Remote Sensing

Remote sensing (RS) is the measurement of a quantity associated with an object by a device not in direct contact with the object. Passive satellite remote sensing is based on instruments or sensors which measure electromagnetic radiation coming from the Earth-atmosphere system. On the other hand active remote sensors ‘throw’ beams of radiation on the Earth-atmosphere system and measure ‘backscattered’ radiation. The back-scattered radiation is converted to geophysical quantities. The intensity of reflected and emitted radiation to space is influenced by the surface and atmospheric conditions. Thus, satellite measurements contain information about the surface and atmospheric conditions. Remotely sensed imagery acquires information in different wavelengths, representing different parts of the electromagnetic spectrum.

EMspecturm

Figure 1:  Electromagnetic spectrum.  Passive and active remote sensing instruments may employ a few to many wavelengths depending on the intention of the sensor design.

Satellite remote sensing observations vary based on several factors including different instruments/sensor, satellite orbits around the Earth, spatial and temporal coverage, and the geophysical quantities derived from the measurements.  Another important component is the quality and accuracy of the quantity retrieved.

So what applications of remote sensing exist for fire and what is the underlying science?  Satellite-based RS is employed for fire weather/season outlook, situational awareness by firefighting operations, modeling, air quality, and post-fire remediation. For our purposes in data assimilation, knowing where the fire is in real-time is a crucial ingredient that can be provided by improved remote sensing capabilities. Active fire detection takes advantage of the excessive black body radiation at typical fire temperatures as combustion releases fuel energy as radiation and emitted gases and aerosols.  The optimal wavelength for satellite-based RS fire detection lies in the 3 to 4 microns (µm) region owing to Planck’s law describing peak radiance for fire temperatures (Figure 2) and very little atmospheric attenuation in this wavelength window.

bbR

Figure 2:  Black body radiation according to Planck’s law.   While typical flaming temperatures emit radiance in the 2.7-3.0 µm, the 4 µm is best to avoid solar contribution and atmospheric attenuation of the signal.

Post-fire burn assessment, including estimating burned area, often uses multi-temporal change detection change in radiometric signal. For example, a commonly used measure, or index, uses the difference in signal between the near-infrared (NIR) and short-wave infrared (SWIR), typically in the 0.8 and 2.0 m range, respectively.  The difference is normalized to account for differences between scenes.  This ratio is referred to as the Normalized Burn Ratio (NBR) and the change in pre-and post-fire estimates is the Differenced NBR (dNBR). See Key et al. (1999) for more details.

NBR = (NIR–SWIR) / (NIR+SWIR) >>> dNBR = Pre NBR – Post NBR

Current research within this project focusing on satellite-based remote sensing seeks to determine the level of accuracy which can be obtained to observe the development of past and real-time fires, and understand the interplay of fuel and fire conditions, fire intensity, and emissions.  A critical component of this work is calibrating and validating RS detections and characterization.  We employ a suite of tools including ground, tower, and airborne-based (UAS and aircraft) sensors to address the cal/val objectives.  The following images highlight field and laboratory cal/val research, as well as end-user (i.e. operational) outreach.

rimfire

Figure 3:  (Top Left) On the ground at the Rim fire, California (September 2013), to conduct fire assessment and user outreach/communication; (Top and Bottom Right) Kruger National Park, South Africa, (August 2014) field work to perform cal/val; (Bottom Left) Prescribed fire in Henry Coe State Park, California, (November 2011) to conduct cal/val of remote sensing estimates.

umdlab

Figure 4: University of Maryland Fire Protection and Engineering fire lab.  Oak leaf burn on load scale with sensors placed overhead and handheld.