ADVANCED FIRE INFORMATION SYSTEM

1
ADVANCED FIRE INFORMATION SYSTEM AFIS I
AFIS is the 1st satellite based, near real time
fire information system developed to fulfill the
needs of both fire fighters and researchers in
Southern Africa.
MODIS fires detected between 1 3 September 2004
MODIS Direct Broadcast reception circle at CSIR
SAC
www.wamis.co.za/eskom/checkboxes/eskom.htm
Satellite Sensors
AFIS Sensor Web Fire Mapper displaying MODIS
fires in Southern Africa. This map is currently
being used by numerous organisations in the SADC
region.
MSG SEVIRI
Aqua Terra MODIS
The South African map displays numerous data
layers such as the ESKOM Transmission lines.
Additional data layers such as National parks,
roads, towns and higher resolution satellite
imagery are also available.

• Geo stationery orbit
• 12 spectral bands
• 4.8 km resolution
• 15 minute repeat cycle

• Polar orbit
• 36 spectral bands
• 1 km resolution
• 6 hourly repeat cycle

The map to the left displays a fire beneath a
power line detected by both satellite systems.
The purple pixels represent MSG and the red
pixels are from MODIS.
Fire Detection Algorithms
MSG Algorithm The current MSG algorithm is based
on the contextual approach as described by Flasse
Ceccato in 1996. The contextual algorithm which
was developed for the AVHRR sensor, makes use of
the 3.9µm and 10.8µm bands to discriminate fire
pixels from background pixels. In summary, the
algorithm calculates an approximation to the
expected background temperature, and then tests a
pixel to see if its temperature exceeds this
background temperature by a statistically
significant degree. The CSIR SAC is currently
working on a new algorithm that exploits the high
update rate of the MSG satellite to derive a
multi temporal fire detection algorithm. This is
achieved by using a Kalman filter to predict the
expected temperature, which can then be used to
identify pixels that deviate significantly from
their expected values. Early results have
indicated significant improvements in detection
accuracy with good comparisons with MODIS
detections.
Email SMS Alert System
GIS Analysis
AFIS has the ability of providing near real time
fire alerts to users in Southern Africa through
email or SMS messages. Alert information is
generated by the GIS analysis scripts that run
for each satellite overpass. Current
example ESKOMs Transmission line managers as
well as their Control Centre receives SMS
messages as soon as a fire is detected within a 5
km buffer of any power line. Information in the
SMS message will include, Latitude, Longitude,
distance to closest pylon, line name, date and
time of the fire occurrence.
A set of shell scripts has been developed to
handle operational GIS analysis requests. These
shell scripts enable the provision of operational
fire information to users within pre defined
areas with the possibility to also define buffer
zones. Current example The shell routines are
currently used to identify fires within a 5 km
buffer around all Transmission lines for South
Africas biggest power utility (ESKOM).
Daily / Weekly Email Report Service
AFIS can also provide email based fire reports
for pre defined areas. Users can specify an area
of not more than 2 x 2 degrees in size and
receive daily and weekly email reports of fires
within the pre defined zone. The advantage of
the email based system is that it doesnt require
the user to log into the web interface but
instead they receive a daily or weekly report in
their inbox. The emails could be used for
generating monthly reports of fire and also in
creating animations. Each email contains a
referenced .JPG image of the pre defined area
displaying all fires detected within the last day
or week. Additionally, a table is provided with
information about each fire located.
MODIS Algorithm The MODIS Rapid Response (RR)
code has been integrated in to the CSIR SAC MODIS
DB system. The RR code ingests MODIS L0 data and
produces a text file containing fire location
information for each overpass. Fire detection is
performed using a contextual algorithm (Giglio et
al., 2003) that exploits the strong emission of
mid-infrared radiation from fires. The algorithm
examines each pixel of the MODIS swath, and
ultimately assigns to each one of the following
classes missing data, cloud, water, non-fire,
fire, or unknown.
Contact pfrost_at_csir.co.za