Geography Reference
In-Depth Information
additional factor to the quantity of this confusion type is the seasonal variation in
the NDVI signals caused by seasonal difference in illumination geometry, which
imitates a phenological cycle (McIver and Friedl 2002 ).
In order to support the capability of remotely sensed data to discriminate
between the various crops, researchers have investigated many alternatives which
have to do with: The sensor-type (e.g., optical or microwave); number of images
(e.g., single-date or multi-date); timing of the imagery; digital processing tech-
niques; or ancillary and spatial data integrating in the classification process (Van
Niel and McVicar 2000 ).
2.5.3 Crop Area Estimation from Satellite-Based Images
Crop area measurement and survey are very common practices in agriculture.
Photo-interpretation of images can give better information than statistical analysis
to evaluate an amount, or area, for a thematic category (Ozdogan and Woodcock
2006 ). Usually, crop area estimation has been achieved with very costly and hard
statistically-based ground surveys that do not determine either the area or the
geographical distribution of individual crops. To overcome or decrease these
drawbacks, remote sensing, either alone or in combination with ground surveys,
were used in crop area estimation (Wardlow and Stephen 2008 ). Obtaining full
efficiency of remote sensing for crop area estimation depends on the landscape
characteristics, especially field size compared with the image resolution, where a
suitable resolution for a specific landscape is realized when the most image pixels
are pure. But, when this relationship is not realized, for example when using
MODIS- or MERIS images especially for landscapes with small fields, then sub-
pixel classification techniques (e.g., pixel un-mixing) can be used (GEO 2010 ).
Remote sensing has not been widely used for crop area estimation, due to the
tradeoff between spatial detail (the scale of the remote sensing data) and area
coverage for each image. In addition, there is the relationship between the spatial
resolution of the remotely sensed data and the agricultural field sizes. Agricultural
fields in most countries in the world are rather small, requiring medium to high
spatial resolution data. However, increases in spatial resolution provide a decrease
in the temporal availability which in turn lowers the chance of clouds-free cov-
erage. Even if the clouds-free suitable spatial resolution data were obtainable, the
increased number of datasets makes the cost high, and the high spatial resolution
sensor covers only small geographical areas at a time. This leads to an additional
problem, the need for atmospheric corrections in automated image digital pro-
cessing and classification, as the required images are often gained at diverse times
during the growing cycle of a crop. Medium spatial resolution data (e.g.,
LANDSAT) may be too coarse in countries with very small cultivated fields (e.g.,
China), but high spatial resolution is more appropriate for use in countries with
large cultivated fields, such as the U.S. (Ozdogan and Woodcock 2006 ). In con-
trast,
lower
spatial
resolution
data
(e.g.,
MODIS)
offer
wide
temporal
and
Search WWH ::




Custom Search