48
Chapter 6
are frequently located in urban or agricultural areas, often far from forested
regions, and therefore the weather measurements may not be sufficiently
representatives of stress conditions in fire-prone areas.
Satellite data, on the other hand, are acquired directly from the vegetation
canopy, and imply an intensive spatial sampling. On this ground, several
studies have shown the usefulness of satellite data to estimate fire danger
conditions (Chuvieco and Martín, 1994; Desbois et al., 1997). However, a
better understanding on the relations between spectral information and the
water content of plants is required. For that purpose, satellite observations at
different scales should be coupled with ground data, to better control
potential sources of noise (landscape patterns, diversity of species, weather
conditions, etc.).
Fire risk estimation requires continuous updating to tackle the temporal
and spatial variability of risk factors. Within a short-term perspective, new
sensors with improved spatial and spectral resolution are required. Currently,
NOAA-AVHRR data are the best choice for fire risk estimation, because this
sensor provides daily observations. However, the atmospheric interference
and radiometric inconsistency of off-nadir observations and cloud cover
reduce the actual temporal frequency of AVHRR images. Future sensors,
such as MODIS, may reduce these difficulties. Ideally, fire risk estimation
would require a sensor with a spatial resolution in the range of 100 to 1000
meters size, with daily observations in the visible, near infrared, middle
infrared and thermal infrared, and with internal calibration sources to assure
temporal consistency.
Another approach to fire risk estimation considers only static factors that
are not altered daily but in long-term trends, such as topography or
vegetation structure. For this latter purpose, remote-sensing systems may
provide fuel type maps, which are critical for fuel management and fire
behavior prediction (Burgan and Rothermel, 1984). Fuel types are defined
according to morphological characteristics of plants (size, volume to height
ratio, density, etc.) which are not easily discriminated with current satellite
systems (Chuvieco and Salas, 1996). Synergism of optical and microwave
data may improve current fuel-type mapping, if understory information may
be derived from them.
2.2
Fire detection
Fire detection implies a great sensitivity to middle infrared radiance, since
this band is very suitable to detect hot targets. Currently, fire detection from
space relies on AVHRR data, which includes a middle infrared band (channel
3). However this sensor is not well adapted to fire detection, because of its low
temporal coverage (1 image every 12 hours) and, specially, because of the
Chapter 6
are frequently located in urban or agricultural areas, often far from forested
regions, and therefore the weather measurements may not be sufficiently
representatives of stress conditions in fire-prone areas.
Satellite data, on the other hand, are acquired directly from the vegetation
canopy, and imply an intensive spatial sampling. On this ground, several
studies have shown the usefulness of satellite data to estimate fire danger
conditions (Chuvieco and Martín, 1994; Desbois et al., 1997). However, a
better understanding on the relations between spectral information and the
water content of plants is required. For that purpose, satellite observations at
different scales should be coupled with ground data, to better control
potential sources of noise (landscape patterns, diversity of species, weather
conditions, etc.).
Fire risk estimation requires continuous updating to tackle the temporal
and spatial variability of risk factors. Within a short-term perspective, new
sensors with improved spatial and spectral resolution are required. Currently,
NOAA-AVHRR data are the best choice for fire risk estimation, because this
sensor provides daily observations. However, the atmospheric interference
and radiometric inconsistency of off-nadir observations and cloud cover
reduce the actual temporal frequency of AVHRR images. Future sensors,
such as MODIS, may reduce these difficulties. Ideally, fire risk estimation
would require a sensor with a spatial resolution in the range of 100 to 1000
meters size, with daily observations in the visible, near infrared, middle
infrared and thermal infrared, and with internal calibration sources to assure
temporal consistency.
Another approach to fire risk estimation considers only static factors that
are not altered daily but in long-term trends, such as topography or
vegetation structure. For this latter purpose, remote-sensing systems may
provide fuel type maps, which are critical for fuel management and fire
behavior prediction (Burgan and Rothermel, 1984). Fuel types are defined
according to morphological characteristics of plants (size, volume to height
ratio, density, etc.) which are not easily discriminated with current satellite
systems (Chuvieco and Salas, 1996). Synergism of optical and microwave
data may improve current fuel-type mapping, if understory information may
be derived from them.
2.2
Fire detection
Fire detection implies a great sensitivity to middle infrared radiance, since
this band is very suitable to detect hot targets. Currently, fire detection from
space relies on AVHRR data, which includes a middle infrared band (channel
3). However this sensor is not well adapted to fire detection, because of its low
temporal coverage (1 image every 12 hours) and, specially, because of the