Assessing Forest Conservation Effectiveness of Natura 2000 Zones in Bulgaria Using Remote Sensing and GIS

Author: Eleonora Mitkova

MSc Geographic Data Science, Birkbeck, University of London

Key Points:

1. Forest conservation is vital for biodiversity and climate resilience
2. Natura 2000 is the world’s largest protected area (PA) network, taking over 34% of Bulgaria’s territory
3. Regional-scale analysis is underexplored
4. Case study: Vitosha & Vrachanski Balkan Nature Parks (Vitosha is the oldest nature park in the Balkans)

Key Tasks:

To evaluate the effectiveness of Natura 2000 zones in conserving and expanding forest cover in Bulgaria.

To compare Protected Areas (PAs) and Non-Protected Areas (non-PAs) using remote sensing and GIS change detection methods

To analyse road network proximity and its influence on forest dynamics.

​

Raster & Vector datasets: Landsat TM (1986, 2003, 2006) & Sentinel-2 MSI (2023), Road network, PAs

Spatial resolution:

• Landsat TM: 30m
• Sentinel-2 MSI: 10m

​

Vrachanski Balkan Nature Park

Methodology

• Processing satellite data: band compositing, segmentation, and NDVI.
• Supervised object-based classification + accuracy assessment.
• Change detection: forest ↔ non-forest transitions.
• Spatial overlays to compare PAs vs. non-PAs.

Vector data

Data acquisition

Data integration

200 m buffers around roads

Clipping & merging datasets

PAs & roads dataset

Change Detection analysis

Roads inside & outside PAs

Raster data

Landsat 4 TM data & Sentinel-2 MSI data

Composite rasters

Mosaic datasets

Image preparation & classification

True Colour composite

NDVI classification

Segmentation & Accuracy assessment

Supervised classification & Confusion matrix

Selection by attributes & Calculation of area

Area of change in km²

Table 1: Categories for image classification

Change detection tool & change detection raster

Geometry calculator & Summary stats

1. Results:

• Forest Dynamics: for the whole Study Area (SA) Forest cover increased from 32% (1986) → 48% (2023)
• PAs: Gains stronger in PAs (45% → 62%); Vitosha reached 72%; Vrachanski Balkan 68% → Confirming the positive impact of PAs
• Non-PAs: Slower recovery (25% → 41%); Indicating possible pressures from agriculture & urban growth
• Road Proximity: Unexpected forest gains near roads in non-PAs → likely afforestation/management initiatives.

Table 2: Change Detection Results for PAs, with % based on total area proportion.

2. Contradicting Results:

• Road Proximity: forest gain in road buffer zones was higher in non-PAs compared to PAs.
• This contradicts previous research linking road proximity with deforestation (Barber et al., 2014) and suggests the possibility of afforestation initiatives around roads outside PAs.

​

Table 3: Change Detection Results for Buffer Zones in PAs.

Conclusions:

• Natura 2000 zones demonstrated higher forest gains compared to non-protected areas (45% → 62% vs. 25% → 41%).

​

• Results indicate that protection status is effective in reducing deforestation and promoting recovery.

​

• Road-buffer analysis revealed unexpected forest gains, suggesting afforestation or management effects rather than uniform degradation.

​

• Limitations: optical sensor resolution prevents detection of selective logging; no ground-truth validation; analysis limited to two parks.

​

• Future research: integration of higher resolution and radar datasets (e.g., Sentinel-1 SAR) and expansion to additional Natura 2000 sites for broader assessment.

Chart 2: Proportion of Forest Area, based on the total area of Buffer Zones in PAs and Non-PAs for both study periods (1986-2003 and 2006-2023).

Chart 1: Proportion of Forest Area, based on the total land area of each area of interest.

Thank you for your attention! �Questions?

Let’s connect!