Agriculture Reference
In-Depth Information
6.2.3 A
PPLICATION
OF
R
EMOTE
S
ENSING
T
ECHNOLOGY
The success of PA depends strongly on efficient and reliable methods for site-specific
field information gathering and processing. Remote sensing (RS) is a powerful tool
for large-scale and rapid data collection and has been widely used in PA including
cotton production. The use of RS in agriculture is based on relationships between
crop biophysical phenomena and their spectral signatures. RS has been used in many
areas of cotton production such as water management, yield prediction, and nutrient
management. Spectral reflectance from image data have often been used to calculate
vegetation indices such as normalized difference vegetation index (NDVI), which is
calculated by dividing the difference between the reflectances at NIR and red bands
by the sum of the reflectances at NIR and red bands, i.e., NDVI = (NIR − red)/(NIR
+ red). Other vegetation indices including reflectance band ratios and individual band
reflectance have also been used for cotton crop management and yield prediction.
NDVI has been used successfully to estimate actual crop coefficients and determine
evapotranspiration for cotton irrigation scheduling (Hunsaker et al., 2005). Detar et
al. (2006) reported that airborne hyperspectral, multispectral, and thermal infrared
RS data could be used to estimate the plant water stress in cotton at full canopy. A
weighted NDVI was found to be well correlated with water stress. Yang et al. (2006a,
2006b) found that both satellite and airborne multispectral imagery could be used
for cotton yield estimation including variability across a field. Wooten et al. (1999)
also used multispectral satellite images to predict yield in cotton. They determined
that both ground observations and yield were correlated with in-season multispectral
satellite images. In subsequent work, Thomasson et al. (2000) found that average cot-
ton yield over a 30-m square area correlated well with the average value of Landsat
image pixels representing the same ground area. Plant et al. (2000) found that NDVI
integrated over time showed a significant correlation with lint yield.
In addition to being an indicator of water stress and yield, reflectance measure-
ments may provide an in-season indication of crop growth conditions such as nutri-
ent deficiency (Thomasson and Sui, 2009). Thenkabail et al. (2000) used spectral
data between 350 and 1050 nm to identify appropriate bands for characterizing
biophysical variables of various crops including cotton. Lough and Varco (2000)
evaluated the relationship between nitrogen (N) treatment level and relative cotton
leaf reflectance, and they found the greatest separation between N treatments in the
550-nm (green) waveband. According to Buscaglia and Varco (2002), cotton leaf
N concentration had a strong correlation with leaf reflectance at 550, 612, 700, and
728 nm. Tarpley et al. (2000) found that reflectance ratios, calculated by dividing
cotton leaf reflectance at 700 or 760 nm by a higher wavelength reflectance (755 to
920 nm) could provide accurate predictions of N concentration. Read et al. (2002)
reported that RS of N status in cotton was feasible with narrow-waveband reflectance
ratios involving a violet-to-blue spectral band and the more commonly studied red-
edge region.
RS could be used for spatially variable insecticide applications in cotton. Willers
et al. (1999) reported a sampling protocol to estimate tarnished plant bug densities
in commercial cotton fields. High-resolution multispectral RS imagery was used in
the protocol to create plant growing status maps for delineation of different sampling
