Geoscience Reference
In-Depth Information
However, this degree of change has traditionally been considered a separate and
even more elusive phenomenon referred to as fire severity.
3.4.2 Severity
Fire severity is a measure of the ecological impact of fire. However, like intensity,
severity is difficult to define in any practical way that can be applied across vari-
ous ecosystems or at various spatial or temporal scales. We can see some of these
difficulties in our previous examples of the fire in the grassland and the fire in the
mixed conifer forest. The grassland fire resulted in complete removal of vegetation
while in the forested example much of the biomass remained intact after the fire.
Therefore, it would seem that the grassland fire was far more severe (and by exten-
sion arguably more intense). However, examining changes caused by each fire, we
see that the grassland fire has few long term effects on the landscape, while some
patches of the forest fire area will carry changes for centuries.
As fire has been increasingly recognized as a natural part of ecosystems rather
than a negative disturbance, it has become even more difficult to determine the sorts
of negative effects we had traditionally defined as “severe.” Because of these sorts
of complications and differences we have tended to look at changes in canopy cover
to estimate severity in forested areas, while we have tended to look at soil changes in
areas without extensive forest canopy. Unfortunately, these two traditions can cause
confusion and they can even cause conflicts for analysts trying to map the severity
of fire events that burn both forests and grasslands.
In the last decade there have been efforts to refine our understanding of severity
and develop both field and remote sensing techniques to better map it. It has also
become increasingly clear that mapping fire severity can be accomplished effec-
tively with satellite imagery if we analyze the fire effects on both vegetation and
soils that are detectable from above. Thematic Mapper (TM) multispectral data sets
have become the standard tools for mapping fire severity (Chuvieco and Congalton
1988
; Collins and Woodcock
1996
; Medler and Yool
1997
; Patterson and Yool
1998
;
Rogan and Franklin
2001
). One of the key capabilities of this type of multispectral
data is the production of band ratio images. TM band ratios have been used to map
fire effects since at least the early 1990s (Lopez and Caselles
1991
).
By the late 1990s, Key and Benson refined a specific band-ratio algorithm that
used TM satellite spectral bands to map and quantify the degree to which areas on
the ground had undergone vegetation removal and soil changes associated with fire
(Key and Benson
1999
; Key et al.
2002
). This Normalized Burned Ratio differenc-
ing algorithm (dNBR, or sometimes simply NBR when using only post fire data)
is now widely used in the U.S. by groups including the National Park Service and
Forest Service. Key and Benson define burn severity as a “scaled index gauging the
magnitude of ecological change caused by fire.” However, the dNBR was devel-
oped for the forests of western North America and more work is required to develop
optimal multispectral algorithms for mapping severity in a broad range of vegetation
