Recent engineering-geology assessment of a proposed damsite in an area of gypsum karst in southwestern Oklahoma has shown that the site is unsuitable and should be abandoned. The proposed Mangum damsite has been investigated and evaluated since 1937 as a potential site for a compacted, earth-fill dam, about 33 m high, that would provide for irrigation water, flood control, and recreation. Initial approval of the site was based only upon its favorable topography, because both proposed abutments are high and form a narrow water gap between them. This decision did not adequately consider the geology, foundation conditions, or water-impoundment capabilities of the site. Abutments would be in the Permian Blaine Formation, consisting here of 60 m of gypsum with thin interbeds of dolomite and shale. The Blaine Formation locally has abundant gypsum-karst features, such as caves, sinkholes, disappearing streams, and springs. Efforts since 1937 have focused on finding a suitable site within the flat-lying Blaine Formation where the karst problems would be minimal.

In 1999, a final assessment was made of the surface geology and the results of coring and pressure-testing of five boreholes (each 40–46 m deep) along the proposed dam alignment. The assessment showed: open cavities, clay-filled cavities, and other karst features are abundant in and near the abutments; and fluid losses (per 3-m interval) ranged from 60–250 L/min in most borehole pressure-tests, and in one borehole the losses were 1,600–5,300 L/min. Engineering measures needed to remediate karstic foundation conditions here would add greatly to the cost of construction, and still would not assure tightness of the reservoir or integrity of the dam. By recommending abandonment of the proposed site, and moving the investigation 11 km downstream to a site with thick shale abutments that are a greater distance apart, I believe a dam can be built at the new location without the potential problems of extensive leakage and possible failure of a dam built upon gypsum karst.