학술논문
Generating a long-term data series of burned area in Eastern Siberia using LTDR data (1984–2016).
Document Type
Article
Author
Source
Subject
*WILDFIRES
*SURFACE texture
*FORESTS & forestry
*LAND use
*STATISTICAL correlation
*DATA
*
*
*
*
*
Language
ISSN
2150-704X
Abstract
The paper presents the results of burned area (BA) mapping in Eastern Siberian forests using Land Long Term Data Record (LTDR) project data. A combined approach that includes an expert assessment of BAs detected by threshold algorithm is proposed. Comparison with Moderate Resolution Imaging Spectroradiometer (MODIS) product (MCD64A1 C6, 2001–2016) and Global Fire Emissions Database (GFED) v.4 data (1995–2016) showed good agreement with correlation coefficients of 0.97 and 0.93 with average relative errors 16.5% and 40.6%, respectively. It is concluded that the use of a combined approach based on expert assessment with additional conditions provides a higher (compared to fully automated algorithms) level of confidence in BA detection, especially over areas with mountainous terrain, complex ground surface texture, vegetation type and/or over areas located at high latitudes, where the occurrence of wildfires is extremely unlikely. [ABSTRACT FROM AUTHOR]