=Paper=
{{Paper
|id=Vol-1152/paper1
|storemode=property
|title=Monitoring Cotton Crop Evapotranspiration Based On Satellite Data
|pdfUrl=https://ceur-ws.org/Vol-1152/paper1.pdf
|volume=Vol-1152
|dblpUrl=https://dblp.org/rec/conf/haicta/BlantaNMS11
}}
==Monitoring Cotton Crop Evapotranspiration Based On Satellite Data==
https://ceur-ws.org/Vol-1152/paper1.pdf
Monitoring cotton crop evapotranspiration based on
satellite data
Anna Blanta1a, Dalezios R. Nicolas1b, Aglaia Maliara1c, and Nicos Spyropoulos2
1
Laboratory of Agrometeorology, Department of Agriculture, Ichthyology & Acquatic
Environment, , University of Thessaly, Volos, Greece, email: aamplanta@uth.gr,
b
dalezios@uth.gr, caglaiamaliara@hotmail.com,
2
Agricultural University of Athens, Department of Natural Resource Development and
Agricultural Engineering, Athens, email: nicosp@hol.gr
Abstract. The water demand to meet seasonal and long-term water needs in
Thessaly, central Greece, is related to historical semi-arid conditions in the
region, which is the main agricultural area of the country. In this paper
irrigation water requirements are assessed through the estimation and
monitoring of crop evapotranspiration ETc for cotton fields in Thessaly.
Remotely sensed data are used to delineate the spatial and temporal variability
of crop coefficient Kc and crop ETc. Cotton crop production is examined for
the years 2007, 2008, 2009 and 2010. Weekly ground based measurements
carried out throughout the growing season and satellite images (Landsat TM)
were processed for the corresponding time period. Satellite data provide the
cover capability of large scale areas and monitoring of crop during growth
stages. Methodology can be applied in large scale areas for the calculation of
Kc and extend to other crops using satellite data. The results are in good
agreement with ground- truth observations.
Keywords: Remote Sensing, Kc, ETc
1 Introduction
Agriculture of any kind is strongly influenced by water availability. In semi-arid
regions, such as Mediterranean, agriculture is already the largest consumer of water
resources. Actual and/or potential evapotranspiration (ET) estimation and monitoring
is important in irrigation scheduling by contributing in rationalizing water needs
during the growing season (Pereira et al, 1999). Monitoring of ET becomes even
more significant when water scarcity combined with drought events cause more
difficulties to agricultural production. Evapotranspiration (ET) is one of the most
significant parameters in agriculture, since it can justify whether water is used
effectively or not. Moreover, ET spatial and temporal variability, in different land
uses can be considered to provide adequate and reliable assessment of water use.
Nevertheless, it is difficult to obtain accurate measures or estimates of ET due to the
____________________________________________
Copyright ©by the paper’s authors. Copying permitted only for private and academic purposes.
In: M. Salampasis, A. Matopoulos (eds.): Proceedings of the International Conference on Information
and Communication Technologies
for Sustainable Agri-production and Environment (HAICTA 2011), Skiathos, 8-11 September, 2011.
1
complexity and variability of meteorological and biophysical components involved
in the process.
Remote sensing methods have already reached a significant level of accuracy and
reliability over the last forty years, thus becoming attractive for ET estimation, since
they have a very high resolution and cover large areas. Remotely sensed models are
currently considered suitable for crop water use estimation at fields as well as
regional scales (Bastiaanssen et al., 1998; D’Urso and Menenti, 1996). In this paper,
evapotranspiration ETc is estimated and monitored in cotton fields in Thessaly,
central Greece, using remotely sensed data. In particular, LANDSAT images are
processed and analyzed in order to compute the Normalized Difference Vegetation
Index (NDVI) and then the crop (cotton) K c coefficient, which is used in
evapotranspiration ETc equation. The method is validated by comparing ET c
estimation using ground-truth conventional meteorological data. The paper is
organized as follows: section 2 describes the study area and the data base, which
includes meteorological and satellite data, as well as agronomic, geographic and
phenological information from selected cotton plots. In section 3 the methodology is
presented including data processing and ET c estimation. Section 4 and 5 shows an
analysis and discussion of results.
2 Study area and Database
2.1 Study area
For the experimental layout the pilot area that was selected is the Pinios river
basin, Thessaly, central Greece, a high agricultural productivity area that produces
high quality products. The region of Thessaly overtakes the central - Eastern
department of continental Greece. It is constituted by the Prefectures Karditsa,
Larissa, Magnesia and Trikala and overtakes total extent of 14036 Km2 (10.6% of
total extent of country). The 36.0% of ground are in a plain, the 17.1% semi-
mountain, while the 44.9% is mountainous. High mountains surround the plain of
Thessaly, which constitutes the bigger plain of country that divided westwards to
Eastern from the river Pinios that is the third bigger river of country. The study area
is under drought conditions. Agriculture is affected by limited availability of water
resources. In this area, intense and extensive cultivation, mostly with water
demanding crops, leads to overexploitation of groundwater. Crop selected is cotton
which composes the main cultivation and is one of the most water demanding crops
in the study area. During spring and autumn the climate is usually not stable and this
has great influence on cotton, as both seasons are very critical for the crop (planting-
harvesting periods). Precipitation is very low during the cotton growing period (April
– September) so that irrigation is needed for the crop water requirements. The
irrigation water comes from rivers by about 46% and from underground water by
about 54%. When rainfall during winter of the previous year is limited, shortage of
irrigating water is apparent.
2
Fig. 1. Study area
Table 1. Irrigated extents in Thessaly
Extents in (thousands hectares) Thessaly
Agricultural ground 3.152
Total of cultivations 3.130
Irrigated 1.672
Percentage % 53
2.2. Database
Monitoring water needs for agriculture in Thessaly requires a combination of field
observations in pilot area, micro-meteorological data and analysis of satellite data. In
this paper the process is described for the computation of crop coefficient Kc, and
crop evapotraspiration ETc for cotton fields in central Creece, a high agricultural
productivity area that produces high quality products.
Conventional data: Ground based micro-meteorological measurements of reporting
period for the pilot area include air temperature, wind speed, humidity, and
precipitation on daily basis, in order to extract reference evapotraspiration (ET o),
crop evapotranspiration ETc, and Kc, for cotton crop for years 2007, 2008 2009 and
2010.
Satellite data: Remote sensing data are used for agriculture monitoring. The spatial
distribution of evapotranspiration is assessed using satellite imagery (Landsat-5 TM)
covering the region of Pinios river basin, which was available during the campaign
for years 2007, 2008 2009, 2010 at the following dates:
3
2007 2008 2009 2010
07/05, 24/06, 30/03, 15/04, 01/05, 02/04, 20/05, 05/06, 21/04, 07/05,
10/07, 26/07, 17/05, 02/06, 18/06, 21/06, 07/07, 23/07, 08/06, 27/08,
27/08, 28/09 04/07, 20/07, 05/08, 24/08 28/09, 03/10
21/08, 06/09
Field data: With regards to field observations, fractional cover and phenological
stages for cotton crop for years 2007, 2008, 2009 and 2010 are recorded. Sampling
started on May and repeated twice per week for the months of May, June, July
August and September. In all pilot areas recorded and measured the followings:
photographs, localization of the fields, crop height, irrigation data and meteorological
data.
3 Methodology
The paper involves monitoring water needs for crop yield through estimation of
evapotranspiration using also satellite data. Methodology includes crop fractional
cover and crop classification, estimation of crop evapotranspiration ETc and crop
coefficient Kc.
3.1 Crop fractional cover and crop classification
Crop fractional cover: One of the factors that determine the crop coefficient Kc is
the crop growth stages. As the crop develops, the ground cover, the height and the
leaf area change. The growing period can be divided into four distinct growth stages:
initial, development, mid-season and late season (FAO, 1998). The initial stage runs
from planting date to approximately 10% ground cover. The development stage runs
from 10% ground cover to approximately 70%. The mid-season stage runs from
effective full cover to the start of maturity. The late season stage runs from maturity
to harvest (FAO, 1998). For the determination of initial and development stages of
cotton for the study area one experimental station is set up. In each experimental
station two polygons are created 93cm*93cm for cotton.
Using ArcGIS software estimated the percentage of ground cover and hence the
initial and development stages.
4
Fig. 2. The process applied to field based on canopy characteristics.
Crop classification: Fifty control points (GCPs) were used for crop classification for
main and most cultivated crops in area as signatures for supervised classification that
was done to ERDAS IMAGINE 9.1. The GCPs were distributed in a uniform manner
along the area of interest. The most common cultivations in Pinios river basin are
cotton, alfalfa, corn, winter wheat (fig 3).
Fig. 3. Crop classification of study area.
5
3.2 Estimation of Crop Evapotranspiration ETc
3.2.1 Computation of Reference Evapotranspiration ETo
Reference evapotranspiration is the rate from a reference, not short of water (FAO,
1998). ETo calculated for years 2007, 2008 2009 and 2010 with ground based
meteorological data of study area according to Penman-Monteith equation:
37 (1)
0.408D( Rn - G ) + g u 2 (e o (Thr ) - ea )
Thr + 273
ETO =
D + g (1 + 0.34u 2 )
where ETo: reference evapotranspiration (mm day-1), Rn: net radiation at the crop
surface (MJ m-2 day-1), G: soil heat flux density (MJ m-2 day-1), T: mean daily air
temperature at 2 m height (0C), u2: wind speed at 2 m height (m s-1), es: saturation
vapour pressure (kPa), ea : actual vapour pressure (kPa), es-ea : saturation vapour
pressure deficit (kPa), Δ: slope vapour pressure curve (kPa 0C-1), γ: psychrometric
constant (kPa 0C-1)
3.2.2 Estimation of crop coefficient Kc
Kc calculated for years 2007, 2008 2009 and 2010 with ground based
meteorological data and field data of study area. Kcinit=0.14 (for cotton) taken from
table. Kc for development stage derived by linear regression using the last value by
initial stage and the first value of mid-season for all years. An indicative equation of
linear regression is y=0.0208*-3.1095.
According to Penman-Monteith equations:
0.3 (2)
æhö
K cmid = K cmid (Tab) + [0.04(u 2 - 2) - 0.004(RH min - 45)]ç ÷
è3ø
where
Kcmid(Tab): value for Kcmid taken from table, u2:mean value for daily wind speed at 2 m
height over grass during the mid-season growth stage, RHmin: mean value for daily
minimum relative humidity during the mid-season growth stage, h: mean cotton
height during the mid-season growth stage.
The same procedure followed for the estimation of Kcend using the daily wind
speed at 2 m height over grass, mean value for daily minimum relative humidity and
mean cotton height for the corresponding late season growth stage.
6
3.2.3 Estimation of crop evapotranspiration ETc
Crop Evapotranspiration under “standard” condition ETc is the evapotransiration
from disease-free, well-fertilized crops, grown in large fields, under optimum soil
water condition and achieving full production under the given climatic conditions
(FAO, 1998). In FAO ET c is calculated as follows:
ETc = K c * ET0 (3)
where Kc and ETo calculated by ground based observations.
3.2.4 Estimation of crop evapotranspiration ETc and crop coefficient Kc based
on satellite data
Selection of satellite images, correction and extraction of reflectance, NDVI, ETc,
Kc maps were produced for years 2007, 2008, 2009 and 2010.
Preprocessing of satellite data: The satellite data pre-processing includes the
atmospheric and geometric correction of the Landsat data. The Landsat-5 TM
satellite images acquired almost every 15 days (2 images per month) for the
cultivation period (May to September) for years 2007, 2008, 2009 and 2010.The
pixel size of images is 30 x 30 m. The atmospheric correction was done using
ATCORE2 model in ERDAS IMAGINE 9.1. Geometric correction of satellite
images performed to software ArcGIS with the use of 12 digital georeferred 1:50000
scale maps that cover spatial the wide area of satellite images. Over eighty ground
control points (GCPs) were used for each image with a third degree polynomial
equation for the geometric transformation. The GCPs were distributed in a uniform
manner along the area of interest. All images were co-registered into the Hellenic
Geodetic Reference System (EGSA’87) using ArcGIS software package.
Image processing: The image processing includes extraction of reflectance, NDVI,
Kc and ETc maps.
Extraction of Reflectance: Reflectance in agriculture describes interaction of light
with soil and crops. Satellite images provide reflectance from the various
components of a crop canopy.
Extraction of NDVI: The development of vegetation indices from satellite images
have facilitated the process of differentiating and mapping vegetation by providing
valuable information about structure and composition. NDVI is exoressed by the
folowing equition:
NDVI=(NIR–RED)/(NIR+RED) (4)
7
For Landsat channel 3 (0.63-0.69) and channel 4 (0.76-0.90) are utilized to calculate
NDVI.
Extraction of Crop Coefficient (Kc): The Kc coefficient integrates the effect of
characteristics that distinguish a typical filed crop from the grass reference, which
has a homogenous appearance and covers completely the ground. The values of Kc
are influenced by crop type, climate, soil evaporation and crop growth stages (Allen
et al., 1998; Bailey, 1990). In the framework of the PLEIADES project, an equation
for Kc estimation was developed for the study area:
K c = 1.15NDVI + 0.17 (5)
Extraction of Crop Evapotranspiration (ETc): Crop Evapotranspiration under
“standard” condition ETc is the evapotransiration from disease-free, well-fertilized
crops, grown in large fields, under optimum soil water condition and achieving full
production under the given climatic conditions (FAO, 1998). In FAO ETc is
calculated as follows:
ETc = Kc * ET0 (6)
where Kc remote sensing and ETo ground based.
3.2.5 Error statistics
The accuracy of the remotely sensed ETc and Kc estimated series is evaluated by
comparison with the corresponding ground-truth ETc and Kc estimations through
four error statistics. The following statistics are employed:
(7)
å (ET - ET )
n
2
cg cs
Eff = 1 - i n=1
å (ET - ET )
2
cg cg
i =1
where equation (7) is the efficiency coefficient, ETcg: ground based values of ETc,
ETcs: satellite based values of ETc, ETcg(uperliing):mean ground based values of
ETc.
8
(8)
å (ET - ET )
k
2
cg cg
RMSE = i =1
K
where equation (8) is the root mean square error (RMSE), ETcg: ground based values
of ETc, ETcg : mean ground based values of ETc, K: number of cases.
(9)
å (ETcg - ETcs )
1 N
BIAS =
N i =1
where equation (9) is the statistical BIAS, ETcg: ground based values of ETc, ETcs:
satellite based values of ETc, N: number of cases.
2 (10)
æ ö
ç
( )( ÷
)
n
ç å ETcg - ETcg ETcs - ETcs ÷
r2 = ç i =1
÷
ç 2 ÷
( ) ( )
n n
å å
2
ç ETcg - ETcg ETcs - ETcs ÷
è i =1 i =1 ø
where equation (10) is the coefficient of determination r2, ETcg are ground based
values, ETcs are satellite based values, ETcg are mean ground based values, ETcs
are mean satellite based values.
4 Results and Discussion
The results are summarized in Table 2 and Figures 4 and 5. Table 2 presents the
error statistics results, namely efficiency coefficient (Eff), RMSE, BIAS and r 2. The
results are considered satisfactory ranging within acceptable levels. Figures 4 and 5
present sample images of the analysis of Kc and ETc, respectively.
The methodology is based on new technologies (GIS combined with satellite data)
for water management. Satellite data provide the cover capability of large scale areas
and monitoring of crop during growth stages. Methodology can be applied in large
scale areas for the calculation of Kc and extend to other crops using satellite data.
New technologies provide easy access to information for all stakeholders (farmers,
Irrigation Advisory Services, Local Organizations of Land Reclamation) while active
participation will be effective with by spatial information and innovative networking
tools.
9
Table 2. Results of error statistics for Kc and ETc
Kc ETc
Statistics
field 1 field 2 field 1 field 2
Eff 0.12 -0.24 0.64 0.74
RMSE 0.35 0.29 0.91 1.15
BIAS 0.08 0.01 0.06 -0.03
R2 0.31 0.16 0.8 0.88
Fig. 4. Kc map of study area (10/072007 and 26/07/2007).
10
Fig. 5. ETc map of study area (10/072007 and 26/07/2007).
5 Conclusions
Crop evapotranspiration ETc is estimated and monitored in cotton fields in
Thessaly, central Greece, using remotely sensed data. In particular, LANDSAT
images are processed and analyzed in order to compute the Normalized Difference
Vegetation Index (NDVI) and then the crop (cotton) Kc coefficient, which is used in
evapotranspiration ETc equation. The method is validated by comparing ET c
estimation using ground-truth conventional meteorological data. The results are in
good agreement with ground-truth observations. The results of the error statistics are
considered satisfactory ranging within acceptable levels.
Acknowledgements: The paper was funded by Pleiades, Smart and Hydrosense EC
projects. Also we would like to thank local farmers and Authorities for their helpful
cooperation.
References
1. Allen, R.G., Pereira, L.S., Raes, D., Smith, M., 1998. Crop evapotranspiration:
Guidelines for computing crop requirements. Irrigation and Drainage Paper No.
56, FAO, Rome, Italy, 300pp.
11
2. Bailey, J.O. (1990). The Potential Value of Remotely Sensed Data in the
Assessment of Evapotransiration and Evaporation. Remote Sensing Reviews,
4(2), 349-377.
3. Bastiaanssen, W.G.M., 1998: Remote Sensing in water resources management:
The state of the art. International Water Management Institute, Colombo, 118pp.
4. Blanta, A., C. Domenikiotis and N.R. Dalezios (2006), “Transferability of the
DEMETER Methodology”, International Conference on: Information Systems in
Sustainable Agriculture, Agroenvironment and Food Technology, 20-23
September 2006, Volos, Greece (in press).
5. D'Urso, G. and M.Menenti, 1996 a. Mapping crop coefficients in irrigated areas
from Landsat TM images. Proc. European Symposium on Satellite Remote
Remote Sensing. Optical Engineering, Bellingham USA, vol. 2585: 41-47.
6. Mplanta, Α., C.Domenikiotis and N.R. Dalezios (2010). “Remotely sensed ETp
and crop coefficient Kc in cotton fields of central Greece”. International Congress
on Information and Communication Technologies in Agriculture, Food, Forestry
and Environment. June 14-18, 2010. Ondokuz Mayis University, Samsun,
Turkey. (pp 116-121)
7. Mplanta, A., C. Domenikiotis and N. Dalezios (2007), “Pleiades-EO-Assisted
Tools For Water Management in Agriculture and Agricultural Decision Support”,
Greece Association Agricultural Engineers, 5th National Congress Agricultural
Engineering, 18-20 October 2007, Larisa, Greece, 345-352 pp.
8. Pereira L. S., Alain Perrier A. and R. G. Allen, 1999.
EVAPOTRANSPIRATION: CONCEPTS AND FUTURE TRENDS. JOURNAL
OF IRRIGATION AND DRAINAGE ENGINEERING / MARCH/APRIL 1999 /
51.
12