Remote Sensing Services

Satellite-Based
Environmental Monitoring

Near-real-time forest cover change detection using multi-spectral satellite imagery and time-series analysis for tropical concession management.

Explore Methodology →
Client Profile

Client

Anonymous Forest Concession in Cameroon

Project Scope

Multi-year deforestation monitoring across 18,500 hectares of tropical moist forest using Sentinel-2 and Landsat time series.

The Challenge

Complex Monitoring Requirements

Tropical forests face rapid, often illegal, deforestation pressures that require precise, temporally dense monitoring to support compliance and certification frameworks.

Cloud Cover

Persistent cloud cover in the Congo Basin limits optical satellite acquisition, requiring sophisticated cloud-masking algorithms and dense time-series compositing.

Small-Scale Disturbance

Selective logging and smallholder agriculture create subtle spectral changes that evade detection by coarse-resolution global products.

Near-Real-Time Need

Certification bodies (FSC, PEFC) and EU Deforestation Regulation require rapid reporting of forest cover changes with documented accuracy metrics.

Methodology

Systematic Processing Pipeline

Our processing chain follows best practices from peer-reviewed remote sensing literature, ensuring reproducibility and accuracy.

1

Atmospheric Correction

Sen2Cor and LaSRC applied to Sentinel-2 and Landsat imagery to convert top-of-atmosphere to bottom-of-atmosphere reflectance.

2

Cloud & Shadow Masking

QA bands and Fmask 4.6 algorithm used to flag clouds, cloud shadows, and cirrus, achieving >95% cloud detection accuracy.

3

NDVI / NDMI Time Series

Monthly composites of NDVI and NDMI indices generated, smoothed using Savitzky-Golay filtering to reduce noise.

4

Change Detection

BFAST algorithm applied to detect abrupt breaks in vegetation time series, with post-classification into deforestation and degradation.

5

Accuracy Assessment

Stratified random sampling (n=500) with aerial validation. Error matrix computed for overall accuracy and Kappa coefficient.

Sentinel-2 MSI (10m) Landsat-8/9 OLI (30m) Hansen Global Forest Watch ESA WorldCover (10m)
Data Visualization

Forest Cover Change Time Series

Interactive D3.js time series showing cumulative forest cover loss and annual disturbance rates from 2000 to 2023.

Figure 1: Annual forest cover change (ha/year) for the concession area, derived from Landsat and Sentinel-2 time series. The trend line indicates a significant (p<0.01) declining rate of disturbance post-2015.

Study Area & Satellite Coverage

Three.js interactive globe showing the Cameroonian study area with Sentinel-2 and Landsat orbit paths.

Figure 2: 3D globe visualization showing the Cameroonian forest concession (red marker) with simulated satellite orbit paths (green) and acquisition swaths.

Project Results

Key Performance Metrics

0
Total Area Analyzed
0
Classification Accuracy
0
Time Series Length
0
Revisit Frequency
Academic Foundation

Peer-Reviewed Literature

Our methodology is grounded in peer-reviewed research from leading remote sensing and conservation science journals.

Hansen, M.C., Potapov, P.V., Moore, R., et al. (2013). "High-Resolution Global Maps of 21st-Century Forest Cover Change." Science, 342(6160), 850-853.

DOI: 10.1126/science.1244693

Reiche, J., Verbesselt, J., Hoekman, D., & Herold, M. (2018). "Improving near-real time deforestation monitoring in tropical humid forests by combining dense Sentinel-1 radar and dense Sentinel-2 optical time series." Remote Sensing of Environment, 204, 147-161.

DOI: 10.1016/j.rse.2017.10.034

Vancutsem, C., Achard, F., Pekel, J.F., et al. (2021). "Long-term (1990-2019) monitoring of forest cover changes in the humid tropics." Science Advances, 7(10), eabe1603.

DOI: 10.1126/sciadv.abe1603
Project Timeline

Execution Phases

Phase 1
Data Acquisition & Preprocessing
Collection of 23 years of Landsat and Sentinel-2 imagery; atmospheric correction, cloud masking, and pixel-level compositing.
Phase 2
Baseline & Training Data
Reference data collection from high-resolution imagery; stratified sample design for accuracy assessment.
Phase 3
Change Detection & Classification
BFAST time-series breakpoint detection; Random Forest classification of change categories.
Phase 4
Validation & Reporting
Accuracy assessment using independent validation samples; production of maps, statistics, and compliance reports.