Environmental Monitoring with Sentinel-2: A Practical Guide for Colombia

Colombia loses approximately 170,000 hectares of forest annually. Regional environmental authorities need monitoring tools that are timely, scalable, and financially sustainable. Sentinel-2, the European Space Agency's Earth observation program, meets all three conditions.

What is Sentinel-2

Sentinel-2 is a constellation of two satellites (2A and 2B) from the European Union's Copernicus program that captures multispectral imagery of the entire Earth's surface freely and openly.

Key specifications

| Parameter | Value | |-----------|-------| | Spatial resolution | 10 m (visible + NIR bands), 20 m (red edge + SWIR), 60 m (atmospheric) | | Temporal resolution | 5 days (with both satellites) | | Spectral bands | 13 bands | | Coverage | Global | | Cost | Free | | Format | GeoTIFF, JP2 |

The 10-meter resolution may seem coarse compared to drone imagery (5 cm) or commercial satellites (30-50 cm), but for regional-scale environmental monitoring it is more than sufficient — and the 5-day revisit period is unbeatable.

How to access the imagery

Copernicus Browser

The most direct access is through the Copernicus Browser. It allows you to:

Search images by location and date.
Filter by cloud cover percentage — critical in Colombia where cloud cover is a constant challenge.
Download individual images or time series.
Visualize vegetation indices directly in the browser.

Copernicus Data Space API

For projects requiring automated download of large volumes, the API allows programmatic queries and downloads. At GEOSAT we use Python scripts that query the API, filter by cloud cover, and automatically download useful scenes for our areas of interest.

Google Earth Engine

Another powerful option is accessing Sentinel-2 through Google Earth Engine (GEE), which allows analysis directly in the cloud without needing to download the data.

Vegetation indices for environmental monitoring

Vegetation indices transform spectral bands into interpretable indicators. The most useful for environmental monitoring in Colombia:

NDVI (Normalized Difference Vegetation Index)

The best-known index. It measures the difference between near-infrared (NIR) and visible red reflectance.

Values between -1 and 1. High values (> 0.6) indicate dense, healthy vegetation.
Useful for: General vegetation cover monitoring, large-scale deforestation detection.
Limitation: Saturates in very dense tropical forests where all vegetation appears the same.

NBR (Normalized Burn Ratio)

Uses NIR and SWIR bands to detect burned areas and assess fire severity.

Useful for: Post-fire monitoring, damage assessment in national parks.
In Colombia: Especially relevant during El Nino events.

NDMI (Normalized Difference Moisture Index)

Combines NIR and SWIR to estimate vegetation moisture content.

Useful for: Early detection of water stress, wetland monitoring, early fire warning.

NDWI (Normalized Difference Water Index)

Designed to identify water bodies and flooded areas.

Useful for: Reservoir level monitoring, flood detection, wetland tracking.

Deforestation detection

The workflow we use at GEOSAT for deforestation detection with Sentinel-2 follows these steps:

1. Reference image selection

Two images (or composites) from different dates are chosen — ideally from the same time of year to minimize phenological differences. Example: January 2024 vs. January 2025.

2. Index calculation

NDVI (or NBR) is calculated for both dates.

3. Change analysis

The recent date index is subtracted from the reference date. Significant negative values indicate vegetation loss.

4. Classification

A threshold is established to separate natural changes (seasonality) from actual deforestation. This threshold is calibrated with field data or high-resolution imagery.

5. Validation

Detected change polygons are validated with higher-resolution imagery (drone, Planet, Google Earth) or field visits.

Land-use change analysis

Beyond point deforestation, Sentinel-2 enables land-use change analysis over time:

Urban expansion: Identifying the growth of built-up areas over rural land.
Agricultural cover change: Detecting conversion from forest to pasture or crops.
Vegetation recovery: Monitoring reforestation programs or areas in natural regeneration.

In our work with CORANTIOQUIA and CORNARE, we have implemented these analyses to generate early warnings of cover change in their jurisdictions, enabling a more agile response to unauthorized interventions.

Limitations in Colombia

Sentinel-2 is not perfect, especially in the Colombian context:

Cloud cover. The main limitation. In regions like Choco or the Amazon, obtaining cloud-free images can be difficult. The solution: use temporal composites (medians over several months) or combine with Sentinel-1 (radar, which penetrates clouds).
Resolution for small parcels. At 10-meter resolution, parcels smaller than 0.5 hectares are difficult to analyze individually.
Automatic classification. Supervised classification algorithms require quality training data, which in many Colombian regions is scarce.

Recommendation for environmental authorities

If your corporation or environmental authority is not using Sentinel-2 as part of its monitoring system, you are leaving on the table a free tool that can detect changes in your jurisdiction every 5 days.

The first step requires no software investment: with Copernicus Browser and QGIS (both free) you can start monitoring today. The next step — automation, alerts, and multitemporal analysis — requires technical capacity that at GEOSAT we are prepared to support.