Tröllaskagi, Iceland -- a classic glacial trough pattern

There are trolls in them thar mountains..........

If you shift your focus from the delicate snow-covered uplands in this image to the trough pattern, you can see how glacier outlet troughs evolve, via greater and greater concentrations of ice in the main troughs and the gradual elimination of smaller valleys or feeder troughs.  Eyjafjordur is the long fjord to the right of centre.

Again, click to enlarge.

David Sugden and I were so impressed by a false-colour image of this same area that we used it for the cover of "Glaciers and Landscape".  Our text that was reprinted many times over more than 20 years of use as a key glacial geomorphology textbook for university students.  The cover designer took some liberties with the orientation........

See also this post, from three years ago:

https://brian-mountainman.blogspot.com/2017/01/brutalised-glacial-discharge-dendritic.html

In our book, David and I discussed the relationships between selective linear erosion, glacier trough development, and divide elimination.  This was a favourite topic for Prof David Linton, one of the great thinkers on glacial processes and landscape evolution.  in the 1960s he turned his attention, in a number of influential articles, to LANDSCAPES of glacial erosion, expanding his field of vision from that of earlier researchers who had concentrated on glacial processes and landforms.  These are two of his articles:

Linton, D.L. (1963), The forms of glacial erosion. Trans. IBG 33, 1–28.

Linton, D.L. (1967), Divide elimination by glacial erosion. In: Arctic and Alpine environments (Wright and Osburn Eds), 241-248.

Even more interesting was his seminal chapter on "Landscape Evolution" in the 1964 volume entitled "Antarctic Research" edited by Priestley and others (pp 85 - 99).

He noted how, in the Antarctic Peninsula region, the efficacy of glacial sculpture varied from place to place depending upon aspect, ice cover, altitude and other factors.  He noted how in some areas there were rounded or gently convex mountain tops emerging from the ice sheet edge or else capped by local ice-caps, while towards the coast, at lower altitudes, there were sharp-crested ridges with jagged or pointed peaks.  Some of these ridges were heavily scalloped, especially at lower levels.  These scallops were interpreted as the remnants of cirques, cut by small independent cirque glaciers and in the catchments of small independent valley glaciers.  So the bigger and higher features were the result of erosion by outlet glaciers evacuating ice from large ice masses, while the smaller and lower peripheral features were seen as the result of "local glaciation" at times of less extensive ice cover, for example during interglacial episodes.  The plateau surfaces were interpreted as relatively old remnants of pre-glacial land surfaces, while the scalloped and sharp-crested ridges, aretes and horns (pinnacled peaks) were thought of as being relatively new. 

Given that the landscapes of West Antarctica and the Antarctic Peninsula have evolved across many glacial and interglacial cycles, Linton  developed a model of glacial landscape evolution in which, close to the ice centres, large outlet glaciers get larger and larger, and dismember the pre-existing drainage divides.  Those first-order divides that do survive become longer and longer, but also narrower and narrower as the scalloping effect of intermittent cirque glaciation runs its course.  That's the mpodel, but in the real world, of course, the irregularity of glacial / interglacial pulses has a major effect on how the landscape will ultimately appear..........

The Linton model of trough expansion and drainage divide elimination as a result of long-continued glaciation.  On the left, the assumed original fluvial (preglacial) fluvial dendritic pattern.  On the right, an expanded outlet trough with truncated interfluves or spurs separating a number of contributing glaciers.  The interfluves are narrowed and scalloped by intermittent cirque glaciation and frost processes to create a landscape of aretes and horns.

I'm not sure how often the Linton model would apply in the real world -- for example in Greenland, Iceland and Antarctica.  The model assumes that the whole of the ice supply feeding into the trough comes from within the original drainage cachment area.  In reality, as a glacial episode runs its course, and an upland area "grows" its own ice cap or ice sheet,  the ice supply into the trough comes from OUTSIDE the original catchment, spilling over and into a trough head.  So initially there is a vastly greater ice supply than the trough can cope with -- so it is greatly enlarged, to a disproportionate extent, chopping off many of the interfluves shown in the right-hand diagram above. That is the situation that applies when properly brutalised dendritic patterns develop -- with a huge main trough and very small flanking feeder troughs with their bases "hanging" well above the floor of the main trough.  This is exactly what is seen in Sognefjord and many other big fjords when bathymetric surveys are undertaken.