DEGRAD

Overview

The DEGRAD project is a research initiative that uses satellite imagery to monitor forest degradation in the Amazon. Unlike DETER’s near real-time alerts, DEGRAD provides a more detailed annual analysis of forest degradation patterns.

This dataset captures:

Forest degradation monitoring: Tracks areas where forests are being damaged without complete clearing
Annual editions: Data released as yearly reports with accumulated observations
Spatial polygons: Detailed geographic boundaries of degradation events
Municipality linkage: Enhanced version links degradation areas to affected municipalities
Historical coverage: Multiple years of degradation monitoring from 2007 onwards

DEGRAD data is valuable for understanding forest degradation as a distinct phenomenon from clear-cut deforestation, important for carbon accounting, biodiversity protection, and understanding transition stages toward complete forest loss.

Data Source and Methodology

DEGRAD monitoring: - Conducted by INPE’s forest monitoring programs - Uses satellite imagery interpretation to identify forest degradation signs - Focuses on selective logging, small-scale agriculture, forest fires, and other degrading activities - Released as annual editions with comprehensive analysis - Limited documentation available (original INPE documentation is sparse)

For information, visit INPE Forest Monitoring.

Important Data Characteristics

Data Organization

Important: DEGRAD data is organized differently than real-time systems. Key points:

Yearly editions: Data is organized by publication year (e.g., “DEGRAD 2016”), not event year
Mixed event years: A DEGRAD edition may contain degradation events from different years
- Example: DEGRAD 2016 edition may include events detected in 2015 or even earlier
Documentation limited: Original INPE documentation is minimal; users should be aware of potential inconsistencies

Spatial Integration

This package enhances the raw DEGRAD data by: - Intersecting DEGRAD spatial polygons with IBGE municipality boundaries (2019 version) - Providing municipality identification for each degradation event - Converting to Simple Features (SF) objects for spatial analysis

Note on CRS: Coordinate system metadata should be verified after loading, as original INPE data sometimes has unclear CRS information.

Available Dataset

degrad (Forest Degradation)

Detailed monitoring of forest degradation across the Legal Amazon.

Coverage: Legal Amazon region with focus on degradation detection
Time period: 2007 onwards (but events within editions may vary)
Spatial unit: Polygons/spatial geometries with municipality identification
Variables: Event date/year, degradation type/cause, area, municipality, state, edition
Data format: Simple Features (SF) spatial objects with geographic boundaries
Use cases:
- Distinguish degradation from complete deforestation
- Analyze forest degradation hotspots
- Understand selective logging extent
- Carbon stock assessment
- Forest fire impacts
- Long-term degradation trends

Function Parameters

1. dataset

Only one dataset is available:

dataset = "degrad"  # Forest degradation monitoring

2. raw_data

Controls whether to download the original data or the processed/enhanced version.

TRUE: Returns raw INPE data with minimal processing
FALSE: Returns treated data with English variable names, municipality identification from spatial intersection, and standardized formatting

raw_data = FALSE  # logical

Recommendation: Use raw_data = FALSE for most applications to get municipality-level information. ### 3. time_period

Specifies which year(s) of degradation events to download.

Available range: Generally 2007-2016 (check current availability)
Format: Single year, vector of years, or range

Important: When you request a year, you get events from that year regardless of which DEGRAD edition they appear in.

time_period = 2015              # single year
time_period = c(2010, 2015)     # multiple specific years
time_period = 2010:2015         # range of years

4. language

Output language for variable names and documentation.

"pt": Portuguese
"eng": English

language = "eng"  # character string

Data Structure

The returned data is a Simple Features (SF) spatial object with:

Spatial column: Geometric polygons representing degradation areas
year: Year when degradation was detected/event occurred
degradation_type: Type of degradation (logging, fire, agriculture, etc.)
area_hectares: Size of degradation event
municipality: Name of affected municipality
state: Brazilian state
edition: Which DEGRAD edition this event appears in
Additional attributes: Quality metrics, confidence levels (vary by edition)

Examples

# download treated forest degradation data from 2010 to 2012
data <- load_degrad(
  dataset = "degrad",
  raw_data = FALSE,
  time_period = 2010:2012,
  language = "eng"
)

Data Notes

Data Organization Complexity

The annual edition structure (e.g., “DEGRAD 2016”) mixed with variable event years within those editions means: - When you request year 2015, you get all detected 2015 events regardless of edition - Some 2015 events may appear in both DEGRAD 2015 and DEGRAD 2016 editions - Duplication is handled in the data loading process

Degradation Types

Common degradation types include: - Selective logging: Commercial timber extraction - Forest fires: Fire damage to forest areas - Agricultural clearing: Small-scale farming expansion - Mining: Degradation from mining activities - Other: Mixed or unclassified degradation causes

(Exact categories vary by edition; verify with your loaded data)

Spatial Considerations

Polygons not points: Each event is a geometric polygon, not a single location point
Municipality intersection: Treated data identifies all municipalities polygon overlaps
CRS verification: Check coordinate system after loading
Geometry validity: Some polygons may have validity issues; use st_is_valid() to check

Data Limitations

Limited documentation: INPE’s original documentation for DEGRAD is sparse
Mixed time periods: Events from different years appear in same edition
Possible inconsistencies: Classification and methodology may vary across editions
Detection limits: Minimum detectable degradation size varies by methodology/edition
Not real-time: This is annual analysis, not near-real-time detection like DETER