STREAM GAGING AND CHANNEL REACH CHARACTERISTICS IN A BREACHED DAM: SCIENCE IN THE PARKS AT THE SOUTH FORK DAM

The South Fork Dam, once among the largest earthen dams in the world, failed on May 31, 1889 in response to over 12 cm of precipitation in 26 hours. This led directly to catastrophic flooding in Johnstown, PA and nearby communities. No modern hydrologic study has been completed on the South Fork, Little Conemaugh River and its watershed, despite enduring debate over the flood and the possible role that modifications to the dam may have played in the event. Within the breach exists an opportune location for a stream gage that encompasses the entire capture area of the former reservoir.

Hydraulic controls at the gage station were explored and mapped via on-site leveling surveys and LIDAR data. The origins of these controls were explored with available historic records. The section control, a transverse gravel bar 75 m downstream, is largely due to a culvert placed by the railroad now running through part of the breach. The control may be augmented upstream by remnant plunge pool topography, as historic photos reveal a characteristic hourglass planform and pool in this part of the breach. In cross-section the original breach resembled a compound weir with a trapezoidal segment underlain by a nearly rectangular segment reaching about 7 m above bed. The latter has since reduced its slope and evolved into two trapezoidal segments, with the upper crest being vertically offset either side of the breach. The lower (north) crest provides a second channel control (which extends to much of the former plunge pool either side of the channel), while the higher crest provides a floodplain surface. This evolution seems to be driven by natural processes.

Stage-discharge data gathered via wading and portable cableway for low and moderate flows were analyzed via iterative methods to begin a rating curve. Scale offset produced by curve fitting matches the surveyed section control elevation (0.15 m below staff gage). Rating slope analysis indicates channel control begins about 0.5 m above effective zero flow. The initial data are consistent with good station control and will enable the next phase of data collection focusing on obtaining higher stage measurements and high-flow extrapolation with HEC-RAS to complete the rating curve and facilitate a subsequent phase of analysis, including watershed modeling and reconstruction of the 1889 flood hydrology.