Vegetation Change & Forest Dynamics in Western Canada

This project investigates vegetation disturbance and forest cover change across two study areas in western Canada using freely available satellite imagery and open-source R tools. It demonstrates a full remote sensing workflow — from raw data ingestion and preprocessing through time series analysis, change detection, and interpretation.

---

Study Areas

Area	Province	Analysis Focus
Yellowhead Country	Alberta	MODIS NDVI time series & BFAST disturbance detection
Williams Lake	British Columbia	Landsat land cover change & forest area dynamics

---

Data Sources

• MODIS MOD13Q1 — 16-day NDVI composite, 250 m resolution, 2001–2020
  • Source: NASA LP DAAC
• Virtual Land Cover Engine (VLCE) — Annual Landsat-derived land cover, 30 m resolution, 1984–2016
  • Source: Hermosilla et al. (2018), Canadian Journal of Remote Sensing

---

Methods

Part 1: MODIS NDVI Time Series (Alberta)

• Loaded and parsed a 460-layer MODIS NDVI time series
• Computed pixel-wise median NDVI composite for baseline visualization
• Extracted average NDVI for two regions of interest (ROIs)
• Analyzed growing season (May–September) NDVI patterns
• Applied BFAST Monitor to detect structural breaks in vegetation greenness and distinguish between permanent loss and recovering disturbances

Part 2: Land Cover Change (BC)

• Loaded a 33-year Landsat land cover time series (12 classes)
• Reclassified land cover into a binary forested / non-forested scheme
• Computed year-on-year lagged differences to identify annual forest gain and loss pixels
• Calculated annual forest gain, loss, and net change in hectares
• Visualized temporal trends in forest area dynamics with ggplot2

---

Key Results

• ROI A (Alberta): BFAST detected an abrupt NDVI decline around 2011 with no subsequent recovery, consistent with a stand-replacing disturbance such as wildfire or clear-cut harvesting
• ROI B (Alberta): A disturbance detected around 2008 was followed by a gradual NDVI increase, indicating natural vegetation recovery
• Williams Lake (BC): Clear cyclical patterns of forest loss (harvesting) and gain (regeneration) are visible across the 33-year record, with net forest change varying considerably year to year

---

---

R Packages Used

Package	Purpose
terra	Raster data loading, processing, and analysis
sf	Vector spatial data
bfast	Break detection in time series (BFAST Monitor)
dplyr / tidyr	Data manipulation
lubridate	Date parsing
stringr	String manipulation for filename parsing
ggplot2	Data visualization

---

---

📚 References

• Didan, K. (2015). MOD13Q1 MODIS/Terra Vegetation Indices 16-Day L3 Global 250m SIN Grid V006. NASA EOSDIS LP DAAC.
• Hermosilla, T., et al. (2018). Disturbance-informed annual land cover classification maps of Canada's forested ecosystems. Canadian Journal of Remote Sensing, 44(1), 67–87.
• Verbesselt, J., Zeileis, A., & Herold, M. (2012). Near real-time disturbance detection using satellite image time series. Remote Sensing of Environment, 123, 98–108.

---