimate the extent of forest structure types (dense, open, scrub) annually within Area of Interest (AoI) and Micro Watershed (MWS) boundaries.
• Use Earth Observation data (Sentinel-2, Landsat) to compute vegetation indices and texture metrics for each year.
• Generate raster outputs at 30m resolution for annual forest structure classification.
• Vectorize raster outputs into field/MWS-level polygons for annual monitoring.
• Publish annual raster and vector outputs as Earth Engine assets.
• Enable visualization and temporal comparison of forest structure changes over multiple years.

Expected Outcome

• Annual raster datasets (30m resolution) showing forest structure classes (dense, open, scrub) per year.
• Annual vectorized polygons with attributes:
  • Year.
  • Forest structure class.
  • Area (ha).
  • Mean NDVI or other structural indicators.
• Earth Engine assets (raster + vector) for each year with metadata (source, date, classification method).
• GEE visualizations for each year with color-coded forest structure maps.
• Validation summary for annual outputs ensuring accuracy, coverage, and consistency.

Acceptance Criteria

Data Acquisition

• Satellite imagery for all years must be clipped to AoI/MWS boundaries.
• Cloud masking applied consistently across years.
• Resolution standardized to 30m.

Raster Classification

• Forest structure classes defined (dense, open, scrub).
• Raster outputs produced for every year in the time series.
• Each year’s raster covers 100% of AoI/MWS.

Vectorization

• Raster outputs converted to vector polygons annually using reduceToVectors().
• Each polygon includes:
  • Year.
  • Class label.
  • Area (ha).
  • Mean NDVI/structural metric.
• Polygons aligned with AoI/MWS boundaries.

Asset Publishing

• Annual raster and vector layers published as Earth Engine assets.
• Metadata includes source, year, classification schema, and processing date.

Implementation Details

Data Sources

• Sentinel-2 (resampled to 30m) and Landsat-8/9.
• Optional: FSI maps for annual validation.

Processing

• Compute NDVI, canopy density, and texture metrics annually.
• Apply classification thresholds (NDVI >0.6 = dense, 0.3–0.6 = open, <0.3 = scrub).
• Clip annual outputs to AoI/MWS boundaries.

Vectorization & Publishing

• Vectorize annual raster outputs.
• Export polygons with attributes including year, class, area, mean NDVI.
• Upload annual raster and vector layers as EE assets.

Visualization

• GEE visualization for each year:
  • Raster color-coded: dense (dark green), open (light green), scrub (yellow).
  • Overlay polygons on MWS boundaries for field-level insights.

Validation

• Spot-check annual polygons against high-resolution imagery.
• Compare yearly outputs with reference datasets to document changes.
• Generate annual validation report: coverage, accuracy, and attribute completeness.

Mockups/Wireframes

No response

Product Name

KYL

Organisation Name

C4GT

Domain

No response

Tech Skills Needed

Python

Organizational Mentor

@amanodt @kapildadheech @ankit-work7

Angel Mentor

No response

Complexity

Medium

Category

Backend