Agriculture and Forestry
Remote sensing is a classical and traditional approach widely
applied in agriculture and agronomy for different purposes
(Atzberger, 2013). Farmers have expressed some requests to
monitor crop conditions in their fields using
UAVs
. In Zhang
et al.
, (2014) a quad-copter equipped with optical and near-
infrared imagery has been used to monitor fertilizer trials,
conduct crop scouting, and map field tile drainage in Ontario,
Canada. The results of a preliminary investigation into the use
of aerial surveillance techniques to estimate weed-patch areas
were presented very early by Thornton
et al.
(1990) using a
low-altitude helium balloon based on imagery for mapping
the wild oat distribution in a wheat field. This platform was
later used in Jensen
et al.
(2007) for detecting attributes in
wheat crops.
Since the beginning of the development of new generations
of
UAVs
, these platforms were considered a well suited tool,
under different configurations, in agriculture and forestry
because of their potential (size, weight, flight speed, altitude)
(Grenzdörffer
et al.
, 2008; Gay
et al.
, 2009). Currently, they
continue offering new opportunities as well as new chal-
lenges (SARS, 2014). Herwitz
et al.
(2004) and Furfaro
et al.
(2005) used early-unmanned aerial platforms in a plantation
of Kauai Coffee Company in Hawaii, equipped with multi-
spectral imagery and a local area network for camera control
and downlinking images.
UAVs
are recently becoming part of
remote sensing applications in agriculture and forestry with
very different and diverse applications; some of them gaining
in performance while being inexpensive compared to tradi-
tional platforms. Zhang and Kovacs (2012) and Stefanakis
et
al.
(2013) identified a research agenda to develop
UAV
systems
and define methods for precision agriculture. In this regard,
Urbahs and Jonaite (2013) proposed the main features for us-
ing
UAVs
in agriculture applications, including weight, flight
duration, flight altitude, payload, and engine.
A categorization of mobile platforms and resulting research
applications were reported in Zecha
et al.
(2013) where
UAVs
’
design and characterization are focused on their use for agri-
cultural tasks. More specific works dealing with the develop-
ment and testing of a
UAV
aerial platform for agricultural tasks
equipped with multispectral cameras can be found in Link
et
al.
(2013). An overview of works involving the development
of technologies, systems, and methods for
UAVs
are examined
and studied for agricultural production management in Huang
et al.
(2013), where limitations of current
UAVs
for agricultural
tasks are reported, as well as future needs and suggestions for
development and application of
UAVs
in agriculture.
In the context of agricultural and forestry, the control of
biophysical variables is of special interest for various pur-
poses, such as chlorophyll and biomass determination for site
specific treatments or forest stands. Several methods and strat-
egies have been developed for the control of such biophysical
variables, as described below. Specifically, Grenzdörffer and
Niemeyer (2011) have used the bidirectional reflectance distri-
bution function (
BRDF
) for computing bidirectional reflectance
properties of plant surfaces. A quad-copter equipped with
four vision-based cameras carefully configured is designed to
cover a field of view from four different perspectives.
Salamí
et al.
(2014) provide a review of
UAVs
used for sens-
ing vegetated areas, including precision agriculture, forest,
and rangeland applications, where sensors, tools, payloads,
and platforms are considered with their categorization. Ap-
plication areas of
UAVs
in agriculture and forestry are quite
diverse, whereas the topics considered in the next section
provide an overview of this.
Crops and Weeds
Crops and weed management in precision agriculture are
two key activities for different purposes, such as yield
estimations, herbicide applications, and pesticide control
resulting in cost savings and minimal environmental impact.
The wider use of
UAVs
in this area consists on the acquisi-
tion of information using sensors onboard, to serve as inputs
to other agricultural systems, such as tractors, that apply agri-
cultural treatments. Nevertheless, an exception is the system
described in Huang
et al.
(2009) where a low volume spray
system is designed to be installed and integrated onboard an
unmanned helicopter to apply crop specific treatments. A
helicopter, powered by two gasoline engines, with rotor diam-
eter of 3 m and maximum payload of 22.7 kg, was equipped
with the sprayer system consisting of boom tubing and
nozzles, spray pump, control box, and spray tank for chemi-
cal with a total weight of approximately 11 kg. The sprayer on
the
UAV
was designed to spray 14 ha of land on a single load
at a low volume spray rate of 0.3 L/ha.
Despite this standalone application, this overview is
focused on the first type, i.e.,
UAVs
equipped with sensors
onboard that provide data for subsequent analysis and treat-
ments when required.
Many applications in crops are oriented to the genera-
tion of maps for monitoring weed infestations and coverage,
biomass estimation, yield prediction, or crop stress. Imaging
maps are commonly georeferenced and ortho-rectified, where
positioning accuracy becomes an important consideration in
map generation.
In this regard, Sugiura
et al.
(2005) developed a system
based on an unmanned helicopter for precise mapping in
maize fields. They applied geometric corrections based on a
real-time kinematic global positioning system (
RTK
-
GPS
), an
INS, and a geomagnetic direction system (
GDS
). An imaging
sensor installed under the fuselage captures images. The large
errors of the
GDS
data, due to a geomagnetic warp surrounding
the helicopter, are corrected based on the parallel crop rows
after imagery was collected.
Cross-pollenization in maize crops was studied in Vo-
gler
et al.
(2009) because of the importance to achieve the
coexistence of conventional and genetically modified maize
with the aim of achieving acceptable yields. The impact of
elevation differences between adjacent donor and receptor
fields on rates of cross-pollenization was analyzed using a
Geographic Information System (
GIS
). Digital images were
captured with a digital still-video camera mounted on an
unmanned helicopter.
Quantification of nitrogen status of rice and winter wheat
were studied in Zhu
et al.
(2009) and Yunxia
et al.
(2005),
respectively, to avoid under/over fertilization. Hyperspectral
imagery was used for computing chlorophyll content to char-
acterize spatial and temporal variation in crop production.
Øvergaard
et al.
(2010) used three radiometers as sensors
to estimate yield in wheat fields and also grain quality. Two of
these instruments are point spectroradiometers covering wave-
lengths ranging from 485 nm - 1650 nm and 350 nm - 2500 nm,
respectively. The third instrument is a hyperspectral imaging
system with wavelengths in the range of 400 nm - 1000 nm.
A spatial stratified random sampling method was applied
for crop area estimation using medium spatial resolution and
UAV
imagery, which is useful for subsequent regional distribu-
tion in specific areas (Pan
et al.
, 2011).
Agüera
et al.
(2011) used two digital compact cameras for
acquisition of
RGB
and
NIR
images onboard a quad-rotor. The
NIR
images were acquired with a camera equipped with an
optical filter that allows the radiation with wavelength greater
than 920 nm, these systems weigh 130 and 250 g, respective-
ly. The aim of this work was to compare an
NDVI
related with
sunflower nitrogen status based on greenness determination
derived from the leaf chlorophyll content.
Costa
et al.
(2012) described an architectural design based
on
UAVs
which can be used in agricultural systems for specific
294
April 2015
PHOTOGRAMMETRIC ENGINEERING & REMOTE SENSING
