523
Soil and Vegetation Moisture Changes for Forest Fire Events
moisture conditions, which enhanced fuel flammability and increased fire activity.
The results reinforce the strong connection between fire occurrence and moisture
availability. As an effective indicator of moisture conditions, the NMDI demonstrates the potential for fire risk monitoring.
21.5 APPLICATIONS OF THE NMDI FOR
ACTIVE FOREST FIRE DETECTION
Soil moisture and vegetation water content can influence the occurrence and behavior of wildland fires. Since the NMDI can simultaneously monitor vegetation and
soil water content, it should provide valuable information about wildland fire conditions. We used the 2007 wildfires in southern Georgia (USA) and southern Greece
to investigate the NDMI’s ability to detect forest fires.
For realistic accuracy assessment, comparison with independent, direct fire observations is necessary. Given the strongly dynamic nature of active fires both in time
and space, collecting an enough set of independent, coincident in situ observations
is logistically difficult (Csiszar et al. 2006). Owing to the lack of suitable field data,
index performance is evaluated by using the MODIS fire products. Performance
measures (overall accuracy, commission error, and fire detection rate) extracted from
the statistical analyses using the confusion matrices are used to verify the capacity of
NMDI for active fire detection.
21.5.1 Study aRea and data
A severe drought in the southeastern United States created record-breaking fire
events along the Georgia/Florida border in 2007. The Sweat Farm Road Fire/Big
Turnaround fire complex began to burn in southeastern Georgia during the afternoon
hours of 16 April 2007, quickly exploded into a major fire, and became the largest
wildfire in Georgia history. The study area was located at latitudes 30.8°N to 31.5°N
and longitudes 82.0°W to 83.1°W.
Throughout the summer of 2007, a series of massive forest fires broke out in
several areas across Greece (http://en.wikipedia.org/wiki/2007_Greek_forest_fires).
The most destructive and deadly fires raged from 23 to 27 August mainly in western
and southern Peloponnese as well as in southern Euboea (Athens News Agency,
2007). This study mainly focused on the Peloponnese Peninsula (36.4°N to 38.4°N,
21.0°E to 23.5°E), where woody savannas, mixed forest, and cropland occupy a great
part of the land area.
The dataset is composed of MODIS L1B calibrated radiance (MOD02, 1 km,
version 5), L1A geolocation data (MOD03, 1 km, version 5), and thermal anomalies,
fires, and biomass burning product (MOD14, 1 km, version 5) acquired over the
study areas for the fire periods. Reflectance from MODIS solar reflective bands 1
(0.62–0.67 µm), 2 (0.84–0.876 µm), 6 (1.628–1.652 µm), and 7 (2.105–2.155 µm) are
used to derive the NDVI and NMDI. The reflectance from MODIS bands 1 and 2,
along with the brightness temperatures derived from MODIS thermal infrared band
32 (11.77–12.27 µm), is employed to flag cloud pixels on the basis of the method
developed by Giglio et al. (2003). The land/sea mask obtained from the MODIS L1A
Soil and Vegetation Moisture Changes for Forest Fire Events
moisture conditions, which enhanced fuel flammability and increased fire activity.
The results reinforce the strong connection between fire occurrence and moisture
availability. As an effective indicator of moisture conditions, the NMDI demonstrates the potential for fire risk monitoring.
21.5 APPLICATIONS OF THE NMDI FOR
ACTIVE FOREST FIRE DETECTION
Soil moisture and vegetation water content can influence the occurrence and behavior of wildland fires. Since the NMDI can simultaneously monitor vegetation and
soil water content, it should provide valuable information about wildland fire conditions. We used the 2007 wildfires in southern Georgia (USA) and southern Greece
to investigate the NDMI’s ability to detect forest fires.
For realistic accuracy assessment, comparison with independent, direct fire observations is necessary. Given the strongly dynamic nature of active fires both in time
and space, collecting an enough set of independent, coincident in situ observations
is logistically difficult (Csiszar et al. 2006). Owing to the lack of suitable field data,
index performance is evaluated by using the MODIS fire products. Performance
measures (overall accuracy, commission error, and fire detection rate) extracted from
the statistical analyses using the confusion matrices are used to verify the capacity of
NMDI for active fire detection.
21.5.1 Study aRea and data
A severe drought in the southeastern United States created record-breaking fire
events along the Georgia/Florida border in 2007. The Sweat Farm Road Fire/Big
Turnaround fire complex began to burn in southeastern Georgia during the afternoon
hours of 16 April 2007, quickly exploded into a major fire, and became the largest
wildfire in Georgia history. The study area was located at latitudes 30.8°N to 31.5°N
and longitudes 82.0°W to 83.1°W.
Throughout the summer of 2007, a series of massive forest fires broke out in
several areas across Greece (http://en.wikipedia.org/wiki/2007_Greek_forest_fires).
The most destructive and deadly fires raged from 23 to 27 August mainly in western
and southern Peloponnese as well as in southern Euboea (Athens News Agency,
2007). This study mainly focused on the Peloponnese Peninsula (36.4°N to 38.4°N,
21.0°E to 23.5°E), where woody savannas, mixed forest, and cropland occupy a great
part of the land area.
The dataset is composed of MODIS L1B calibrated radiance (MOD02, 1 km,
version 5), L1A geolocation data (MOD03, 1 km, version 5), and thermal anomalies,
fires, and biomass burning product (MOD14, 1 km, version 5) acquired over the
study areas for the fire periods. Reflectance from MODIS solar reflective bands 1
(0.62–0.67 µm), 2 (0.84–0.876 µm), 6 (1.628–1.652 µm), and 7 (2.105–2.155 µm) are
used to derive the NDVI and NMDI. The reflectance from MODIS bands 1 and 2,
along with the brightness temperatures derived from MODIS thermal infrared band
32 (11.77–12.27 µm), is employed to flag cloud pixels on the basis of the method
developed by Giglio et al. (2003). The land/sea mask obtained from the MODIS L1A