Modeling fire danger in data-poor regions: a case study from the Russian Far East

Abstract

Wildland fire is a widespread phenomenon that affects how many ecosystems function and often threatens human life and property. Development of fire danger rating systems, aimed at identifying critical periods of high fire danger at early stages of its occurrence, is an important step for proactive fire and resource management. Development of such systems relies on long-term records of fire occurrence as well as numerous data sources for supporting information, but accurate and spatially explicit information is not available in many regions of the world affected by fire. Global satellite systems are becoming a major source of information for data-poor regions. The present paper describes a framework for modeling fire danger at a regional scale using publicly available data sources and global satellite imaging. It details a fuzzy logic-driven fire danger model developed for the Russian Far East using remotely sensed data. Fire activity recorded by the MODIS (Moderate Resolution Imaging Spectroradiometer) active fire product was analyzed for 2001–05 as a function of various parameters. Model performance was evaluated against 2006 data. Fire danger was evaluated within the model using the ordered weighted averaging approach with fuzzification. The model produces three scenarios. All output model scenarios provide a meaningful representation of fire danger levels in the region with the ‘trade-off’ scenario being the most applicable to mapping fire danger during low fire activity seasons.