±1.12 (He
et al.
, 2007; Jaggi
et al
., 2006; Lantinga
et al.
, 1999;
Juarez
et al.
, 2009; He and Guo, 2006; Mitchell
et al.
, 1998;
Misra and Misra, 1981). This aggregate range based on limited
sampling compares well with the worldwide
LAI
compilation of
2.5 ±2.98 (
n
=28) accomplished by Scurlock
et al.,
(2001). Both
the average site-specific and worldwide ranges compare well
with the average
LAI
based on
LAD
of 3.33 ±1.04 (
n
= 17) found
in this study (Table 1).
LAD
(or
KM
) is less often reported, par-
ticularly for grasslands. We found very few reported
KM
ranges
of 0.5 to 1 for grass-clover (Lantinga
et al.
, 1999), 0.28 to 0.46
for a managed grassland (Jaggi
et al.
, 2006), and from 0.74 to
1.72 for three grass species under full sun to 50 percent shade
(Barro
et al.
, 2012). Our average
KM
average and range (±1 std)
of 0.69 ±0.20 is consistent with those reported ranges.
Bottom PAR Prediction of Marsh Biomass
Although single
PAR
measurements at the bottom of the marsh
canopy may approximate marsh biomass in certain cases
(Whitbeck and Grace, 2006), success requires canopy struc-
ture to remain fairly constant throughout the target marshes
(Figure 3). Our research shows that derivation of canopy
density and orientation parameterized as
LAI
and
LAD
better
account for the biomass vertical structure (Figures 2b and 5c).
Although improvement is needed in representing the marsh
structure, results confirm the strategy for improvement will be
providing more
LAI
and
LAD
vertical detail.
Use of Constant Light Extinction Estimates
The use of a constant (
LAD
= 0.5) extinction coefficient provided
high reproduction of the
PAR
vertical attenuation in the marsh.
As in the comparison to the bottom
PAR
relationship to total
biomass, however, the
LAI
based on a constant extinction coeffi-
cient was not significantly (p <0.1) related to total biomass (Fig-
ure 5b). In contrast,
LAI
based on an optimized
LAD
had a low
(R
2
= 0.2) but significant (
p
<0.1) relationship to total biomass
(Figure 5c). Furthermore, constant
LAD
based
LAI
’s were more
tightly centered in a narrow range between (3 to 4.5
LAI
) in
contrast to the more dispersed distribution of
LAI
’s based on an
optimized
LAD
(Figure 5a). Similar to bottom
PAR
comparison,
these results confirm that derived canopy structure indicators,
LAI
and
LAD
, better reflect the highly varied marsh structure,
and substantiate the need to account for the marsh orientation.
Conclusions
Although vegetation indexes (
VI
) provide good estimates of the
live and dead composition in coastal marshes, many of these
marshes exhibit complex structures that are not well repre-
sented by
VI
variance. Further, bulk biomass measures do not
provide the detail necessary for understanding biophysical
processes and function of these marshes in sustaining the
coastal resource and as effected by the forces acting on the
coastal resource. In order to overcome these limitations, we
present an approach for producing the spatiotemporal repre-
sentation of the three-dimensional marsh structure from
PAR
field measures without the necessity of user estimates of the
marsh leaf-stem orientation. The parameterization of the
PAR
profiles is based on standard equations and published field
collection procedures. The independent leaf area index (
LAI
)
and leaf angle distribution (
LAD
) reproduced the
PAR
profiles
with 99 percent accuracy fulfilling the critical necessity of ful-
ly recreating the marsh structure as represented by the
PAR
pro-
file. Results showed that bottom
PAR
recordings do not account
for biomass variability and that use of constant
PAR
extinction
estimates does not well represent structural variability in these
highly complex coastal marshes. Although more substantia-
tion is needed, results confirm that derived canopy structure
indicators,
LAI
and
LAD
, better reflect the highly varied marsh
structure, and substantiate the need to account for the marsh
orientation. Results also confirm that the strategy for improve-
ment will be obtaining more
LAI
and
LAD
data from a wider
range of marshes in field efforts coordinated with remote sens-
ing data collections. The application of
LAI
profiles and
LAD
in
these complex wetland systems offers a more meaningful rep-
resentation of the marsh that will improve the understanding
and representation of biophysical function, and importantly,
provide variables more amenable to remote sensing mapping.
Acknowledgments
We thank Dr. Stephen McNeill of Landcare Research Infor-
matics Team for his insightful and constructive review. This
research was supported in part by the National Aeronautics
Space Administration (
NASA
) Grant No. 11-TE11-104 and was
carried out in collaboration with the Jet Propulsion Labora-
tory, California Institute of Technology, under a contract with
NASA
, and by US Geological Survey Hurricane Sandy Supple-
mental Funds. Any use of trade, firm, or product names is for
descriptive purposes only and does not imply endorsement
by the US Government.
Appendix 1. Model Flow Section
LS is the leaf area index per layer, KM is the optimized ex-
tinction coefficient, PAR is the light fraction at depth (IDPTH).
ITKM was set to one for the calculations of KM. Goudriann
(1977) used a value of three. ITKM=1 limits the adjustment of
DELTA. Note: Depth increases from canopy top to bottom.
Figure 5. (a) Site-average total LAI based on the optimized LAD and on a constant LAD of 0.5, (b) The total biomass values predicted from
a regression of the site total biomass values listed in Table 1 and the LAI values based on a constant LAD of 0.5, and (c) The same as in
5b but with LAI values based on the optimized LAD. Calculated slopes for the observed versus predicted total biomass (Figures 5b and
5c) equaled one and in neither plot did the bias differ significantly from zero (p ≤0.1).
PHOTOGRAMMETRIC ENGINEERING & REMOTE SENSING
October 2015
813
