CHAPTER NO.

3 CROP DISCRDmfATION AND YIELD IIONITORilfG

► Crop Dtaerlmlnatlon (differentiation/ claulflcation) :-

• Crop cllacrlmlnatton is a necessary step for most agricultural monitoring

systems in which differentiate/classification of crops as per season,
phonological stages, pigmentation, spatial arrangement/ architecture as well as
spectral features is called as crop discrimination.

• Crop discrimination is usually an important step for development and

management of crop monitoring systems.

• Since remote sensing technology is used for crop discrimination, both the
theory and the technological tools have been in constant development, this has
led to a remarkable increased in the range of applications and scope of crop
discrimination techniques.
 management of crop monitoring systems.

• Since remote sensing technology is used for crop discrimination, both the
theory and the technological tools have been in constant development, this has
led to a remarkable increased in the range of applications and scope of crop
discrimination techniques.
•
• Crop discrimination types using remote sensing techniques are based on the
characterization and understanding of the electromagnetic behavior of target.

• This behavior depends on the wavelength at which the crop observation is

performed, due to this, remote sensors for Earth observation are designed to
operate at wavelengths in which the response of each target (crop type) is
characterized and known, facilitating on this way the target identification in the
scene.
► It is important to note that the electromagnetic response of the
different parts of a crop cover not onty depend on the wavelength
used for observation, it also
shows variations depending on the season, angle of incidence of the
sensor, crop' s features, ilumlnation Intensity, weather phenomenon
and topography among other extemal factors.
► Recentty, GIS and RST (Remote sensing Technology) has come up
with a capable role in agricultural research, predominantly In crop
yield prediction in addition to aop suitabilty studies and site specific
resource allocation.
► Role of GI to discriminate different crops at various levels of
classfflcation, monitoring crop growth and predctlon of the aop yield.
 Remote sensing (RS)
~ Remote sensing (RS) is an efficient technology and worthy source of
earth surface information, as it can capture images of reasonably
large area on the earth.
~ Due to advancement in the sensor technologies, there is availabiity
of high spatial as well as spectral resolutions imageries and also non-
imaging spedroradiometer.
~ With the use of these imaging and non-Imaging data, we can easily
characterize the different species.
~ Different crops shows distinct phonological characteristics (plant lfe
cycle events) and timings according to thel1 r nature of gennlnation,
tilerlng, flowering, boH formation (cotton), rlipenlng etc.
~ Even for the same crop and growing season, the duration and
ma~itude of each phonological stage can differ between the
varieties which introduce data variability for crop type discrimination.
~ Agricultural crops are significantly better characterized, classified,
modelled and mapped using hyper spectral data.
 Feature Extraction
► Feature Extraction is the process of defining image characteristics or
features which effectively provides meaningful information for image
interpretation or classification called as feature extraction.
► Goals of Fearure extraction are
a) Effectiveness and efficiency in classification
b) Avoiding redundancy/extra space of data
c) Identifying useful spatial as well as spectral features
d) Maximizing the pattern of crop discrimination
► For crop discrimination, spatial features are useful.
► Crops are planted in rows, either rrultiple or single rows as per crop
type to maximize yield.
► Different spatial arrangement of the crops gives better spatial
Information but it requires high spatial resolution images.
► In spatial image classification, spatial image elements are combined
with spectral properties in crop classification decision.
 Role of Texture in Classification
~ Texture means information in spatial (space) arrangement of colors/
intensity of images
~ In general. it is possible to distinguish between the regular textures
manifested by man made objects from the Irregular manner that
natural objects exhibit texture.
~ Hence, the texture characteristics/classification can be used for
discriminate crops.
~ For crop discrimination, remotely sensed data, conventional texture
analysis and grey level co-ocrurrence matrix (GLCM) methods are
used.
Grey Level Co-Occurrence Matrix {GLCM)

~ In GLCM method describing the grey value relationships.

~ GLCM Is a statistical method of examining texture.
~ GLCM can be viewed as a two dimensional hlstogam
(rectangles) of the frequency with which pairs of 'l&Y level
pixels/Images occur In a given spatial relationship.
~ GLCM method is a way of extracting texture features
~ Image composed of pixels each with an intensity (a specific gray
level), the GLCM is a tabulation of how often dfferent
combinations of '18Y levels co-occur in an image or image
section.
 Local Binary Pattern (LBP)

• LBP is powerful feature for texture classification.

• Local Binary Pattem (LBP) is an effective texture which thresholds
(applied for all pixels of the Image) the neighboring pixels (smallest
unit of Image) desa1ptor for Images based on the value of the
current ~xel.
• Local Binary Pattem (LBP) is a simple yet very efficient texture
operator which labels the pixels of an image by thresholding the
neighborhood of each pixel and considers the result as a binary
number.
• Due to discr r has become a
popular apJ: .,_.,_.. _ -
--.-•-
• -:.:.~ -
C-bl
,....._ ,.
- - . - ..

1
 Spectral features for crop classification
~ Spectral characteristics of green vegetation have very noticeable
features.
~ Visible portion of the spectrum are determined by the pigments
contained in the plant.
~ Chlorophyll absorbs strongly in the blue (0.4-0.5 nm) and red (0.68
nm) regions also known as the chlorophyll absorption bands.
~ Chlorophyll is the primary photosynthetic pigment in green plants.
~ The spectral reflectance signature has a dramatic increase in the
reflection for healthy vegetation at around 0.7 nm.
~ In the near infrared {NIR) between 0.7 nm and 1.3 nm, a plant leaf will
naturally reflect between 40 and 60 %, the rest is transmitted with
only about 5 % being absorbed.
~ Three strong water absorption bands are noted at around 1.4, 1.9 and
2.7 nm can be used for plant-water estimation.
 1. Band selection

• Band selection is one of the important steps in hyper spectral

(hundreds or thousands of bands) remote sensing.
• There are two types of spectral bands (make only pass desired
wavelength) viz., multispectral i.e. 3 to 10 bands and
hyperspectral i.e. 100 or 1000 bands remote sensing
• There are two conceptually different approaches of band
se lection like unsupervised and supervised.
1

• Due to availability of hundreds of spectral bands, there may be

same values in several bands which increases the data
redundacy (additional space).
• To avoid the data redundancy and get distinct features from
available hundreds of bands, we have to choose the specific
bands by studying the reflectance behavior of crops.
 2. Narrowband Vegetation Indices
• Spectral indices :- Assume that the combined interaction between a
small numbers of wavelengths is adequate to describe the biochemical or
biophysical interaction between light and matter.
• Hyperspectral system :- is their ability to create new indices that integrate
wavelengths not sampled by any broadband system and to quantify
absorptions that are specific to important biocliemical and biophysical
quantities of vegetation.
• Vegetation properties measured with hyperspectral vegetation indices
(H\71).
• HVI can be divided into three main categories
1. Structural properties :- These properties include green leaf biomass, LAI
and Fraction absorbed photo synthetically active radiation {FPAR)
2. Biochemical properties :- It includes water, pigments (Chlorophyll,
carotenoides, amthocyanins), other nitrogen rich compounds {protein)
and plant structural materials (Lignin ancf cellulose)
3. Physiological and stress indices:- It measure delicate changes due to a
stress induced change in the state of xanthophylls changes 1n chlorophyll
content, change in soil moisture.
4. Narrowband vegetation indices:-
../ can be used as potential variables for crop type discrimination .
../ Best vegetation indices of different category to discriminate the crop
types which are greenness/leaf pigment indices, Chlorophyll red edge
indices, light use efficiency indices and leaf water indices
 Importance of hyperspectral Remote sensing
• HRS is an advanced tool that provides high spatial/spectral resolution
data from a distance, with the aim of providing near-laboratory-quality
radiance for each picture element (pixel) from a distance.
• Recent advances in remote sensing and geographic information has led
the way for the development of hyperspectral sensors.
• Hyperspectral remote sensing, also known as imaging spectroscopy
(study of visible light), is a relatively new technology that is currently
being investigated by researchers and scientists with regard to the
detection and identification of minerals, terrestrial vegetation, and man-
made materials and backgrounds.
• Hyperspectral remote sensing combines imaging and spectroscopy in a
single system which often includes large data sets and require new
processing methods.
• Hyperspectral data sets are generally composed of about 100 to 200
spectral bands of relatively narrow bandwidths (5-10 nm), whereas,
multispectral data sets are usually composed of about 5 to 10 bands of
relatively large bandwidths (70-400 nm).
• Hyperspectral Remote Sensing (HRS) and Imaging Spectroscopy (IS),
are two technologies that can provide detailed spectral information from
every pixel in an image.
 Yield Monitoring and Mapping
► Yield monitoring :- is the estimation of crop yield well before the
harvest at regional and national scale is imperative for planning at
micro-level and predominantly the demand for crop insurance.
► Crop yield estimation is plays a significant role in economy
development.
► Currently, it is being done by extensive field surveys and crop
experimentation.
► Yield monitoring equipment was introduced in the early 1990s and is
increasingly considered a conventional practice in modern agriculture.
 Different approaches and technologies used
for yield monitoring (crop inventory)
1. Aerial Photography :-
~ There are two distinct aspects of yield estimation a) forecast of yield
based on characteristics of the plant or crop b) Estimates of the crop
yield based actual weight of the harvest crop
~ After World War II, various researchers used the emerged concept of
Aerial Photography for optimized use of resources for agriculture and
crop inventory.
~ Black and white photography has been used for crop identification,
primarily based on ground appearance, photographs of selected fields
and growing seasons of crops.
2. Multispectral scanners :-
~ MSS - have 0.3-14 nm or more bands
~ MSS have certain advantages and disadvantages compared to
photography.
~ Ability to differentiate wheat from other agricultural crops using MSS
data 1n a computer format.
~ An important consideration in the task of species identification is the any
stage of growth of the crop.
 Potential Applications of Yield maps/mapping
• Yield maps represent the output of crop production.
• On one hand this information can be used to investigate the
existence of spatially variable yield limiting factors.
• On the other hand, the yield history can be used to define
spatially variable yield goals that may allow varying inputs
according to expected field productivity.
• Yield maps are one of the most valuable sources of spatial data
for precision agriculture.
• In developing these maps it is essential to identifying the points/
locations where get the higher and lower crop yields.