Dynamic Characteristics of Correlation between Cropland and Forest Active Fires and Population Density in Mainland Southeast Asia

Abstract

Anthropogenic activities including slashing and burning (or swidden agriculture), illegal logging, and agricultural residue burning, are important drivers and driving mechanisms for the occurrence and development of global active fires especially in the tropics. In the past, limited by the access to the global datasets of gridded population density and consistent active fires, the research on the correlation between the occurrence and development of active fires and human activities was relatively insufficient. It is of great significance to clarify the relationship between human and fire for the sake of revealing the nature of fire regime, estimating carbon emissions and public health management. With Moderate Resolution Imaging Spectroradiometer (MODIS) Collection 6 (C6) active fire products provided by the Fire Information for Resource Management System (FIRMS) and LandScan Global Population Database (e.g., population density) developed by the Department of Energy's Oak Ridge National Laboratory (ORNL) during 2003-2019, the dynamic characteristics of spatial and temporal correlations between occurrence intensity of agricultural and forest active fires and population distribution in Mainland Southeast Asia were quantitatively revealed through GIS-based fishnet (1km) statistical analysis and bivariate spatial correlation. The results show that: (1) Nearly 80% of the "fire-affected grids" in Mainland Southeast Asia were positively correlated with population distribution during the study period. Grids with medium positive correlation and above (r>0.4) accounted for approximately 75%, mainly distributed in areas with low population density (<25 persons/km2) and covered by forest, including central and eastern Thailand, eastern and western Myanmar, northern Laos and northeastern Cambodia. (2) Cropland "fire-affected grids" in Mainland Southeast Asia during 2003-2019 have more population than forest fire-affected grids. More than 90% of the agricultural fire grids and nearly 3/4 of the counterpart of forest fire were positively correlated with population distribution. (3) The grids of strong positive correlation within agricultural fire area were mainly distributed in areas with population density of 25~100 persons/km2, such as southern Myanmar, and central and eastern Thailand. In contrast, the corresponding grids within forest fire area were mostly seen in areas with population density below 25 persons/km2, which include eastern and western Myanmar, northern Laos, and northeastern Cambodia. (4) The correlation between occurrence intensity of agricultural and forest fires and population distribution during 2003-2010 was significantly higher than that during 2011-2019, while the annual average population increased obviously in cropland and forest fire grids in the second period.

Keywords

• active fire
• cropland fire area
• forest fire area
• population density
• fishnet analysis
• mainland southeast asia