Wildfire Impact: Latakia, Syria - 2025

This repository analyzes wildfire impacts in Latakia, Syria using:

• Sentinel-2 Level-2A imagery from a STAC API
• FIRMS active fire detections
• Spectral indices (NDVI, NDMI, and NBR)
• Static and animated visual outputs for comparison over time

The workflow is centered around the notebook wildfire_impact_latakia.ipynb, with reusable helper scripts in this project root.

Project Components

• build_s2_timeseries.py: queries Sentinel-2 from STAC, masks invalid pixels using SCL classes, and builds monthly composites plus index time series.
• calc_spec_indices.py: index formulas (NDVI, NDMI, NBR).
• fetch_firms.py: pulls FIRMS data in 5-day API windows and merges results.
• plotting.py: static plotting helpers for FRP, index trends, and before/after/now map comparisons.
• *_aoi.geojson: AOI boundaries.
• *_time_series.csv, *_camp_forecast.csv: previously generated outputs.

Prerequisites

• Python 3.10+
• Access to a STAC endpoint serving sentinel-2-l2a
• NASA FIRMS API key

Setup

1. Create and activate a virtual environment.

Windows (PowerShell):

python -m venv .venv
.\.venv\Scripts\Activate.ps1

2. Install required packages.

pip install pandas geopandas shapely pystac-client xarray odc-stac dask python-dotenv matplotlib seaborn pillow ipython

3. Create a .env file in the repository root:

AWS_STAC_API_URL=https://your-stac-api.example.com
MAP_KEY=your_firms_map_key

Environment variables used by scripts:

• AWS_STAC_API_URL: used in BuildS2TimeSeries to open the STAC client.
• MAP_KEY: used by fetch_firms.py for FIRMS API requests.

Typical Workflow

1. Define AOI and date range.
2. Build monthly Sentinel-2 composites and indices with BuildS2TimeSeries.extract_time_series(...).
3. Fetch FIRMS detections with fetch_firms(...).
4. Plot static comparisons with helpers in plotting.py.

Minimal Usage Example

from build_s2_timeseries import BuildS2TimeSeries
from fetch_firms import fetch_firms

builder = BuildS2TimeSeries()

# Option A: use AOI bounds from a GeoJSON.
# Option B: create a small bbox around a point.
bbox = builder.define_bbox(lat=35.0, lon=36.0, rad=500)

ds = builder.extract_time_series(
		aoi_bbox=bbox,
		aoi_name="example_aoi",
		start_date="2025-01-01",
		end_date="2025-12-31",
)

firms_df = fetch_firms(
		source=["VIIRS_SNPP_NRT", "VIIRS_NOAA20_NRT"],
		bbox=bbox,
		start_date="2025-07-01",
		end_date="2025-08-01",
)

Outputs

• xarray.Dataset from extract_time_series(...) containing:
  • monthly red, green, blue
  • monthly NDVI, NDMI, NBR
• FIRMS DataFrame with duplicate detections removed

Notes

• Cloud and invalid pixel masking is based on Sentinel-2 SCL classes.
• Sentinel-2 search currently uses eo:cloud_cover < 20.
• Monthly values are aggregated using median and interpolated to monthly starts.

Todo

• Add a pinned requirements.txt.
• Add tests for index calculations and basic STAC/FIRMS input validation.