\TION
re, accurate
ng as much
tudy of the
elationship
ral satellite
operational
ind surface
ce from the
yd accuracy
| emissivity
1al Oceanic
hrough the
nent of the
op fields in
the accurate
monitoring
However, a
rrors in the
f.
; the kinetic
other hand,
as the skin
ave thermal
owever, the
surface with
nieli 1999).
surface and
bed in three
servative on
icularly with
id the wind
the relation
modynamics
onitoring of
t. night-time
yy useful for
jective is to
International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Vol XXXV, Part B7. Istanbul 2004
observe the nocturnal LST decreasing and the variation in
regard to the air temperature.
The data were acquire by visible, near infrared and thermal
otbital scanning with “Advanced Very High Resolution
Radiometer” (AVHRR) instruments of the “National Oceanic
and Atmospheric Administration” (NOAA) of "Polar
Operational Environmental Satellites” (POES) system.
Currently, in orbit we have morning and afternoon satellites,
which provide global coverage six times daily. Because of the
polar orbiting nature of the POES system, these satellites are
able to collect global data on a daily basis for a variety of land,
ocean, and atmospheric fine structure applications.
Many researchers have studied the relationship between
emissivity and some experimental data (van de Griend & Owe,
1993) or theoretical vegetation parameters (Valor & Caselles,
996). In this study, the authors applied Valor & Caselles,
(1996) algorithm for the surface emissivity estimation derived
from measures of NDVI.
2. MATERIAL AND METHODS
This study area cover twelve (12) groud-based stations
distributed over the Rio Grande do Sul State with 281.731.64
Km’, at the South Brazil, delimitated by following cordinates:
Longitudes .49?4222"W and 57738'34"W. Latitudes
27*04'49"'S and 33?44'42"'S.
The study area presents particularly features of relief (canyon)
and crop fields covering to the north and it is often covered by
shrub and grassland vegetation in the other sites.
The satellite data used in this study with the highest elevation
angle overpasses in the Winter season of the 2002, were
received, analysed and selected from the permanent available
files at the Centro Estadual de Pesquisas em Sensoriamento
Remoto e Meteorologia (CEPSRM-UFRGS). The CEPSRM is
located at coordinates: 51?11'35'"'W and 30*06'39"'S.
The Table 1 present the geographical position of the 12 ground
based-stations under shelter installed 1.5 meters above the
surface in the Rio Grande do Sul State, Brazil.
Table 1. Geographical position of the stations
Ground-based stations location
Station Latitude Longitude Altitude (m)
| Bajé 31^ 20'S 54? 06"W 212
2 Bom Jesus 28? 40'S 50? 26"W 1046
3 Caxias 29° 10°S 51° 12’°W 817
4 E. do Sul 30? 32'S 829 31^ W 432
5 Farroupilha 29° 14’S 51? 26'W 783
6 Irai 272.41!S 53° 14’W 716
7 Porto Alegre 30° 05’S 51° 10°W 3
8 Quarai 30? 23'S 56° 26’W 112
9 Santa Rosa 27° 51'$ 54° 25°W 277
10 S. Vitoria 339 31'S 53° 21’W 23
1! Säo Luiz 289 23'S 54° 58°W 231
12 Taquarí 29? 48'S 5]? 49"W 54
It were selected 4 (four) night-time images from NOAA-15 and
16 with two overpasses in July 15 and September 04 dates, on
High Resolution Picture Transmission (HRPT) format as shown
in Table 2.
Table 2. Selected night-time images for LST data retrieval.
157
AVHRR data Date Overpass (GMT)
02196a16 July/15 04:59
02196a15 July/15 10:18
02247a16 September/04 05:41
02247a15 September/04 10:31
The determination of the LST variation analysis requires the use
of data obtained from satellites together measurements taken on
ground-based stations (in situ) measurements.
The main steps undertaken were: a) clear-sky satellite
overpasses selection; b) georreferencing data; c) monthly
maximum NDVI data genneration; d) emissivity data retrieval;
€) computing of the 9 pixels mean LST over the ground-based
stations position; f) generate of a scatterplot of LST variation
against Tair variation data for each day.
Formerly, it was applied in the software ERDAS-Imagine
version 8.5 the orbital algorithm for control points calculation
based on the efemerids data provided by NOAA for the orbital
route. This type of georreferencing with the Spheroid and
DATUM WGSS64 for high elevation overpasses often present an
mean accuracy of 2 pixels (2200m) for this region. Latter, it was
undertaken a second georreferencing with a number of control
points based on the well known location of water bodies in the
dimension of few pixels. The obtained results in the
georreferencing are quite good and necessary in the same way.
The ERDAS-Imagine 8.5 software was used to transform the 10
bits radiometric records, i. e. 1024 digital counter, in
temperature data. This module permit to transform digital
counters value in radiance and to get brightness temperature
data through Planck’s equation (Gusso & Fontana, 2003).
However the calculation based on the black body's theory
temperature by Planck's equation is complex numerically
because it needs to consider the spectral response function of
temperature to the radiance in differents wavelengths. Details of
the physical fundamentals of thermal interpreting signals by the
brightness temperature from the observed radiance in remote
sensing theory, are very well discribed at Sullivan (1999).
2.1 Land surface temperature
At any time, the value taken by the surface temperature Tskin
results from the balance between the various forms of energy
exchanges at the surface: net radiation, convective sensible and
latent heat fluxes and soil conduction heat flux. Because the
long wave radiation losses of the surface Tskin (LST) and Tair
both decrease during the night and reach their minimum values
generally just before sunrise (Lagouarde and Brunet, 1993).
Tskin usually remains slightly lower than Tair.
The radiative cooling at night and the difference Tair — Tskin
are greater the more important as the longwave downward
atmospheric radiation is low. Low values of soil thermal
conductivity, roughness and wind speed act in the same
direction, by limiting the energy exchanges between the surface
and the air above. However in most cases the night-time
temperature differences Tair — Tskin does not usually exceed 3
degrees Celsius (Oliveira, 1997). Values greater than 4 or 5
Kelvin are rarely found and correspond to very stable
atmospheric situations (Lagouarde and Brunet, 1993). Also, the
maximum surface temperature reached shortly after solar noon
strongly depends on surface conditions. For instance, on very
dry bare soils the difference may reach 20 K (Seguin et al, 1982
