False Positives: Not every detection is a fire

Introduction

Not every detected heat source is an actual fire. False positives in our Wildfire Solution refer to detections where non-fire heat sources (like solar panels, industrial sites etc.) are mistakenly identified as potential fires. 

OroraTech’s WFS provides tools to help users distinguish genuine fire threats from false positives, ensuring that decision-makers have the most reliable data available.

The main features we recommend are:

• Cluster Confidence: Indicates how many satellites and algorithms have detected and processed the heat source; higher confidence suggests a more reliable detection.
• Fire Confidence: Shows the likelihood that a cluster represents an actual fire, based on weather, fire intensity, satellite confirmations, land type and other variables. To learn more read this section.

What Are Common False Positives?

Common sources of false positives include:

• Industrial sites and machinery.
• Volcanic activity or geothermal hot spots.
• Reflections from materials like solar panels, steel roofs, or water bodies.
• Naturally heated surfaces like rocks or dry, harvested fields.

These detections can occur in both rural and urban settings, with certain types of locations (e.g., cities, industrial zones) being more prone to false positives.

Minimizing False Positives with Fire & Cluster Confidence

1. Use Cluster Confidence to Confirm Detection Reliability

Cluster Confidence is determined by the number of satellites and algorithms that identify a heat source. A higher confidence level generally indicates that the heat source is significant enough to warrant further attention.

It represents the overall confidence that the cluster is a valid heat source and is primarily influenced by the number of independent hotspot detections by satellite and algorithms.

To serve the purpose of  early detection we developed OroraTech's Fire Confidence for effective decision making at an early stage of a potential fire explained below ⬇️

2. Use Fire Confidence to Assess Likelihood of Fire

Fire Confidence helps assess whether a cluster is likely to be a wildfire. This metric accounts for:

• The heat intensity of the source.
• The environment (e.g., forests vs. urban areas).
• Confirmations from multiple satellites.
• Land Cover type & flammability (how easily the surrounding area can burn)
  
  

_{Example of high Fire Confidences compared to Clusters with low cluster confidence levels}

Understand Location-Based Patterns

Determining the context of a detection can also help evaluate its legitimacy:

• Forest and Grassland Areas: Hotspot detections in these areas are more likely to represent actual wildfires due to the high flammability of vegetation.
• Urban or Industrial Zones: Detections in these areas are more susceptible to false positives, often triggered by buildings, machinery, or solar panels.

                                                    _{Clusters located in the middle of a forest tend to be real fires.}

Conclusion

While false positives are an inherent part of wildfire detection, OroraTech’s Wildfire Solution provides powerful tools to help users confidently interpret and verify detections.