184 - The Basics of Satellite Oceanography. 3. Remote Sensing of The Sea

Lecture 3
Remote Sensing of the Sea
IoE 184 - The Basics of Satellite Oceanography. 3. Remote Sensing of the Sea
Remote sensing of the sea includes:
1. Sensor calibration
2. Atmospheric correction
3. Positional registration
4. Oceanographic sampling for "sea truth"
5. Image processing
6. Oceanographic applications of satellite remote

sensing
Compare satellite remote sensing and the
traditional sources of oceanographic information:
Remote sensing is better than traditional methods:
1. Synoptic view, because satellites collect huge amount of

information much exceeding the data collected by contact
oceanographic observations;
2. Satellite observations cover wide areas of the World Ocean
hardly accessible for field observations.
Problems of remote sensing:
1. The parameters measured by the satellites cannot be directly

attributed to traditionally measured oceanographic
characteristics;
2. Some satellite observations (ocean color and infrared) are
more sensitive to unfavorable meteorological conditions than
traditional oceanographic methods.
Each oceanographic equipment should be calibrated both before

and after deployment. In the case of satellites we need to take
into account the following:
1. The stress of launch;
2. High vacuum of outer space;
3. The power limitations on board the satellite, often resulting

in gradual deterioration in the power supply on the
satellite;
4. No opportunity of retrieving the instrument for periodic

recalibration in the laboratory.
On some scanners, part of the

scan views a reference target,
a lamp of known brightness for
the visible wavelength
scanners, or a black body of
measured temperature for
thermal IR sensors. In this way
gradual drift of the sensor can
be detected and corrections
made in the data analysis.
Some sensors use the moon as a natural

object with constant optical characteristics.
The sensors look at the ocean surface through another medium,

the atmosphere. The atmosphere is opaque to electromagnetic
radiation at many wavelengths, and there are only certain
wavelengths through which radiation may be fully or partly
transmitted.
The following compounds of

the atmosphere change its
transmittance:
• Gas molecules themselves

• Water vapor
• Aerosols (soot, smoke, air
pollution, etc.)
• Suspended particles of dust
• Water droplets in the form
of clouds
~40% of sunlight is reflected by clouds

~20% of sunlight is absorbed by the atmosphere
~40% of sunlight is absorbed by Earth’s surface
Ray 1 - the useful signal;
Ray 2 - the radiation leaving the
sea which is absorbed by the
atmosphere;
Ray 3 - the radiation, which is
scattered by the atmosphere out
of the sensor field of vision.
Ray 4 - the energy emitted by the
constituents of the atmosphere;
Ray 5 - the energy reflected by
scattering into the field of vision
of the sensor;
Ray 6 - the energy which
previously left the sea surface
but from outside the field of view.
Atmospheric pathways of
electromagnetic radiation between
the sea and the satellite sensor.
The ocean area within the IFOV

emits rays 1+2+3
Rays 4, 5, and 6 reach the

sensor without having left the
sea surface in the field of view,
and therefore constitute
extraneous "noise" on top of
the signal.
The sensor receives rays

1+4+5+6
The complete atmospheric

correction should result in the
sum of rays 1+2+3.

In the case of optical sensors

Ray 2 is absent.
On the contrast, in thermal IR

sensors Ray 2 is important:
the cool atmosphere absorbs
radiation (Ray 2) and re-
emits it with lower
temperature characteristics
(Ray 4).
The main strategies of atmospheric correction:

1. No separate attempt of atmospheric correction, instead we
calibrate each scene with ground data.
2. A universal atmospheric correction based on an average

model of atmospheric effects.

3. Using different wavelengths, assuming that certain channels
are unlikely to have any upwelling radiation from the sea. In
this way we process each pixel of the image.

4. An atmospheric (microwave) sounding sensor can be
mounted on the same satellite as an oceanographic sensor.
100
9 0
Without atmospheric correction, each

8 0
7
6
0
0
S a n P e d ro 0 2 /2 7 /2 0 0 4
50
scene can be calibrated with ground

40
T o ta l S u s p e n d e d S o lid s ( m g /L )
30
20
data, but the slope of correlation for 10
each scene is unique.

9
8
7
6
5
0 1 2 3 4
n L w 5 5 5 (m W c m -2  m -1 s r-1 )
Increased atmospheric pathlength resulting from oblique

viewing.
The oblique view results in looking through longer path length
of atmosphere than for nadir viewing. This feature is used in
atmospheric correction. By viewing the same piece of sea
twice, through different lengths of atmosphere, an objective
estimate of atmospheric effect can be made.
2. Atmospheric correction - Cloud detection
Cloud cover is a main obstacle

for satellite imagery in visible
and infrared spectral bands.
Clouds are transient
atmospheric features that
consist of small ice and liquid
water particles with dimensions
from under a micrometer to a
few millimeters, resulting from
water condensation and
freezing.
Cloud properties vary with height. In the visible and infrared part of
spectrum, the liquid water and ice crystals contained in the clouds
scatter and absorb radiation, so that thick clouds make it impossible
to view the surface. At any time, clouds cover almost two-thirds of the
globe.
In both ocean color and SST, the first step of the

procedure of atmospheric correction is to determine
if every oceanic pixel in the image under
investigation is cloud-free.
The SeaWiFS (8 optical

channels) cloud detection is
most primitive.
It is assumed, that the water-

leaving radiance of near-
infrared wavelength is near
zero. As such, the pixels with
a reflectance greater than a
preset threshold are
classified as clouds.
In AVHRR and MODIS the cloud detection is based

on two factors:
1) the clouds are colder and more reflective than the ocean surface;
2) for spatial scales of order 100 km, the ocean surface, in contrast
to clouds, is nearly uniform in temperature and reflectance.
Three kinds of tests are used:

1) “Threshold” tests eliminate pixels that are more reflective or
colder than the ocean surface.
2) “Uniformity” tests examine the variance of temperature or
reflectance in a rectangular array of pixels.
3) The retrieved SSTs are compared with climatology and with SSTs
retrieved using alternative algorithms; e.g., according to the
“unreasonableness” test, SST must be within the range from
-2ºC to +35ºC.
Positional registration means the identification on a map of the

place to which a remote-sensed measurement refers. The problem
of knowing where the satellite was when a measurement was
made depends on type of sensor, first of all its spatial resolution.
An approximate
estimation of the
satellite position
can be obtained
from the time of
observation.
However, the
precision of this
estimation is
within few
kilometers.
AVHRR radiometer on NOAA satellite

Positional registration means the identification on a map of the

place to which a remote-sensed measurement refers. The problem
of knowing where the satellite was when a measurement was
made depends on type of sensor, first of all its spatial resolution.
An approximate
estimation of the
satellite position
can be obtained
from the time of
observation.
However, the
precision of this
estimation is
within few
kilometers.
AVHRR radiometer on NOAA satellite

Often the "ground control points" are used. However, the ground
control points can be used mostly in the coastal zones. The
problem of distortion of the image results from oblique viewing of
the spherical earth surface. During processing each pixel of the
image should be attributed to geographical coordinates.
In recent satellites
more precise
estimation of the
position is obtained
using the signals
of GPS (Global
Positioning
System) satellites.
Most sophisticated method of position registration is used in

TOPEX/Poseidon radar-altimeter. It is based on Doppler effect.
When the sensor is moving

from the signal the
frequency it receives
“decreases”.
When the sensor is moving

toward the signal the
frequency it receives
“increases”.
DORIS system determines the position of TOPEX/Poseidon

satellite orbit to within a few centimetres.
The technique used (known as orbit

determination), consists of locating a
satellite in relation to about fifty ground
control points on the Earth's surface.
The main problem is that in general

the remote-sensed characteristics of
the sea change on a much shorter
time scale than those on the land.
Using for this purpose the overpasses
of the satellite should be done
carefully. In some cases it is
impossible (e. g., altimeter measuring
swell waves).
In other case (SST or water color

measured few hours one after
another) we can compare overpasses
of the satellites.
The strategy of collecting of samples is very important. The samples
must span as wide range of data values as possible. Typically,
transects across the gradients are used.
M O D IS n L w (5 5 1 ) (W /m 2 /µ m /s r )
Spatial resolution of
3 4 .5
0 4 /1 3 /2 0 0 1
the sensor is
important as
3 4 .4
10
compared with
3 4 .3
9 spatial variability of
8
the measured
parameter, because
3 4 .2 7
3 4 .1 5 the value measured

4 within a point may
not be
3 4 .0 3
2
3 3 .9
1
representative of
0 the average
3 3 .8
parameter within
3 3 .7
-1 1 9 .8 -1 1 9 .6 -1 1 9 .4 -1 1 9 .2 -1 1 9 .0 - 1 1 8 .8
the whole pixel
measured by the
satellite.
Comparing satellite observations and “sea truth” data we should

keep in mind that the data collected by contact methods can be
not more precise than remotely-sensed data.
In practice, the satellite data and “sea truth” data are nothing but
two data arrays collected by different methods.
5. Image processing
Level of data processing
Level 0 - Raw data received from satellite, in standard binary

form;
Level 1 - Image data in sensor coordinates, containing individual

calibrated channels;
Level 2 - Derived oceanic variable, atmospherically corrected

and geolocated, but presented in sensor coordinates;
Level 3 - Composite images of derived ocean variable resampled

onto standard map base and averaged over a certain
time period (may contain gaps);
Level 4 - Image representing an ocean variable averaged within

each grid cell as a result of data analysis, e.g.,
modeling.
5. Image processing
The “raw” information measured by the sensor onboard the

satellite (raw radiance counts from all bands as well as spacecraft
and instrument telemetry) is transmitted by radio-signal and
received by the ground station.
These data are called “Level 0” data.
Level-1 Data Products
Level-1 products contain all the Level-0 data, appended calibration

and navigation data, and instrument and selected spacecraft
telemetry that are reformatted and also appended.
5. Image processing
The radiances are measured at different wavebands, called

“channels”.
Different channels provides information on different properties of
the Earth’ surface.
One method of analysis is when the images observed at different
wavebands can be combined to result in a “true color image”.
5. Image processing
The radiances are measured at different wavebands, called

“channels”.
Different channels provides information on different properties of
the Earth’ surface.
One method of analysis is when the images observed at different
wavebands can be combined to result in a “true color image”.
5. Image processing
At this MODIS image

of the Mississippi
River delta you can
see clouds, coastline,
river, the zones of
phytoplankton bloom
and pollution in the
coastal ocean, etc.
5. Image processing
True color images are an important source of information about

natural disasters like these wildfires in California in autumn 2003.
5. Image processing
Each pixel of Level-2 data contains geophysical values (e.g., sea

surface temperature, surface chlorophyll concentration, etc.)
estimated from the radiances measured by the satellite
Each Level-2 product is generated from a corresponding Level-1

product.
Level-2 data are derived from the Level-1 raw radiance counts by
applying the sensor calibration, atmospheric corrections, and the
algorithms specific for each kind of geophysical value (e.g., bio-
optical algorithms for water color data, etc.).
5. Image processing
Example of Level 2 data:

MODIS Sea Surface Temperature, 2000 December 6, 17:05
5. Image processing

MODIS Surface Chlorophyll Concentration, 2000 December 6, 17:05
5. Image processing

MODIS Total Suspended Solids , 2000 December 6, 17:05
5. Image processing
Level-3 means geophysical parameters observed during a certain

period and interpolated on a global grid. For The SeaWiFS the
periods of Level 3 data are:
- one day,
- 8 days,
- a calendar month, or
- a calendar year.
For other satellites these periods can be different.
The spatial resolution of the global grid can be:

1 degree (360 x 180 grid);
18 km (2048 x 1024 grid) - MC SST;
9 km (4096 x 2048 grid) - SeaWiFS, Pathfinder SST v.1-4;
4.5 km (8192 x 4086 grid) - MODIS, Pathfinder SST v.5.
5. Image processing
SeaWiFS Level 3 chlorophyll image, 1997 December 8 (daily)
5. Image processing
SeaWiFS Level 3 chlorophyll image, 1997 December 11 – 18 (8-day)
5. Image processing
SeaWiFS Level 3 chlorophyll image, 1997 December 1 – 31 (monthly)
5. Image processing
The satellite data are disseminated via

internet.
Users select the images in online databases

and either download or order data files.
Typical satellite images are very big (e.g.,
one MODIS image is about 250-300 Mb).
To enable the users to have a brief look at

each image before selecting it low-
resolution “browse” images are often
produced. If the area of interest is free from
clouds, the user orders the data file,
downloads it, and works with it using
appropriate software.
5. Image processing
Data format.
Many types of satellite information are stores in Hierarchical Data

Format (HDF).
HDF is a cross-platform file format for storing a wide variety of

scientific data. This public-domain open standard was created by
the National Center for Supercomputing Applications at the
University of Illinois Urbana-Champaign (NCSA).
A typical HDF file might contain a dataset, data table, descriptions

of data, images produced from the data and other related
information. It can be processed using special software, such as
Noesys, MATLAB, IDL, etc.
5. Image processing
All good software packages are commercial.
To understand the basic features of HDF files you can use free
program HDFExplorer from the Internet site
http://www.space-research.org/
1. Select <<Download>>
2. Fill out the form with your name, etc.
3. Click <<Submit>>
4. Store the HDFExplorerSetup.exe file on the hard drive of your
computer
5. Double-click HDFExplorerSetup.exe to install HDFExplorer.
MAC users may instead of HDFExplorer use HDF Viewer from the
web-site
http://www.hdfgroup.org/hdf-java-html/hdfview/index.html
5. Image processing
Download from the site
www.obee.ucla.edu/test/faculty/nezlin
From the section Lecture 3 “Remote Sensing of the Sea”
two example files:
MO36MWN2.sst4.zip and
C1978341012416.L2_BRS.hdf.zip
Uncompress these files using WinZip and open them in

HDFExplorer. On the left you see the content of the file. Clicking
“+” expand the structures.
5. Image processing
MO36MWN2.sst4.zip
contains data on
sea surface
temperature (SST4)
collected by MODIS
Terra satellite.
The dataset sst4_mean contains the data array. Double-click it to see

the content.
Let us analyze the content of the dataset at the example of one grid
node (x=1; y=150).
The grid node with column = 1 and row = 150 contains the value
31706. To understand its meaning double-click on the records
Scale_type, Slope and Intercept.
5. Image processing
You see the following:
Scale_type =
Y=Slope * x + Intercept;
Slope = 0.01
Intercept = -300
31706 * 0.01 – 300 = 17.06
Double-click Units and Name.
Now you see that it is the temperature of the ocean surface

measured in Degrees C.
5. Image processing
Double-click Start Year, Start Day,
End Year, End Day.
You see that the data were

collected from 2000, Julian Day
336 (December 1) to 2000 Julian
Day 344 (December 9).
Double-click
Northernmost Latitude
Southernmost Latitude
Westernmost Longitude
Easternmost Longitude.
You can see that the data array covers the entire Earth surface from
90S (i.e., -90) to 90N and from 180W (i.e., -180) to 180E.
Number of Columns and Number of Lines indicate the grid size:

1024 x 512.
5. Image processing
Right-click <sst4_mean> and select “Discrete Map View”.
5. Image processing
HDFExplorer transforms arrays into images using primitive method with

standard options. Other software can help you make a graphical
representation of the HDF file content.
5. Image processing
Open the file C1978341012416.L2_BRS.hdf and expand the structures
clicking “+”.
5. Image processing
This file contains CZCS chlorophyll
image; the snapshot was obtained
in the western Pacific.
8-bit Raster Image 1 contains the

image of surface chlorophyll
concentration.
8-bit Raster Image 2 indicates the

snapshot location.
Point the mouse cursor at the Raster Image 1, right-click and select
“View”. You see the values from 0 to 255, i.e. bytes.
Point the mouse cursor at the Raster Image 1, right-click and select
“Image View”. You see the graphic representation of the data array
made using the Image Palette.
5. Image processing
The records Start Year, Start Day,
End Year, End Day indicate that the
image was obtained in 1978, Julian
Day 341 (December 7).
Northernmost Latitude,
Southernmost Latitude,
Westernmost Longitude and
Easternmost Longitude indicate
the approximate location of the
image.
Latitude and Longitude structures indicate the exact geographic

locations of selected pixels of the image. These arrays can be used by
the software like MATLAB Mapping Toolbox to produce the image in
any geographical projection you wish.
5. Image processing
Double-click the records
Scaling Equation,
Base,
Slope, and
Intercept.
You see:
Base = 10;
Slope = 0.012;
Intercept = -1.4;
Scaling Equation =
= “Base**((Slope*brs_data) + Intercept) = chlorophyll a”
We can check some values within the range 0-255 and see that
the brs_data = 100 results in 0.630957…,
the brs_data = 200 results in 10.00000…, etc.
Double-click Parameter and Units. You see the description of the data
“chlorophyll a concentration” and “mgm^-3”.
Oceanographic applications of satellite remote
sensing include:
1. Visible wavelength "ocean color" sensors
2. Sea surface temperature from infrared scanning radiometers
3. Passive microwave radiometers
4. Satellite altimetry of sea surface topography
5. Active microwave sensing of sea-surface roughness
These sensors operate in the

visible part of the
electromagnetic spectrum,
measuring electromagnetic
radiation emitted by the sun
and reflected by land and
ocean surface.
The color of the Earth’

surface, especially the
color of the ocean, results
primarily from biological
processes.
Measuring the absorption

and backscattering
characteristics of ocean
surface, we can estimate
the concentrations of
different kinds of matter
suspended in seawater,
including phytoplankton
cells.
2. Sea surface temperature from infrared radiometers
Infrared sensors measure

electromagnetic radiation
within the band 1-30 µm,
emitted by the ocean surface
and resulting from the
temperature of the upper sea
layer.
2. Sea surface temperature from infrared radiometers
The near-infrared and infrared radiation is processed to sea surface

temperature (SST). The most important SST sensors are Advanced
Very High Resolution Radiometer (AVHRR) on NOAA satellites,
MODIS, GOES geostationary satellites, and some others.
3. Passive microwave radiometers
Passive microwave radiometers operate at electromagnetic
wavelengths 1.5–300 mm (i. e., the frequency 1–200 GHz).
Their advantage is the comparatively long wavelength, which is not

sensitive to scattering by the atmosphere or aerosols, haze, dust,
or small water particles in clouds. So, the microwave sensors are
all-weather devices. This principle advantage is countered by the
fact that thermal emission is very weak at these longer
wavelengths. To overcome noise levels a large field of view must
be received; that results in low spatial resolution (25–150 km). So,
these observations are used for studies of heat balance of the
ocean.
The emissivity of the sea at microwave frequencies varies with the

dielectric properties of sea water (including salinity) and the
surface roughness. Hence, the development of this technique in
future can enable the measurements of surface salinity.
4. Satellite altimetry of sea surface topography
Satellite altimeters are radars,

which transmit short pulses toward
the earth beneath them. The return
time of the pulse after reflection at
the earth's surface is measured,
and this yields the height of the
satellite. The most important are
ERS-1, ERS-2, TOPEX/Poseidon,
and Jason-1 satellites.
Synthetic Aperture Radar (SAR)
Synthetic aperture
radar (SAR) is based
on the comprehensive
analysis of
contribution from
individual points to the
signal received when
the sensor is at a
particular point. The
result is very high
resolution.
Synthetic Aperture Radar (SAR)
SAR images
enable the analysis
of small-scale and
mesoscale eddies,
river plumes,
oil slicks,
ice packs, etc.

184 - The Basics of Satellite Oceanography. 3. Remote Sensing of The Sea

Uploaded by

Document Information

Original Description:

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

184 - The Basics of Satellite Oceanography. 3. Remote Sensing of The Sea

Uploaded by

Copyright:

Available Formats

Lecture 3

Remote Sensing of the Sea

4. Oceanographic sampling for "sea truth"

6. Oceanographic applications of satellite remote

Remote sensing is better than traditional methods:

1. Synoptic view, because satellites collect huge amount of

Problems of remote sensing:

1. The parameters measured by the satellites cannot be directly

Each oceanographic equipment should be calibrated both before

1. The stress of launch;

2. High vacuum of outer space;

3. The power limitations on board the satellite, often resulting

4. No opportunity of retrieving the instrument for periodic

On some scanners, part of the

Some sensors use the moon as a natural

The sensors look at the ocean surface through another medium,

The following compounds of

• Gas molecules themselves

~40% of sunlight is reflected by clouds

The ocean area within the IFOV

Rays 4, 5, and 6 reach the

The sensor receives rays

The complete atmospheric

In the case of optical sensors

On the contrast, in thermal IR

The main strategies of atmospheric correction:

2. A universal atmospheric correction based on an average

Without atmospheric correction, each

scene can be calibrated with ground

data, but the slope of correlation for 10

each scene is unique.

Increased atmospheric pathlength resulting from oblique

Cloud cover is a main obstacle

In both ocean color and SST, the first step of the

The SeaWiFS (8 optical

It is assumed, that the water-

In AVHRR and MODIS the cloud detection is based

Three kinds of tests are used:

Positional registration means the identification on a map of the

AVHRR radiometer on NOAA satellite

Positional registration means the identification on a map of the

AVHRR radiometer on NOAA satellite

Most sophisticated method of position registration is used in

When the sensor is moving

When the sensor is moving

DORIS system determines the position of TOPEX/Poseidon

The technique used (known as orbit

The main problem is that in general

In other case (SST or water color

3 4 .1 5 the value measured

Comparing satellite observations and “sea truth” data we should

Level 0 - Raw data received from satellite, in standard binary

Level 1 - Image data in sensor coordinates, containing individual

Level 2 - Derived oceanic variable, atmospherically corrected

Level 3 - Composite images of derived ocean variable resampled

Level 4 - Image representing an ocean variable averaged within

The “raw” information measured by the sensor onboard the

These data are called “Level 0” data.

Level-1 Data Products

Level-1 products contain all the Level-0 data, appended calibration

The radiances are measured at different wavebands, called

The radiances are measured at different wavebands, called