Adapting Forest Monitoring During Conflict Using Satellite Data & Geostatistical Modeling In Sudan

Sudan's National Forest Inventory (NFI) is the foundation of the country's ability to monitor its forests and report on carbon stocks to the United Nations Framework Convention on Climate Change (UNFCCC). When conflict intensified across the country in 2023, the NFI field crews were prevented from accessing some of Sudan's most forested states, including South and West Kordofan, East and South Darfur, and the Blue Nile, leaving large and critical gaps in the NFI sampling design.

Without data from these areas, Sudan risked being unable to produce a credible Forest Reference Emission Level (FREL), putting its REDD+ commitments at risk. To address this, the Forest Reference Level (FRL) team used Earth Observation data as a rigorous substitute for missing field measurements.

They combined NASA GEDI lidar data (2019–2023), which provides direct measurements of forest height and biomass at 1 km resolution, with a 2023 National Forest/Non-Forest Probability Map. These datasets were integrated with existing NFI plots using a Bayesian geostatistical model (Hunka et al., 2025) to estimate biomass and forest volume across the entire country, including conflict-affected areas.

The results were strong. The model explained 72% of variation in forest volume and 69% in biomass, with low bias and solid validation. This allowed Sudan to produce its first wall-to-wall national biomass map at a 5 km × 5 km grid, showing clear ecological patterns from drier central regions to more forested areas in the south and west. These outputs provided the emission factors needed to complete the FREL and continue UNFCCC reporting with confidence.

Beyond reporting, the work established a practical and transferable approach. The code is publicly available, the method is reproducible, and the process strengthened national capacity in geostatistics and Earth Observation. Sudan’s experience offers a useful model for other countries facing conflict or limited data.

Sudan's National Forest Inventory (NFI) relies on field plot visits across the country to measure forest extent, health, and carbon stocks, data essential for international climate reporting. Since conflict intensified in 2023, large parts of Sudan became physically inaccessible to field crews, particularly the richest forest areas in South and West Kordofan, East and South Darfur, and the Blue Nile states. This left critical gaps in the NFI sampling design, threatening the accuracy and credibility of Sudan's national forest information. Without complete data, Sudan risked being unable to submit a credible Forest Reference Emission Level (FREL) to the UNFCCC, undermining its REDD+ commitments and international climate obligations. The FREL requires two components, activity data (land use/cover change from satellites) and emission factors (carbon stocks from NFI field data) and the conflict was compromising the emission factors component.

Two EO data products filled gaps where field plots were inaccessible. First, NASA GEDI (2019–2023) provided direct, laser-based measurements of canopy height and above-ground biomass, summarized at 1 km resolution across Sudan. This offered a physically robust estimate of biomass to replace missing NFI emission factor data.

Second, the 2023 Forest/Non-Forest Probability Map identified the likelihood of forest cover at pixel level, helping constrain biomass estimates to areas with ≥10% probability of forest presence. The forest change probability map was developed on the SEPAL (System for Earth Observation Data Access, Processing and Analysis for Land Monitoring) platform of the Food and Agriculture Organization of the United Nations (FAO). The map shows the likelihood of forest change for each pixel across Sudan and was generated using a stack of multi-sensor satellite data. This probability map enabled the stratification of the landscape into areas with varying probabilities of forest change, thereby improving the efficiency and accuracy of the sampling framework used to estimate forest change areas. The final area estimates were derived from visually interpreted samples, where land-use (LU) and transition class information were collected using the Collect Earth application of FAO. These interpreted samples served as the primary dataset for generating statistically robust, stratified area estimates of forest change for Forest Reference Level (FRL) reporting, commonly referred to as activity data in carbon accounting.

Together, these datasets provided spatially continuous, independent information on forest structure and extent. Combined with available NFI plots in a Bayesian geostatistical model, they enabled wall-to-wall biomass and volume estimates across Sudan, including conflict-affected regions, resulting in the country’s first complete national biomass map.

The analysis relied on three complementary data sources: GEDI LiDAR observations (primary biomass signal), the national forest probability map (spatial context), and Sudan’s NFI field plots (ground truth for model training and validation). EO data extended the reach of field measurements rather than replacing them. FAO also supported activity data for the FREL, capturing land-use change over time to complement the emission factors derived from this integrated approach.

Sudan kept its national forest monitoring alive during a time of conflict, using satellites to fill in the gaps where field teams couldn't go, and producing the country's first complete biomass map.