Modeling Above-Ground Biomass in
Tallgrass Prairie Using Ultra-High Spatial
Resolution sUAS Imagery
Chuyuan Wang, Kevin P. Price, Deon van der Merwe, Nan An, and Huan Wang
Abstract
We examined the relationship between tallgrass above-ground
biomass (
AGB
) and
NDVI
from ultra-high spatial resolution
multispectral imagery collected by small unmanned aircraft
systems (
sUAS
). This study was conducted at the Tallgrass
Prairie National Preserve in Chase County, Kansas. Results
show that
NDVI
values computed from
sUAS
imagery explained
up to 94 percent of the variance (p <0.01) in
AGB
measure-
ments. The model coefficient of determination (r
2
) decreased
with increasing aircraft flight altitude suggesting image spa-
tial resolution is a key factor influencing the strength of the
relationship. A scaling-up approach from small-scale
sUAS
im-
agery to broad-scale, digital aerial imagery collected at 1,200
meters by a piloted aircraft was used to provide
AGB
model
estimates across the entire 4,500 ha of the Preserve. Spectral
reflectance data measured by spectroradiometer were also
used to identify three optimal regions of the spectrum that
have the highest significant correlations with tallgrass
AGB
.
Introduction
The tallgrass prairie ecosystem is native in the Great Plains of
central North America that is among the most species diverse
and productive warm-season grassland ecosystems (Knapp
et al
., 1998; Knapp and Seastedt, 1998). The tallgrass prairie
once covered a large portion of the American Midwest, but to-
day less than 4 percent of the North American tallgrass prairie
remains due to intensive cultivation, over grazing, and urban
sprawl. It has become one of the most endangered and rarest
ecosystems in the world (Samson and Knopf, 1994; Steinau-
er and Collis, 1996). The largest contiguous area of tallgrass
prairie now can only be found in the Flint Hills that mostly
resides in the State of Kansas. The existing and converted tall-
grass prairie, together with other types of prairie ecosystems
in the Central Great Plains, covers about 40 percent of Kansas
and produces billions of dollars in grazing livestock each year
(USDA, 2007). It is therefore essential to both nature conser-
vation and agriculture in Kansas.
Above-ground biomass (
AGB
) is an important ecological in-
dicator for understanding the tallgrass prairie ecosystem health.
It is of critical importance to the proper management and
understanding of climatic and anthropogenic influences on the
tallgrass prairie ecosystem in the Central Great Plains (Lauen-
roth, 1979; Seaquist
et al
., 2003; An
et al
., 2013). Accurate
AGB
measurements and estimations can be used to predict secondary
production in a tallgrass ecosystem, such as livestock yield, and
to better understand the rangeland properties so that conserva-
tion of tallgrass prairie natural resources could be improved.
Various means of measuring and predicting grassland
AGB
have been developed, but they all have certain level of
limitations. The traditional method involving field-based
measurements is very time-consuming and labor-intensive.
Examples of studies involving traditional approaches can be
found in Briggs and Knapp (1995); Briggs
et al
. (2005); Reed
et al
. (2005); Nippert
et al
. (2006); Heisler-White
et al
. (2008);
Craine
et al
. (2010); and La Pierre
et al
. (2011). The challenge
of using these more traditional field-based biomass harvest
methods is that it is only practical for a relatively small
geographic area. Predicting
AGB
for the entire tallgrass prairie
ecosystem is not possible using these methods.
Physical-based models that incorporate climatic attributes
and environmental factors as variables, such as annual mean
precipitation, mean air temperature, mean soil temperature,
soil water-holding capacity, and topographic position, to
predict grassland biomass and productivity have also been in-
vestigated (Towne and Owensby, 1984; Sala
et al
., 1988; Briggs
and Knapp, 1995; Nippert
et al
., 2006; La Pierre
et al
., 2011).
Physical-based models may work well for other ecosystems,
but mixed results and variable findings have been reported in
the cited studies that focus on the tallgrass prairie ecosystem
in the Central Great Plains (An
et al
. 2013). It is because native
tallgrass species in the Central Great Plains have adapted to
severe drought conditions and can extract soil moisture at con-
siderable depths (Brown and Bark, 1971), thus rendering drier
conditions less of an influencing factor depending on the site
location (Weaver and Albertson, 1943). Also the conversion
of climatic attributes and environmental factors to spatial and
temporal variables is not error free. Therefore, those important
variables in physical-based models are not necessarily the best
predictors for tallgrass prairie in the Central Great Plains.
For reasons discussed above, some scientists have turned
to the use of satellite remotely sensed data to measure and
predict grassland
AGB
and productivity at the regional and
global scales (Tucker, 1985; Asrar
et al
., 1985; Wylie
et al.
,
Chuyuan Wang is with the Department of Geography, Kansas
State University, Manhattan, Kansas.
Kevin P. Price is with the Department of Agronomy, Kansas
State University, Manhattan, Kansas; and the Department of
Geography, Kansas State University, Manhattan, Kansas (kev-
).
Deon van der Merwe is with the College of Veterinary Medi-
cine, Kansas State University, Manhattan, Kansas.
Nan An and Huan Wang
are with the Department of Agrono-
my, Kansas State University, Manhattan, Kansas.
Photogrammetric Engineering & Remote Sensing
Vol. 80, No. 12, December 2014, pp. 1151–1159.
0099-1112/14/8012–1151
© 2014 American Society for Photogrammetry
and Remote Sensing
PHOTOGRAMMETRIC ENGINEERING & REMOTE SENSING
December 2014
1151
