Geography Reference
In-Depth Information
Factors controlling the spectral responses of the vegetation and its reflectance
measurements include many natural and technical parameters, such as: atmosphere
conditions (e.g., the quantity of occurring sunrays and the proportion of water
vapor, change reflectance from plant canopies) (Gao and Goetz
1992
); soil
background (Mickelson et al.
1998
); wind (Lord et al.
1985
); viewing angle
(Galvao et al.
2004
); the altitude of the sensor from plant canopies; and the amount
of light.
There is an important relationship between the available images for an individual
study area and the plant growth stages, where the growth stage determines which
images are suitable for separation between the crops spectrally. So, learning the
phenological details about the crops of interest to an individual study area may be
required. These phenological details refer to the natural vegetation calendar or a crop
calendar. Data for these calendars can be obtained from: literature of previous
ecological studies; meeting with qualified field-oriented ecologists; in state or
regional bureaus engaged with natural resource management in the region; or from
field-work based observation and measurements (e.g., Spectrometer measurements).
Single-date captured remotely sensed data would be inadequate for primarily
vegetated areas described by large temporal changeability and typical spatial
patterns of highly frequent land cover changes between vegetation canopies.
Multi-date remote sensing would be able to cover this problem: when specific data
might not be suitable to separate individual LULC-classes, the use of another
acquisition date might prove more appropriate for classification. Therefore, the use
of the total multi-temporal information gives us a better separation between sev-
eral classes, and consequently, more classification accuracy (see Fig.
5.23
). Crop
phenology understanding is very important in crop monitoring and classification
(Chen et al.
2008
).
2.5.2 Crop Discrimination from Satellite-Based Images
The most frequently practiced utilization of remote sensing for agriculture is the
identification of crop types and then classification (Van Niel and McVicar
2000
),
where crop discrimination is a critical and difficult first step for most agricultural
observing activities. The capability of remotely sensed data to identify crop class
makes it promising to classify and estimate each crop area, and so calculate the
relevant statistics automatically that can used as inputs to crop production fore-
casting models (Blaes et al.
2005
). The application of remote sensing for dis-
crimination between agricultural crop classes and internal crop characteristics has
been widely studied throughout the past decade (Senay et al.
2000
; Blaes et al.
2005
; Satalino et al.
2009
). Most of these researchers have focused on increasing
classification accuracy through the development of several techniques and meth-
ods. In contrast, only small studies have been presented on determining the best
time(s) to obtain images in order to distinguish different crops (Van Niel and
McVicar
2004
).
Search WWH ::
Custom Search