It seemed apparent from the Heli animation that the tension setting used for the spline interpolation was resulting in a surface that was too ``stiff'' to respond quickly enough to adequately represent the bottom, particularly in places where the channel is narrow.

However, lowering the tension setting indiscriminantly, while it might provide a better fit in some places, could result in overshoots, in which the surface reacts so quickly in complex bottom locations that it shoots above the water level, creating islands where none actually exist.

A series of 10 interpolations were sequentially performed in one small bottom location through a range of tension settings, and the residual errors examined using virtual push-pins as before. It was found that a tension setting of 130, along with a smoothing factor of 0.5, produced a bottom surface which appeared to minimize the residual errors. The entire bottom of the Clinch River was re-interpolated using these parameters.

Another animation was produced to examine the residuals for all portions of the Clinch bottom, as before. The spatial trend in residuals which was evident at narrow places in the Heli animation, where 2 rows of red pins surround the centerline of the bottom and are flanked by 2 rows of green pins at each shoreline, is no longer present with the new interpolation.

Moreover, the visualization shows that these settings also provide a good fit for most of the center bottom, as well as the broader, shallower wide sections of the river bottom. These shallower, broader locations are the places where the water is slowing, and potential deposition of contaminants is likely to occur.

The improved splined surface seems be representative of the general bottom shape as indicated by available depth data. If a more accurate bathymetric surface is required, further investment should be made in obtaining more depth data rather than tweaking the spline tension parameter.