Satellite Remote Sensing

We work with both “raw” satellite data and processed products, including on the development of algorithms to convert observations made from Earth orbiting satellites into information related to wildfire activity.  Satellites allow wide-overview observations of a region or even the entire globe, as frequently as ever 15 minutes from geostationary orbit, as well as providing striking geographic detail of particular areas using very high spatial resolution data.

We have been particularly involved in the development and evaluation of methods to detect active fires and estimate their “fire radiative energy” [FRE] and “fire radiative power” [FRP] from satellites. We use these metrics to estimate fuel consumption and fuel consumption rate in fires, and from this information you can calculate the emission of chemical  species to the atmosphere.  Examples of this and other type of satellite-based work are provided below.

Linking Earth Observation data to Climate and Landcover Processes

We have developed an archive of fires information for Indonesia stretching back to the 1980’s, conducted in collaboration with CIFOR and others interested in SE Asian fire history.  Below are two animations from raw AVHRR GAC imagery of the Island of Borneo from July to Dec 1997 (top image) and Jan to June 1998 (lower image). Pixels with fires burning within them have higher midwave IR brightness temperatures, appearing brighter than normal (sometimes even white). Clouds are cold (dark).  It is clear that there were huge areas of burning in the southern part of Borneo in 1997, and more towards the east in 1998. This was due to a mixture of human disturbance of the landscape and an El Nino related drought, whose exact consequences are being further investigated.  You can read more about this in Biogeosciences (2012).

July to Dec 1997

Jan to June 1998  

Satellite Fire Radiative Power Products

We have been involved in the development of FRP products from both polar orbiting and geostationary sensors, primarily as a way to estimate wildfire fuel consumption, carbon, trace gas and aerosol emissions to the atmosphere. As part of the EUMETSAT Land Surface Analysis Satellite Applications Facility (LSA SAF) we have developed an operational FRP product from the Meteosat series of satellites (Meteosat 8, 9 and 10) that carry the SEVIRI sensor.  You can find the product page here, including details of how to obtain the product and its format and data content.  The product is separated into four areas of the SEVIRI imaging disk, which can be combined if required.  Multiple products can be used to estimate fuel consumption rate and its changes over time, for example in the region of the Sahel over a two week period as shown at far right below (colours represent burning occurring in different landcover classes).

Further work in this area includes the development of the active fire detection and FRP product from the new SLSTR instrument, to be launched onboard ESA’s Sentinel-3 satellites from around 2014 (SLSTR is the instrument with the curved black Earth-viewing openings shown in the figure).  SLSTR is a follow-on instrument to the very successful (A)ATSR series, primarily used for very accurate mapping of sea surface temperature (SST).  SLSTR has a number of enhancements, including a much wider swath and dedicated “fire” measurement spectral channels, but still offers similar radiometric accuracy as the forerunner instruments.  We have worked to model the signature of fire as it will be measured by SLSTR ( figure below), and have developed a prototype fire detection and FRP characterisation algorithm for the sensor.  In the absence of real SLSTR data until the satellites launch, we have instead tested the algorithm on MODIS data, and have compared the results to the current Collection-5 MODIS Active fire algorithm (called ‘MOD14’).

 

 

Example of linking Earth Observation and Atmospheric Modelling

We also work to use satellite remote sensing in the modelling of smoke emissions to the atmosphere, both for the entire globe and for the monitoring of individual “extreme” events such as the 2010 Russian wildfires. We are helping to develop the operational ‘Atmospheric Service’ of the European Global Monitoring or Environment and Security (GMES) programme (www.gmes-atmosphere.eu), currently as part of the Monitoring Atmospheric Composition and Climate (MACC) EU FP7 project.  Here we collaborate with many groups, such as the European Centre for Medium Range Weather Forecasting (ECMWF) who operate the modelling system.  The service will provide data on atmospheric composition for recent years, the monitoring of present conditions, and short term (few days) forecasts of key atmospheric constituents, including those emitted by biomass burning. Further details can be found in Biogeosciences (2012).

An example event was  the Station Fire” that started August 26, 2009 just a few miles from NASA’s Jet Propulsion Laboratory (JPL) shown at left.  Around this time, the “snapshot” MACC “daily fire product” shown below, produced from the Global Fire Assimilation System (GFAS) developed as part of the GMES programme, depicts a map of daily worldwide fires FRP per unit area, as derived from satellite data.  Circled is what appears to be the Station Fire, seen even at this 0.5 deg grid cell size (Note that the GFAS system is now using 0.1 deg (10 km) gridding).  Below the FRP map is a 4-day animated map of  atmospheric carbon monoxide (CO) sourced from biomass burning.  You can see many large plumes, including from the Station fire. Maps for many atmospheric species are available at www.gmes-atmosphere.eu.

 

Another widely reported group of large fires burned around Moscow in August 2010. These fires involved the burning of boreal peat as well as forest, which seems to have produced thicker smoke than normal. At top below is a “true colour” NASA “snapshot” image from the MODIS sensor of these fires, courtesy of Jeff Schmaltz of the MODIS Rapid Response Team at NASA GSFC. The locations of fires can be spotted through the smoke (since smoke is partly transparent to mid-wave IR data used to identify fires).  The fires located due to their high brightness temperatures are highlighted in red.  Moscow can be seen at far left, and is clearly enveloped in thick smoke.  The smoke thickness can be quantified via its so-called “aerosol optical depth” (AOD), and the single-day map from the MACC system below (at bottom) shows the superimposition of both the fires FRP (red to yellow – indicating different rates of fuel consumption and smoke aerosol emission) and the modelled AOD of the smoke that is calculated as having been emitted from them (shades of blue). Please goto the GFAS web pages to find out more about this system, and how to obtain these types of real-time or archived data products.