hanan abou ali , donna delparte , michael griffel · 4 14000 5 24400 6 38200 7 65700 8 28800 9...
TRANSCRIPT
RESEARCH POSTER PRESENTATION DESIGN © 2015
www.PosterPresentations.com
Lebanon has traditionally been a major potato producer, 451,860 tons in
2014, and exporter where 60% of its production is distributed within the
Arab region, to the UK and Brazil and has potato make up 30% of the total
agricultural exports. The purpose of this study is to promote precision
agriculture techniques in Lebanon that will help local farmers in the central
Bekaa Valley with land management decisions. The European Space
Agency’s satellite missions Sentinel-2A, launched June 23rd 2015, and the
Sentinel-2B, recently launched on March 7th 2017, are multispectral high
resolution imaging systems that provide global coverage every 5 days. The
Sentinel program is a land monitoring program that includes an aim to
improve agricultural practices. The imagery is 13 band data in the visible,
near infrared and short wave infrared parts of the electromagnetic spectrum
and ranges from 10-20 m pixel resolution. Sentinel is freely available data
that has the potential to empower farmers with information to respond
quickly to maximize crop health. Due to the political and security conflicts
in the region, utilizing satellite imagery for Lebanon is more reasonable
and realistic than operating Unmanned Aircraft Systems (UAS) for high
resolution remote sensing. During the 2017 growing season, local farmers
provided detailed information in designated fields on their farming
practices, crop health, and pest threats. In parallel, Sentinel-2 imagery was
processed to study crop health using the following vegetation indices:
Normalized Difference Vegetation Index, Green Normalized Difference
Vegetation Index, Soil Adjusted Vegetation Index and Modified Soil
Adjusted Vegetation Index 2. As most Lebanese farmers inherit their land
from their parents over generations, most still use traditional farming
techniques for irrigation, they make decisions based on prior generations’practices, which is no longer compatible with changes in climatic
conditions in the region. Normalized Difference Water Indices are
calculated from satellite bands in the near-infrared and short-wave infrared
to provide a better understanding about the water stress status of crops
within the field. Preliminary results demonstrate that Sentinel-2 data can
provide detailed and timely data for farmers to effectively manage fields.
Despite the fact that most Lebanese farmers rely on traditional farming
methods, providing them with free crop health information on their mobile
phones and allowing them to test its efficiency has the potential to be a
catalyst to help them improve their farming practices.
ABSTRACT
Study Area
METHODS RESULTS
CONCLUSIONS & FUTURE WORK
IMAGE SOURCES
. Planet Team (2017). Planet Application Program
Interface: In Space for Life on Earth. San Francisco,
CA. https://api.planet.com
. Copernicus Sentinel data [2017].
. Esri, DigitalGlobe, GeoEye, i-cubed, USA FSA,
USGS, AEX, Getmapping, Aerogrid, IGN, IGP,
swisstopo, and the GIS User Community
Hanan Abou Ali: [email protected]
(1) Idaho State University – Department of Geosciences, Pocatello, ID 83209 (2) Idaho National Laboratory – Department of Biofuels and Renewable Energy Technology, Idaho Falls, ID 83415
Hanan Abou Ali(1) , Donna Delparte(1) , Michael Griffel(2)
UTILIZING SENTINEL-2 SATELLITE IMAGERY FOR PRECISION AGRICULTURE OVER POTATO FIELDS IN LEBANON
Table 1. Field AreasFigure 1. Study Area, Tal Znoub, Bekaa, Lebanon -
Basemap source: Planet Team (2017).
Vegetation Index Formula Reference
Normalized
Difference Vegetation
Index
𝜌𝑁𝐼𝑅 − 𝜌𝑅𝜌𝑁𝐼𝑅 + 𝜌𝑅 (Rouse et al. 1973)
Soil Adjusted
Vegetation Index
𝜌𝑁𝐼𝑅−𝜌𝑅𝜌𝑁𝐼𝑅+ 𝜌𝑅 +𝐿 + (1+L) (A. R. Huete 1988)
Normalized
Difference Water
Index
𝜌𝐺𝑟 𝑛 − 𝜌𝑁𝐼𝑅𝜌𝐺𝑟 𝑛 + 𝜌𝑁𝐼𝑅 (S.K. 1996)
Field_ID Area (m2)
1 6000
2 11900
3 19200
4 14000
5 24400
6 38200
7 65700
8 28800
9 320000
10 35400
11 30400
27 38300
28 31500
29 17200
30 18800
33 50800
Table 2. Band Math
The study area is located in Tal Znoub in the southern western part of the
Bekaa Valley in Lebanon. It lies northern of Quaroun Lake and is along the
path of the Litani River. It is located at 4 km north northeast of the city of Jeb
Jannine which is the capital of the West Bekaa. The overall area of the site is
462,267 m2 which is divided into sub fields as shown in Table 1.
CONTACT INFO
The Sentinel – 2A and Sentinel – 2B imagery are processed using
the open source software QGIS for atmospheric correction via the
Semi-Automatic Classification Plugin. The software takes level –1C Sentinel imagery metadata and individual bands and converts
the imagery from “Digital Count” to “Reflectance Values” to be
able to run indices and perform the needed analyses.
After the scenes are corrected in QGIS, the needed bands (Near
Infrared: Band 8, Red: Band 4 and Green: Band3) were imported
into ArcMap for processing of vegetation indices (Table 2). In
order to increase efficiency, using raster calculator, the various
indices’ formulas were built into a tool using model builder in
ArcMap and the output raster datasets were saved into a specific
geodatabase for data management purposes.
After the processing of all indices on the fields, “Zonal Statistics
as Table” tool in ArcMap was used to summarize the values
obtained. Various statistics were calculated and exported as an
excel sheet, these statistics included the following information for
each field: minimum value, maximum value, mean and standard
deviation
Figure 2. Normalized Difference Vegetation Index Figure 3. NDVI – Zonal Statistics
Figure 4. Soil Adjusted Vegetation Index
Figure 5. Soil Adjusted Vegetation Index – Zonal Statistics
Figure 6. Normalized Difference Water Index Figure 7. Normalized Difference Water Index – Zonal Statistics
Through the various indices processed over Sentinel – 2 data,
satellite imagery proved that it is a reliable source for analysis and
for precision agriculture applications. All the indices that were
used showed compatibility with one another as well as with data
provided from the farmer in Lebanon. The peak of the season
based on the imagery analysis was in late May through mid-June
which is based on the farmer’s input just before the crops were
harvested in late June. In addition, the water index showed that the
water content was not uniform throughout the fields and that is
related to the fact that the field is not at one level which according
to the grower influences how the water is distributed around the
field.
Despite the limitations in this project, it still has long ways to
come. While the only dataset used for this was Sentinel – 2,
processing Planet RapidEye and 4Band data will allow the more
accurate analysis due to the almost daily coverage over the study
area as opposed to much less with Sentinel – 2. In addition,
Planet data provides a higher resolution and having more data
will allow the option to run regression models on the data which
was not feasible with this limited dataset here so this is the next
step.
CITATIONSHuete, A. R. 1 . A Soil-Adjusted Vegetatio I dex SAVI .
Remote Sensing of Environment 25(3): 295–309.
Rouse, J. W., R. H. Hass, J.A. Schell, and D.W. Deering. 1973.
“Monitoring Vegetation Systems in the Great Plains with ERTS.” Third Earth Resources Technology Satellite (ERTS)
symposium 1: 309–17.
S.K., McFeeters. 1 6. The Use of the Nor alized Differe ce Water Index (NDWI) in the Delineation of Open Water
Features. International Journal of Remote Sensing 17(7):
1425–32.