FIRES: A Fire Reconnaissance System for Small Satellites
(IFFN No. 11 – July – 1994)
Up to now the AVHRR (Advanced Very High Resolution Radiometer) on NOAA satellites is the main satellite sensor used for detection of active wildfires. Because NOAA AVHRR was not devised for fire detection it has serious drawbacks. The main disadvantages are
saturation of the IR channels (the 3-5m m channel saturates already at 50° C) and
the low repetitive coverage of the same area
Planned future remote sensing systems, such as MODIS-N on EOS, only partly overcome these disadvantages. But these drawbacks can easily be overcome with dedicated specialized systems on small satellites operating on optimized orbits. For instance, with an adequate exposure control (via integration time or aperture) of the main sensor, saturation can be avoided. And a special satellite system consisting of 4 to 6 satellites could provide a good spatial and time coverage for forest fire detection. In the following the concept of a pre-operational satellite system will be described, which demonstrates the feasibility and usefulness of an operational system for vegetation fire reconnaissance. This pre-operational mission should be included in the remote sensing activities of the IGBP/IGAC/BIBEX programs.
The devised multisensor system as the main payload of a small satellite specialized for fire detection shall be equipped with a substantial computing power (on-board intelligence) in order to carry out an important part of the data processing. The system user should obtain the information he wants very fast (i.e. within a few minutes after the detection) and in highly compressed, user-friendly form. This means that the system must have the ability to reduce the data substantially by means of autonomous decision making and control. To guarantee this, a new smart sensor system, consisting of a fore field sensor for area of interest definition, a controllable main sensor for (precise) IR radiation measurements in the areas of interest , and a sensor “brain” for autonomous decision making and control, fire parameter estimation and fire classification, is proposed.
To demonstrate the feasibility and usefulness of a future operational small satellite system for fire reconnaissance, the proposed pre-operational FIRES satellite (Fire Reconnaissance System for Small Satellites) is not only devised for the pure detection of fire events in large areas, but also for its (more or less) precise location, the assessment of its extent (in space and time) and type, and the timeliness of providing this information to local authorities.
Fires to be detected are mainly forest and other vegetation fires, but also fires in industrial and power plants, including nuclear power plants (especially in remote areas or if kept secret), and fires of oil wells, platforms, pipelines, tankers and so on. Heat emitting forms of volcanic activity (smoke plumes, glowing avalanches, lava outflows, lava lakes) can be detected too. Furthermore, the sensor system can contribute to other remote sensing tasks related to heat emission.
Secondary objectives of FIRES are the estimation of vegetation damage and atmospheric pollution due to fire and the assessment of the regrowth of vegetation on burned areas.
In order to fulfill the main objectives of the system one needs on-board:
thematic data reduction and generation of thematic maps,
orbit determination and attitude measurements,
geocoding of fire parameters and thematic maps.
These challenging objectives are not solved up to now.
The Principal Structure of the Multi-Sensor System
The proposed smart sensor system consists mainly of three parts:
I. Fore Field Sensor: a forward looking sensor with
large swath width
low geometric resolution
few spectral channels (mainly in the 3-5m and 8-12m atmospheric windows)
Its main task is the definition of areas of interest by hot spot detection and the coarse estimation of fire parameters within the areas of interest.
II. Main Sensor: a multi sensor with
low swath width
medium to high geometric resolution
several channels from VIS to TIR
a controllable line of sight (in order to get directed to the areas of interest)
Its main task is the investigation of the areas of interest in order to determine fire and smoke plume parameters and to produce corrected image data (thematic maps) of fire regions. Further tasks of the main sensor are related to the secondary and other objectives of FIRES.
III. Sensor Brain: The sensor brain as the component of sensor intelligence uses modern computer architecture for the
Real-time processing of the Fore Field Sensor signals for defining the areas of interest and coarse fire parameter estimation
Real-time determination of the control information for the Main Sensor for
directing the line of sight to the areas of interest
exposure control to avoid saturation of the Main Sensor
Near real-time processing of the Main Sensor signals for
determination of fire and smoke plume parameters
fire type classification
(Geometric) correction of area of interest images
Evaluation of data from position and attitude measuring systems for geocoding of the data products.
Feasibility of Fire Reconnaissance from Space
First estimations (for a sun-synchronous orbit of 888km height) of signals to be detected by two IR sensors at 3.7m m and 8.5m m (for the parameters see chapter 5) show that small (sub-pixel) fires can be discriminated from the background, demonstrating the feasibility of forest fire detection with a moderate technological effort.
Tab.1.: Estimated minimum resolvable fire size independence on the temperature (for D*=1010cmÖHz/W)
Estimated minimum resolvable fire diameter at the fire temperature T =
for the fore field sensor, ground pixel size = 1420m
for the main sensor, ground pixel size = 265m
Fig. 2. Mission architecture of FIRES in relation to international local users with direct reception of high-level data products
Tab.2.: Preliminary orbit parameters and satellite characteristics
Satellite baseline characteristics
circular integer orbit
±0.05° per axis
X-band and UHF
The preliminary results presented in Table 1 show that the fire detection capability of the system is very good. Using the two IR channels, it is possible to estimate the fire temperature and the (sub-pixel) area of the fire. First investigations show that the necessary algorithms for this can be easily implemented on-board and can be made very fast (using look-up tables). This means that it is possible to generate user-friendly data products on-board in near real-time.
If it becomes possible to provide correct position data of the events (e.g. with a precision of some hundred meters) by on-board evaluation of the data from orbit position and attitude measuring systems, then the ambitious goals of FIRES can be fulfilled satisfactorily.
The preliminary orbit parameters and satellite characteristics are pointed out in Table 2. The planned launch date is 1998 for a pre-operational phase.
The peculiarities of this mission are not only the fire detection but also the following features:
first system which is dedicated to reconnaissance and remote sensing of fire,
the on-board orbit determination and navigation,
the on-board data processing for classification and thematic data reduction,
the on-board geocoding of data products,
the transmitting of high-level data products to the final local user.
Figure 1 shows the mission architecture. It should point out the inexpensive ground segment dedicated to defined user groups of a small satellite mission which are in direct contact with the satellite. They should get high-level data products from the satellite. Geocoded numerical data products without image information should be received by operational users with handy receivers and the geocoded image data are foreseen for fire management, fire modelling and local authorities.
From:Herbert Jahn and Klaus Brieß Address: Deutsche Forschungsanstalt für Luft- und Raumfahrt e.V. (DLR) Institut für Weltraumsensorik Rudower Chaussee 5 D – 12489 Berlin