navigation image map


1-1: The two landmarks that best tie the TM and MSS subscenes together are the river, with its distinctive bends, and an arrowhead-shaped mountain (Catawissa Mountain) south of the river. Approximately 20% of the MSS subscene (and less than 10% of the full scene) coincides with the TM subscene. What TM shows alone is some area north of the river. BACK

1-2: Farming! Farms occupy most of the land in the valleys. The chief crops are corn, soybeans, some wheat, oats, hay (alfalfa and timothy), pumpkins and squash. In the June Landsat TM subscene, most of the barren fields (light buff) are really in crop: corn, the mainstay of the region but being planted late still only about a foot or less high and hence not contributing much to the field signature. Two typical views of the countryside north of Bloomsburg, with a mix of farmlands and woodlands, appear below (these areas host a number of Amish farms).

Farms and wooded hills near Jerseytown, PA

An Amish farm near Lake Chillisquaque.

1-3: Although the image gives little sense of verticality or relief, this is the Catawissa Mountain. Geologically it is a syncline. Its peak elevation is 641 m (1940 ft), making it the highest mountain in the TM subscene we are studying. It rises some 1400 ft above the Susquehanna River and about 900 ft above the flatlands just to its north. It is covered largely by deciduous trees, with scatterings of hemlocks. Here is a view of the mountain from about 10 km (6 miles) to the west:

Catawissa Mountain in the distance; a local commercial orchard in front left.

1-4: You probably selected lumbering, that is, cutting down the trees for wood suited to various uses. But, there is no sign of any extensive clearcutting (see later question). Forest harvesting is a minor use, although locals chop down trees for wood-buring stoves. A few small companies take down certain tree species selectively in the woods to sell for pulp but this does not change the tree density enough to affect the vegetation signature. Lumbering is more active in the Appalachian Plateau to the north of this subscene, up to the New York state border. The major use in the Bloomsburg area is for hunting: this is a significant preoccupation with many of Pennsylvania's citizens - a tradition more than 200 years old. Deer, turkey, and ruffed grouse (the state bird) are the principal animals sought during hunting season. Much of the wooded areas are not only state forests but are part of the widespread State Game Lands system that is especially widspread in the northern half of the State. One popular hunting ground is Knob Mountain near Orangeville about 12 km (8 miles) northeast of Bloomsburg (long green "prong" in upper right of the full subscene; this is an anticlinal mountain).

Knob Mountain near Orangeville, PA.

1-5:In the TM subscene, the area in question is a light tan. This is exposed soil or dirt. In 1987 this land was being cleared. By mid-1988, the aerial photo shows a very large building on this land, surrounded by a black top, typical of a parking lot. This is now the Columbia Mall, which opened for business in 1989. Note that there is a cloud just to the west in the TM subscene.BACK

1-6:There is a second cloverleaf where I-80 is crossed by State Route 487. Just to its north, about 1/2 mile is an elongate lake visible in the TM subscene and more so in the aerial photo. The photo shows a dam on its west (left) end. This is an artificial lake used as a water source and for fishing, mainly by the residents of the village of Lightstreet just to its east. In the aerial photo there is also a light blue area less than a mile west. This is a sand excavation pit in which water has now formed a shallow lake.BACK

1-7: Bloomsburg lies astride a broad ridge known as Turkey Hill (I actually saw turkeys on it once!), an anticlinal structure. The Interstate runs through the small valley carved by Fishing Creek. Where Fishing Creek turns south along Route 42, the hill to its east is the highest locally. Most of the terrain north of the Interstate is rolling hills, much of which is open farmland. The dark spotches are isolated strands of hemlock trees; some woodlands (not leafed yet) to the north have more hemlocks. The bright surface on the Susquehanna is sun glint. There are no buildings tall enough to cast shadows that would suggest time-of-day, but shadows associated with the bridges at Bloomsburg and the smaller Catawissa to the southest (probably too thin to show on your screen or in the printout) are consistent with mid-morning.

This is a good time to reproduce here part of the 1:24000 7 1/2 minute topographic quadrangle map made in 1953 (hence the town is smaller) that covers part of Bloomsburg and areas to its north. Compare the contour patterns with your assessment of the topography derived from your stereo inspection.

Part of 1:24000 topographic map that includes the northern half of Bloomsburg.

1-8:At the juncture, the land is flat and open to the east and north of the Fishing Creek. This is partly why the Fairgrounds were located there but houses in this joint floodplain are occasionally flooded. To its south, there is a sharp cliff rising to uplands on the Susquenanna's west side. South of Bloomsburg the river abutts against a steep sloping bluff topped by farmed uplands towards Catawissa Mountain. But note that homes have been built along the side which have a striking view of Bloomsburg when trees are winter-bare (the whole setting reminds me of my visit to Heidelburg, Germany on the Neckar River). Farmland to the south is higher than that in much of the land a few miles north of Bloomsburg. The alternating light-dark patterns, found throughout the full subscene, are a prime example of contour farming, in which the field is plowed parallel to lines of equal elevation and commonly two crop types (one usually corn) are planted side by side. You will see an good example later on the ground and will be challenged to locate it in the image.

This next photo was taken from Route 487 on the northeast edge of Bloomsburg overlooking the valley of the Susquehanna, its south bluffs, and the mountain areas to the south which are largely game lands (my home is hidden beyond the foreground trees).

The Valley of the Susquehanna River southeast of Bloomsburg; State Game Lands occupy much of the mountains beyond.

1-9:The cluster of white geometric patterns denotes the roofs (light gravel) of several of the factories (one makes rugs for all General Motors cars) in the Bloomsburg industrial area. The two ovals are race tracks (one for practice) connected with the Fairgrounds. The University is rather difficult to find: best to start with the false color aerial photo. Follow the prominent Route 11 into town, where it jogs abruptly at angle of about 120 °, then turns again as Main Street. Just to the right of that second turn is a cluster of buildings, some with light - others with dark roofs. This is the campus. In stereo you will see it as occupying a hill, so it overlooks town. Route 11 is now (and probably in the '80s) macadam (dark asphalt) with several short patches of concrete. The short strip to the south is the single runway at the local apart. The bridge is all concrete - from roadway to its underparts, as you saw in the view of Bloomsburg from across the river.

This aerial oblique photo (Courtesy: Bloomsburg University) shows the Main campus of BU looking northeast (the ridge ending abruptly in the background is Knob Mountain).

Bloomsburg University's Main Campus; Upper Campus on hill in upper left.

1-10: The blackish pattern to the left of A is Lake Chillisquaque (Indian name), the largest lake in the TM subscene. It has a rock-earth dam at its south end used to build up the lake from local stream inflow. Its curved length is nearly two miles. It was constructed by the Pennsylvania Power and Light Company (PP&L) as a storage area for water used elsewhere, but they have turned it into a splendid recreational area for boating, fishing, and nature trails (it is the best birding spot for waterfowl). Here is a photo:

Lake Chillisquaque at the bend of its two arms.

1-11: In the TM subscene, the area left of B shows what appears to be water, dark ground, perhaps a building or so, and smoke. You might have guessed it to be a power plant - an electricity generator which uses coal for fuel. This is one of the PP&L plants for central Pennsylvania. Here are two views:

Pennsylvania Power and Light plant near Washingtonville, PA; water in foreground is an artificial lake used in the coal-heated generation of steam.

Closer view of the Power Plant.

1-11:The borough of Danville (it's not a town, remember), another old river hamlet that has grown to more than 6000 residents. Let's take a look at it:

Danville, looking north across the new bridge.

You are seeing Danville from the south end of the brand new bridge (opened in July 2000) that leads to the residential community of Riverside. Next to it is the 100+ year steel bridge in process of being dismantled. The "Town" Hall like building is the county seat for Montour County. In the far background, against the hills, is the yellowish main building of the Geisinger Medical Center.


1-12: To try to locate the field, go just below the red D. There is a large elongate field with a darker slightly curved band, and then a broader field to its south, and a small grove of trees. This seems to be the area shown in the photo but in the year 2000, the planting pattern is somewhat different in that there are several narrow green fields (alfalfa?) alternating with the now dry stalks of corn. BACK

1-13: Several of these dark green patches are Christmas Tree Farms, as they are called locally. A farmer must plant them and then wait a number of years (at least 5) to cut them down for sale across the east. Starting in November, one might see a number of long flatbed trucks or ones with gatelike sides going south or east with a load of Xmas trees for the metropolitan areas of the East Coast. In both Landsat and aerial images, they have the dark signatures of evergreens but because of their regular spacing in rows, a fair amount of soil makes a mixed pixel so that their recognition is difficult. Here is a picture of one of these patches:

Christmas tree farm on the southwest slope of Catawissa Mountain.

1-14:From F you will see some farms and fields and then a long ridge with about the same elevation. Thus:

Little Mountain and farmlands near Numidia, PA.

At point G as you look west you should see a small lake. There is a second larger one further west. These are dammed to make reservoirs for water supply and for wildlife. This is the view of the closer, smaller lake; it is partially filled with aquatic plants:

Small reservoir lake in valley between Little and Big Mountain (ridges on either side of breached anticline).

This bears on the third question: where are the higher elevations. If you look at the lower of the two above pictures, you will see a ridge on the south side. The water is in a valley in the central part of the green pattern; it is most unlikely to have been developed at the top of a ridge, unless water were hauled up by thousands of trucks - for no apparent good purpose. What the TM subscene fails to show is the true nature of the relief in the green pattern. It is not a single ridge, highest in the center, but is two bounding ridges, with a valley in between. The north ridge is known as Little Mountain; the south ridge is Big Mountain. Geologically, this is a breached anticline - an upfold in which the resistant rock units (sandstones) at the top were eroded away, exposing softer rocks (shales) that were then rapidly eroded downward (breached) to form a valley with the ridges on either side staying higher because they are composed of the hard sandstone. This situation is a strong argument for having stereo capability to resolve topographic uncertainties.


1-15:This ground photo should tell you the answer:

Electric power lines running eastward, here crossing Route 42 at Aristes, PA.

These are, of course, power lines. When constructed, their right-of-way is cleared for about 100 feet to allow easy access for maintenance. This particular line's pathway is being somewhat overgrown and may require later re-clearing.BACK

1-16: You're in coal country now. The south side of Big Mountain is held up by the Devonian Oriskany sandstone that dips south. The rocks to its south are younger - Mississippian and Pennsylvanian. The latter contains the Coal Measures which because of low-grade metamorphism have converted to anthracite - the premium coal in this country because of its high BTU content. The area around I has been extensively strip mined, as seen here:

Coal and overburden rock in spoil banks in the now largely abandoned borough of Centralia, PA.

Spoil banks - a mix of coal and shale - cover the surface. There are many trees with whitish bark in the photo: these are birch, which are acid soil lovers and grow profusely where the strip wastes weather into acid (low pH) residue; they lose their leaves early in the Fall.

This coal waste area is locally "famous" (or perhaps "imfamous"). The nearest town is/was Centralia, about 1500 people whose lives depended on coal mining. There were a number of underground mines at one time but high costs of extraction led to their abandonment. Then, one year someone through waste down a mine shaft. This eventually started a fire by spontaneous combustion. The fire followed a thick coal seam, burning its way westward. As the coal turned to ash, the space created led to instability and collapse, affecting the surface and the row houses in the village. It got so bad that the Federal and State governments forced evacuation of Centralia (a few intrepid diehards remain) and resettlement of most of the population. Today, as you drive through this area, especially during a rain, there are lines of steam (groundwater reaching the fire) that mark the current advance of the burning. This may take hundreds of years to put out (attempts to douse the fires have failed) and threaten Mount Carmel to the west (at the very edge of the enlarged TM scene, below the large lake).BACK

1-17: This is a "gimme". It's another power line. You will notice other, more irregular lines in the forests associated with Catawissa Mountain and the Game Lands in the mountains attached to it. These are dirt roads used for access by hunters and forest rangers (to aid in fighting possible forest fires).BACK

1-18:Go to the red G and move diagonally (about 45°) to the upper right. There is an irregular polygon whose tone in the enlarged subscene is lighter green than its surroundings. Several logging roads lead up to it. At least one other similar road ends at a lighter green patch. These are clearcuts now undergoing reforesting. Look around also for several small areas of green that is darker than the general color of the forest lands. These are hemlock stands. The black specks throughout the green tones are slope shadows or even tree shadows. BACK

1-19: K marks the borough of Berwick, similar in size to Bloomsburg. It's economy is largely industrial - small factory operations. Many retired workers have stayed there. There is great civic pride in Berwick High's football team - it has been state champs in its division several times in recent years, in one of which it won USA Today's national championship. The bridge leads to Nescopeck across river. If you look northeast from the bridge, you will note two long, narrow uninhabited islands. The Susquehanna is a very shallow river over much of its length, until it nears the Chesapeake Bay. Boating is restricted to canoes and flat-bottomed row boats because propellers can be ruined when the river is low. Here is a ground photo of the island:

Island in the Susquehanna River just upstream from the bridge into Berwick, PA.

Further upstream, past L on the labeled TM image, the valley of the Susquehanna becomes a typical area of farms between ridges, as shown in this image:

Photo taken from the Council Cup cliff showing the Susquehanna Valley, here a farmland, east of Berwick.

1-20: First, the municipalities are a richer blue than normal for a TM 2,3,4 = BGR version. This is because Bloomsburg, for example, is somewhat brighter in TM Band 1 than in Band 2, and thus contributes more to the blue band in the composite. Second, the Susquehanna River is red instead of the usual blue or bluish-black. I've never seen this before. To be red, Band 4 should have a lightish tone in the river, which it doesn't. Sediments could to this, but they would be even more effect in Bands 1-3, which they aren't (if you look at the natural color rendition, the river is black). I don't have an explanation for this anomaly.BACK

1-21:Four general thermal patterns are of note: 1) the forested mountains are cooler (variably dark), but there is a little variation in tone which relates to differences in tree density and type (some areas have been reforested); no effect of differing elevations is obvious; the decreased temperature may well be from evapotranspiration cooling; some of the black fringe may be partial shadow; 2) most of the lowlands are comparatively warm, especially the "barren" fields; 3) the three municipalities - Danville, Bloomsburg, and Berwick - are bright, i.e., warm; this is the natural "heat island" characteristic of urban areas; and 4) the Susquehanna River is moderately cool (in the daytime) relative to the nearby land, as is typical of water (its temperature near the surface could be, say, 75° while the land in mid-morning might be 82° in mid-June. BACK

1-22: At first glance, the ratio 4/2 image seems almost like a density slice image with just a few gray levels. All vegetation - forests, grasses, crops - are very bright (the image was strongly stretched) and crop-early or barren fields and urban areas are dark. This is a good vegetation discriminator. But, there is a bit more variability in the scene which can be brought out by applying an IDRISI color palette to the image, as seen here:

Four color expression of the patterns made by ratioing (dividing) TM Band 4 by TM Band 2.

In this version, there are four shades of green that roughly separate four types of vegetation, two or possibly three of which are on the mountains and one (or two) in the farmlands. The black tones are associated with bright fields and with urban areas. It is much harder to pinpoint the nature of the red tones. Many seem to be where fields or open areas with some vegetation are located. They generally correlate with a pinkish-gray color in the natural color renditions.BACK

1-23: Green dominates the subscene, suggesting that the NDVI measure is very sensitive to vegetation. The lighter greens are scattered in the hill forests and more prevalent in the farmlands. Yellows correspond to open, near-barren fields. Towns are also yellow. The reds are found in the rivers and lakes.BACK

1-24: This image and its subset are a classic example of colorful remote sensing products that look at first glance as though they are rich in information. Yet, on closer scrutiny some of the color patterns do not seem to correlate closely with either the raw imagery, i.e., before special enhancement procedures are applied, or with field experience. This PCA 142 composite does single out certain things effectively. The urban centers are distinctive. The vegetation on the mountains is well identified, and reds in the valley usually relate to woodlands and groves. The purples generally define the bare soils and the blue-greens associate with gray and/or (?) pinkish fields. The Susquehanna is red again but the small streams feeding it are in black. Inspection of the enlarged image shows that in a local area there may be a considerable patchwork of colors, including yellows and oranges, often with black borders, that may indicate the edges or boundaries (as between fields) are being highlighted. To sum up, this PC composite has notable correlations with reality in places but also seems to have a capricious or random component of patterns that defies easy interpretation. It is well to judge that remote sensing computer-processed products can assist in understanding surface, features, and conditions but often (especially with lower resolution) introduce or leave ambiguities. These produces are not necessarily a total solution.BACK

1-25: Several classes or categories are obvious. For instance, clouds and their shadows; the river; the forest; the municipalities and clusters of houses comprising villages. The forest appear to have three levels of green: 1) deeper green for thick forest growth (including hemlocks and maybe the Christmas Tree Farms but the latter are too small and isolated to be picked out reliably in the scene); 2) areas of medium green with a brownish tone - these could be second or multiple growth stands of trees developed long after they were timbered; 3) open areas of lighter green - most likely more recently cleared forests now undergoing regeneration. There seems to be four categories of fields (designated as such because of their location and shapes), based on color in the natural color subscene: 1) bright toned (whitish) fields, likely bare or with very early stage crops (e.g., corn); 2) gray fields, nature unknown but fairly common; 3) light green fields, presumably crops such as wheat or hay, that are at a more advanced stage of growth; and 4) pinkish-brown fields whose status is not deciphered (since this is a 1987 image, no supporting ground truth can be retrieved and thus no specific crop types can be chosen). The areas of strip mining in the lower right of the subscene are another discrete class. From the above, you probably can assemble 12 meaningful classes. But, note that all are land cover types rather than landforms (still, the mountains can be inferred from the knowledge that continuous forest cover occurs mainly on the larger structural units). Now return to see what the writer actually selected.BACK

1-26: 1) The natural color rendition of the TM subscene does a good job of displaying the different classes - those that are features larger than about 30 m in dimension. A skilled photointerpreter can generally identify and give names to most of the spatial/color patterns in this rendition. Classification is done mainly to specifically assign descriptive names to the classes, display each class in some separating color, and thus produce a map of the area shown; 2) Most of the classes chosen are straightforward, although some may seem a bit "artificial", i.e, the fields which are given names based on their image colors rather than what they are actually being used for or supporting (crops) at the time of image acquisition. One pair of classes should be renamed (the IDRISI program is unforgiving in this respect, not allowing editing of class names): the class called CLEARCUT is valid in the sense that once, a long time ago, it represented extensive forest removal that has since grown back to forest stands similar to the class FOREST; it probably should have been named REFOREST; the class REGROWN refers to more recent reforestation after later clearcutting; 3) Overall, the visual reaction one should get in looking at the classification (especially the printout version I made) is that the classes chosen and their locations make for a fairly successful classification (I would guess 80% or better accuracy. This is not surprising when the signatures are plotted out from IDRISI's SigComp program, as shown below (the legend colors are a default and cannot be rearranged to match the colors in the legend).

Spectral signatures for 10 of the 12 classes set up for Supervised Classification; abscissa lists from left to right Bands 1 through 7; ordinate show DN range from 0 to 255.

Most of the classes (Pond and Shadow were dropped in this plot) could be separated just from TM Bands 1-3 data; the remaining four improve the separability. None of the classes chosen could be mistaken for any other on the basis of signature. Four of the five vegetation classes all have similar TM4 values but WHITEFLD is much lower - its signature suggests it is truly barren (not yet planted or with just emergent crop). The towns - large and small - are well separated. RIVER is singled out; the large and small lakes in the valley between the mountain ridges near the image bottom classify as RIVER as well; the class POND was chosen from a single training site near Lightstreet and its color (yellow) shows only there. The various field classes seem correctly placed but note that the classification could not resolve those fields that consist of alternating strips following contours (they are plotted as a single color pattern so some misclassification occurs here). The reds of STRPMINE are properly placed in the Centralia area. But small patches of red occur where they shouldn't (e.g., around Elysburg; see map); this is probably caused by a land use class that resembles the wasted land of the strip mines. Note, too, that the PP&L power plant near Lake Chillaquaque is shown in brown and red: it, in fact, is similar to both a town and waste areas. Both small streams and some roads are shown in black (they were too thin to allow training sites to be drawn and are thus prone to misclassification). Shadows, in black, may be correct where associated with mountains but the few clouds in the scene are classed as UNKNOWN (first black in legend), since they were not selected as signature, and their shadows are blue (similar in signature to RIVER). To summarize, the classification came out better than I expected.


1-27: Just north of meeting of the two branches, and on the north side of the North Branch is Northumberland (whose greatest claim to fame is that James Priestly, the discoverer of oxygen as an element, lived there). Across river is the largest urban center, Sunbury, PA. On the west side of the combined branches is the borough of Shamokin Dam, mainly a very long strip of malls and businesses. Although its name is cut off the edge of the road map, the next town south along the river is Selinsgrove, home of Susquehanna University, a private liberal arts college. None of these really stands out in the L-band radar image, but all are in light tones (many building reflectors). Strangely, they are not clearly differentiated from their surroundings in the radar color composite, where they appear pink. Sunbury, however, shows little pink despite it being the largest community. Sometimes, radar composites can be misleading. The purples are mostly open fields. BACK