Image interpretation quizz - Table of contents
Test your interpretation skills!
Question 1
Question
This Landsat Thematic Mapper image segment has a
band => colour assignment of:
- R,G,B => 1,2,3
- R,G,B => 2,3,1
- R,G,B => 5,4,3
Choose the band-to-colour assignment (A, B or C, above) that fits this Landsat TM satellite image colour composite.
Did You Know?
Notice the very rough and convoluted shoreline in the pink areas in the top half of the image and how it contrasts with the relatively smooth shoreline in the green areas in the bottom part of the image. This is a sure indication that the rocks of the Canadian shield are exposed in the pink areas, while there is substantial overburden (soil) in the green areas.
Image Information
eastern end of Great Slave Lake, Landsat 5 TM
-
Answer to question 1
This Landsat Thematic Mapper image segment has a
band => colour assignment of:- R,G,B => 1,2,3
- R,G,B => 2,3,1
- R,G,B => 5,4,3
The answer is C.
A. No.
Band 1, which is the shortest wavelength band of the TM has the best water penetration of these three bands and would show sediment best. If assigned as red (as in this choice), it would depict water in red tones.
B. No.
Band 2, which is green-sensitive, would depict (green) vegetation in its assigned colour of red. But we see that the lowland vegetation (bottom portion of image) is shown as green.
C. Yes.
Of these three bands the lowest wavelength band (3), which has the best volume reflection from water assigned as blue, indeed shows water as blue. Green vegetation is most reflective (of these three bands) in band 4, and with its green assignment does show vegetation as green.
Question 2
Question
The diagonal white line at the top right of the image is:
- paved road
- contrail
- hydro line
- gravel road
It takes careful observation and logical reasoning to figure this one out.
Image Information
eastern end of Great Slave Lake, Landsat 5 TM
-
Answer to question 2
The diagonal white line at the top right of the image is:
- paved road
- contrail
- hydro line
- gravel road
The answer is B.
A. No.
A road wouldn't be as straight as this line, nor would it cross water bodies without a bridge!
B. Yes.
An aircraft condensation trail is not limited to land or water. The parallel contrail shadow just above it confirms this feature and also gives a mechanism for calculating the altitude of the contrail (the date and time of the image identifies the sun angle and the horizontal offset of contrail and shadow can be measured from the image).
C. No.
Usually a hydro line is visible because the right-of-way has been cleared of vegetation. But it would not be visible while crossing water bodies, as it is here.
D. No.
The same reasoning goes here as for a paved road. A road wouldn't be as straight as this line, nor would it cross water bodies without a bridge!
Did You Know?
By knowing the time that the satellite image was recorded, we can determine the sun angle (above the horizon). With that parameter, as well as the horizontal distance on th image between an object (such as the contrail) and its shadow, we can calculate the height of that object above the Earth's surface. This is the relationship:
h = height of the object above the Earth's surface
d = horizontal distance between object and shadow
θs = angle of the sun above the horizonh = d * tan θs
Question 3
Question
The areas of water in this image are mostly a speckled grey tone, but with large sections of black. These black areas are:
- Oil slicks on the water surface.
- Shadows from the hills on the islands.
- Wind sheltering by the land areas.
Choose A,B or C
Did You Know?
SEASAT was launched on June 26, 1978. A massive short circuit in the satellite electrical system on October 10, 1978 ended the mission. During its lifetime, fourteen Earth orbits were completed each day at an altitude of 800 km. Data was received at the CCRS receiving stations from July 9, 1978 to October 9, 1978.
Image Information
Strait of Georgia, SEASAT / SAR
-
Answer to question 3
The areas of water in this image are mostly a speckled grey tone, but with large sections of black. These black areas are:
- Oil slicks on the water surface.
- Shadows from the hills on the islands.
- Wind sheltering by the land areas.
The answer is C.
A. No.
It would take an extraordinary amount of oil and an incredible current and wave action to distribute oil among all those islands in the pattern seen here.
B. No.
There is no evidence of very tall hills on the islands. The incidence angle from the satellite would not be blocked to that extent by even very tall hills or mountains. The shape of the black water areas do not conform to prospective hills on the islands.
C. Yes.
The brighter areas of water are due to wind-generated waves which cause a brighter radar reflection. Where the land shelters the water surface from wind, the lesser wind speed reduces the magnitude of the waves and the water surface acts like a mirror, reflecting the radar beam in a specular manner - away from the radar antenna. This appears as areas of little or no radar return on the image: a black tone.
Question 4
Question
The radar beam that illuminated this scene came from the:
- Top
- Bottom
- Right
- Left
Choose A, B, C or D
Hint
The sides of hills that face towards the incoming radar beam are more illuminated than the sides facing away.
Image Information
Strait of Georgia, SEASAT / SAR
-
Answer to question 4
The areas of water in this image are mostly a speckled grey tone, but with large sections of black. These black areas are:
- Top
- Bottom
- Right
- Left
The answer is D.
A. No.
You're just guessing, aren't you?
B. No.
There is absolutely no evidence for this!
C. No.
Another wild guess!
D. Yes.
Evidence for this comes from a number of image features. At the top, as the hills are illuminated by the radar, they show a brighter reflection from the radar-facing side which are the left side slopes. At the bottom right are two long piers jutting out into the water. These piers show a radar shadow to their right, which indicates radar illumination from the left.
Did You Know?
- The linear structures seen here are called internal waves.
- Ferries on their way to and from the islands can be seen on the water.
- These piers are the ferry (left) and bulk cargo (right) terminals at Tsawwassen, BC.
Question 5
Question
The features in this radar image are:
- Hills with no apparent vegetation on them.
- Convection cells expressed on the ocean surface.
- Lines of sand dunes in a very arid part of the world.
- Glaciers in the northern part of Antarctica.
Image Information
Labrador Sea RADARSAT-1 / SAR
-
Answer to question 5
The features in this radar image are:
- Hills with no apparent vegetation on them.
- Convection cells expressed on the ocean surface.
- Lines of sand dunes in a very arid part of the world.
- Glaciers in the northern part of Antarctica.
The answer is B.
A. No.
The sharp outlines to the features do not correspond to known hill shapes.
B. Yes.
This large area is experiencing extensive "convective instability", caused by the air-sea temperature difference. There are sharp changes where the airmass is rising or falling, and such changes are repeated frequently over a few hundreds of square kilometres. Regions of rising and falling air modulate the surface wind speed, which influences the water surface roughness which in turn modulates the radar backscatter. The mottled, checkerboard pattern of light and dark areas indicates just how changeable the air movement is in this area on this occasion.
C. No.
Sand dunes would be more similar in shape and size, instead of the wide variety seen here.
D. No.
Just another wild guess!
Canada's RADARSAT is particulary adept at imaging atmospheric events which "imprint" themselves on the ocean surface. A good knowledge of atmospheric physics and satellite microwave techniques can lead the interpreter to deducing a great variety of such phenomena with unprecedented detail.
Did You Know?
Near-surface wind speed over the ocean can be measured from calibrated radar images. The higher the wind speed, the more roughness to the ocean surface and therefore the greater the radar backscatter.
Question 6
Question
Find the one bridge in this image segment
To find the bridge, you first have to recognize which patterns are land and which are water!
Where does the land begin and the water end? In most satellite images, the land-water boundary is quite distinct. Here, the shoreline can be quite obvious, if you realize that the water is very turbid. The high sediment load makes the surface water a stronger than usual reflector (the applied enhancement exacerbates this). Use the edge of the green vegetation to help you find the shoreline.
Image Information
North West River, Labador
LANDSAT TM
-
Answer to question 6
Did You Know?
There are interesting patterns on both the land and water portions of this image. The land-based patterns (bottom left) are due to wetland vegetation types and distribution. The water-based swirling patterns are due to the confluence of three large rivers, and show up because of the very high suspended sediment loads carried by these rivers.
Question 7
Match the band combinations with their respective images.
Landsat TM Band Combinations
- 3, 2, 1
- 3, 4, 5
- 4, 3, 2
- 5, 4, 3
When looking at image "C": the dark brown areas in the middle of the image are forest fire scars, the white areas at the bottom of the image are snow-covered mountain tops, the light blue line across the middle of the image is a river. There is a diagonal line of small clouds (and cloud shadows) across the top right of the image. The small black areas are lakes.
Image Information
Mackenzie River, N.W.T.
LANDSAT TM
Hints
3, 2, 1
This combination is the only one that includes band 1 with its blue sensitivity. Therefore, this combination of bands and colour assignments is the closest that we can come to making a real-colour simulation.
TM band 1 is the band most affected by atmospheric scattering, so it shows clouds, haze and smoke the best.
3,4,5
The red-sensitive band (#3) that has a little water penetration ability is coded as red in this band combination. The other two bands have no significant water penetration ability, so any sediment-laden water body should appear as red.
4,3,2
This is the simulation of the traditional of false-colour infrared film which shows healthy vegetation in shades of red./
5,4,3
The red-sensitive band (#3) which has a little water penetration ability is coded as blue in this band combination. The other two bands have no significant water penetration ability, so any sediment-laden water body should appear as blue.
Landsat 5 Thematic Mapper spectral band definitions:
Band 1: 0.45-0.52 micrometres (blue-green)
Band 2: 0.52-0.60 micrometres (green)
Band 3: 0.63-0.69 micrometres (red)
Band 4: 0.76-0.90 micrometres (near infrared)
Band 5: 1.55-1.75 micrometres (near infrared)
Band 6: 10.4-12.5 micrometres (thermal infrared)
Band 7: 2.08-2.35 micrometres (shortwave infrared)
-
Answer to question 7
A - 3, 4, 5
B - 3, 2, 1
C - 4, 3, 2
D - 5, 4, 3
text version
Satellite images: Mackenzie River, N.W.T., LANDSAT TM
All four illustrations represent a Landsat-TM image of the Mackenzie River, NWT. The user is asked to identify which band combination is represented on each of the four colour images.
- Image A: The image has a predominant green component with red tones in the water bodies due to sediment.
- Image B: The color combination is very close to the natural colors.
- Image C: The image has a predominant red component notably in vegetated areas.
- Image D: The image has a predominant green component with some blue tones in the water bodies due to sediment.
Did You Know?
Each of these images (A, B, C, D) is made from the same TM data set. Only the band assignments differ.
Question:
What is the number of possibilities for selecting three different bands of the Landsat 5 Thematic Mapper's 7 available bands and coding them as the three primary colours?Answer:
7 x 6 x 5 = 210 different combinations!
Question 8
Which satellite image (A, B, C or D) corresponds to the oblique airphoto?
It is very different to observe the Earth's surface from directly above or from an oblique angle. This difference is made even stranger when the two views are at substantially different scales. Try to wrap your mind around this difference by matching small-scale vertical satellite images to large-scale oblique airphotos.
Note: The red arrow in the satellite image shows the location and direction that the airphoto was taken.
Airphoto 1
Satellite Images
Hint for Scene 1
Don't let the three striped smokestacks distract you. From the satellite's vantage point directly above, they are inconsequential. Look for the shoreline features, such as the unusually-shaped pier.
Airphoto 2
Satellite Images
Hint for Scene 2
The golf course fairways make for unique shapes but also consider the adjoining street structure and nearby surface water.
Airphoto 3
Satellite Images
Hint for Scene 3
You have streets, forest and lakes combining to make unique shapes in this image.
Image Information
Halifax harbour and vicinity, Nova Scotia
LANDSAT TM and airphoto
-
Answer to question 8
Airphoto 1 AirphotoSatellite Images A Satellite image: LANDSAT TMB Satellite image: LANDSAT TMC - correct image Satellite image: LANDSAT TMD Satellite image: LANDSAT TMHint for Scene 1
Don't let the three striped smokestacks distract you. From the satellite's vantage point directly above, they are inconsequential. Look for the shoreline features, such as the unusually-shaped pier.
Airphoto 2 AirphotoSatellite Images A Satellite image: LANDSAT TMB Satellite image: LANDSAT TMC Satellite image: LANDSAT TMD - correct image Satellite image: LANDSAT TMHint for Scene 2
The golf course fairways make for unique shapes but also consider the adjoining street structure and nearby surface water.
Scene 3
Airphoto 3 AirphotoSatellite Images A Satellite image: LANDSAT TMB - correct image Satellite image: LANDSAT TMC Satellite image: LANDSAT TMD Satellite image: LANDSAT TMHint for Scene 3
You have streets, forest and lakes combining to make unique shapes in this image.
Did You Know?
Two slightly different angles of view of the same object, from two airphotos or satellite images, mimics the pair of views that we get from our eyes, and therefore offers the possibility of stereo (3-dimensional) viewing!
Question 9
Winnipeg, Manitoba
Below is a LANDSAT image of Winnipeg. In it much can be learned about the city. There are three rivers, large medium and small. Grass is light green and trees are dark green. Hard areas (non vegetated) are pink to purple depending on the nature of the material and shadowing. Water is black.
Find examples of the following features in this LANDSAT TM image of Winnipeg.
Airport: The criss-crossing, linear shapes should make this easily identifiable.
River: Its dark in tone and sinuous in shape. Another easily identified target
Business district: The colouring identifies it as non-vegetated. Note the small areas of shadow.
Rail yards: Very dark-toned, unvegetated, linear in shape but quite wide.
Ponds: Small, rounded shapes. Very dark in tone (so not vegetated).
Residential: An organized pattern of mostly thin, linear shapes close together.
Industrial: A jumble of small rectangular shapes of varying tones, but all un-vegetated.
Rail road corridor: Long, thin, linear shapes which cut through neighbourhoods at odd angles. And they point to railroad yards.
Farm fields: Large rectangular shapes; some of them covered with uniform vegetation.
Park / forest: From the tone and hues, this is obviously vegetated, but the irregular, mottled shapes indicate that it is natural, rather than man-made.
Golf course: The green and yellow tones indicate lush vegetation of some sort. The linear shapes within suggest that its designed by humans rather than by nature.
Tank farm: In a non-vegetated setting, the white dots seem to be arranged in a regular pattern.
Image Information
Winnipeg, Manitoba
LANDSAT TM (bands 5, 4, 3 = RGB)
-
Answer to question 9
Did You Know?
To verify the interpretation of a satellite image, it is common to inspect selected sites and targets on the ground. This is often called "ground truthing". Yet, when the target is distributed over a large or inaccessible area, then ground truthing is more conveniently performed by aerial inspection or aerial photography! Interpretation of small-scale imagery such as that from the NOAA AVHRR sensor has been known to be ground truthed using the larger scaled Landsat imagery! Perhaps the term "ground truthing" needs to evolve to something more appropriate.
Question 10
A histogram is a graphical way of representing statistics related to image brightness. A one dimensional histogram, like the ones shown here, depict the amount of image area (vertical axis) that corresponds to particular brightness values (horizontal axis). Analysis of the histogram of an image allows the interpreter to objectively identify the pixel population of the image as classified into brightness levels.
How well do you know your histograms? Match the histogram to each image.
Image 1
Hint for Image 1: Notice that most of the image is bright, with a small amount of mid-grey and very dark areas.
Image 2
Hint for Image 2: There appears to be a full range of greys in this image.
Image 3
Hint for Image 3: There are two distinct shades of mid-grey in this image, each represented by the two large fields. The light-toned road in the bottom of the image has a small relative area and a narrow range of grey levels.
Image 4
Hint for Image 4: The grey levels in this image are all mid-level, with a complete absence of lighter shades.
Image 5
Hint for Image 5: More than half of the area in this image has a narrow range of dark tones representing the water. The land contains a wide range of mid to high tones.
Image 6
Hint for Image 6: Most of the area of this image is taken with a wide range of mid to high tones representing the land. On the water side, a small fraction of the image is covered with a narrow range of dark tones.
Image 7
Hint for Image 7: Most of the area of this image is taken with a wide range of mid to high tones representing the land. On the water side, a small fraction of the image is covered with a narrow range of mid-grey tones.
Did You Know?
A histogram or "frequency distribution" of an image allows the analyst to:
- detect some possible problems in the processing of a digital image,
- plan appropriate enhancements for the visual interpretation of an image,
- plan suitable classification schemes for an image or image portion,
- plan suitable ground truthing in support of image classification,
- prepare for appropriate mosaicking methods of several images,
- and more!
Image Information
single band images
-
Answer to question 10
Image 1
The matching histogram is C
Image 2
The matching histogram is B
Image 3
The matching histogram is D
Image 4
The matching histogram is B
Image 5
The matching histogram is B
Image 6
The matching histogram is A
Image 7
The matching histogram is C
Question 11
Do you know your image enhancements?
Visual interpretation of images often make use of image enhancement to make certain features more prominent and simpler to recognize. But the specific enhancement to be used depends on the feature to be interpreted and the nature of the imagery.
Original Image
Text Version - Original
Original
This illustration shows an image of Truro (N.S.) in grey tones and acquired by LANDSAT-TM. We can see cropland, rivers, forest and urban areas. The image lacks contrast voluntarily.
Which enhanced image below (A, B, C or D) is the result of:
1. contrast stretch enhancement
2. high pass filter enhancement
3. intensity slicing enhancement
Text Version - Item A
Item A
This illustration shows the original image after contrast enhancement. This type of enhancement produces an image that emphasizes areas of fine spatial detail.
Text Version - Item B
Item B
This illustration shows the original image after contrast enhancement. Grey tones are stretched in this type of enhancement; dark areas are even darker, and bright areas brighter.
Text Version - Item C
Item C
This illustration shows the original image after contrast enhancement.
Text Version - Item D
Item D
This illustration shows the original image after contrast enhancement The image is in color.
Hint for Contrast Stretch:
From the CCRS Glossary:
"A process in which the range of brightness levels of an image is expanded to use the full brightness range of the recording film or display device."
The "stretched" image shows dark areas even darker, and bright areas brighter.
Hint for High Pass Filter:
From the CCRS Glossary:
"A type of spatial filter that uses a convolution filter to produce an image that emphasizes areas of fine spatial detail (e.g. edges)."
The enhanced image should look sharper/crisper, with better defined lines.
Hint for Intensity Slicing:
This process involves substantially reducing or grouping the number of discrete grey levels from (say) 256 to 16. The resulting grey levels can be displayed as unique shades of grey or as discrete colours.
Image Information
Truro, Nova Scotia, Landsat TM
-
Answer to question 11
A2. high pass filter enhancementB1. contrast stretch enhancementCD3. intensity slicing enhancementDid You Know?
An enhancement is any process that improves the interpretability of the original image. That means that simply bringing the image closer to your eyes could be an "enhancement", if you are looking for fine detail. Alternately, moving the image further away can make larger image patterns more obvious, so that would be an "enhancement".
Question 12
Can you read a 2-D histogram? Test your ability!
The graphical plots represent two-dimensional histograms or the frequency of occurrence of brightness levels in TM bands 3 and 5 for various sites in the accompanying image.
Match the histograms below (blue letters) to the marked sites on the image.
Did You Know ?
A spectral signature can have as many dimensions as there are individual spectral bands recorded. The more and narrower the bands, the better a feature's "signature" is identified. With modern sensors, it is quite common to have hundreds of spectral bands, each very narrow and contiguous with neighbouring bands. These are "hyperspectral" sensors.
Hints
Site 1: There is a large variation in the green tones and very little in the red, so the histogram should be quite spread out in the vertical axis but not much in the horizontal.
Site 2: Compared to site 1, this site shows a much more limited range of green tones, so the histogram spread will be noticeably less in the vertical direction. The average red values appear to be the same as for site 1.
Site 3: This site shows a very narrow range of dark tones, especially in the green, and therefore one would expect to see a histogram with very limited spread, close to the origin.
Site 4: At this location, the range of brightnesses is large, with a high mean value in both red and green. One would expect to see a large spread in the histogram, with the mean point well away from the origin.
Site 5: The range of brightness for this feature is very narrow, so the histogram will show a very small plot. The average green value is very low, while the red is very high.
Site 6: The green brightness range is very large here, but the red range is not. The mean green value is very high, much more than for the red. As a result, the histogram should be much wider in the vertical than in the horizontal dimension and the central point in the histogram well offset in the vertical direction.
Site 7: There is a good range of green and red tones at this site but the mean value of the red is noticeably higher than for the green. Thus the histogram should be located much closer to the horizontal axis than the vertical.
Image Information
Vancouver, British Columbia, Landsat TM
-
Answer to question 12
Can you read a 2-D histogram? Test your ability!
The graphical plots represent two-dimensional histograms or the frequency of occurrence of brightness levels in TM bands 3 and 5 for various sites in the accompanying image.
Match the histograms below (blue letters) to the marked sites on the image.
A - image section #3
B - image section #2
C - image section #4
D - image section #7
E - image section #5
F - image section #6
G - image section #1
Note: The image is colour-coded as: Thematic Mapper bands 3 5 7 = R G B and the two dimensional histograms are shown as intensity of band 3 along the horizontal axis and of band 5 along the vertical axis.
Did You Know ? A spectral signature can have as many dimensions as there are individual spectral bands recorded. The more and narrower the bands, the better a feature's "signature" is identified. With modern sensors, it is quite common to have hundreds of spectral bands, each very narrow and contiguous with neighbouring bands. These are "hyperspectral" sensors.
Hints (back to image)
Site 1: There is a large variation in the green tones and very little in the red, so the histogram should be quite spread out in the vertical axis but not much in the horizontal.
Site 2: Compared to site 1, this site shows a much more limited range of green tones, so the histogram spread will be noticeably less in the vertical direction. The average red values appear to be the same as for site 1.
Site 3: This site shows a very narrow range of dark tones, especially in the green, and therefore one would expect to see a histogram with very limited spread, close to the origin.
Site 4: At this location, the range of brightnesses is large, with a high mean value in both red and green. One would expect to see a large spread in the histogram, with the mean point well away from the origin.
Site 5: The range of brightness for this feature is very narrow, so the histogram will show a very small plot. The average green value is very low, while the red is very high.
Site 6: The green brightness range is very large here, but the red range is not. The mean green value is very high, much more than for the red. As a result, the histogram should be much wider in the vertical than in the horizontal dimension and the central point in the histogram well offset in the vertical direction.
Site 7: There is a good range of green and red tones at this site but the mean value of the red is noticeably higher than for the green. Thus the histogram should be located much closer to the horizontal axis than the vertical.
Page details
- Date modified: