Without further ado, we now present a two image (September 1984) Landsat TM mosaic (Band 5) that includes the study area and mountains to its south down to the California border. The terrane boundaries, redrawn from a regional map prepared by the U.S. Geological Survey, are overlaid in white.

At first glance, and especially with your monitor limitations in mind, you may decide that differences in topographic expression are hard to pick out. But, to illustrate that these are really there for some terranes, look at the two in the upper left listed as Sixes River and Elk. These are clearly unlike, and, as we shall see in detail later, are the best case supporting the terrane-terrain hypothesis. Other terranes with apparent differences relative to their neighbors include: Rogue Valley, Smith River, and Yolla Bolly. Sixes River and Pickett Peak look too similar to justify any distinctions from each other. Both Smith River and Yolla Bolly terranes also display evident differences in topography within themselves. Several areas within the Smith River terrane near its upper left (northwest) border are apparently high and have more broadly spaced divides: these coincide with a lithologic unit known as the Josephine ophiolite which consists largely of metamorphosed basalts or greenstones that represent parts of oceanic crust that failed to subduct. Consider also three arbitrarily selected segments of the Yolla Bolly terrane designated later in this survey as YbNorth (next to Snow Camp upper label); YbCentral (next to Dry Butte Label); YbSouth (next to upper Smith River label). Your eye should easily pick up visual differences in the appearance of the terrains (e.g., look for ridge or valley spacings). We will quantify these differences later.

Let's take a closer look at the northwest section of the mosaic, in this enlarged image:

Let your eyes adjust to this image and then try to draw imaginary boundaries around parts of it that seem separable. Once more, the Sixes River-Elk pair of terranes leap out as different. Now, we enlarge this area even more:

It is almost self evident that the upper and lower halves on the right and center of this enlargement are conspicuously different. They are indeed separable visually on the basis of topography.

The boundary between Sixes River (north) and Elk terranes runs horizontally across this map at approximately the middle. Clearly, the Elk terrane is on average notably higher than the Sixes River terrane.

Next, using a map (not shown) that plots the mountain ridges between valleys, initially drawn from contours on individual 15' topographic quadrangles and then merged into a single regional map, the total length of ridges per unit area (each a square mile) can be calculated. This is an indirect measure of ridge density. The average of two subunits in the Elk terrane is 5.16 (miles/sq mile) whereas that for the Sixes River is 3.29. This suggests that the ridges will be closer-spaced in the Elk than in the Sixes River terrane, a conclusion borne out by simply looking at the enlarged Landsat images.

A third difference is not readily apparent from the images. But when the regional map of ridges is analyzed in terms of average direction of trend or orientation of ridges in square mile sampling units (this was done by a special computer program), rose diagrams for each terrane indicate strikingly dissimilar patterns:

Most ridges in the Elk terrane trend northwest in contrast to the northeast trend dominating the Sixes River terrane. Reasons for this will be given later (for now, accept the speculation that both ridges and stream valleys reflect structural control).

These ridge measurements, when made from topographic maps, are relatively unbiased, i.e., with careful work ridges that are drawn as bisectors of crest-proximate contours should all be recorded . Thus, the inherent problem in space imagery and aerial photography of missing some linear features because of sun directional bias is eliminated. Re-examination of the Landsat imagery already seen will indicate that the shadowing in this high relief terrain owing to the morning Sun direction from the southeast does impose a strong bias towards a northeast emphasis of those ridges trending in that general direction. But, with judgment by the observer looking at enlarged Landsat imagery, many of the ridges in other orientations can be detected using tonal patterns related to moderated illumination differences along slope pairs not optimally oriented towards easy discrimination. Still, some will inevitably be missed.

Stereo pairs should reduce this bias effect. Two SPOT images made with the tiltable HRV scanner (see Section 3) pointed at different angles and look directions on different dates allowed ridges to be mapped directly from the imagery. Here is one of the (15 m resolution) panchromatic images (60 km on a side), covering the Gold Beach, Pickett Peak, and segments of Sixes River and Yolla Bolly terranes:

Although the higher resolution leads to more detail, the image appears "flat", with lack of strong shadowing inhibiting the expression of much of the topography. When viewed stereographically, the ridges associated with mountains of various heights do stand out well enough to foster tracing them on to a map.

This next map shows all ridges mapped in both SPOT and Landsat images in red; green denotes extensions of ridges selected from Landsat alone; and blue records images found only in SPOT. Surprisingly, the use of SPOT does not seem to improve ridge detection by much; perhaps the Klamath terrain is well suited to effective ridge depiction even under the morning sun conditions.

The same cannot be said for Seasat radar imagery. This digitally correlated SAR image, taken in August of 1978 over part of the same areas as the SPOT image , indicates the sensor to be looking east and as such there is a pronounced slant-range distortion that results in foreshortening layover. A notable bias is introduced such that the bulk of ridges oriented NNE are highlighted and many others completely missed. A data product in this form is not suited to orientation analysis.

Look now at two terranes - Rogue Valley and Yolla Bolly - that are similar appearing at their northern end, as evident in this enlargement (with a locator sketch map) (we will not comment on the Sixes River and Snow Camp terranes north of the Canyonville fault):

Both terranes are bounded by a right lateral strike-slip fault (i.e., the northern block has move right relative to slippage left in the lower block) near Canyonville. The terranes themselves are juxtaposed along a large thrust fault inclined downward to the east, resulting where the Yolla Bolly terrane was pushed under the Rogue Valley. At a quick glance, the ridge spacing and relief for the mountains in each terrane near this fault seem similar. The ridge orientations of the RV terrane are more northward than those of YB.

One difference is not evident from visual inspection but comes out when measurements are made from the Canyonville topographic sheet. The highest elevation in each township section (1 square mile) was read from that map. When plotted as a histogram (frequency of maximum elevations), this distribution results:

Allowing for the difference in the number of data points involved (67 vs 42), there still is a notable variance between height distributions. In this topo sheet, the average maximum elevation in the Rogue Valley terrane is 3226 feet (993 m) whereas in the Yolla Bolly terrane it is 2774 ft (854 m). Treating these two sets of measurements as possibly representing two distinct populations, a statistical device known as Student's t test gave a value of 5.31 which is significant (real difference) at the 99% confidence level. This dual distinction is surmised to result from differences in dominant lithologies: the Yolla Bolly is characterized by siliceous shales; the Rogue Valley includes more resistant rock types.

---

[an error occurred while processing this directive] [an error occurred while processing this directive]