Remote Sensing Tutorial Examination - Lecture Material - Completely Remote Sensing tutorial, GPS, and GIS -
Remote Sensing Tutorial Examination

This is your first of two major interpretive Exams in this Tutorial, although subsequent Sections will have a fair number of questions of their own. What is different from earlier questions is that most here will be associated with images and maps included in this special subsection, which was drawn from the writer's (NMS) 1982 Landsat Tutorial Workbook. They all relate to a single regional scene that includes much of south-central Pennsylvania and in several images small parts of Delaware and Maryland. Most images are made by the Landsat MSS (remember, Band 4 = green part of the spectrum; Band 5 = red part; and 6 and 7 = Near Infrared). Two thermal and two radar images, made by other sensor systems, are used as well. Some of the image products are computer-derived outputs such as spatial filtered images and classifications.

We begin by showing first a satellite view of the entire state, made by the AVHRR sensor on one of the meteorological satellites. For the fun of it, locate Pittsburgh and Philadelphia (they appear in a lighter blue tone).

AVHRR image of the state of Pennsylvania.

Then, again to set the region we will be working in into context, we will look at much of the eastern half of the state in terms of its topography, using a color rendition of elevations (red is higher) made from the national DEM (Digital Elevation Model) data. Fit subsequent scenes into this as a reference; also fit it into the above AVHRR image.

DEM elevation map of part of Pennsylvania

Now, we start with an MSS 5 image, reproduced on your screen at a nearly 1:1,000,000 scale - the standard size in images distributed by NASA and other agencies and sold commercially. Below it is the corresponding MSS 7 image. And beneath that is a false color composite of the same scene. All were acquired on October 11, 1972, a day after a major storm had hit much of the northeast as far south as southern Virginia. After the sky cleared, the air was especially crisp and dry, cutting down the scene-degrading effects of moisture.

MSS Band 5 image, Oct. 11, 1972, of south-central Pennsylvania, including Harrisburg.
MSS Band 7 image of the same area in Pennsylvania.
False color composite of the October, 1972 Harrisburg Landsat scene.

1-1: Using all three images, but especially the Band 7 image, with the help of an atlas or AAA Road Map, locate the following towns and other features: Harrisburg, York, Gettysburg, Hanover, Lancaster, Hershey, Lebanon, Reading, Allentown, Easton, Wilmington, DE, the Susquehanna River, the Delaware River; the Juniata River. Also, look hard for the small (15000+) town, Bloomsburg, PA ,which is the current home of the writer (NMS), where this question is being written. ANSWER

1-2: What other towns, rivers, or natural features and points of interest in this scene can you list?ANSWER

1-3: What band(s) seems best for picking out the towns and rivers in each of the images? What is MSS 5 especially good at bringing to your attention. ANSWER

To give you a feel for the typical appearance of the mountain ridges portion of the image, glance at this ground photo taken about 15 km (10 miles) north of the Blue Mountain - the first major ridge encountered going north from the Lebanon Valley. The area photographed, on a February 1999 day when the trees are bare, is along Interstate 81 and lies near the center of the Landsat image.

View from Interstate 81 near Pine Grove, PA of a ridge in the Folded Appalachians.

Now, look at the black and white print of the geological map below that covers most of this scene, plus some surrounding areas.

View from Interstate 81 near Pine Grove, PA of a ridge in the Folded Appalachians.

1-4: Fit the major rock unit patterns to the Band 5 image. Comment on the fit. ANSWER

1-5: Comment on what may be unusual about (the color in) the false color composite. ANSWER

1-6: You may have noted that there is a strip of notation at the bottom of both the individual band images and the false color composite. This is on nearly all renditions of Landsat images regardless of the printed size. We will reproduce this strip, enlarged a bit, just below. Make an educated guess as to what the different groupings of letters/numbers mean. ANSWER

The notation located at the bottom of a Landsat MSS image.

Next, familiarize yourself with this October 23, 1975 false color composite of the same area.

False color composite of the October 23, 1975 Landsat MSS scene that includes Harrisburg.

1-7: What overall is different and why? Account for the blue in the Susquehanna River. ANSWER

1-8: Now examine the full MSS false color image of this region as seen on June 8, 1977. Account for the large areas in blue. Take note of the dark grayish-black patches on some of the ridges; there are also some blackish areas in the valleys between ridges, primarily in the upper third of the image. We will take note of these special features later in this exam. Note the course of the Susquehanna River in the lower third of the image. At several places the river widens noticeably. Can you explain this. ANSWER

June 8, 1975 false color composite of the Landsat MSS scene containing Harrisburg.

1-9: Below is a scene taken on July 14, 1977. Describe the principal difference between this and the June 8th image. Look also in the vicinity of Harrisburg at the Susquehanna River. It is marked by a blue color on its left side; why? ANSWER

Same area of Pennsylvania around Harrisburg, imaged as a false color composite of a July 14, 1977 overpass.

1-10: A Landsat 2 MSS Band 5 scene, taken on January 23, 1979, is reproduced below. Comment on differences you note in this image from those we've examined before. How might one distinguish snow from clouds? ANSWER

Winter view of the Harrisburg scene, Landsat MSS Band 5, January 23, 1979.

At this point, we will zero in on a single localized region, around Harrisburg, to get a better feel for the detail that can be picked out in Landsat MSS imagery (would be much sharper for Landsat TM images with better than 2 times higher resolution but a TM image was not available to the writer).

But first, to get up close and personal, look at this ground scene taken from the I-83 bridge looking northeast at downtown Harrisburg. The bridge carries a railroad track across the Susquehanna River. At the far left skyline is a part of Blue Mountain and the water gap through which the river flows. Find the dome of the State Capitol building.

Downtown Harrisburg, viewed from the Interstate-83 bridge across the Susquehanna River.

We start our survey from above with a false color aerial photo (scale 1:142,000) taken from an altitude of 50000 feet by a camera mounted on an RB-57's lower bay, on February 5, 1974.

Harrisburg and surrounding towns and countryside, in a false color aerial photo.

1-11: Peruse this image. Locate Harrisburg first, then using an Atlas pinpoint other towns, major roadways, and the meandering (curving) river. Mention anything else you find interesting. ANSWER

1-12:Knowing that the standard Landsat full scene (7.25 inches wide at its base, representing 115 miles across) has a scale of 1:1,000,000, calculate the width in miles of this aerial photo scene. ANSWER

Just below is a subscene from the October 11, 1972 Landsat MSS false color composite which was enlarged to about the same scale as the RB-57 scene above. It was made more blue to look more like the aerial photo.

Subscene made from Oct. 1972 Landsat MSS, with boundaries and area covered similar to the high altitude aerial photo.

1-13: Compare this subscene with the RB-57 photo, noting what can be found in both and what is not "seeable" in the Landsat version. ANSWER

Still another Harrisburg subscene is this July 1977 false color composite.

A July 1977 false color subscene of the Harrisburg area.

1-14: Comment on anything unusual or different in the above image. ANSWER

We will now look closer at details present in enlargements, concentrating on the area around the Chesapeake & Delaware Canal, south of Elkton, MD. (This is at the lower right corner of the October 1972 scene.) The first of two images below is a 512 x 512 pixel subscene from Band 5 of the July 1977 image set. Beneath it is an aerial photo, at a scale of 1:60,000 that contains farmlands included in the Landsat image; this was taken in May of the same year.

Area near Elkton, MD extracted from the 1972 MSS Band 5 scene that includes Harrisburg to the northwest.
Aerial photo (1:60,000) which includes part of the above Elkton, MD area subscene.

1-15: Can you locate the aerial photo in the Landsat subscene? Watch out - the field patterns have changed between the two dates for an obvious reason. ANSWER

We are ready to change directions to concentrate on a prime use of imagery of photo-like nature: change detection. We are going to utilize several subscenes that have been cut from the upper half of the full images we have already encountered. Consider first this June 8, 1977 Band 7 image.

Part of a Band 7 (Near IR) image taken on June 8, 1977 that covers much of the upper third of the Harrisburg scene.

1-16: One thing that is obvious: widespread patches of almost black features, found almost entirely in valleys between ridges. Do you have any idea what these represent? Hint: they are related to a geological material that has immense commercial value; this material is in beds in rocks of Pennsylvanian age. ANSWER

1-17: There are dark patterns, one elongate, the other as three spots, near the tips of arrows G and H. What do you think these are? Can you find them in corresponding local areas in the October 1972 MSS Band 7 image? What's going on? ANSWER

Now, inspect the scene below which is a July 8, 1973 MSS 7 subscene of the same area just looked at.

Essentially the same area covered in the previous image, this time another Band 7 view obtained in July of 1973.

1-18: What's new and different? Identify the features at arrow tips of the letters indicated (the rightmost arrow is labelled "I"). Features such as at "K" are a new phenomenon, in which the ground is responsible for the reflectance that gives rise to a medium gray tone (that's a clue!) ANSWER

Pennsylvania was hard hit by the phenomenon at "K" again in 1977. The next illustration is a set of six subscene panels taken from an area north of Williamsport, PA (just outside the full scenes we have been looking at; here that town is the stretched out blue pattern near the top). For now, don't read the black print between the two panel rows.

Subsets of the Williamsport, PA area consisting of Landsat false color composites acquired on the dates indicated in 1977.

1-19: Since you now know what "K" is from the previous question, comment on the progression of the feature that appears dark brownish-gray in these panels. ANSWER

We will now transition into a second phase of questioning in which you will be asked to analyze and comment on some computer-generated imagery produced in this study of south-central Pennsylvania.

First, is a panel of four subscenes of an area that includes Hanover, PA (at the upper left edge, west of the reservoir [thin black pattern]) about 20 miles south-southwest of York, PA. Passing diagonally across each subscene is a series of nearly parallel straight lines. These are also visible in most of the full scenes. They are low ridges carved from more resistant units in the Precambrian metamorphic complex that makes up this part of the Piedmont. Here is the scene:

Landsat black and white images of a subscene near York, PA, processed as described in the next paragraph

The upper left image is a MSS Band 7 rendition (stretched) of a June 1978 scene; the lower right is a linear stretch of a Band 7 of a February 1978 scene in which the ground was covered with snow. The upper right consists of a high bandpass filtering by using a Fast Fourier transform algorithm applied to the June subscene whereas the lower left is also a high bandpass filter version of Band 5, with Cubic Convolution resampling.

1-20: Which of the four subscenes seems best/worst for picking out linear features? Compare the Fast Fourier and Cubic Convolution processing for enhancing linear features and sharpening details. Make a guess as to the identity (nature) of the irregular feature (in dark gray to black) that lies just south of the reservoir lake.ANSWER

Now, back to Harrisburg and Gypsy Moth defoliation. Look at this June 27, 1977 subscene. The defoliation appears as very dark gray patches on the ridges.

False color subset from the June 27, 1977 Landsat MSS image here showing an area of Blue Mountain near Harrisburg.

There are several ways to focus on this defoliation while ignoring other features/classes in the image. One is to calculate the Vegetation Index for the image. At least six different formulae have been proposed for this VI. We will use the simplest, namely, a ratio of MSS7/MSS5. When this is done there will be a range of DNs that could be shown as gray levels. But, another way is to produce a color mask image (a variant of density slicing). All non-vegetated features are set to a single DN value (masking them out) and vegetated areas with signatures characteristic of trees and forest are assigned different colors. This is the result, with masked out areas shown in yellow-gray, healthy forest in green, defoliated forest in red, and stressed vegetation in purple.

Colorized Ratio 7/5 image showing healthy forest as green, gypsy moth defoliation in red, and stressed trees in purple.

1-21: What is the advantage of this VI image? Where may confusion occur? ANSWER

This same scene (and date) has been subjected to Principal Components Analysis. The first four component images are shown below (but, in an error in layout, the first component is B and the second, A, with C and D the third and fourth components respectively.

Black and white renditions of four of the Principal Component images derived from the June 27, 1977 Landsat MSS bands; see text for PC identification.

1-22: Describe the main gray patterns in each component image. ANSWER

A color composite can be made from any three of these images. Generally, the fourth component image is bland and blurry and is not used. Here below is a false color image made using PC 2 = red; PC 1 = green; PC 3 = blue.

Color composite of the June 1977 scene assigning PC2, PC1, and PC3 to Red, Green, and Blue respectively.

1-23: What does this PC image show that is less obvious in the standard false color composite shown above?ANSWER

We are ready now to embark on Supervised Classifications of 4 areas within the original October, 1972 full MSS scene that we have examined at the beginning of the test. We'll start with this 8 class map of the Harrisburg subscene. This, and subsequent classifications were all done on the IDIMS computer system at NASA Goddard.

Supervised Classification of the  Harrisburg October 1972 Landsat image; here subseted to show the areas near that city.

1-24: Evaluate this classification. ANSWER

Onward to the anthracite coal belt. We present first a standard false color composite of the subscene that shows folded mountain ridges from Hazleton (T) southeast from Blue Mountain (lowest N) out into the agricultural lands (A) of the Great Valley itself. Most of the brighter blue is mine waste. Defoliation was moderate in this July 8, 1973 scene, being represented by the blue-gray tones.

False color composite of a July, 1973 Landsat MSS data subset which includes Hazleton, PA southward into the Lebanon Valley

A 1:78,000 scale black and white aerial photograph covers part of this Landsat subscene.

Aerial photo of the area near Hazleton.

1-25:Try to fit this aerial photo into the Landsat scene. Name the towns (use an Atlas) you find. Is the mine refuse well displayed in the photo. ANSWER

Let's look ahead to Section 16, in which products from the spacecraft Terra are introduced. One sensor, MISR, has yielded this near-natural color image of the Pennsylvania scene plus the New York-New Jersey areas to the east.

MISR image of New Jersey and eastern Pennsylvania.

1-26:Is there anything special about the areas in the coal country we have been looking at? Why is this speciality ambiguous?ANSWER

An Unsupervised Classification of the July 8th 1973 subscene near Hazleton was made on a General Electric Image-100 interactive processing system. Only four cluster classes were specified. After they were blocked out as patterns, they were then named, using the writer's general knowledge of the region. Here it is:

Unsupervised classification of the land near Hazleton, PA.

1-27: Evaluate this classification, compared with your expectations from the fcc image for this date. Use the lakes (shown in black) to relate the class map to the image. ANSWER

The next two classifications treat the area around the Chesapeake and Delaware Canal and Chesapeake Bay headwaters that occupy the lower right corner of the October, 1972 image.

The first of these is a Supervised Classification (right) of the subscene extracted from the July 1977 full scene shown below on the left as a false color image. The classes were chosen by the writer following a field trip to the area and discussions with the local Soil Conservation Service representative for New Castle County, DE.

False color rendition of the Upper Chesapeake Bay (left) and a Supervised Classification of this subscene (right).

1-28: Which seems more abundant, corn or soybeans? If you consult a map of the area, you might conclude that most of the larger lavender patterns associate with towns and villages. But what about smaller clusters of lavender; what might they be? Give a reason why so many of the fields do not have straight boundaries, as is typical, nor are rectangular. ANSWER

The last classification shows categories of water purity in the Chesapeake Bay headwaters in northern Maryland around Havre de Gras and Aberdeen. The date of image acquisition was May 28, 1978. This is the result:

Sediment in water of the Chesapeake Bay near its northern end, at levels of turbidity determined by a classification using field data collected from a boat.

The mouth of the Susquehanna River is at center top. The Chesapeake Bay itself is the drowned segment of the lower Susquehanna which once emptied in Virginia near Norfolk but has been inundated by rising sealevel after the last glacial retreat. The land here is rendered in one color (except for white wetlands) by masking, which involves assigning all land pixels to a single value shown as black.

1-29: How do you think the turbidity in the Bush River was determined (assessed)? ANSWER

We are nearing the end of this "exam". Now we turn our attention to two other types of remote sensors - those that produce active microwave - radar - images and those that measure thermal properties . These sensor systems are treated in Sections 8 and 9 respectively. The following is just a preview.

For thermal imagery covering the Pennsylvania region and some neighboring states, we will turn to the satellite known as HCMM (Heat Capacity Mapping Mission; see Section 9). This operated during 1978 and produced scenes 700 km wide in three modes: Day Visible; Day (thermal) IR; and Night IR.

Look first at a Day IR image (September 26, 1978) in which the bottom half includes much of the region around Harrisburg that we have been studying. Dark tones correspond to "cool"; light to "warm".

HCMM daytime thermal image of part of Pennsylvania into surrounding states; September 26, 1978.

1-30:First locate yourself within this image relative to the landmarks you have come to know. Why are the ridges cool (dark) and the valleys warm (light-toned)? There is a long, canoe-shaped white area in the right center; why does it appear to be so hot? While it is not in the region you have been studying, speculate on the very dark patterns in the upper center. ANSWER

Then, look at this Night IR image, obtained on June 11, 1978.

HCMM nightime thermal image of part of the East Coast including some of Pennsylvania; June 11, 1978.

1-31: Why do the ridges and rivers and other water bodies appear to be warmer (lighter toned) than the valleys? Can you find the Harrisburg area? ANSWER

Finally, we take just a momentary peak at two radar images that include the Harrisburg area. Because interpretation of radar imagery requires considerable understanding of how the sensor operates and the images are formed (which we will learn about in detail in Section 8), we will only show you the two examples. We suggest you locate these images in the framework of the Landsat scenes and pick out any special features and landmarks of which you are already familiar. The first image is made from an aircraft-mounted SLAR system (K-Band), as it flew over the terrain on July 22, 1966. The second was taken from space by the Seasat SAR (L-Band) on September 28, 1978. In that image, the location of Harrisburg should be obvious (its buildings are strong reflectors).

K-Band SLAR (aircraft) radar image of the area just west of Harrisburg.
L-Band Seasat SAR radar image of an area that includes Harrisburg (white patch).

1-32: Compare the aircraft SLAR image with the Seasat image. Make note of any obvious differences between the two and suggest reason(s) why. ANSWER

Now, we will close by asking you to look at these two images that contain Harrisburg.

Landsat Harrisburg A.
Landsat Harrisburg B.

1-33: We challenge you to determine what evidence you note that indicates the time of the year when each of these Harrisburg subscenes was imaged. ANSWER

Well, if you tried the "exam" and have clicked back to the bottom, we expect you found it rather challenging, but hopefully not tedious. Grade yourself, but by merely working through it, you deserve an "A". In the rest of the Tutorial we will not again face such an extensive sequence of questions until the Final Exam (Section 21) - but in nearly all Sections in between there will be just enough queries to help you become better acquainted with the important concepts and features associated with remote sensing interpretations.