An empirical evaluation of glacial trough cross-section shape is perfo
rmed on seven vertical cross-sections in three Sierra Nevada valleys g
laciated during the late Quaternary. Power and second-order polynomial
functions are fitted by statistical regression. Power functions are v
ery sensitive to subtle valley-bottom topographic features and require
precise specification of the valley-bottom-centre location. This depe
ndency is problematic given under-representation of valley bottoms by
conventional contour-sampling methods, and the common alteration of va
lley-bottom morphology by non-glacial processes. Power function expone
nts vary greatly in response to these and other non-genetic factors an
d are not found to be reliable indicators of overall valley morphology
. Second-order polynomials express overall valley shape in a single ro
bust function. They are applied to both bedrock- and sediment-floored
glacial valleys with negligible statistical bias except where side-slo
pes are stepped or convex-upward or where valley form is asymmetrical.
They can describe alluviated or severely eroded valleys, and can obje
ctively identify individual components of polymorphic valleys, because
valley bottom and centre locations need not be specified. Mathematica
l expressions of parameters useful for geomorphic measurements and gla
ciological modelling are analytically derived from the polynomials as
functions of the three polynomial coefficients. These parameter equati
ons provide estimates of valley side-slopes, mean and maximum depth, m
idpoint location, width, area, boundary length, form ratio and symmetr
y.