Comparative evaluation between NOAA AVHRR, MODIS, Feng Yun 1D, and LandSat 5 TM images working as a

1
Comparative evaluation between NOAA AVHRR, MODIS,
Feng Yun 1D, and LandSat 5 TM images working as a
satellite constellation for burned areas
detection on Paraná Medio Flooding Valley in
Argentina.
2008 Satellite Direct Readout Conference
Researchers
Arq. Carlos Cotlier1
Agrim. Benito Vicioso2
Agrim. José Belaga2
Tec. Diego López1
1 - Centro de Sensores Remotos - Universidad
Nacional de Rosario - Argentina
2 - FCEIyA - Universidad Nacional de Rosario -
Argentina
Asociación de Universidades Grupo Montevideo
Facultad de Ciencias Exactas, Ingeniería y
Agrimensura
2
Abstract.
Bushes and natural vegetation are burned on the
islands belonging to the Paraná River Flooding
Valley to obtain soft grass for cattle raising,
burning has been intensified in the last years,
destruction of a wild wetlands with unique
characteristic are done. Because of the unique
characteristics, this flora and fauna reservation
should be protected by monitoring against
indiscriminate burning. The use of NDII
(Normalized Difference Infrared Index) and BAI
(Burned Area Index) indexes are applied for the
analysis of the affected areas and images were
created with infrared bands combinations.
3
NOAA and Feng Yun Burned Detected Areas.
4
Applied Methodology.
Using images from different satellite sensors
over a same study area we tried to determine if
the data information of a satellite constellation
allow to detect which ones are optimum by means
of their spatial or spectral resolution.
a)
A LandSat 5 TM image was used (February 18, 2006)
to determine the fire focus location with the
most accurate precision, a main focus with a big
smoke column was chosen.
With aid of the high temporal resolution of
NOAA-AVHRR, Feng Yun 1D and MODIS, the next day
images (February 19, 2006) were used to check the
burned areas and analyze the sensors different
spectral response.
b)
5
Utilized Satellites.
FENG YUN Series (Wind and Cloud)
NOAA Series
North Oceanic and Atmospheric Administration US
Department of Commerce USA.
China Meteorological Administration Peoples
Republic of China.
1.1 Km Pixel Resolution
1.1 Km Pixel Resolution
6
Utilized Satellites.
Terra-Aqua
LandSat 5 TM
National Aeronautics and Space AdministrationUSA
National Aeronautics and Space Administration USA
30 m Resolución Pixel
500 m Resolución Pixel
7
Reception Antenna.
8
Capture and Decoding System.
9
02/18/2006 LandSat 5 TM Image Analysis.
Latitude Longitude
-33º 39 34.8 -59º 18 41.9
Analyzed central pixel
10
02/18/2006 LandSat 5 TM Image Matrixes.
11
NOAA - Feng Yun -MODIS - LandSat comparative
table.
12
Bow Tie effect on MODIS images.
In contrast to other scanning sensors like e.g.
AVHRR, MODIS is observing within one scan ten
lines of 1km spatial resolution (40 lines of 250m
resolution and 20 lines of 500m resolution,
respectively). Due to this unique feature, the so
called panoramic "bow tie" - effect occurs at the
border of each scene. In the figure, a schematic
layout of the "bow tie" effect is presented.
Three consecutive MODIS scans each consisting of
ten 1km lines. Due to the panoramic "bow tie"
effect, the scans are partially overlapping at
off
Nadir angles. The first and third scan are
represented by the light grey grids, while the
second scan is shown in black.
13
Index Calculation for Specific Area Detection.
NDII (Normalized Difference Infrared Index)
NOAA-AVHRR
MODIS
Feng Yun 1D
LandSat 5 TM
BAI (Burned Area Index)
NOAA-AVHRR and Feng Yun 1D
MODIS
14
19/02/2006 MODIS Image Analysis.
Latitude Longitude
-33º 39 34.8 -59º 18 41.9
Analyzed burned area central pixel.
Radiometric value 0.170700
Latitude Longitude
-33º 44 43.8 -59º 16 55.9
Analyzed healthy vegetation central pixel.
Radiometric value 0.290331
15
19/02/2006 MODIS Image Matrixes.
9x21 Matrix Subset - 7 Band
9x21 Matrix Subset - 2 Band
9x21 Matrix Subset - 1 Band
9x21 Matrix Subset - 32 Band
16
19/02/2006 Feng Yun 1D Image Analysis.
Analyzed burned area central pixel.
Latitude Longitude
-33º 39 34.8 -59º 18 41.9
Radiometric value 0.0948339
Latitude Longitude
-33º 44 43.8 -59º 16 55.9
Analyzed healthy vegetation central pixel.
Radiometric value 0.109571
17
Feng Yun 1D 19/02/2006 Image Matrixes.
9x21 Matrix Subset - 6 Band
9x21 Matrix Subset - 2 Band
9x21 Matrix Subset - 1 Band
9x21 Matrix Subset - 4 Band
18
19/02/2006 NOAA-AVHRR Image Analysis.
Analyzed burned area central pixel.
Latitude Longitude
-33º 39 34.8 -59º 18 41.9
Radiometric value 0.125630
Latitude Longitude
-33º 44 43.8 -59º 16 55.9
Analyzed healthy vegetation central pixel.
Radiometric value 0.152622
19
NOAA-AVHRR 17 19/02/2006 Image Matrixes.
9x21 Matrix Subset - 3 Band
9x21 Matrix Subset - 2 Band
9x21 Matrix Subset - 1 Band
9x21 Matrix Subset - 4 Band
20
CONCLUSIONS.
An optic satellite constellation proved to be
optimum for fire damage analysis when satellite
revisit dates allow it. With the aid of low and
medium spatial resolution, a more precision size
of the burned areas and possible assess of the
damage caused was determined.
1
The use of a LandSat TM 5 image with a 30 meters
spatial resolution to determine the main fire
focus and with the help of lower spatial
resolution images but with higher daily and
hourly revisit as NOAA, Feng Yun and MODIS allows
a better and more accurate data information in a
24 hours period after fire detection. Care should
be taken not confusing the burned areas with
water bodies (pure water) or any other surface
that can confuse .
2
21
CONCLUSIONS.
The question remaining is how to work with this
constellation methodology when LandSat revisit
time is 16 days? In this case we recommend the
use others satellite images like IRS, SPOT (30
meters resolution), SAC-C (with a 175 meters
spatial resolution), CBERS (25 meters
resolution), IKONOS, Quick Bird (one meters
resolution), etc.
3
Each year more countries are putting in orbit
more medium resolution satellite, in the near
future countries like Nigeria, China, Israel,
France and Korea has or are going to have more
MIR bands satellite with a spatial resolution of
30 meters or lower.
4
22
CONCLUSIONS.
If no commercial satellite match with their
revisit time, MODIS can be used because of the
daily revisit time but with a spatial resolution
of 500 meters.
5
Finally we can conclude that the use an optic
satellite constellation when revisit dates allow,
it is optimum for fire damage analysis. Allowing
with the use of Medium and Low spatial resolution
platforms allow to determine an accurate size of
burned areas and possible assess of the damage
with more precision.
6