Active Fires

Active Fires

Overview

The Active Fires KPI detects and visualises fire events that occurred during the analysis period in the area surrounding the site, showing their intensity and distance from the site. Information is presented via an interactive map: the size of the dots indicates the extent of the area affected by fire, while the colour represents the intensity level (FRP, Fire Radiative Power). A greater number of fires with high intensity indicates an area with high fire risk, with potential impacts on biodiversity, air quality, and human safety.

Detection is based on satellite thermal anomalies (hotspots) detected by VIIRS (Visible Infrared Imaging Radiometer Suite) sensors aboard the Suomi-NPP (NOAA/NASA) and NOAA-20 satellites, with near-real-time updates (latency of a few hours).

This KPI is a binary presence/absence indicator: it has no graded quality levels (A--E). The KPI is visible only if at least one fire was detected in the selected period.

---

Calculation Methodology

Active fire detection is based on the identification of thermal anomalies in the SWIR (Short-Wave Infrared) band of the VIIRS sensor (Schroeder et al., 2014). The algorithm compares the temperature of each pixel with neighbouring pixels; if the difference exceeds an adaptive threshold, the pixel is classified as a hotspot (potential active fire).

Variables recorded for each hotspot:

• FRP (Fire Radiative Power, MW) -- fire intensity; correlates with combustible biomass and emission rate
• Confidence -- detection reliability (low, nominal, high)
• Estimated area -- surface area affected by the thermal anomaly
• Distance from site -- geodesic distance from the study area

Processing pipeline:

1. GEE collection merge of Suomi-NPP and NOAA-20 VIIRS data
2. Per-collection .max() compositing on the confidence band, then collections added together
3. Bicubic smoothing at 30 m scale
4. Water masking (ESA Land Cover class 80)
5. Cluster detection via 8-connectivity labelling on non-zero pixels
6. Per-cluster metadata extraction: area (ha), centroid, min/max/mean confidence, distance from image centre

Visibility logic: the KPI is visible only if at least one fire has been detected in the selected period in the "Active Fires" maps. In the absence of events, the KPI is not displayed.

Data Sources

Near-real-time updates: 3--6 hour latency from satellite overpass.

---

What You See on the Platform

Maps > Active Fires > "Active Fires" (Map Layer)

Map Layer. An interactive map overlay showing fire hotspot clusters detected during the analysis period, colour-coded by confidence/intensity level.

Purpose: Answers the question "Where have fires occurred near this site, how intense were they, and how large an area did they affect?"

Description: The map displays coloured dots positioned at the centroid of each detected fire cluster. Dot size is proportional to the estimated fire area (hectares). Dot colour represents the confidence/intensity level, ranging from yellow (low) to dark red (very high). Users can hover or click on individual dots to see metadata: area in hectares, confidence level, FRP, and distance from the site centre. The site boundary is shown as a reference polygon. If no fires were detected in the selected time period, the layer is empty and the KPI is hidden.

How it's calculated: Each dot corresponds to a fire cluster identified through 8-connectivity labelling of VIIRS hotspot pixels. The cluster's centroid is plotted; its area is computed from the connected pixel count at 30 m resolution; its colour is determined by the maximum confidence value within the cluster (0 = No Fire, 1 = Low, 2 = Medium, 3 = High, 4 = Very High).

Legend:

Interpretation example: If the map shows three clusters within 5 km of the site -- two small yellow dots (low confidence, < 1 ha) and one large dark-red dot (very high confidence, ~15 ha) -- the site faces a significant fire risk from a large, intense fire nearby, while the two smaller detections may be lower-reliability signals or minor burn events.

---

Bibliography

1. Schroeder, W., Oliva, P., Giglio, L., Csiszar, I.A. (2014). The New VIIRS 375 m active fire detection data product: Algorithm description and initial assessment. Remote Sensing of Environment, 143, 85--96.

2. Giglio, L., Schroeder, W., Justice, C.O. (2016). The collection 6 MODIS active fire detection algorithm and fire products. Remote Sensing of Environment, 178, 31--41.

---

Detailed Methodology

See the Calculation Methodology section for the core computation. Additional processing details are documented here for expert users.

Technical Notes

• The KPI is visible only if at least one fire has been detected in the selected period; otherwise it is completely hidden from the platform interface.
• Unlike most other KPIs, Active Fires has no A--E quality classification. It is a binary presence/absence risk indicator, with severity communicated visually through FRP intensity and event count.
• The VIIRS algorithm can generate false positives from hot surfaces such as industrial plants, gas flaring, or volcanic activity. The "nominal" and "high" confidence thresholds reduce but do not eliminate false positives.
• The 375 m spatial resolution of VIIRS enables detection of small-scale fires, a significant improvement over previous MODIS sensors (1 km resolution).
• Near-real-time updates have typical latency of 3--6 hours from satellite overpass, making this one of the most temporally responsive KPIs on the platform.
• Fire clusters are detected using 8-connectivity pixel labelling after water masking, meaning adjacent hotspot pixels are grouped into a single event with aggregated metadata.
• This KPI is related to the Climate Change and Aridity risk KPIs: areas experiencing drought stress and elevated temperatures are more susceptible to fire events.