module name/title remote sensing applications in agriculture
TRANSCRIPT
Component-I(A) - Personal Details
Component-I (B) - Description of Module
Role Name Affiliation
Principal Investigator Prof.MasoodAhsanSiddiqui Department of Geography,
JamiaMilliaIslamia, New Delhi
Paper Coordinator, if any Dr. M P Punia Head, Department of Remote
Sensing, Birla Institute of
Scientific Research, Jaipur
Content Writer/Author (CW) Kaushal Panwar Senior Research Fellow, Birla
Institute of Scientific Research,
Jaipur
Content Reviewer (CR) Dr. M P Punia Head, Department of Remote
Sensing, Birla Institute of
Scientific Research, Jaipur
Language Editor (LE)
Items Description of Module
Subject Name Geography
Paper Name Remote Sensing, GIS, GPS
Module Name/Title Remote Sensing Applications in Agriculture
Module Id RS/GIS 15
Pre-requisites
Objectives Student will get to know RS analysis helps in agriculture
field.
Student will acquire skill how to study data and its
algorithms in agriculture sector.
Student will be equipped with knowledge to study further
about the applications.
Remote Sensing Applications in Agriculture
Outline
Introduction about role of remote sensing in agriculture and impact on
Indian economy.
Remote Sensing application in agriculture.
Crop production forecasting.
Assessment of crop damage and crop progress.
Horticulture, Cropping Systems Analysis.
Environmental Impact Assessment of Agricultural System.
1.
Keywords
Non-point source pollution.
Impact of Climate Change on Agricultural System.
Biophysical parameter retrieval and process modeling.
Demonstration of technique for forewarning Pests and Diseases.
National Mission
National production forecast (FASAL).
Crop Identification Technology Assessment for Remote
Sensing (CITARS).
Conclusion.
Introduction
Earth observations (EO), comprising satellite, aerial, and in situ systems, are
increasingly recognized as critical tool for studying and monitoring of natural
resources and unfolding the intricate behavior of the complex earth’s dynamic
processes. The Remote Sensing research has evolved as multidisciplinary theme
dedicated to developing the applications of remote sensing technology of Land
Ocean and atmosphere addressing geologic, botanic, and hydrologic issues at
national, regional, and site-specific scales.
Agriculture plays dominant role in economy of almost every nation. Whether
agriculture represents a substantial trading industry for an economically strong
country or simply sustenance for a hungry, overpopulated one, it plays a significant
role in almost every nation. The production of food is important to everyone and
producing food in a cost-effective manner is the goal of every farmer, large-scale
farm manager and regional agricultural agency. A farmer needs to be informed to
be efficient, and that includes having the knowledge and information products to
forge a viable strategy for farming operations. These tools will help him
understand the health of his crop, extent of infestation or stress damage, or
potential yield and soil conditions. Commodity brokers are also very interested in
how well farms are producing, as yield (both quantity and quality) estimates for all
products control price and worldwide trading. The policy makers are interested in
harnessing the best available tools for optimizing the resource use, minimizing the
damage/losses and ensuring the societal benefit. Most important component of
such decisions is the agricultural and allied information at best possible resolution
of spatial and temporal scales. Remote sensing techniques come handy for
achieving such an objective. The diverse agro climatic conditions give rise to
cultivations of large number of crops. The information on long term sustainability
vis-à-vis environmental impacts demands seasonal and yearly information of
variant as well as invariant resources. The diverse crop growing conditions coupled
with uncertainties of climate situation demands much more finer crop information
needs on temporal and spatial scale.
During the last two decades remote sensing and GIS techniques are applied to
explore agricultural applications such as crop identification, area estimation, crop
condition assessment, soil moisture estimation, yield estimation, agriculture water
management, agro meteorological and agro advisories. The application of remote
sensing in agriculture, i.e. in crops and soils is extremely complex because of
highly dynamic and inherent complexity of biological materials and soils (Myers,
1983). However, remote-sensing technology provides many advantages over the
traditional methods in agricultural resources survey. The advantages include,
i) capability of synoptic view
ii) potential for fast survey
iii) capability of repetitive coverage to detect the changes
iv) low cost involvement
v) higher accuracy
vi) Use of multispectral data for increased information, and so on.
Now almost all part of the electromagnetic spectrum is utilized for agricultural
applications at various scales. Crop growth and associated factors change
dynamically and need continuous monitoring. The different sub themes of
agriculture are shown in fig. 1 Remote sensing (RS) technology has potential to
estimate crop area and forecast productivity at district and regional level due to its
multispectral, large area and repetitive coverage. The following sections explain
the developments taken place during the last three decades with special reference
to agricultural remote sensing in India.
Fig. 1: Remote Sensing Applications in Agriculture
Source: https://nrsc.gov.in/Agriculture
Remote Sensing Application in Agriculture
Crop Production Forecasting
Fig.2: Crop Production Forecasting Graph
Source: http://apal.org.au/2015-pome-fruit-crop-forecast/
Agricultural crop identification and area estimation has been the focus ever since
the inception of civilian RS program in the U.S. in the early 1960s. Some of the
early studies conducted were experiments such as Crop Identification Technology
Assessment for Remote Sensing (CITARS) and Large Area Crop Inventory
Experiment (LACIE). Exploring the use of remote sensing for agricultural
application in India started with the use of multi band and colour infra red (CIR)
aerial photographs as early as 1974-75. Further knowledge on crop signature was
gathered through scientifically designed field experiments using multi band
radiometer under the Indian Remote sensing Satellite-Utilisation Programme (IRS-
UP) on:
i) Crop production forecasting
ii) Crop stress detection, and
iii) Crop yield modelling. Many of these studies have led to the operationalisation
of the methodology and conduct of national-level projects.
Launch of the Indian Remote Sensing Satellites (IRS-1A, IB & 1C) carrying linear
imaging self-scanning sensors (LISS I,II &III) provided a much-required impetus
to agricultural applications. IRS-1C carried onboard a unique combination of three
sensors viz., (i) Wide Field Sensor (WiFS) with 188m spatial, two spectral bands –
red and near infrared, 810km swath and a repetivity of 5 days, (ii) Linear Imaging
Self scanning Sensor (LISS-III) with 23.5m spatial resolution in the green red and
near infrared region, and 70.5 m in the middle infrared region, and 140 km swath,
and (iii) Panchromatic (PAN) camera with 5.8m spatial resolution, 70km swath
and stereo capability. The launch of RISAT has filled another dimension to the
agricultural remote sensing as all weather capability of data is now reality from
Indian satellite.
Crop production forecasting comprises identification of crops, acreage estimation
and forecasting their yield. Crop identification and discrimination is based upon
the fact that each crop has a unique reflectance pattern in different parts of the
electromagnetic spectrum which is termed as spectral signature. The general
spectral response of a crop canopy in the visible and NIR region is characterised by
absorption in the 0.35 to 0.5 um and 0.6 to 0.7 um regions (due to chlorophyll
pigments), high reflectance in the green region (around 0.54 um), a steep increase
in the reflection in the 0.7 to 0.74 um and high reflectance in 0.74 to 1.3 um region
due to internal cellular structure of the leaves. The absorption at 1.45, 1.95 and 2.6
um spectral bands is due to leaf water content. The varying response of the crops
stems from the fact that various factors such as type of crop, stage of the crop,
canopy architecture, per cent ground cover, differences in cultural practices, crop
stress conditions, background soil/water etc., contribute to the composite response.
Each crop has its own architecture, growing period, etc. thus enabling
discrimination through remote sensing data. If there are two crops with similar
spectral signatures on a given date (confusion crops), multidate data are used to
discriminate them Vigour of the crop is manifested in the absorption in the red and
reflectance in the near infrared spectral regions. It has been observed that the ratio
of near infrared to red radiance is a good indicator of the vigour of the crop. Some
of these properties are utilised in crop identification, crop condition assessment and
yield forecasting. The broad procedure used for crop acreage estimation is shown
in illustration below. The data is analysed applying Maximum Likelihood
Supervised classification technique (other classifiers are also used), where limited
field information called ‘Ground truth’ is used to generate the training signature. In
case of complete enumeration data for analysis was selected by overlaying the
boundary mask of the area over the remote sensing images. When it was extended
to large area stratified sampling technique was developed where area was first
divided with a grid representing sampling frame size (5 X 5 km) and then data
belonging to selected sample segments is extracted (20%) are analysed. The
samples are randomly drawn proportionate to size of each stratum. Area estimate is
made from the proportion of crop present in the sample.
Fig. 3: Procedure For Acreage Estimation using Sampling Techniques
Source: self
Since the space technology has advanced and variety of sensors of different spatial,
spectral and temporal resolutions is available and there is a continual need for crop
information throughout the growing period (Fig. 3), frequent monitoring is feasible
at various scales. Realizing the importance of multiple source information like
weather, econometric and field survey towards a robust approach for multiple
forecasts of a number of crops, a new concept was formulated: FASAL
(Forecasting Agricultural output using Space, Agro- meteorology and Land based
observations) (fig. 4). Implementation of FASAL was initiated in 2007-08.
Fig. 4: Information need and sources in frequent crop monitoring
Source: https://www.crcpress.com/GIS-Applications-in-Agriculture/Pierce-
Clay/p/book/9780849375262
Assessment of crop damage and crop progress
Damage to crop due to moisture stress is a common occurrence in rain fed rice
growing region. The characteristic backscatter profile of rice using temporal SAR
data is useful in characterising the crop condition as normal and sub-normal. Flood
is a common phenomenon in many rice-growing regions, particularly in monsoon
season. Temporal SAR data is found not only to map flood affected rice fields, but
also to model duration and degree of submergence. Complete submergence of rice
at any given period of growth lower the backscatter. The degree of submergence
was modelled with reference to crop height and its deviation from the reference
normal growth profile. The model thus can detect the completely submerged fields
as well as partially submerged fields. It is well established that sowing dates have a
significant effect on crop biomass and yield. Temporal SAR data is used to
categorise the fields as normal, late and very late sown which is additional
component of crop assessment need that enables identifying the reasons for delay.
Similar efforts have been made which for deriving the regional variation of rice
growth profile using optical data (fig. 6)
Fig: 5a.crop growth pattern in feb 2016
Fig: 5b. Crop growth pattern in April 2016
Fig: 5c.crop growth pattern in oct 2016
Fig. 6: Rice spectral profile of different regions in India derived using multidate
optical data
Source: Remote Sensing of the Environment, John R. Jensen
O – OBSERVEDM - MODELLED
J&k
TNAPPUNJAB
Crop yield models
Fig.7: Crop yield model
Source: https://www.slideshare.net/grssieee/luoigarss20112385ppt
Out of the two constituents of crop production, namely crop acreage and crop
yield, the assessment of the latter is most complex because of the high variability
involved. The information on crop yield is an important input for production
estimation shown in fig(7). Every crop genotype has certain yield potential, which
can be achieved (to an approximation) in experimental field with optimal
conditions. However, in the real world, the crop yield is conditioned by various
parameters like soil, weather and cultivation practices, like date of sowing,
irrigation and fertiliser. Crop yield is also influenced by biotic stresses like disease
and pest. While the variability of the weather explain most of the annual variability
over a short period of time, the cultivation practices and new varieties explain most
of the variability over a period of 10 to 20 years. For longer periods of time,
climatic changes or soil improvements or degradation are the main factors
influencing the crop yield However, the fact that all these factors are
interdependent makes the yield assessment a more complex task. Hence, one way
of forecasting of the yield is understanding the variability in the above parameters
and defining their relationship with the final crop yield. Satellite based remote
sensing provides a suitable alternative for crop condition and yield assessment/
forecasting, as it gives a timely, accurate, synoptic and objective estimation of
various crop parameters. The time series based trend and arima models were
developed in the beginning of CAPE project based on district-wise yield data of
DES which were used to compute production by multiplying with the RS derived
area estimates. Further agro meteorology model, spectral models and combination
of these models were tested and used for a variety of conditions, crops and regions.
The agro meteorology inputs were predominantly significant rainfall at fortnightly
intervals, minimum and maximum temperatures etc that would form part of
correlation weighted regression model. The RS input, NDVI (single date or derived
from profiles integration) in such models were used for model development. The
rice biomass has high correlation with SAR backscatter and therefore the yield.
The crop vigour is an indication of crop yield. The vigour of crop can be assessed
using vegetation indices derived from different parts of the spectrum. The
normalized vegetation index is one such index which represents the green biomass
of the plant. The NDVI can be directly correlated to the yield of the crop and this
relation can be used for estimating yield. A variety of models involving
combination of factors of weather and spectral parameters have been developed
and used in conjugation with remote sensing derived acreage for providing
production estimates. Plant growth simulation models have been used for
monitoring crop growth, health shown in fig(8) and predicting yield .However,
their use in large areas has been limited because most plant growth models were
developed at the field scales and the performance of the models is not so
satisfactory when they are extended from field to regional scales.
Fig.8: Crop Stress Detection
Source: http://www.cornandsoybeandigest.com/soybeans/detect-crop-stress-
thermal-maps
Remote Sensing Applications - Horticulture
India is bestowed with varied agro-climate which is highly favourable for growing
a large number of horticultural crops such as fruits, vegetables, root tuber,
ornamental, aromatic plants, medicinal, spices and plantation crops like coconut,
areca nut, cashew and cocoa. India is the largest producer of fruits (49.36 MT) and
second largest producer of vegetables (93 MT) in the world. Horticulture occupies
about 12 per cent of the total cultivated area in the country, and contributes about
25 per cent of the total agricultural export. Remote sensing technology helps in
generation of crop Inventory of major horticultural crops, site suitability analysis
for expansion/introduction, infra-structure planning for post harvest requirement,
disease detection and precision planning for horticulture. The general approach
involves the use of high resolution/high temporal data (LISS-III) for identifying the
crop of interest and relevant collateral information (i.e soil, water, climatic,
infrastructure etc) and processing for logical clustering for decision-making.
Feasibility studies demonstrating the remote sensing technology in horticultural
sector have been carried out. Inventory of orchards like apple, grape, mango,
coconut, banana and vegetables like potato, onion has been carried out in different
agro climatic regions of the country.
The post-harvest infrastructure planning and optimization of cold store facility for
post-harvest management of potato has been demonstrated. Apart from these early
trends in national winter potato production from the country is regularly brought
out to infer about National/state production prospects and identifying areas with
significant change.
Cropping Systems Analysis
A cropping system is defined as the cropping pattern and its management to derive
benefits from a given resource base under a specific environmental condition. This
requires identification of crops and areas where changes in cropping patterns are
desirable. This calls for an initial step of creating an updated database of the
present cropping systems of the country and simulate the long-term effects, taking
into consideration the resource base and agro climatic condition. Satellite remote
sensing (RS) and Geographical Information System have a crucial role to play in
this direction. The multidate satellite data is helpful in deriving seasonal cropping
pattern, sowing pattern, crop rotation, efficiency indicators and other related
parameters.
Environmental Impact Assessment of Agricultural System
Fig. 9:Concept of Environmental Impact Assessment of Agricultural System
Source: Environmental Impact Review.v.28
The agriculture has transformed from simple sustenance objective to intensive-
commercial form thus depleting and degrading the environmental resources.
Pressure on high production has led to intensification of agriculture. Intensive
Agriculture, long term sustainability and quality of natural resources, thus is matter
of compromise and concern. Agricultural is a major reservoir and transformer in
global cycles of carbon, nitrogen and water. Intensive agriculture leads to erosion
of soil resources, loss of biodiversity, alienation of ecological niches, temporary
imbalance in soil microbial functioning, associated long-term effects on microbial
processes and changes in biogeochemical Cycles. Agricultural activities contribute
about 70% of all anthropogenic N2O emissions and about 65% of all anthropogenic
CH4 emissions. Nutrient leakage from agriculture is a prime cause of degradation
of groundwater, surface waters and estuarine and coastal marine systems, and via
the atmosphere affects other terrestrial systems. Nitrate contamination of
groundwater is common in agricultural areas around the world. Some of the
specific components include fertilizer and pesticide residual toxicity, plant/soil
metabolic exudates such as methane/nitrous oxide in the immediate micro/macro
environment. Methane and nitrous oxide form important components of such an
interface. The GHG pattern which is also available using sensors on board
satellites is also being studied in detail which is clearly shown in fig(8).
The methodology was developed to generate total annual methane emission map
from the rice areas of India and its temporal pattern taking into consideration the
diverse conditions under which the rice is grown. The methodology was developed
for the variety components on use of RS data for stratification, spatial and temporal
sampling strategy, development of indigenous method of sampling and up-scaling
of methane. Results showed that the major stratum emerged as the rain fed drought
prone with 42.8 per cent of total rice lands (wet season) and found in many states.
Livestock constitutes an integral component of Indian agriculture. India possesses
the world’s largest total livestock population of 485 million, which accounts for
about 57% and 16% of the world’s buffalo and cattle populations, respectively. A
detailed state/ district-level methane emission inventory for different livestock
categories was made using the country-specific and Indian feed standard based
methane emission coefficients, which are based on IPCC guidelines, and the latest
available livestock census. The total methane emission including enteric
fermentation and manure management has been estimated as 11.75 Tg for the year
2003 (Chhabra et al. 2008, 2009). Enteric fermentation accounts for ~92% or 10.65
Tg of the total, while manure management contributes only 8% or 1.09 Tg.
Non-point source pollution
Agriculture has been identified as the largest contributor of non-point source (NPS)
pollution of surface and ground water systems globally (Thorburn et al., 2003).
Fertilizers, which are used as important inputs in agriculture to supply essential
nutrients like nitrogen (N), phosphorus (P), and potassium (K) also, serve as a
major non-point source pollutant. An integrated methodology was developed for
quantification of different forms of nitrogen losses from rice crop using remote
sensing derived inputs, field data of fertilizer application, collateral data of soil and
rainfall and nitrogen loss coefficients derived from published nitrogen dynamics of
kharif and rabi seasons. The nitrogen losses through leaching in form of urea-N,
ammonium-N (NH4-N) and nitrate-N (NO3-N) are dominant over ammonia
volatilization loss. The study results indicate that nitrogen loss through leaching in
kharif and rabi rice is of the order of 34.9% and 39.8% of the applied nitrogenous
fertilizer in the Indo-Gangetic plain region. This study provides a significant
insight to the role of nitrogenous fertilizer as a major non-point source pollutant
from agriculture.
Impact of Climate Change on Agricultural System
Fig.10: Impact of Climate Change on Agriculture
Source: http://www.mdpi.com/2071-1050/7/7/8437/htm
Climate change is one of the most discussed topics of the last two decades shown
in the fig(10). It impact on agricultural systems with inputs from multiple sources
were studies with simulation models. The CropSyst and water balance models
along with the climate forecasts using GCMs, RCM and statistical downscaled
model (under different scenario) for understanding the impact of climate changes
on agricultural systems were studied. The impact of climate change as projected by
the RCM-PRECIS (A1B scenario) on the rice-rice system of West Bengal showed
yield reduction ranging from 7.88 to 18.84 %.The adaptation study in rice-rice
system was early sowing by a period of 12 days to increase the yield in 2020 and
compensate the yield reduction in 2050. Analysis of the climatic extreme events
under climate change scenario of HadCM3 through different temperature and
precipitation indices was carried out.
Biophysical parameter retrieval and process modeling
The satellite derived biophysical products is one of the key developments taken
place during last two decades. The investigations on deriving these products were
carried out in India using Indian and other sensors. The NDVI, fAPAR, insolation,
LAI, LST etc are some of them on which R & D were carried out. The Leaf Area
Index (LAI) is a key biophysical variable used by plant physiologists and
modellers’ for estimating foliage cover and plant growth and biomass. The
regional modeling of growth processes such as evapotranspiration (ET) and net
CO2 assimilation require retrieval of some core variables such as land surface
temperature (LST), leaf area index (LAI), albedo and soil moisture etc.. Field-scale
(local) and regional-scale (agro-climatic zone) non-linear empirical models are
developed for wheat leaf area index (LAI) based on normalized difference
vegetation index (NDVI) using IRS 1D LISS-III.
Demonstration of technique for forewarning Pests and Diseases
The remote sensing pests and diseases has remained a challenge due to the
complexity of occurrence, overlapping with other factors, varying magnitudes and
subtle manifestation. The mustard aphid (Lypaphis erysimi) infestation models
have been developed from near-surface air temperature and relative humidity from
sounder data (e.g. TOVS), and sowing dates (Bhattacharya et al, 2007) and
validated through a collaborative study with National Research Centre on
Rapeseeds and Mustard (NRCRM), Bharatpur, Rajasthan. These models were later
up scaled and extrapolated using SPOT-VGT to map aphid onset dates (Dutta et al,
2008, Fig.).A new methodology of multi-stage tracking of Sclerotinia rot
(Sclerotinia sclerotiorum) disease in a large mustard growing region over in
Bharatpur district has been conceptualized and demonstrated.
National Missions:
FASAL: National production forecast
Fig.11: Concept of FASAL
Source:https://nrsc.gov.in/Forecasting_Agricultural_Output_using_Space_Agro-
meteorology_Land_based_Observations
Multiple in-season wheat forecasts using multi-date IRS WiFS data and weekly
weather variables at meteorological sub-divisions in major wheat producing states
of India are being made at national level shown in fig(11). Seven major wheat-
producing states, Uttar Pradesh, Punjab, Madhya Pradesh, Haryana, Rajasthan,
Gujarat and Bihar form the study area. These states account for more than 90% of
LandObservations
MULTIPLE IN-SEASON FORECAST
Pre-Season
Early-Season
Mid-SeasonState
Pre-HarvestState
Pre-HarvestDistrict
Cropped area Crop condition
Crop acreage
Crop yield
Revised Assessing Damage
Crop area &
Production
Crop area &
Production
the wheat production. A three level stratification of the area based on agro-climatic
zones, an agricultural area and crop proportion is made. A grid of 5X5 km is
overlaid on the satellite images and fifteen per cent random sample (each of size
5*5 km) is selected within a state. The sample segments are classified using in-
season ground truth and a hierarchical (decision rule based) classifier. The state
level acreage estimates are then statistically aggregated to arrive at national level
wheat acreage estimates. The methodology for state level yield forecasting is
multiple regression models based on temperature, using a correlation weighted
regression approach. The National level acreage and yield estimates are then
combined to provide National Production Forecast.
Investigations using space borne SAR data started with limited use of data from
JERS, SIR C sensors. However, the possibility of examining space borne radar
data for large area agricultural application was realised with the successful launch
of ERS-1 Synthetic Aperture Radar (SAR) (SAC, 1997). Due to the problem of
cloudy weather during rainy season, Kharif rice crop production estimates in the
major rice growing states are being generated using multi-date Synthetic Aperture
Radar data (fig. 11a). Rice growing environment or management practices ensures
that there is standing water beneath the canopy at least for a short duration during
crop season. This information is used to characterize the rice crop on a temporal
domain (fig. 11b). The rate of change and direction of change of SAR response
aids in building decision rule for classification of rice pixels. A stratified random
sampling approach for each state is adopted for acreage estimation with a sample
size of 5*5 km. A fifteen per cent sampling fraction and in-season ground truth
information of the selected sample segments are used. The segments are classified
using a decision rule classifier followed by statistical aggregation of state level
acreages to national level rice acreage estimate. The statistical relationship between
yield and rainfall during the cropping season is used for yield forecasts. The
district level models are combined with acreage and production forecasts for the
country is made.
Fig. 12a: Three date colour
composite of Scan SAR data (
R:G:B: Date1:Date2: Date 3) over
the study district showing distinct
signature of rice (b,c,d,e,f) in
different growth stage and other
land cover classes (a: water, h:
forest and i: urban).
Fig. 12b: Temporal pattern of rice crop
Source:https://www.researchgate.net/publication/301310407_Special_Issue_Extre
me_Weather_Events_and_Indian_Agricultu.
Crop Identification Technology Assessment for Remote Sensing (CITARS)
CITRAS (Crop Identification Technology Assessment for Remote Sensing) was an
experiment to quantify the crop identification performance achievable with several
automatic data processing classification technique .It was conducted from April
1973 to April 1975.The five specific objectives of CITARS were:
a
b c d
e
f
g
h
i
-25
-20
-15
-10
-5
0
5
-24 0 24 48 72 96 120 144 168
Days ( July 1=1)
Ba
ck
scatt
er
co
-eff
icie
nt
(dB
)
Rice Water Urban Homestead Other Crops
Temporal backscattering profile of Rice & Non-Rice Area
1
2
3
1 45 6 7
2 3 4 5 6 7
Field-preparation Puddling Transplanted Vegetative Peak-vegetative Heading Maturity
-25
-20
-15
-10
-5
0
5
-24 0 24 48 72 96 120 144 168
Days ( July 1=1)
Ba
ck
scatt
er
co
-eff
icie
nt
(dB
)
Rice Water Urban Homestead Other Crops
-25
-20
-15
-10
-5
0
5
-24 0 24 48 72 96 120 144 168
Days ( July 1=1)
Ba
ck
scatt
er
co
-eff
icie
nt
(dB
)
Rice Water Urban Homestead Other Crops
Temporal backscattering profile of Rice & Non-Rice Area
1
2
3
1 45 6 7
2 3 4 5 6 7
Field-preparation Puddling Transplanted Vegetative Peak-vegetative Heading Maturity
1) To assess the effects of Landsat data acquisition during the corn and soybean
growing season on crop-identification performance.
2) To access the effect of differing geographical locations having differing soil,
weather, management practice, crop distributions, and field sizes on crop-
identification performance.
3) To quantify the variation in crop-identification performance using the differing
automatic data processing (ADP) classification procedure.
4) To test the ability to extend training signatures, selected within the test area, to
train the classifiers in other areas.
5) To access crop identification benefits to be derived from classifying with
multiple Landsat data acquire multi-temporally during the crop growing season.
Technical Approach
Six major tasks were completed during CITARS they were:
i) Design of sampling scheme in Illinois and India for corn and soybeans
using Landsat MSS data.
ii) Acquisition and preparation of a Landsat-1 data set with ancillary data
sufficient to support the experimental objectives and design.
iii) Computer added processing of this data set with the selected
classification algorithms and procedures.
iv) Quantification of the crop-identification performances to evaluate the
ability of these procedures to satisfy agricultural applications
requirements.
v) Statistical analysis to quantitatively evaluate the impact of major factors
known to affect crop identification performance.
vi) Interpretation of the results to ascertain the underlying factors responsible
for the result and to drawn inference as to the status of the technology as
it relates to agricultural applications.
Conclusions
Remote sensing applications of agriculture expanded into different domains and
further many of them grew to higher levels of maturity during last twenty-five
years. The crop production forecasting for example started from experimental
stage and moved up to operational stage. The newer applications of agriculture
and its environment assessment were also explored. With availability of SAR
sensors, monitoring of crop during kharif season became a reality. India is a
global leader in agricultural applications of remote sensing and has carried out
capacity building on not only Indian scientist, but also of scientists of other
countries in India as well as outside. The specific tools and techniques have
been developed to cater to above needs and operationalisation. Almost all
sensors spanning the entire range of EMS used in RS application have been
studied and host of them have been showcased. The emphasis in future should
be on products and services sector encompassing more decipherable and ready
to used knowledge based RS products. The agro ecosystems analysis and
climate change impacts would be the focal theme in which variety of
components can come from RS data. The horticulture and site specific
management would demand much more complex algorithms and service
oriented products. The RS and communications technologies would fuse in the
future to deliver the near-real time service to all stakeholders of agriculture.