How satellite-based digital MRV monitoring works
Digital MRV (dMRV) uses Earth observation, repeatable processing, and documented QA to support monitoring, reporting, and verification for nature-based and forest carbon programs. The workflow below is how Kumi Analytics typically structures an engagement so outputs stay defensible for technical reviewers and registries.
Define the area of interest and monitoring objectives
Confirm project or portfolio boundaries in a consistent CRS, agree reporting cadence (e.g. annual vs event-driven), and list the carbon and environmental indicators required (land cover change, biomass, disturbance alerts, or water quality context). Document exclusions, leakage belts, and reference regions where applicable so every later step maps back to the same spatial frame.
Assemble the multi-sensor Earth observation stack
Select optical, SAR, and elevation sources appropriate to cloud regime, canopy density, and required revisit. For humid tropics, combine SAR with optical composites; for coastal blue carbon, add tidal and water optical constraints. Align all layers to a common grid or vector AOI so time-series comparisons are consistent across sensors.
Pre-process imagery and apply quality controls
Apply radiometric calibration, atmospheric correction where needed, cloud and shadow masking, and seasonally aware compositing. For SAR, apply speckle-aware workflows and terrain correction. Record QA masks and compositing rules so analysts can explain why any date was accepted or rejected during verification.
Run models for land cover, change, and/or biomass
Execute classification, change detection, or biomass estimation models trained or calibrated for the target biome. Fuse multi-sensor features where that improves stability. Produce geospatial layers plus tabular summaries (area by class, change events, biomass statistics) tied to the agreed legend and units.
Validate against field data and quantify uncertainty
Compare model outputs to field plots, inventory data, or independent reference where available. Report confusion metrics, bias checks, and confidence intervals appropriate to the methodology. When field data is sparse, document assumptions and conservative defaults so reviewers can see how uncertainty was handled.
Package audit-ready maps, tables, and methodology notes
Deliver map products, change narratives, and CSV/GeoTIFF exports suitable for validation and verification workflows. Include a short methodology note covering data sources, versions, processing dates, and QA logic—mirroring what appears in structured data and on this page for consistency.
Operationalize monitoring cycles and stakeholder review
Schedule recurring runs aligned to credit issuance or risk reviews, set alert thresholds for disturbances where needed, and capture reviewer feedback as inputs to the next cycle. Maintain versioned outputs so each reporting period can be traced without overwriting prior evidence.
For definitions of terms like dMRV, NDVI, or REDD+, see the Technical Glossary. For common buyer questions, see the FAQ or explore Nature Based Solutions Intelligence and KACSAT.