Laboratory Measurements of Plant Drying:
Implications to Estimate Moisture Content from
Radiative Transfer Models in Two Temperate Species
Sara Jurdao, Marta Yebra, Patricia Oliva, and Emilio Chuvieco
Abstract
The estimation of live fuel moisture content (
LFMC
) is neces-
sary for fire danger assessment. Several studies have suc-
cessfully used satellite imagery to estimate
LFMC
, both using
empirical and simulation approaches (Yebra et al., 2013). The
latter are based on Radiative Transfer Models (
RTM
). They are
generally more robust and easier to generalize, but they rely
heavily on the proper parameterization. Since some of the
input parameters are associated with different physiological
processes, a better understanding of how those parameters
co-vary is necessary for constraining the simulation scenarios,
thus avoiding combinations of parameters that are unlikely
to occur (for instance, in temperate ecosystems, it is unlikely
to find simultaneously high values of leaf chlorophyll and
low values of leaf moisture).To improve parameterization of
RTM
models for
LFMC
estimation, we conducted a laboratory
experiment to measure trends in leaf and canopy variables of
two tree species broadly distributed in Eurosiberian climates:
Beech (Fagus sylvatica L.) and pedunculate Oak (Quercus
robur L.). Measurements of
LFMC
, equivalent water thick-
ness (
EWT
), dry matter content (
DMC
), chlorophyll (C
a+b
), leaf
area index (
LAI
), leaf angle distribution (
LIDF
), crown height
to width ratio (
CHW
) and plant reflectance were performed.
Significant positive correlations were found between
LFMC
and
EWT
(Rs >0.5), and negative ones were found between both pa-
rameters and C
a+b
(Rs <-0.3).
LFMC
and
EWT
were positively re-
lated to
DMC
and
LAI
, with lower correlation coefficients for the
latter. The effect of moisture variation in spectral reflectance
was also analyzed using two indices: the spectral angle (
SA
)
and the root mean square error (
RMSE
).The former contrib-
uted the most to the estimation of
LFMC
variations. Spearman
correlation coefficients (Rs) between
SA
and
LFMC
were 0.656
and 0.554 for F. sylvatica and Q. robur, respectively; while
for
RMSE
and
LFMC
they were 0.366 and 0.430, respectively.
Introduction
Fuel amount and fuel moisture conditions play a remarkable
role in the complexity of fire effects (Henry and Yool, 2002).
Fuel moisture content (
FMC
) affects fire ignition and propaga-
tion, as higher
FMC
slows the rate of fuel combustion (combus-
tibility) by decreasing ignitability (increase in ignition time
and decrease in ignition potential), and sustainability (dura-
tion of flaming combustion: Nelson, 2001). As a consequence,
FMC
is a critical component of a fire risk (Chuvieco
et al.,
2012; Yebra
et al.,
2013).
FMC
is defined as the ratio between
the water mass and the dry mass of the sample:
FMC
W W
W
w d
d
=
−
*100
(1)
where
W
w
and
W
d
are the wet and dry weight of the sample,
respectively.
FMC
has been estimated from several methods, including
field campaign sampling, meteorological indices and remote
sensing data. Field sampling provides a simple estimation of
FMC
, and involves collecting leaves and small branches that
are weighed, oven dried, and weighed again to compute the
water content by difference of weights. It is a simple and accu-
rate method but very time consuming and site specific (Chuvi-
eco
et al.
, 2004). Meteorological indices are based on weather
observations and are commonly used in standard fire danger
assessments (Aguado
et al.
, 2007). These indices provide reli-
able estimations of moisture conditions for dead fuels (Castro
et al.
, 2003), but are less reliable for live fuels, as plants use
different strategies to adapt to summer water scarcity.
Satellite images provide a spatially comprehensive and
cost-effective alternative to estimate live
FMC
(
LFMC
). A num-
ber of papers have been recently published on this (see a good
summary in Yebra
et al
., 2013). These works can be divided
into two large groups: empirical and simulation. The former
use statistical models that relate satellite-derived reflectances
or spectral indices to concurrent field measurements of
LFMC
(Cheng
et al.
, 2006; Chuvieco
et al.
, 2004; Dennison
et al
.,
2005;Fensholt and Sandholt, 2003; Qi
et al.
, 2012; Roberts
et
al
., 2006). Empirical models are simple to compute and may
work well regionally, but they are difficult to extrapolate to
regional or global scales (Riaño
et al.
, 2005b). The alternative
approach relies on simulating a set of
LFMC
scenarios based on
Sara Jurdao is with the Environmental Remote Sensing
Research Group, Department of Geography and Geology, Uni-
versity of Alcalá.Calle Colegios 2, 28801 Alcalá de Henares,
Spain
).
Marta Yebra is with CSIRO Land and Water, GPO Box 1666,
Canberra ACT 2601, Australia, and Australian National
University, College of Medicine, Biology and Environment,
Canberra ACT 2601, Australia.
Patricia Oliva is with the Department of Geography, Universi-
ty of Maryland. 4321 Hartwick Road, College Park, MD 20742.
Emilio Chuvieco is with the Environmental Remote Sensing Re-
search Group. Department of Geography and Geology, Univer-
sity of Alcalá.Calle Colegios 2, 28801 Alcalá de Henares, Spain.
Photogrammetric Engineering & Remote Sensing
Vol. 80, No. 5, May 2014, pp. 451–459.
0099-1112/14/8005–451
© 2014 American Society for Photogrammetry
and Remote Sensing
doi: 10.14358/PERS.80.5.451
PHOTOGRAMMETRIC ENGINEERING & REMOTE SENSING
May 2014
451
