Statistical analysis of active fire remote sensing data: examples from South Asia

Area of Science:

• Environmental Science
• Remote Sensing
• Spatial Statistics

Background:

• Active fires, primarily forest and vegetation fires, significantly impact ecosystems by releasing aerosols and greenhouse gases.
• These fires pose environmental threats in Nepal, Bhutan, and Sri Lanka, damaging flora and fauna and emitting toxic gases.

Purpose of the Study:

• To analyze the spatial behavior of active fires over one year in Nepal, Bhutan, and Sri Lanka.
• To assess the environmental impact of these fires using advanced statistical models and remote sensing data.
• To predict fire radiative power (FRP) and vegetation burning potential in unsampled areas.

Main Methods:

• Utilized NASA's MODIS remote sensing data for fire detection and analysis.
• Applied spatial statistics, including autoregressive models (Spatial Durban, Lag, Error, Manski, Kelegian Prucha), to study fire incidence.
• Employed spatial interpolation and kriging techniques to predict fire radiative power (FRP).

Main Results:

• Identified and analyzed the spatial patterns of active fires across Nepal, Bhutan, and Sri Lanka.
• Determined the best-performing autoregressive model based on pseudo R-squared values for fire behavior analysis.
• Successfully predicted fire radiative power (FRP) and vegetation burning potential using spatial interpolation.

Conclusions:

• Spatial statistics provide valuable insights into active fire behavior in data-scarce regions.
• Remote sensing data combined with statistical modeling can indirectly assess the impact of forest fires.
• This study offers a country-specific perspective crucial for environmental management in South Asia and other developing regions.