|
<br />f
<br />
<br />-
<br />
<br />Boundary Conditions and Flow Patterns in a Mountain River
<br />Andrew J, Miller', Diane M,L, Mas, Student Member ASCE', James A, Smith',
<br />Wei-Hao Chung'
<br />
<br />A.Mlnet
<br />F1nod flow patterns in mountain valleys may be influenced by longitudinal
<br />variations in channel and valley morphology and by temporal variations in
<br />bouoc!a1y conditions, This study utilizes a 2-<1imensional depth-averaged fmite-
<br />element model (RMA2) to explme the sensitivity of flow patterns to valley bends
<br />and constrictions and flood-induced geomorphic changes at a site along the South
<br />Branch Potomac River in West Virginia.
<br />Study Area
<br />The South Branch of the Potomac River flows through the Appalachian
<br />Highlands of West Virginia. Width of the valley floor alternates between narrow,
<br />bedrock-constrained reaches and broader reaches with more extensive f10ndplain
<br />development, The longest canyon along the South Branch, known as the Smoke
<br />Hole. has an average gradient of about ,008 and average width less Ihan ISO m,
<br />Two major stoons in the last 50 years have had major geomorphic impacts on Ihe
<br />South Branch as it flows through the Smoke Hole, An intense convective storm in
<br />June 1949 mobilized scores of debris avalanches in tributary valleys and along Ihe
<br />walls of the main valley. Some of these were large enough to divert the river
<br />against the opposite valley wall, Deposits left hehind at the mouth of Redman Run
<br />(Ihe site of this study) fooned a new bottomland about 70 m wide and several
<br />hundred m long bordering the left valley wall. Subsequently the record flood of
<br />November 1985 (Miller and Parkinson, 1993) removed these deposits complelely,
<br />shifted the main channel back toward the left valley wall, left hehind a large gravel
<br />bar extending downstream along the right valley wall, and reworked much of Ihe
<br />floodplain surface inunediately downstream (see Miller, in press, fig, I for
<br />photol1fllphs),
<br />
<br />I Associate Professor. DcparUnent or Geography, University of Maryland. Baltimore. MD
<br />21228,5398
<br />
<br />2Graduate Student. ) Assistant Professor, Department of Civil Engineering and Operations
<br />Research, Princeton University, Princeton, NJ 08544
<br />
<br />762
<br />
<br />.
<br />
<br />BOUNDARY CONDITIONS & FLOW PATTERNS
<br />
<br />-763
<br />
<br />Miller (in press) suggests that flood-induced geomorphic changes may have a
<br />feedback effect on flow patterns. altering the shape of the water surface and
<br />thereby altering stage-<lischarge relations as well as spatial distributions of
<br />vel?"ity, and she~ stress, The analys~s was based on 2-d flow modeling applied to
<br />an Idealized strmght canyon reach W3th a large debris-fan constriction, The
<br />present paper i~ ~ preliminary effort,to simulate flood flow patterns using realistic
<br />boundary condihons collected at a SIte where major flood-induced changes have
<br />been observed,
<br />, Field, survey ?ata collected after the 1985 flood were supplemented with
<br />mfonnahon ~btamed from pre- and post-flood aerial photographs in order to
<br />consll1lCt, fiOlte-element representations of the study reach, Contour maps
<br />(contour mterval 0.5 m) of canyon topography from befme and after the
<br />November 1985 flood ~fig. I) are based on these data. Preliminary worlc reported
<br />here mvolves a companson of results from steady-state 2-d flow simulation of
<br />pre- and post. flood conditions at ~ single discharge, The discharge value chosen
<br />for the upstre,am boundary cond'hon, 1415 cms, is slightly less than the peak
<br />dIScharge esttmate<! for theI985 flood from a U.S, Geological Survey slope-area
<br />measurement at a Sl'." on the South Branch upstream of the Smoke Hole, The
<br />~ater'surface elev~hon c?osen f~r the downstream boundary condition, 606,80 m
<br />larbllrary datum), IS conSIstent WIth the elevations of surveyed high-water marks
<br />left behmd by the 1985 flood,
<br />"'low modelinp
<br />
<br />The flow model used in this study is RMA2, a two-dimensional finite-element
<br />1I\l~el based on a dep~-!,veraged foon of the Navier-Stokes equations for free
<br />~urface now. ~MA2 IS IDcorporated in the U,S, Aony Corps of Engineers'
<br />I ABS-2 modeling system (Thomas and McAnally, 1990), Model ourput includes
<br />Illlw depth and x- and y- components of depth-averaged velocity at each node in
<br />Ih, mesh" Turbulent energy losses are represented using an eddy-viscosity teon
<br />III modelhng the ,Rey~olds stresses, In general, the value chosen for the eddy-
<br />~~~O~IlY,coefficlent IS the ~mallest value that will allow a convergent solution.
<br />. a~n~ng ~ n values are assigned for individual mesh elements allowing spatial
<br />vanatlon In roughness.
<br />
<br />, Mesh conslructi,on was accomplished using the FastT ABS program (Boss
<br />(IIrporalton and Bngham Young University, 1993). (See Mas et al tht's
<br />\ulumc for d 'I ' "
<br />. '.. m~re eta! on mesh construction and related topics). The "marsh
<br />(h:I1I~ms weUmg and drying option was utilized in this study.
<br />. I or the m~el runs reported here, spatial variations in roughness were not
<br />;:l~lorporaled In.the boundary conditions files. A constant Manning's n value of
<br />"1;1~;as.'peCtfied for all channel and nondplain elements. with a higher value
<br />, I I 11 asslgn~ ~o some of the steeper valley walls. The comparison of pre- and
<br />II' \. ood conditions therefore examines only the influence of topographl'c
<br />~ l;tnge:S.
<br />
<br />..
<br />
<br />111
<br />
|