Detection and Monitoring of Wildfires by a Constellation of Small Satellites with Infrared Sensor Systems

1. Introduction

Fire activity is a global phenomenon characterized by strong spatial and temporal variability.  Wildfire is an important ecosystem disturbance with varying return frequencies, resulting in land cover alteration and change, and atmospheric emissions on multiple time scales. Catastrophic wildfires in the last few years have again stressed that sound fire management decisions (including justification of a possible severity ranking) on the deployment of fire fighters and of limited technical equipment on the ground, rely very much on timely and detailed information on location, intensity, direction and rate-of-spread of the fire fronts.

In general, information on wildfire activity is used for (Csiszar 2008):

• global change research, estimating atmospheric emissions and developing periodic global and regional assessments, (i.e. quantification of the fire impact on climate),
• fire management and ecosystem management planning,
• operational purposes (preparedness and wildfire suppression),
• development of informed policies.

Satellite earth observation is appropriate to provide repetitive data at spatial and temporal scales necessary for detection and monitoring of wildfires. Existing and currently planned meteorological and environmental satellite sensor systems are able to provide useful coarse scale data sets for: (a) fire danger assessment, (b) fire occurrence detection, and (c) post-fire assessment. Unfortunately, the relative coarse spatial resolution of these data sets – with a pixel size larger than 1km for detecting electromagnetic radiation in the mid and thermal infrared wavelength bands - is insufficient to accurately locate individual fire fronts and to assess their strength.  However, it is these attributes that are very important for fire management decisions.

2. Some Relevant Definitions

An international expert team in (Csiszar 2008) defined the “Fire Disturbance Essential Climate Variable” which includes Burned Area as the primary variable and two supplementary variables. The “Fire Disturbance Essential Climate Variable” is one of the Essential Climate Variables (ECV`s) defined by the Global Terrestrial Observing System (GTOS).

Burned Area is defined as the area affected by human-made or natural fire and is expressed in units of area such as hectare (ha) or square kilometre (km²). Information on Burned Area, combined with other information (combustion efficiency and available fuel load) provides estimates of emissions of trace gases and aerosols. Measurements of Burnt Area can be used as a direct input (driver) to climate and carbon-cycle models.

Active Fire is the location of burning at the time of the observation and is expressed in spatial coordinates (or by an indicator of presence or absence of fire in a spatially explicit digital raster map, such as a satellite image). Detection of active fires provides an indication of regional, seasonal and inter-annual variability of fire frequency or shifts in geographical location and timing of fire events. Active fire information is also required by some algorithms used to generate burned area products. Detection of active fires can also serve as part of the validation process for burned area products.

Fire Radiative Power (FRP) is the rate of emitted radiative energy by the fire at the time of the observation and is expressed in units of power, such as Watts (W). The methodologies to derive FRP use physical-empirical approaches to derive rates of total emitted radiative energy from narrow-band, unsaturated radiance measurements. There is a strong empirical relation between FRP and rate of combustion, allowing CO₂ emission rates from a fire to be estimated from FRP observations. Multiple FRP observations can in principle provide estimation of the total CO₂ emitted during the fire through estimating time-integrated Fire Radiated Energy.

The ratio the FRP (W) of a single fire front line to its length (m) is called Fire Line Strength (W/m). The fire line strength is a measure of the burning severity, which is directly related to the fire impact on the ecosystem. Estimates of the fire line strength derived from data obtained by a satellite or an aircraft sensor enable the distinction, for instance, between cleaning fires and devastating fires.

3. Active Fire Satellite Information - Currently Available Products and Gaps

Table 1 shows currently available major satellite derived Active Fire Products with Fire Radiative Power (FRP) information (Csiszar 2008). This is a service provided by currently available sensor systems, such as, the MODerate resolution Imaging Spectro-radiometer (MODIS) on the polar orbiting satellites “Terra” and “Aqua” or the Spinning Enhanced Visible and InfraRed Imager (SEVIRI) on the geostationary satellite Meteosat-8.

Table 1: Currently available major satellite derived Active Fire Products with Fire Radiative Power (FRP) information (Csiszar 2008).

Unfortunately, these products only marginally meet the requirements defined by the Global Climate Observation System (GCOS) Implementation Plan. Considering this situation, the Committee on Earth Observation Satellites (CEOS) identified the following information deficiencies (Csiszar 2008):

• no global products at the specified 250m spatial resolution and daily observing cycle exist,
• product continuity and consistency between products derived from the various sensors remains unresolved.

4. Germany’s Bi-spectral Infrared Detection (BIRD) Mission

The primary mission objective of the Bi-spectral InfraRed Detection (BIRD) satellite, which was piggyback launched in a 570 km circular sun-synchronous orbit on 22 October 2001, was detection and quantitative analysis of high-temperature events (HTE) such as wildfires and volcanoes (Briess et al 2003). In 2002 - 2004 BIRD very convincingly demonstrated the potential of unsaturated fire data obtained with an earth observation system with a spatial resolution of 200 - 250m for:

• the derivation of Active Fire Products with Fire Radiative Power (FRP) information and the estimation of the Fire Line Strength of individual fire fronts,
• the comparison of Active Fire Products, such as FRP, derived from different satellite sensors using data nearly obtained simultaneously.

The principal BIRD imaging payload includes the Bi-spectral Infrared Camera with channels in the Mid-Infrared (MIR: 3.4µm - 4.2µm) and Thermal Infrared (TIR: 8.5µm - 9.3 µm) spectral ranges, and the Wide-Angle Optoelectronic Stereo Scanner WAOSS-B with a nadir channel in Near-Infrared (NIR: 0.84µm – 0.90µm) spectral range. (The two other off-nadir stereo channels of WAOSS-B are not used for hotspot detection). The ground resolution of the BIRD nadir channels is 185m in the NIR and 370m in the MIR and TIR. The NIR, MIR and TIR channels have the same sampling step of 185m due to an over-sampling by a factor of 2 of the MIR and TIR data.

A unique feature of the BIRD push-broom type MIR and TIR sensor channels is the real-time adjustment of their integration time (Skrbek and Lorenz 1998). If the real-time sensor on-board processing indicates that some detector elements are saturated (or close to saturation) during the regular exposure, a second exposure is performed within the same sampling interval with a reduced integration time. The data of both exposures are merged during the on-ground processing. This “intelligent” procedure preserves a 0.2°K radiometric resolution for pixels at normal temperatures and eliminates detector saturation over high temperature targets (Skrbek and Lorenz 1998).

4.1 Example of MODIS-BIRD Data Comparison

Comparing near-simultaneously collected data from BIRD and lower spatial resolution sensors such as MODerate Imaging Spectro-radiometer (MODIS) on the US environmental satellite “Terra” provides a useful metric by which the enhanced performance of BIRD in terms of fire detection and characterization can be judged (Wooster et al 2003).

Figure 1 shows an area west of Lake Baikal, Russia, observed nearly simultaneously, i.e. within a 30-minute interval, by MODIS/“Terra”and BIRD on 16 July 2003. The comparison clearly shows that only the BIRD data allow an estimation of fire front attributes relevant for fire mangers.

Figure 1: Fragments of forest fire images at Baikal, obtained by MODIS and BIRD on 16 July 2003.  Images show the MIR image from each sensor, and detected hot clusters are colour coded according to their Fire Radiative Power (FRP) in Mega Watts (MW). The arrows in the MODIS image fragment show false alarms

Table 2 shows fire attributes of the numbered forest fire fronts, which were extracted from the BIRD image in Fig. 1 (right image) using dedicated algorithms.

Table 2: Attributes of numbered forest fire fronts in the BIRD image fragment in Fig. 1 (right) of the Lake Baikal area in Siberia, Russia obtained on 16 July 2003.

5. Required Classes of Satellite Instruments to Provide Active Fire Datasets

The provision of Active Fire information products requires data records of appropriate multi-spectral imagery, for example, through the following classes of instruments and satellites (Csiszar 2008):

• SEVIRI-class instruments, extended to a full set of geostationary meteorological satellites, providing frequent but lower spatial resolution observations,
• MODIS-class observations which should be extended by future “Aqua” and “Terra” type developments through the Visible Infrared Imaging Radiometer Suite(VIIRS) – as foreseen in the future US National Polar-orbiting Operational Satellite System (NPOESS) program,
• Future BIRD-type instruments, required for higher spatial resolution data acquisition with reduced spatial coverage, to augment the more frequent but lower spatial resolution datasets described above for detecting missing smaller and weaker fires.

A BIRD type sensor instrument was successfully piggy back launched on board of the German first Technologie Erprobungs Traeger (TET-1) into a sun-synchronous orbit on 22 of July 2012.

6. On-Orbit Verification ofOn-board Fire Detection and Analysis

Timelyprovision of information is essential to support decision-making of fire managers in fire suppression planning, crew mobilization and movement. Therefore, on-board processing of fire front attributes, including geo-referencing and their direct transmission to the user on the ground is a challenging task for small satellites, but it should be technically feasible.

Key procedures for on-board fire detection and analysis are pre-processing and extraction of fire attributes.

Pre-processing includes:

• Radiometric correction (using system correction files),
• Inter-channel co-registration (using system correction files),
• Geo-referencing (using on-board navigational information).

The fire detection and analysis extraction of fire attributes includes:

• Background classification for threshold adaptation: land, water, clouds, sun glints,
• Hotspot detection (based principally on the BIRD algorithm),
• Consolidation of hot pixels in hot clusters,
• Extraction of attributes of hot clusters, such as coordinates, FRP and, optionally, fire line strength, effective fire temperature and area.

On-orbit verification of on-board fire detection and analysis will be conducted on Germany’s Berlin InfraRed Optical System,(BIROS) small satellite, which is planned to be launched in 2013.  The BIRD mission, the successful launch of TET-1, and the planned on-board extraction of fire front attributes on BIROS are three important milestones on the way to the creation of a prospective Fire Monitoring Constellation.

7. Need of a Fire Monitoring Constellation Based on Small Satellites

The demand by fire mangers for repeated, fast, and detailed information for all fire fronts in the areas of surveillance is highly justified.   Wildfire activity is at maximum in the afternoon and optimum wildfire observation requires two consecutive data takes at 13 -15 and 16 - 18 clock local time. These local observation times are not convenient to the majority of earth observation missions.

The essential fire attribute information, obtained twice during the peak period of wildfire burning will not be delivered by existing or planned satellite sensors, because their spatial resolution is too coarse and most of these systems have local observation times before noon.  Therefore, a dedicated Fire Monitoring Constellation (FMC) must be implemented to secure (i) a spatial resolution of 200-300m and (ii) optimum observation times for fire detection and monitoring at afternoon.

The FMC must resolve the deficiencies identified by CEOS (Csiszar 2008, chapter 3) by provision of:

• Active Fire Products at the specified 250m spatial resolution within a daily observing cycle,
• continuity and consistency between Active Fire Products derived from the FMC and existing (see Table 1) and planned meteorological or environmental satellite sensors with fire-adapted IR bands, such as, the Sea Land Surface Temperature Radiometer (SLSTR) on Sentinel-3 and the IR sensor of Meteosat Third Generation (MTG) - with coarser spatial resolution (greater than 1km).

The FMC should consistof four small satellites of the BIRD/TET/BIROS class in low earth orbit and be equipped with compact and intelligent IR sensors and on-board data processing.  Active Fire Products from a FMC will be of great value also for fire ecologists, health organizations and national administrations worldwide (Oertel 2005).

8. References

1. I. Csiszar (Univ. of Maryland), ECV T13: Fire Disturbance, Assessment of the Status of the development of standards for the Terrestrial Essential Climate Variables, Global Terrestrial Observation System (GTOS) – draft report, Rome, (2008).
2. K. Briess, H. Jahn, E. Lorenz, D. Oertel, W. Skrbek and B. Zhukov, Remote Sensing Potential of the Bi-spectral InfraRed Detection (BIRD) Satellite. Int. J. Remote Sensing, 24, 865-872 (2003).
3. W. Skrbek and E. Lorenz, HSRS – An infrared sensor for hot spot detection. Proc. SPIE, 3437, 167–176 (1998).
4. M. Wooster, B. Zhukov and D. Oertel, Fire radiative energy release for quantitative study of biomass burning: derivation from the BIRD experimental satellite and comparison to MODIS fire products, Remote Sens. Environm., 86, 83-107 (2003).
5. D. Oertel (German Aerospace Center - DLR), ECOFIRE – Study on Scientific Assessment of Space-borne High Temperature Event Observing Mission Concepts, (ESA/ESTEC Contract 17690/30/NL/FF) Berlin, (2005)

[1]With contributions from: Olivier Arino (European Space Agency), Raffaella Geraci (United Nations Food and Agriculture Organization), Louis Giglio (Science Systems and Applications Inc., USA), Johann G. Goldammer (Global Fire Monitoring Center, Max Planck Institute for Chemistry, Germany), William de Groot (Natural Resources Canada, Canadian Forest Service), Christopher O. Justice (University of Maryland, USA)

Shobha Kondragunta (National Oceanic and Atmospheric Administration, USA), Elaine Prins (University of Wisconsin, Madison – Cooperative Institute for Meteorological Satellite Studies - Consultant, USA), Reuben Sessa (United Nations Food and Agriculture Organization), Kevin Tansey (University of Leicester, UK), Martin Wooster (King’s College London, UK).

[1]With contributions from: Boris Zhukov, Eckehard Lorenz, Volker Tank, Michael Hess, Stefan Voigt, Franz Schreier, Thomas Holzer-Popp, Doris Hiepe (German Aerospace Center /DLR/, Germany), Martin Wooster (Department of Geography King’s College London /KCL/, University of London, United Kingdom), Florian Siegert (Remote Sensing Solution GmbH /RSS/,Germany), Johannn G. Goldammer (Global Fire Monitoring Center(GFMC/, Max Planck Institute for Chemistry, Germany), Susana Martinez (Ingeneria y Servicios Aerospaciales S. A. /INSA/, Spain), Hendrik Lübberstedt (OHB-System /OHB/Germany).

Authors

Dieter Oertel, Astro- und Feinwerktechnik Adlershof GmbH, Albert-Einstein-Str. 12, D-12489 Berlin, Germany, Tel.: +49 172 9450074, Fax: +49 35477 51778, e-mail: kdoertel@t-online.de

Eckehard Lorenz & Winfried Halle, Deutsches Zentrum für Luft- und Raumfahrt, Rutherfordstr. 2, D-12489 Berlin, Germany

Reprinted with permission, United Nations Office for Outer Space Affairs. Original content published from the booklet, "Geoinformation for Disaster and Risk Management."