Google earth Engine PK

🌍 Vegetation Drought Monitoring in Gilgit, Pakistan (2010–2025) 🌱

Using MODIS MOD13Q1 (250m NDVI, 16-day) data and Google Earth Engine, we analyzed vegetation health and drought severity across Gilgit from 2010 to 2025.

🔎 Key steps in the analysis:
✅ Monthly NDVI composites were generated for the entire study period.
✅ Long-term climatology (mean & standard deviation) was calculated for each month.
✅ Standardized NDVI anomaly (z-score) was derived to detect vegetation stress.
✅ Drought severity was classified into No Drought, Mild, Moderate, Severe, and Extreme.
✅ Time-series analysis was performed to track NDVI and anomalies over 15 years.
✅ Final drought severity maps and statistics were exported for decision support.

📊 Applications of this work:

Monitoring drought impacts on agriculture 🌾

Supporting water and land resource management 💧

Climate change resilience planning in mountain ecosystems 🏔️

Early warning systems for food security 🍽️

This approach highlights how remote sensing and cloud computing (GEE) can provide scalable and timely insights for sustainable natural resource management.

🌍 Soil Erosion Risk Mapping in Swat Valley using RUSLE (Revised Universal Soil Loss Equation)
This work applies the RUSLE model in Google Earth Engine to estimate potential annual soil loss (t/ha/yr) across the Swat region. Soil erosion is a major environmental challenge in mountainous areas like Swat, where steep slopes, rainfall intensity, and land-use practices accelerate land degradation.
🔎 Model Inputs
R-Factor (Rainfall erosivity): Derived from CHIRPS daily rainfall data (2000–2024).
K-Factor (Soil erodibility): Extracted from SoilGrids silt content.
LS-Factor (Slope & Topography): Computed from SRTM DEM, combining slope steepness and flow accumulation.
C-Factor (Cover management): Estimated from MODIS NDVI (2023) representing vegetation cover.
P-Factor (Conservation practices): Based on ESA WorldCover land-use classes and their corresponding management factors.
🖥️ Outputs
Spatial distribution maps for each factor (R, K, LS, C, P).
Final Soil Loss (A-Factor) map showing estimated erosion risk across Swat.
Interactive histograms and statistics summarizing soil loss trends.
📊 Key Insights
High erosion risk areas are strongly linked to steep slopes, sparse vegetation, and high rainfall zones.
Agricultural and bare soil regions contribute significantly to soil loss, highlighting the need for conservation practices.
Mean annual soil loss (t/ha/yr) was computed for the region, providing a quantitative benchmark for policy makers.
🌱 Why It Matters?
Soil erosion not only reduces agricultural productivity but also causes siltation in rivers, landslides, and ecological imbalance. Mapping erosion hotspots helps in planning sustainable land management, reforestation, and soil conservation strategies in fragile mountain ecosystems like Swat.

🌍 District-wise Carbon Stock Estimation in Punjab, Pakistan (2024–2025)

This analysis uses Sentinel-2 imagery (Nov 2024 – Mar 2025) to estimate vegetation carbon stocks across Punjab at the district level.

✅ Workflow Highlights:

NDVI derived from Sentinel-2 bands (B8, B4).

Converted NDVI → Above Ground Biomass (AGB) (tons/ha).

Estimated Carbon Stock = AGB × 0.5.

Classified into Carbon Zones:

🌱 Low (0–25 tons/ha)

🌿 Medium (25–50 tons/ha)

🌳 High (>50 tons/ha)

✅ Outputs:

Spatial maps of NDVI, Carbon Stock, and Carbon Zones.

District-level statistics of mean carbon stock.

Bar chart visualization for easy comparison.

Highlighted districts with high carbon reserves (>50 tons/ha).

📊 This approach helps in:

Monitoring vegetation health & biomass.

Supporting climate change mitigation and carbon credit programs.

Guiding policy decisions for sustainable forest/agro-ecosystem management in Punjab.

🗺️ Landslide Detection using Google Earth Engine (Gilgit Region, 2022)

This workflow integrates multi-source remote sensing data to detect potential landslide-affected areas.

🔹 1. AOI (Gilgit) – A shapefile is loaded and centered for analysis.
🔹 2. Timeframes – Pre- and post-event periods are defined for both Sentinel-1 SAR and Sentinel-2 optical datasets.
🔹 3. Terrain Data – SRTM DEM is used to calculate slope, as steep terrain is more landslide-prone.
🔹 4. Sentinel-1 Analysis – VV backscatter difference highlights sudden surface changes (soil/rock displacement).
🔹 5. Sentinel-2 Analysis – NDVI difference indicates vegetation loss due to slope failure.
🔹 6. Risk Masking – A threshold-based mask combines:
• VV drop (< -0.7)
• NDVI drop (< -0.03)
• Slope > 10°
Result = Potential landslide zones 🧱
🔹 7. Visualization – Maps show input layers, masks, and final detected landslides.
🔹 8. Export – Detected landslide zones are exported as GeoTIFF.
🔹 9. Ground Truth Sampling – Random landslide (1) and non-landslide (0) points generated for validation.
🔹 10. Validation – Confusion matrix used to assess classification accuracy.
🔹 11. Output – Ground truth samples exported as CSV.
🔹 12. Legend – A custom interactive legend added for better interpretation.

⚡ Key Insight: By combining radar (Sentinel-1), optical (Sentinel-2), and DEM slope information, this approach provides a reliable method to detect landslides in mountainous regions like Gilgit.

🌍 This can support disaster risk management, hazard mapping, and resilience planning.